Apparatus for applying characteristics using quantum computer generation technology

Abstract

The invention relates to an apparatus for generating a property associated with a chemical product. A representation providing unit provides a molecular structure representation, and a division unit divides the molecular structure representation into two or more sub-representations based on geometric division and / or electronic division. A solution determination unit causes the quantum computer to generate a solution of the sub-representations, and causes the classical computer to generate solutions of the remaining sub-representations. An intermediate characteristic calculation unit generates an intermediate characteristic, which is an electronic characteristic of the electronic structure, by combining the solutions of these sub-representations. The generation unit generates a characteristic based on the intermediate characteristic.

Description

Technical Field

[0001] This invention relates to an apparatus, method, and computer program product for generating technical application characteristics associated with chemical products using a quantum computer. Furthermore, this invention relates to a system for generating technical application characteristics using the apparatus. Background Technology

[0002] Quantum computers are typically a completely new type of computing system that allows the use of the special behavior of quantum mechanical systems to perform computations that ordinary computers cannot complete in any reasonable amount of time, under certain circumstances. Summary of the Invention

[0003] The object of this invention is to provide apparatus, methods, systems, and computer program products that allow for improved efficiency and effectiveness in performing calculations on chemically relevant problems using quantum computers, i.e., reducing computation time and required computational resources while achieving the same quality of results. Furthermore, the chemical industry sells chemical products with specific technical application characteristics. At the heart of any chemical product is a chemical molecule or combination of chemical molecules; for example, a chemical product comprises one or more chemical molecules that determine its properties. The properties of chemical monomers or oligomers can be determined, for example, by extrapolation, to determine the properties of chemical polymers.

[0004] Chemical products, comprising one or more types of chemical molecules, are highly complex real-world systems containing multiple interacting electrons that constitute the properties of the chemical molecules and thus the properties of the chemical products. Therefore, it is extremely important in calculations to capture the complexity of chemical molecules by reflecting the types of atoms, the chemical bonds between atoms of the same and / or different types, the three-dimensional structure of the atomic arrangement, the interactions between multiple atoms in this three-dimensional arrangement, and especially the subatomic structure and electron density constituted by electrons. These highly complex molecular structures are typically described through electronic structure problems, for example, using the Schrödinger equation employing Hamiltonian operators. To allow for the solution of such complex chemical applications, approximation methods can be used. However, with approximation methods, the properties derived from electronic correlations may not be adequately reflected in most systems. Therefore, it is necessary to include as many electronic correlations as possible from real-world systems to obtain results that are as close as possible to the real-world technical application properties of chemical molecules. This is particularly important for the chemical industry, which produces chemical products based on the generation, properties, and production formulations of chemical molecules to generate corresponding real-world chemical products with the corresponding properties.

[0005] Calculations of the properties of chemical products are essential for designing experimental setups, where the number of experiments that produce real-world chemical molecules with desired properties often depends on the accuracy of the calculations and the accuracy of the properties determined from those calculations. Achieving highly accurate calculations can be an NP-hard problem—meaning that it would take classical computers decades to solve. Therefore, generating such accurate solutions is impossible in many real-world scenarios for developing practically relevant chemical molecules with enhanced or novel properties. This leads to more intensive, inefficient laboratory work and increases the time required to develop new chemical molecules.

[0006] Given the environmental impact of chemical development and the pace at which the chemical industry must adjust its chemical framework and chemical products, it is advantageous, particularly through more accurate solutions to the technical application characteristics of chemical products, to improve the development of new or enhanced chemical products. Embedding quantum computing into this development cycle may offer a solution to this problem. However, the embedding of quantum computing is itself a challenge. Therefore, the devices, systems, methods, and computer program products disclosed herein allow for more efficient and effective use of quantum computing to enhance the development of new or enhanced chemical products.

[0007] By first dividing the molecular structure representation into two or more sub-representations associated with the geometric and / or electronic partitions of that molecular structure, solving the corresponding sub-representations using quantum and classical computers, and generating intermediate properties of the molecule's electronic structure by combining these solutions, and by utilizing one or more of these generated intermediate properties to generate technical application properties (which are measures of chemical and / or physical quantities associated with the molecule), quantum computers can be effectively and efficiently utilized in real-world applications. In particular, quantum computing can be advantageously embedded in workflows for developing new or enhanced chemical products, making chemical product development more efficient and effective. Furthermore, generating technical application properties that can withstand objective testing due to their measurable output allows for further refinement and fine-tuning of the utilization of the computations and quantum computers used. This enables improvements in the efficiency and effectiveness of chemical design processes.

[0008] In a first aspect of the invention, an apparatus is provided for generating technically applicable properties associated with a chemical product, wherein the chemical product is associated with one or more molecular structures, wherein the apparatus comprises: a) a molecular structure representation providing unit for providing a molecular structure representation associated with the molecular structure of the chemical product; b) a partitioning unit for partitioning the molecular structure representation into two or more sub-representations, wherein the partitioning is associated with a geometric and / or electronic partitioning of the molecular structure represented by the molecular structure representation; c) an intermediate solution determining unit for enabling a quantum computer to generate and provide a quantum computing solution for at least one of the two or more sub-representations, and for enabling a classical computer to generate and provide a classical computing solution for the remaining one or more sub-representations; d) an intermediate property calculation unit for generating one or more intermediate properties associated with the molecular structure by combining the solutions of the two or more sub-representations, wherein the intermediate property is an electronic property of the electronic structure of the molecular structure; e) The feature generation unit is used to generate a technical application feature associated with the chemical product based on one or more intermediate features, wherein the technical application feature refers to a measurable chemical and / or physical quantity associated with the molecular structure; and the providing unit is used to provide the technical application feature.

[0009] Because the molecular structure representation is divided into two or more sub-representations associated with the geometric and / or electronic parts of the molecular structure, where one of these sub-representations is computed by a quantum computer while the others are computed by a classical computer, the specific advantages of quantum computers (e.g., the possibility of efficiently representing and computeing highly entangled quantum states) can be leveraged to compute the more relevant parts of the molecular structure representation (e.g., those involving multiple interactions or having a higher impact on computations of technological application properties) with very high accuracy. Furthermore, the less complex parts of the problem that are less relevant to the solution can then be computed with sufficient accuracy on a classical computer, and these parts are still taken into account when generating intermediate properties. Therefore, the computational resources of a quantum computer can be focused on the parts of the molecular structure representation that should be solved with higher accuracy, thereby improving the accuracy of the solution to the overall problem (e.g., computations of technological application properties). Furthermore, by dividing the molecular structure representation into parts solved by a quantum computer and parts solved by a classical computer, and then combining these parts to generate intermediate properties, and then calculating the technology application properties based on the intermediate properties, the resources of quantum computers can be utilized more efficiently. This allows for increased problem complexity and / or size, and enables the calculation of real-world application properties relevant to practical applications in the context of chemical products.

[0010] Technical application characteristics associated with a chemical product refer to measurable chemical and / or physical quantities related to the molecular structure relevant to the technical application of the chemical product. Technical application characteristics can refer to any characteristic of a chemical product that allows for the assessment of the technical suitability of the corresponding chemical product provided after its production. Preferably, technical application characteristics are associated with and / or include any of the following: chemical reactivity, spectral and spectroscopic properties, molecular properties, activation energy and reaction energy of a predetermined chemical reaction. Preferably, technical application characteristics include at least one of mechanical properties, spectral properties, physicochemical properties, chemical properties, and biological properties. Typically, mechanical properties can refer to any of the following: adhesion, tensile strength, stiffness, hardness, shrinkage, elongation, crack tear, tear strength, resilience, compressibility, abrasion, spillage, morphology, tactile properties, fracture stress, elongation at break, particle size, and filling density. Spectral properties typically include any of the following: tinting, turbidity, opacity, transparency, reflectance, appearance, absorption, scattering, color strength, hue, color saturation, color intensity, cloud point, extinction, optical density, spectrum, refractive index, IR spectrum, Raman spectrum, NMR spectrum, ESR spectrum, and UV / Vis spectrum. Furthermore, physicochemical properties can refer to any of the following: density, viscosity, K-value, molar mass, dispersibility, molar mass distribution, particle size distribution, solubility, partition coefficient, interfacial properties, surface tension, dispersibility, storage stability, odor, segregation, condensation, electrical conductivity, capacitance, surface area, flow time, vapor pressure, VOC, solids content, hygroscopicity, magnetism, miscibility, thixotropy, phase transition properties, glass transition temperature, corrosion inhibition, solvent separation, aggregation, self-heating capacity, impact sensitivity, loss on drying, response angle, electrostatic charge, minimum film-forming temperature, charge density, electrostatic multipole moment, and thermal conductivity. Chemical properties may include any of the following: reaction thermodynamics, reaction kinetics, chemical resistance, reaction time, demolding time, growth, hard / soft segment content, crystallinity, reaction temperature, reaction pressure, decomposition, thermal decomposition, photodegradation, acidity, pKa, pH, moisture / water content, flammability, combustion rate, auto-ignition, flash point, formation of flammable gases, fire reactivity, deflagration rate, residual monomer count, by-product formation, degree of polymerization, salt content, temperature resistance, oxidation properties, reduction properties, reactivity, ash content, non-volatile matter content, stability, chelating ability, calorific value, and saponification value.In addition, biological characteristics may include any of the following: biodegradability, biological resistance, particularly resistance to pathogenic viruses, bacteria, fungi, plants or animals or developmental stages of said pathogens, tolerance to environmental parameters, such as drought tolerance, tolerance to enzyme degradation, such as protease resistance, lipase resistance, amylase resistance, hydrolase resistance, resistance to pest control agents, toxicity, biotransformation, ecotoxicology, sensitization, particularly allergenicity, bacterial count, enzyme activity, substrate specificity, cofactor dependence, product specificity, substrate and / or product inhibition, dissociation constant, Michaelis kinetics, activity / stability under or in different of the following: pH, temperature, pressure, organic solvent concentration, carrier formulation, encapsulated formulation; distribution in the environment, compartmentalization, bioaccumulation, biological exposure LD50, mutagenicity.

[0011] The molecular structure representation providing unit is adapted to provide a molecular structure representation associated with at least one of one or more molecular structures of a chemical product. Specifically, the molecular structure representation providing unit may refer to a storage unit on which a molecular structure representation is already stored. However, the molecular structure representation providing unit may also include an input unit, for example, which a user may use to instruct the molecular structure representation to be provided to the molecular structure representation providing unit.

[0012] Molecular structure representations depict the electronic and / or geometric structure of molecules in the digital domain. Therefore, molecular structure representations represent real-world objects in the digital domain, whereby they can represent all or only predetermined aspects of the corresponding real-world object. For example, molecular structure representations can include SMILES representations, international chemical identifiers, and 3D contours, embedded model representations, Cartesian model representations, or molecular diagram representations.

[0013] A partitioning unit is configured to divide the molecular structure representation into two or more sub-representations. This partitioning unit can be configured to automatically partition the molecular structure representation into sub-representations based on one or more application cases (e.g., one or more molecular categories, available quantum computing resources, etc.) based on one or more predetermined rules and / or on a machine learning-based model trained on multiple molecular structure representations and separately labeled partitioned parts. However, the partitioning unit can also be configured to partition the molecular structure representation during human-computer interaction, where the partitioning unit provides a corresponding interface and guides the user in partitioning the molecular structure representation. For example, the partitioning unit can provide the user with a corresponding human-readable representation (e.g., a graphical representation or a chemical formula representation). The user can then be prompted to identify the parts of the molecular structure representation as sub-representations based on one or more predetermined criteria. For example, the partitioning unit can prompt the user to identify the most relevant part of the molecular structure representation, and then the second most relevant part, and so on. The partitioning unit can then use one or more additional criteria to verify whether the parts identified by the user are suitable for further processing. For example, the partitioning unit can verify whether the most relevant molecular structure identified by the user can be computed on available quantum computing resources. For example, if the selected most relevant molecular structure is too large to be computed on available quantum computing resources, the partitioning unit can accordingly notify the user and prompt them to select a new (e.g., smaller) most relevant molecular structure representation. Furthermore, additional criteria (such as expected computational accuracy, the technical application characteristics to be computed, etc.) can be considered as other standards. Additionally, the partitioning unit can also be a receiving unit that receives the partitions of molecular structure representations, for example, from a database that already stores partitions or from a corresponding user interface where the user provides the partitions without further interaction with the partitioning unit.

[0014] This partitioning is associated with the geometric and / or electronic partitioning of the molecular structure represented by the molecular structure representation. A sub-representation can represent the geometric and / or electronic portions of the molecular structure. These portions may overlap or may be separate. A sub-representation can also refer to a representation of the molecular structure with varying degrees of accuracy. Thus, a sub-representation can represent a complete molecular structure that is less accurate than the provided molecular structure representation. In embodiments, a sub-representation can refer to the most relevant portion of the molecular structure, and a sub-representation can refer to a complete molecular structure that is less accurate than the provided molecular structure representation. In embodiments, sub-representations have mathematically different scales. For example, a sub-representation can include different numbers of variables, different numbers of relevant variables, or other parameters that determine the mathematical scale of the mathematical problem. Preferably, one or more sub-representations computed in quantum computing are smaller than one or more sub-representations computed on a classical computer. However, in some cases, a sub-representation computed on a quantum computer can also be larger than at least one of the sub-representations computed on a classical computer, particularly where highly sophisticated quantum computers, such as fault-tolerant quantum computers, are available.

[0015] Geometric partitioning refers to dividing a molecule based on its geometric representation, taking into account the geometry of the molecular structure. For example, geometric partitioning may include selecting portions of a molecule that should be represented by sub-representations, including the corresponding atoms and the bonds between them. For instance, such sub-representations could refer to molecular cuts, which are portions of the molecular structure cut away from the rest of the molecular structure by virtually severing one or more molecular bonds between the molecular cut and the rest of the molecular structure. Therefore, geometric partitioning focuses on specific geometric portions of the corresponding molecular structure. Electronic partitioning refers to dividing a molecular structure based on its electronic structure. The electronic structure of a molecule refers to the electronic configuration describing each electron in the corresponding orbitals of a molecule or atom. For example, this electronic structure can be represented by corresponding molecular electronic orbitals. Partitioning a molecular structure based on its electronic structure involves dividing the corresponding molecule's electronic orbitals. An example of this partitioning is dividing the molecular orbitals into active and inactive spaces.

[0016] An intermediate solution determination unit is configured to enable a quantum computer to generate and provide a quantum computation solution for at least one of two or more sub-representations, and to enable a classical computer to generate and provide a classical computation solution for the remaining one or more sub-representations. To enable the quantum computer to generate and provide solutions for the corresponding sub-representations, the solution determination unit can be configured to generate control data to control the quantum computer to generate and provide quantum computation solutions for the corresponding sub-representations. Specifically, the solution determination unit can be configured to enable the quantum computer to perform quantum mechanical calculations based on the corresponding sub-representations such that the result of the quantum mechanical calculations indicates a solution for the corresponding sub-representation, and then generate the solution based on the result of the quantum mechanical calculations. The apparatus may include an optional conversion unit that provides a conversion from the corresponding sub-representation to a representative operational description, and the solution determination unit may be adapted to provide the representative operational description to the quantum computer so that the quantum computer performs quantum mechanical calculations based on the determined representative operational description. The representative operational description of the problem refers to a series of instructions that can be used to control the hardware of the quantum computer to perform a series of quantum operations on the quantum elements of the quantum computer to compute the corresponding problem. However, if the device does not include a conversion unit, the solution determination unit can be adapted to provide the corresponding sub-representation to the quantum computer, wherein a portion of the quantum computer system (e.g., the control unit of the quantum computer system) can be configured to provide a conversion from the corresponding sub-representation to a corresponding operational description, which provides operations that can be executed on the quantum computer to compute solutions to the corresponding sub-representation. Specifically, the solution determination unit can be communicatively coupled to the quantum computer or the control unit of the quantum computer, enabling the quantum computer to perform quantum mechanical calculations and obtain results indicating solutions to the corresponding sub-representations. However, the communicative coupling can also be an indirect coupling, for example via a storage unit such as a cloud storage unit, where the corresponding control data for the quantum computer to perform quantum mechanical calculations is stored and then sent to the quantum computer.

[0017] A quantum computational solution to a given electronic structure is an observable value of one or more measurements that have been transmitted and computed on a quantum computer. Specifically, observable values ​​can be represented by the states of the quantum elements of the quantum computer after a quantum computational calculation of the corresponding electronic structure problem. Examples of such observable values ​​are, for instance, energy, reduced density matrix, etc.

[0018] Furthermore, the intermediate solution determination unit is configured to enable a classical computer to generate and provide classical computational solutions for one or more sub-representations that have not been computed by the quantum computer. Specifically, the intermediate solution determination unit can be configured to prepare and control the classical computer using known methods for enabling the classical computer to solve the corresponding sub-representations. The solutions for one or more sub-representations can be any kind of contribution of one or more sub-representations to the solutions for the molecular structure representation. For example, a solution can refer to any contribution of the inactive space to the electronic structure problem of a chemical product. For example, the solution determination unit can be configured to determine the solutions for one or more sub-representations based on corresponding known algorithms (such as Hartley-Fock or density functional theory (DFT)). In a preferred embodiment, the solutions for one or more sub-representations are determined based on the solutions of the sub-representations computed on the quantum computer. Therefore, in this step, the influence of different parts of the molecule on each other can already be considered. In a preferred embodiment, self-consistent field (SCF) solutions are determined for one or more sub-representations. Therefore, the solutions for one or more sub-representations are a better approximation of the overall solution.

[0019] Since quantum computing is typically based on the electrical structure representation of the corresponding problem, intermediate solution determination units can be configured to generate electronic structure representations of one or more sub-representations among the corresponding sub-representations. Generally, electronic structure representations can be provided in any form that allows for the determination of the electronic structure problem to be solved and defines the quantities of the electronic structure problem and the interactions between these quantities. Preferably, the electronic structure representation refers to a mathematical description of the electronic structure problem to be solved, for example, a mathematical description of the electronic structure problem using electronic structure Hamiltonians. However, the electronic structure representation can also refer to any other explicit symbolic form of the corresponding electronic structure problem. Typically, the electronic structure representation is associated with the molecular structure of a chemical product. In particular, the molecular structure of the chemical product associated with the electronic structure representation can define the electronic structure problem and the quantities of the electronic structure problem. For example, if the chemical product refers to a specific molecule whose technical application characteristics should be determined, then the corresponding molecule and its structure define the quantities and interactions of the quantum mechanical electronic structure problem indicated by the electronic structure representation. Preferably, depending on the specific chemical product and / or the quantum mechanical electronic structure problem, the electronic structure representation includes or indicates atomic positions, basis sets, charges, and spin multiplicity.

[0020] Generally, electronic structure representation refers to the determination of intermediate properties associated with determining the electronic structure of atoms, molecules, crystals, or amorphous solids. For example, electronic structure representation can be used to determine the high-energy ground state of atoms and / or molecules. Based on the solutions of the electronic structure representation, especially based on intermediate properties, other measurable properties of atoms, molecules, crystals, or amorphous solids can be derived. Typically, electronic structure representation can be mathematically represented using different classes of underlying electronic structure Hamiltonians. Examples include Born-Oppenheimer Hamiltonians, non-Born-Oppenheimer Hamiltonians, Hamiltonians with additional single-electron potentials representing, for example, electrostatic fields, two-component Hamiltonians including spin-orbit coupling, and fully relativistic four-component Dirac Hamiltonians, etc.

[0021] Optionally, the apparatus may further include or utilize a conversion unit configured to convert the electronic structure representation into a representative operational description indicating a sequence of operations to be applied to the quantum elements of the quantum computer. Typically, the conversion unit may be part of the apparatus, but may be omitted from the apparatus, and may be part of the quantum computer, particularly the part of the quantum computer's control unit that automatically converts the provided electronic structure representation into a corresponding operational description. The conversion unit is configured to determine, in particular, the sequence of operations that enables the execution of quantum mechanical computations on the quantum computer. Typically, operations are performed by the quantum computer by manipulating the states of the quantum elements of the quantum computer, where quantum elements can refer to all elements in the quantum computer used to simulate the electronic structure representation, such as the quantum elements that form the qubits of the quantum computer. Methods for determining such operations for a problem to be computed on the quantum computer are generally known and can be used by the conversion unit to determine the operational description of the electronic structure representation of the sub-representation. For example, quantum phase estimation (QPE) or variational algorithms (such as variational Hamiltonian simulation (VHA) or variational quantum eigenvalue solver (VQE) with unit coupled cluster (UCC) simulation) combined with Jordan-Wigner or Bravyi-Kitaev transform and CZ algorithm or similar algorithm, and FSIM network algorithm using low-rank decomposition for unitary evolution can be used to transform and prepare the first part of the problem description for computation on a quantum computer. Examples of this transformation can be found in, for example, the following articles: “Quantum Computational Chemistry”, S. McArdle et al., Rev. Mod. Phys. 92, (2020); “Quantum Algorithms for Quantum Chemistry and Quantum Materials Science”, B. Bauer et al., Chem. Rev. (2020); and “Quantum Chemistry in the Age of Quantum Computing”, Y. Cao et al., Chem. Rev. (2019).

[0022] The intermediate property calculation unit is configured to generate one or more intermediate properties associated with the molecular structure by combining solutions from two or more sub-representations. The solution calculation unit is configured to generate solutions for the electronic structure representation by combining solutions from the sub-representations. Specifically, quantum computation solutions of the sub-representations computed in a quantum computer and classical computation solutions of the sub-representations computed in a classical computer are computationally combined to generate intermediate properties associated with the chemical product. Typically, the calculation of solutions to the sub-representations based on the results of quantum mechanical calculations can be performed by the intermediate solution determination unit before or during the combination.

[0023] Intermediate properties are electronic characteristics of the electronic structure of a molecular structure. Intermediate properties can be any characteristic that can be derived from solving the electronic structure representation of the molecular structure. Examples of intermediate properties can refer to the ground-state energy of a molecule or a general electronic system. In particular, this allows for the generation of reaction end products, thermodynamic properties, and kinetic properties as technical application characteristics within the context of determining the ground-state energies of all molecular species present in a chemical reaction. This understanding and reaction characteristics can then be used again to optimize chemical production processes, determine the microstructure of polymers, optimize material properties, etc. Furthermore, intermediate properties can also refer to the multipole moments of chemical products. Such calculations can be related to determining technical application characteristics of chemical products related to electrical and other technical application characteristics (e.g., the dielectric behavior of chemical products), and also to determining characteristics arising from intermolecular interactions that strongly depend on the polarity of the chemical product—the range of which can be from solubility and compatibility with certain media to ionic complexation behavior or effects on spectral characteristics (e.g., color). Determining the value of this intermediate property based on the results of solving the sub-representation is based on the corresponding intermediate property that should be determined and further on the information provided by the solution of the sub-representation.

[0024] The feature generation unit is configured to generate technical application features associated with a chemical product based on one or more intermediate features. This feature generation unit can be configured to utilize known relationships, such as physical or chemical relationships, between the generated one or more intermediate features and the technical application features. Alternative feature models based on machine learning can also be utilized. Preferred embodiments are described below.

[0025] The generated technical application characteristics can then be provided. For example, as a result, the technical application characteristics can be provided to a user interface or database. Preferably, the method includes providing the technical application characteristics to monitor and / or control a technical application including the chemical product, preferably, a chemical reaction including the chemical product and / or the synthesis of the chemical product. Therefore, the provision of the technical application characteristics is configured to be directly usable for the corresponding control and / or monitoring of the application. For example, the provision may include implementing the technical application characteristics into a corresponding control / monitoring system for the application. The provision may include providing the technical application characteristics in a corresponding data format, and providing information on how to use the technical application characteristics to control and / or monitor the application. The method may include providing control signals based on the technical application characteristics to monitor and / or control a technical application including the chemical product. The technical application can be any application within a technical context that includes the chemical product. The technical application can be a chemical reaction including a chemical product (e.g., a catalyst). The technical application can be a synthetic process for synthesizing a chemical product. The technical application can be a product design screening process that includes a screening process for determining one or more properties of a chemical product. The technical application characteristics can be provided for further processing in the technical application, particularly in a chemical application. Technical application characteristics can be provided to a production and / or experimental control system that controls the chemical production and / or experiment utilizing the chemical product, wherein the production and / or experimental control system can be configured to use the technical application characteristics to control and / or monitor the chemical production and / or experiment. The technical application characteristics can be associated with the control and / or monitoring processes and / or experimental processes of the chemical production associated with the chemical product. A providing unit can be configured to provide technical application characteristics for controlling and / or monitoring the chemical production processes and / or experiments associated with the chemical product. Technical application characteristics can be provided such that they are configured for the control and / or monitoring processes of the chemical production processes and / or experiments associated with the chemical product. For example, technical application characteristics can be used to control the production process associated with the chemical product to, for example, take into account the corresponding technical application characteristics with respect to one or more process parameters or with respect to the production of the chemical product, such that the corresponding target technical application characteristics are met. For example, monitoring can refer to verifying whether the chemical product actually meets the generated technical application characteristics, and if verification fails, modifying the production process or adjusting the generation accordingly, as described above. In one embodiment, the device further includes a control unit for providing control signals based on the generated technical application characteristics to monitor, control, produce, and / or process the chemical product. The provided control signals may refer to any signal from which actions for controlling, monitoring, producing, and / or processing the chemical product can be derived.

[0026] In an embodiment, the device may further include a target characteristic determination unit configured to a) receive a target technical application characteristic, and b) compare a technical application characteristic generated for the molecular structure of a chemical product with the target technical application characteristic, wherein the provision includes: if the comparison indicates that the generated technical application characteristic meets the target technical application characteristic within a predetermined allowable range, then providing the chemical product as the target chemical product. The device may further include an iteration unit configured to control iteration for determining the target chemical product based on the comparison, wherein the iteration includes a) if the comparison indicates that the generated technical application characteristic does not meet the target technical application characteristic within a predetermined allowable range, then providing a new molecular structure representation of the new chemical product, and b) repeatedly generating technical application characteristics based on the new molecular structure representation until a termination criterion is met or the target chemical product is determined. The chemical product may be any product produced in a chemical industrial process and may include one or more substances. The target technical application characteristic may be a measurement characteristic of an unknown chemical product, wherein the target molecular structure is the molecular structure of the unknown chemical product. In this case, the device is configured to provide an indirect measurement for measuring the molecular structure of the unknown chemical product. The target technical application characteristic can be a characteristic suitable for a predetermined technical application of a chemical product, and the target molecular structure is associated with a chemical product suitable for the predetermined technical application. In this case, the device is configured to provide a design process for identifying a chemical product for the predetermined technical application.

[0027] In an embodiment, the partitioning includes: identifying the most relevant molecular structure portion regarding the technical application characteristic based on the molecular structure representation; and partitioning the molecular structure representation such that a sub-representation is associated with the most relevant molecular structure portion and can be solved in quantum computing, wherein the intermediate characteristic computation unit is configured to enable the quantum computer to generate and provide a quantum computing solution for the sub-representation associated with the most relevant molecular structure portion. The most relevant molecular structure portion is the portion of the molecular structure whose influence on the corresponding technical application characteristic exceeds a predetermined threshold; for example, a change in the characteristics of the most relevant molecular structure portion causes a change in the determined technical application characteristic exceeding the predetermined threshold. Identifying the most relevant portion may include identifying at least a portion of the electronic structure representation associated with a correlation score higher than a correlation score associated with at least one other portion of the electronic structure representation, wherein the correlation score characterizes the association between two or more portions of the molecular structure representation and the technical application characteristic.

[0028] In an embodiment, the most relevant molecular structural portion is associated with the most relevant geometric portion of the molecular structure, and the partition refers to a geometric partition. Preferably, the partition includes generating an electronic structure representation for the most relevant geometric portion as a sub-representation to be computed on a quantum computer, wherein the electronic structure representation is associated with the electronic structure of the most relevant geometric portion of the molecular structure.

[0029] In embodiments, the most relevant molecular structural portion is associated with the most relevant electronic structural portion of the molecular structure, and the partition refers to an electronic partition. Preferably, the sub-representation of the most relevant molecular structural portion indicates an active space comprising a portion of the electronic structure associated with the molecular structure, and wherein one or more sub-representations indicate an inactive space comprising another portion of the electronic structure associated with the molecular structure. Preferably, the active and inactive spaces of the electronic structure representation refer to subspaces of the Hilbert space of the electronic structure representation. In particular, the basis of the Hilbert space of the electronic structure representation may be formed by electron orbitals and mathematically represented by the antisymmetric tensor product of electron orbitals, and the inactive and active spaces are defined by the electron orbitals they refer to. However, for example, if the electronic structure representation is formulated in another space, the active and inactive spaces may also be defined by other mathematical representations. At least one sub-representation is preferably defined based on the respective active or inactive space, for example, on the respective basis of the respective space, but may also include terms related to orbitals representing the respective other space, for example, terms related to electrons or electron orbitals forming the basis of the respective other space. Typically, sub-representations of electronic structure representations are determined based on the chemical product and the molecular structure to be described by the electronic structure representation. Specifically, since the molecular structure of the chemical product defines the quantities of the electronic structure representation and the relationships between these quantities, the molecular structure also defines how these quantities and relationships are described with respect to the sub-representation of the electronic structure representation. In particular, since the chemical product defines the electronic orbitals in the electronic structure representation, and especially which orbitals have the strongest correlations, these correlations should be considered more accurately in order to accurately solve the electronic structure representation of the molecular structure of the chemical product; therefore, the chemical product determines how to define the sub-representation. Specifically, sub-representations indicating the active space of the electronic structure representation are provided such that these sub-representations allow the use of quantum computers to calculate with high accuracy the most application-relevant parts of the electronic structure of the chemical product. In most cases, sub-representations indicating the active space are provided such that these sub-representations are based on the electronic orbitals most relevant to the specific application because these electronic orbitals exhibit the strongest correlations in the molecular structure of the chemical product during chemical reactions and / or change during the specific application, for example, participating in bond breaking and / or bond formation processes during the chemical reactions of the chemical product's molecular structure. Furthermore, the intermediate properties of chemical products depend on the active and inactive spaces. In particular, the intermediate properties and technical application properties depend on the electronic structure represented by the sub-representations indicating the active space and the sub-representations indicating the inactive space, and therefore depend on the solutions of the sub-representations of the molecular structure representation.

[0030] In an embodiment, the partitioning includes determining the most relevant geometric portion of the molecular structure, generating an electronic structure representation of the most relevant geometric portion, and determining the most relevant electronic structure portion within the electronic structure representation of the most relevant geometric portion as a sub-representation, such that the sub-representation can be solved in quantum computing. The intermediate characteristic computation unit is configured to enable the quantum computer to generate and provide a quantum computation solution for the sub-representation associated with the most relevant electronic structure portion of the most relevant geometric portion. Therefore, in this embodiment, geometric partitioning and electronic partitioning are combined to combine the benefits of both partitioning for particularly complex cases. For example, to calculate the possible binding sites of complex molecules (such as proteins), if the electronic structure of the relevant geometric portion of the binding site is calculated entirely, that portion may still be too large to be processed on an available quantum computer. However, in such cases, additionally partitioning the most relevant geometric portion into its electronic portion, particularly into orbitals involved in binding and those not involved in binding, can also utilize the benefits of quantum computing for this problem, thereby generally providing more efficient computation for such complex molecular structures. Preferably, the most relevant electronic structure portion of the most relevant geometric portion indicates the active space of the electronic structure of the most relevant geometric portion.

[0031] In embodiments, the one or more intermediate properties are associated with one or more of the following: electron density, electrostatic multipole moment, electronic state energy, excited state properties, partial charge obtained through population analysis of a single electron density matrix, and the derivative of the ground state or excited state energy. Intermediate properties may include excited state energy, and technical application properties may include one or more spectral properties of the molecule. Intermediate properties may include the ground state energy and optionally the electron density of one or more molecules, and technical application properties may include the chemical reactivity of the chemical reaction in which one or more molecules participate, such as the reaction free enthalpy or free activation enthalpy. Intermediate properties may include one or more properties that can be used to train machine learning algorithms, and technical application properties may include one or more properties used as labels when training machine learning algorithms.

[0032] In embodiments, the applied technical characteristics are associated with one or more spectral properties of the molecular structure, wherein the generated intermediate properties are associated with the electronic state energies of the molecular structure. Spectral properties can include any property associated with the spectrum of the molecule. Spectral properties can be characteristics of the molecule's spectrum. For example, spectral properties can include UV / Vis spectra, infrared spectra, transition energies, transition strengths, oscillator strengths, transition intensities, Frank-Condon curves, and electronic vibrational spectra. Spectral properties can refer to NMR spectral properties, which can be obtained from nuclear screening through the molecule's electronic environment. For example, intermediate properties can refer to the ground or excited state energies of the corresponding molecular structure. The spectral properties of the molecular structure can then be calculated based on the ground or excited state energies or based on their derivatives.

[0033] In embodiments, the technical application characteristics are associated with the thermodynamic and / or kinetic reaction characteristics of a chemical reaction involving one or more chemical products, wherein one or more intermediate properties are generated for the molecular structure of the chemical product involved, and wherein the one or more intermediate properties are associated with one or more electronic state energies. For example, intermediate properties can be generated for each molecular structure involved in the chemical product. One or more intermediate properties of the chemical product involved can be generated using the steps defined above for generating intermediate properties of the chemical product. In this embodiment, the method may include receiving a representation of the molecular structure of the chemical product involved in the chemical reaction. The method may also include receiving a representation of the chemical reaction (e.g., a chemical reaction equation or other representation), and deriving the chemical product involved and the corresponding molecular structure from the chemical reaction representation.

[0034] In a preferred embodiment, determining the thermodynamic and / or kinetic properties of a chemical reaction includes determining the vibrational zero-point energy and vibrational partition function of the corresponding molecular structure based on one or more electronic state energies as intermediate characteristics, wherein these thermodynamic and / or kinetic properties are further determined based on the vibrational zero-point energy and the vibrational partition function. The vibrational partition function can encode how probabilities are distributed among the different vibrational states of the molecular structure based on the individual energies of those states. The vibrational partition function can represent the sum of Boltzmann probabilities of vibrational states in which a normal mode of molecular vibration can be found. The vibrational partition function can be calculated within a harmonic approximation method. The vibrational partition function can represent the principal components of molecular energy and entropy. Determining the vibrational zero-point energy and vibrational partition function of the corresponding molecular structure based on one or more electronic state energies may include deriving one or more additional quantities from the one or more electronic state energies, and determining the vibrational zero-point energy and / or vibrational partition function based on the derived quantities. Specifically, the second derivative of the energy with respect to the nuclear coordinates is determined (e.g., by Hessé), and the vibrational spectrum in the harmonic approximation method can be derived from this second derivative as the basis for determining the vibrational zero-point energy and the vibrational partition function. Further, translational and rotational partition functions can be determined, and thermodynamic and / or kinetic reaction characteristics can be further determined based on these functions. Typically, the translational and rotational partition functions can be determined without utilizing the electronic state energy determined as an intermediate characteristic by a quantum computer. Further, for cases where at least one of the involved chemical products is in solution, it is preferable that one or more intermediate characteristics generated for at least one involved chemical product in solution are further determined based on electron density, preferably including electron density. Preferably, in this case, determining the thermodynamic and / or kinetic reaction characteristics of the chemical reaction further includes: i) determining the shielding charge density on the surface of the corresponding molecular structure based on electron density; ii) determining the chemical potential of the corresponding molecular structure in the corresponding solvent based on the shielding charge density; and iii) determining the Gibbs free energy of dissolution based on the chemical potential, wherein the Gibbs free energy of dissolution is used to determine the thermodynamic and / or kinetic reaction characteristics of the chemical reaction. For the molecular structure in the solvent, the shielding charge density can be an imaginary charge distribution that arises due to the presence of the solvent and exerts a shielding effect on the charge distribution of the dissolved molecular structure at a certain distance from the atomic nucleus where the assumed "molecular surface" is located. This distance can be element-type related or depend on, for example, an electron density threshold. The solvent can be described by its dielectric constant to be used to calculate the shielding density. The Gibbs free energy of dissolution can be used to account for the influence of the solvent when generating, for example, the reaction Gibbs free energy or the activation Gibbs free energy as thermodynamic and / or kinetic reaction characteristics, respectively.

[0035] In an embodiment, the device further includes an alternative characteristic model providing unit, configured to provide a machine learning-based alternative characteristic model, wherein the machine learning-based alternative characteristic model is trained based on a historical training dataset to generate the technical application characteristic based on one or more intermediate characteristics, wherein the characteristic generation unit is configured to determine the technical application characteristic by providing one or more intermediate characteristics as input to the alternative characteristic model and receiving the technical application characteristic as output to the alternative characteristic model. For example, as a historical training dataset, the alternative characteristic model can be trained using intermediate characteristics and technical application characteristics known from a variety of different chemical products. In particular, the alternative characteristic model can be trained for a specific application case (e.g., for one or more specific types of chemical products, preferably small molecules, polymers, etc.). The alternative characteristic model can also be trained for specific application cases, such as specific technical application characteristics, such as spectral, thermodynamic, and / or kinetic reaction characteristics. Furthermore, specific application cases can be combined. However, a more general alternative characteristic model can also be trained by correspondingly expanding the training dataset (e.g., by integrating different types of chemical products). Alternative feature models based on machine learning can be any machine learning-based algorithm, such as neural networks, decision tree algorithms, random forest algorithms, graph neural networks, etc.

[0036] In an embodiment, the device further includes a) a training data generation unit configured to generate a training dataset for a characteristic model based on one or more determined technical application characteristics and / or intermediate characteristics associated with one or more molecular structures; b) a training unit configured to train the characteristic model based on the generated training dataset to determine one or more technical application characteristics and / or intermediate characteristics based on predetermined input data; and c) providing the trained characteristic model. The input data may include at least one of the following: one or more technical application characteristics, one or more intermediate characteristics, and one or more molecular structure characteristics that can be derived from a molecular structure representation. The device for generating training data for the characteristic model allows for the generation of high-quality training data within a reasonable timeframe using reasonable computational resources. High-quality training data also allows for the use of specific training algorithms, which can operate with a reduced amount of training data, provided the training data quality is high. Reducing the amount of training data used to achieve a predetermined level of accuracy for the characteristic model further allows for a reduction in the energy and computational resources that must be used to train the characteristic model. Technical application characteristics and / or intermediate characteristics can then be generated using the trained characteristic model without utilizing the complex algorithms employed by quantum computers or devices. This further reduces the use of computing resources and the time spent generating technical application features and / or intermediate features. The feature model can be based on any machine learning-based algorithm. Preferably, the feature model is a regression model, such as a neural network, graph neural network, etc. The specific algorithm used can be determined based on the available training data, particularly the amount of available training data.

[0037] In another aspect of the invention, a system for generating technical application characteristics associated with chemical products is proposed, wherein the system comprises: a) the apparatus as described above; b) a quantum computer communicatively coupled to the apparatus; and c) a classical computer coupled to the apparatus, wherein the apparatus is configured to control the quantum computer and the classical computer to generate quantum computing solutions and classical solutions, respectively.

[0038] In another aspect of the invention, a system for determining molecular structures based on spectroscopic measurements is provided, wherein the system comprises: a) a measurement spectrum receiving unit for receiving spectra and / or one or more spectral properties of an unknown molecular structure; b) a molecular structure receiving unit for receiving one or more molecular structure representations; c) an apparatus as described above, configured to generate one or more spectral properties based on the molecular structure; and c) an iteration control unit for comparing one or more spectral properties of the generated molecular structure with the measured spectra and / or one or more spectral properties, wherein if the comparison indicates that the generated one or more spectral properties satisfy the measured spectra and / or one or more spectral properties within a predetermined permissible range, the molecular structure is provided as associated with the measured chemical product. Preferably, if the comparison indicates that the generated one or more spectral properties do not satisfy the measured spectra and / or one or more spectral properties within a predetermined permissible range, the iteration control unit is configured to begin generating new one or more spectral properties based on the new molecular structure and repeat the process until a predetermined termination criterion is met or the molecular structure is provided as associated with the measured chemical product. The association with the measured chemical product should be understood as being at least a part of the measured chemical product. For example, the measured chemical product may include or may be composed of the provided molecular structure. The received one or more molecular structures may be received as user input or may be stored in a database. The received one or more molecular structures may include a list of candidate molecular structures to be used for searching for molecular structures associated with the measured chemical product. The iterative control unit may be configured to provide new molecular structures from this list. The system can be considered as performing an indirect measurement of molecular structure using the measured spectrum and / or one or more spectral properties. Typically, a problem in spectroscopy is that many measured spectra are too complex to directly allow the derivation of the underlying structure from the spectrum. This hinders the determination of molecular structure within a chemical product using easily measurable spectra. The inventors have discovered, in particular, that the above method allows the generation of one or more spectral properties of complex molecular structures within a reasonable time and using reasonable low classical and quantum computing resources, making it possible to determine chemical products using measured spectra even in complex cases.

[0039] In another aspect of the invention, a method for determining molecular structure based on spectroscopic measurements is proposed, wherein the method includes a) receiving a spectrum and / or one or more spectral properties of an unknown molecular structure, b) receiving one or more molecular structure representations, c) generating one or more spectral properties based on the molecular structure according to the above method, and c) comparing the generated one or more spectral properties of the molecular structure with the measured spectrum and / or one or more spectral properties, wherein if the comparison indicates that the generated one or more spectral properties satisfy the measured spectrum and / or one or more spectral properties within a predetermined permissible range, then the molecular structure is provided as associated with the measured chemical product.

[0040] In another aspect of the invention, a system for determining a catalyst for a chemical reaction is provided, wherein the system comprises: a) a target chemical reaction rate receiving unit for receiving a target chemical reaction rate for the chemical reaction; b) a molecular structure receiving unit for receiving one or more molecular structure representations as candidate catalysts and one or more molecular structure representations of molecules participating in the chemical reaction; c) an apparatus as described above, configured to generate a reaction rate for a candidate catalyst based on these molecular structure representations; d) an iteration control unit for comparing the reaction rate of the generated candidate catalyst with the target reaction rate, wherein if the comparison indicates that the target reaction rate is met within a predetermined permissible range, the candidate catalyst is provided as a target catalyst; and e) a providing unit for providing the target catalyst for controlling and / or monitoring the chemical reaction and / or for synthesizing the candidate catalyst. For example, the chemical reaction rate of the determined candidate catalyst can be compared with the target chemical reaction rate, and based on the comparison, i) the corresponding candidate catalyst can be determined as the target catalyst, or ii) a new candidate catalyst can be provided, and the reaction rate determination according to the invention can be repeated based on the new candidate catalyst. The system can receive one or more molecular structures as user input or store them in a database. The received molecular structures can include a list of candidate catalysts to be used in the search for a target catalyst. The reaction rate can be determined based on the activation free enthalpy, which can be determined using the methods described above. An iterative control unit can be configured to provide new candidate catalysts from this list. This system allows for the search for catalysts in a computer-resource-efficient manner, enabling improvements to chemical reactions, particularly by identifying catalysts with higher reaction rates to increase chemical reaction efficiency. Typically, when searching for new catalysts, the complexity of the molecular structures involved in the chemical reaction is a problem. This hinders the search for new catalysts and thus the improvement of chemical reaction efficiency to reduce resource usage and waste. The apparatus and method described above allow for the discovery of new catalysts for chemical reactions within a reasonable timeframe and with reasonably low classical and quantum computing resources, enabling improvements in chemical reaction efficiency. In an embodiment, the method can be used to identify candidate catalysts from a variety of candidate catalysts that satisfy a target reaction rate within a predetermined allowable range, and to select the candidate catalyst with the highest reaction rate from the identified candidate catalysts. Therefore, the system allows for the selection of the most suitable catalyst from the list for use in the reaction. Providing a target catalyst may include providing control data for controlling and / or monitoring chemical reactions and / or for synthesizing the target catalyst.Control data can be configured to control the corresponding synthesis system used to synthesize the catalyst, and / or control and / or monitor the corresponding reactor undergoing the chemical reaction, for example, controlling and / or monitoring one or more feeds to the reactor to feed a corresponding amount of catalyst into the reactor. In embodiments, generating reaction rates includes generating reaction rates for a plurality of possible reactions involving candidate catalysts, and further determining a target catalyst based on the reaction rates. In particular, a target reaction rate can be provided for a desired reaction involving the catalyst, and another target reaction rate can be provided for an undesired reaction involving the catalyst, wherein the target catalyst can be determined based on the target reaction rates. In particular, preferably, the target catalyst includes a reaction rate for the desired reaction that is higher than the target reaction rate for the desired reaction and a reaction rate for the undesired reaction that is lower than the target reaction rate for the undesired reaction.

[0041] In another aspect of the invention, a method for determining a catalyst for a chemical reaction is provided, wherein the method includes a) receiving a target chemical reaction rate of the chemical reaction, b) receiving one or more molecular structure representations as candidate catalysts and one or more molecular structure representations of molecules participating in the chemical reaction, c) generating a reaction rate of the candidate catalyst based on the molecular structure representations according to the above method, c) comparing the reaction rate of the generated candidate catalyst with the target reaction rate, wherein if the comparison indicates that the target reaction rate is met within a predetermined permissible range, the candidate catalyst is provided as a target catalyst, d) providing the target catalyst for controlling and / or monitoring the chemical reaction and / or for synthesizing the candidate catalyst.

[0042] In another aspect of the invention, a system for determining a chelating agent for a predetermined metal is provided, wherein the system comprises: a) a target chemical complex formation constant receiving unit for receiving a target chemical complex formation constant for the binding of a metal with a chelating agent; b) a molecular structure receiving unit for receiving one or more molecular structure representations as candidate chelating agents and one or more molecular structure representations of a corresponding metal; c) an apparatus as described above, configured to generate a complex formation constant of the chelating agent and the metal based on the molecular structure representation; d) an iterative control unit for comparing the complex formation constant of the generated chelating agent with the target complex formation constant, wherein if the comparison indicates that the target complex formation constant is satisfied within a predetermined allowable range, the candidate chelating agent is provided as the target chelating agent; e) a providing unit for providing the chelating agent for controlling and / or monitoring a chemical reaction and / or for synthesizing the chelating agent. The chemical complex formation constant can be determined based on the reaction free enthalpy, which can be determined using the method described above. The chemical complex formation constant of the identified candidate chelating agents can be compared with the target complex formation constant, and based on this comparison, i) the corresponding candidate chelating agent can be identified as the target chelating agent, or ii) a new candidate chelating agent can be provided, and the determination of the complex formation constant according to the invention can be repeated based on the new candidate chelating agent. The received one or more molecular structures can be received as user input or stored in a database. The received one or more molecular structures can include a list of candidate chelating agents to be used for searching for the target chelating agent. The iterative control unit can be configured to provide new candidate chelating agents from this list. This system allows for the search for chelating agents in a computer resource-efficient manner, which allows for improvements in chemical reactions, particularly by searching for chelating agents that bind as strongly as possible to a predetermined metal and bind as little as possible to other metals. Typically, when searching for new chelating agents, the problem often lies in the complexity of the molecular structures involved in the chelating agent. This hinders the search for new chelating agents that bind effectively to a specific target metal and thus improve the efficiency of chemical reactions to reduce resource use and waste. The aforementioned apparatus and method overcome this problem and allow for the discovery of new chelating agents for metals within a reasonable timeframe and with reasonably low classical and quantum computing resources, thereby improving the efficiency of chemical reactions. In embodiments, the method can be used to identify candidate chelating agents from a variety of candidates that satisfy a target complex formation constant within a predetermined allowable range, and to select from the identified candidate chelating agents the one with the highest complex formation constant for a predetermined metal. Therefore, the system allows for the selection of the most suitable chelating agent from this list for use in the reaction.Providing a target chelating agent may include providing control data for controlling and / or monitoring the chemical reaction utilizing the target chelating agent and / or for synthesizing the target chelating agent. The control data may be configured to control and / or monitor the corresponding synthesis system or the corresponding reactor performing the reaction, for example, controlling and / or monitoring one or more feeds to the reactor to feed a corresponding amount of the target chelating agent into the reactor.

[0043] In another aspect of the invention, a system for determining a chelating agent for a predetermined metal is provided, wherein the system comprises: a) a target chemical complex formation constant receiving unit for receiving a target complex formation constant for the binding of a metal with a chelating agent; b) a molecular structure receiving unit for receiving one or more molecular structure representations as candidate chelating agents and one or more molecular structure representations of a corresponding metal; c) an apparatus as described above, configured to generate a complex formation constant of the chelating agent and the metal based on the molecular structure representation; c) an iterative control unit for comparing the complex formation constant of the generated chelating agent with the target complex formation constant, wherein if the comparison indicates that the target complex formation constant is satisfied within a predetermined allowable range, the candidate chelating agent is provided as the target chelating agent; d) providing the chelating agent for controlling and / or monitoring a chemical reaction and / or for synthesizing the chelating agent.

[0044] In another aspect of the invention, a method, particularly a computer-implemented method, for generating technical application characteristics associated with a chemical product, wherein the chemical product is associated with one or more molecular structures, wherein the method comprises: a) providing a molecular structure representation associated with the molecular structure of the chemical product; b) dividing the molecular structure representation into two or more sub-representations, wherein the division is associated with a geometric and / or electronic division of the molecular structure represented by the molecular structure representation; c) causing a quantum computer to generate and provide a quantum computing solution for at least one of the two or more sub-representations, and causing a classical computer to generate and provide a classical computing solution for the remaining one or more sub-representations; d) generating one or more intermediate properties associated with the molecular structure by combining the solutions of the two or more sub-representations, wherein the intermediate properties are electronic properties of the electronic structure of the molecular structure; e) generating technical application characteristics associated with the chemical product based on the one or more intermediate properties, wherein the technical application characteristics refer to measurable chemical and / or physical quantities associated with the molecular structure; and f) providing technical application characteristics associated with the control and / or monitoring processes of a chemical production process associated with the chemical product.

[0045] In another aspect of the invention, a computer program product for generating technical application characteristics associated with a chemical product is provided, wherein the computer program product is configured to cause the apparatus described above to perform the method described above.

[0046] It should be understood that the apparatus, method, system and computer program product described above have similar and / or identical preferred embodiments, particularly as defined in the dependent claims.

[0047] It should be understood that the preferred embodiments of the present invention may also be any combination of the dependent claims or the above embodiments and the corresponding independent claims.

[0048] These and other aspects of the invention will become apparent and will be illustrated with reference to the embodiments described below. Attached Figure Description

[0049] In the attached diagram:

[0050] Figure 1 This demonstrates the state representation of qubits used in quantum computing devices.

[0051] Figure 2 A schematic example of a quantum computing device using qubits as the computing unit is shown.

[0052] Figure 3 A schematic example method is shown for generating control signals to perform operations on a quantum computing device and for processing measurement signals from a quantum computing device.

[0053] Figure 4 A schematic example of a hybrid system including classical and quantum computing devices is shown.

[0054] Figure 5 A schematic example of a superconductor-based quantum computing device is shown.

[0055] Figure 6 A schematic example of a quantum computing device based on trapped ions is shown.

[0056] Figure 7 An embodiment of a system for generating properties associated with chemical products is illustrated schematically and exemplary.

[0057] Figure 8 A flowchart illustrating, and exemplarily demonstrating, is provided for a method of generating properties associated with a chemical product.

[0058] Figure 9 A flowchart illustrating, and exemplarily demonstrating, further details of a method for generating properties associated with a chemical product, and

[0059] Figures 10 to 12 A preferred application of the method is illustrated schematically and exemplary. Detailed Implementation

[0060] The following section first briefly introduces the general principles of quantum computers and their computational performance. Furthermore, the general principles can also be found in the following literature: "Quantum Computation and Quantum Information: 10..." th Anniversary Edition [Quantum Computing and Quantum Information: 10th Anniversary Edition]”, MANielsen and IL Chuang (2010).

[0061] Classical computing devices use transistor-based processors. Each transistor has two controllable states, 1 or 0, representing a digital binary or bit. To execute operations on a classical computing device, human-readable program code is translated into machine-readable instructions by a compiler. Machine-readable instructions are control signals for each transistor, such as voltage settings. The representation of machine-readable instructions can include binary or hexadecimal representations. Based on these machine-readable instructions, operations are executed on the processor of the classical computing device.

[0062] Quantum computing is a relatively new method of computation that uses quantum effects, such as superposition and entanglement, to perform certain calculations more efficiently than classical digital computers. Unlike digital computers, which represent information in bits (e.g., "1" or "0") as described above, quantum computing devices (i.e., quantum computers) use qubits (quantum bits) to represent information. Quantum computing devices are based on quantum elements that follow quantum mechanics, such as superconductors, ions, atoms, quantum dots, photons, particle spin, bosons, etc. These quantum elements can be manipulated in a controlled manner to perform operations.

[0063] Although qubits and their manipulation can be described according to their mathematical properties, each such qubit can be implemented in any of a variety of different ways in physical quantum elements. Examples of such quantum elements include superconducting materials, trapped ions, photons, optical cavities, single electrons trapped in quantum dots, point defects in solids (e.g., phosphorus donors in silicon or nitrogen-vacancy centers in diamond), molecules (e.g., alanine-vanadium complexes), or any medium exhibiting qubit behavior, which includes quantum states and transitions between them that can be controlled to sense or detect.

[0064] Typically, for any given physical quantum element that implements a qubit, any property of that physical unit can be chosen to implement the qubit. For example, if an electron is chosen to implement a qubit, the x, y, or z components of the electron's spin degree of freedom can be chosen as properties of that electron to represent the state of the qubit. For any particular degree of freedom, the physical quantum element can be controllably placed into a superposition or entangled state, and measurements can then be taken at the chosen degree of freedom to obtain a reading of the qubit value.

[0065] Compared to transistors in classical computing devices, each quantum element in a quantum computing device can not only take the form of a ground state... or Furthermore, arbitrary superposition of these ground states can also be used, such as states. The state of each quantum element is represented by the state of a quantum bit (qubit), such as... Figure 1 This is illustrated by the two-dimensional simplification. To represent this state, Dirac notation is commonly used in quantum mechanics. In Dirac notation, states in an n-dimensional complex vector space (such as Hilbert space) are represented using braket notation, for example, In conventional terminology, the superposition of "0" and "1" states in a quantum computing device can be represented as... The states "0" and "1", or bits, of a classical computing device are analogous to the ground state of a quantum computing device. and Or a quantum bit. Value This indicates that the quantum bit will The probability of being measured in a given state, and the value This indicates that the quantum bit will The probability of being measured in a state. If there is more than one qubit, two or more qubits may become entangled. Entanglement means that the state of one qubit depends on the state of at least one other qubit, and vice versa. Furthermore, in an entangled state, the individual qubits can no longer be considered as individual qubits. Typically, in quantum computers... A register of 10 qubits can be in a superposition of ground states simultaneously, and A classical bit register can only be in a single ground state at a time. Therefore, compared to classical computing devices on quantum computing devices, it is possible to manipulate and process data simultaneously. Each ground state enables exponential intrinsic parallelism.

[0066] To perform operations on a quantum computing device, the computational methods used to solve a given problem can be translated into qubit manipulations, which in turn can be translated into control signals for manipulating the qubits. The representation of machine-readable instructions can include common quantum mechanical representations of operations in Hilbert space. Different representations of the qubit states can be chosen depending on the specific implementation of the quantum computer. Any state preparation on a quantum computing device can be represented by manipulations acting on the qubit states. Manipulations can be translated into control signals to control corresponding parts of the quantum computer, depending on the type of quantum computing device used. Thus, based on manipulations acting on the qubit states, operations can be performed on the quantum equivalent of a classical processor (as part of the quantum computing device).

[0067] In gate-based quantum computer systems, manipulations acting on the state of a qubit can typically be one or more qubit operations. A qubit operation can change the state of a qubit, for example, changing it to a state similar to... Figure 1 The vectors shown The rotation corresponds to a specific superposition. For example, in a superconducting quantum computer, this can be achieved by microwave pulses, or in a trapped-ion quantum computer by irradiating ions with a laser beam. Multi-qubit operations can generate entanglement between two or more qubits. For example, in a superconducting quantum computer, this can be achieved by connecting qubits via an intermediate electrical coupling circuit, or in a trapped-ion quantum computer by controlling the collective vibration of trapped ions.

[0068] Typically, to prepare manipulations for solving a given problem, the corresponding quantum mechanical representation of the problem is converted into qubit manipulations, which are then performed to prepare a solution to the given problem. After the predetermined solution is prepared—that is, after operations are applied to the qubits of the quantum computer—a projection measurement is performed on all individual qubits, returning either 0 or 1 for each qubit. This projection typically occurs at the qubit level. In the characteristic basis, this characteristic basis is also used to define the computational ground states "0" and "1" of a qubit. This means that only one state can be measured simultaneously as "0" and "1". The product of Pauli operators can be directly transformed into operators of such operators. On quantum computing devices, this measurement is achieved by applying a series of hardware-specific readout protocols (including control pulses) that manipulate readouts and monitoring the response to those pulses. For example, a superconducting qubit can be coupled to a hardware resonator. Measuring the shift in the resonator frequency allows the determination of the qubit's state, as this shift depends on the state of the coupled qubit. For example, in the case of trapped ions, optical readout can be used; for instance, if the ions emit light, the qubit's state is 1, and if the ions do not emit light, the qubit's state is 0, and vice versa. In this way, qubits can be used to implement logic circuits or gates, as in classical computing devices.

[0069] exist Figure 2 The image shows a schematic example of a quantum computer. Figure 2 The quantum computing device 100 shown includes: a quantum register 104 configured to perform quantum computation, a manipulation section 106 configured to manipulate the quantum register (particularly the quantum elements forming qubits), and a readout section 108 configured to collect measurement signals from the quantum register 104 to read out the qubits after a quantum mechanical computation. Specifically, the manipulation section 106 provides manipulation signals for manipulating the quantum register, wherein these manipulation signals are generated based on received control signals determined based on corresponding operations to be performed on the qubits. In some embodiments, a feedback loop may be provided between the manipulation section 106 and the measurement section 108. In the case of a gate-based quantum computer, quantum computation, compared to classical computation (where a measurement cycle provides the state of a transistor), involves performing multiple measurement cycles to provide a probability density or probability of the qubit state.

[0070] The quantum register 104 can be based on different quantum elements representing qubits. In some embodiments of gate-based quantum computers, qubits can be implemented as quantum elements by photons. Such an optical quantum computing device may include a laser that generates photons provided to a waveguide. A beam splitter may be provided for manipulating the photon state based on a manipulation signal, such as a mechanical rotation applied to a mirror. In this embodiment, the measurement section 108 may be a photon detector, and the measurement signal may be a photon.

[0071] In other embodiments of gate-based quantum computers, qubits can be implemented using the electronic states of ions trapped in a magnetic field. In this case, the manipulation section 106 can utilize a laser, and the manipulation signal can provide control laser pulses. Furthermore, in this case, the readout section 108 can be a photon detector combined with the readout laser pulse, and the measurement signal 102 can be a photon. Other qubit implementations can be based on superconductors as quantum elements, semiconductor materials with anyons as quantum elements, etc.

[0072] Figure 3 A schematic exemplary method is shown for generating control signals to perform manipulation on a quantum computing device and for processing measurement signals from the quantum computing device. In most embodiments of quantum computing devices known to date, the control signals for the quantum computing device are prepared on a classical computing device, and the measurement signals provided by the quantum computing device are further processed on the classical computing device. However, as quantum computing devices mature, other embodiments are conceivable. In the following examples, a quantum computer refers to a gate-based quantum computer, and manipulation refers to the operation on the quantum elements of the quantum computer.

[0073] To generate control signals for performing operations on the quantum computing device, in step S10, a problem to be solved using the quantum computing device is preferably provided in a mathematical description. For example, such a problem may include a mathematical description based on the electronic structure of a material to determine material properties. Other problems may include optimization problems and associated objective functions. In step S12, based on the problem to be solved, an operational description of the problem or subproblems may be generated, wherein the operational description includes operations to be applied to the qubits of the quantum computer to solve the problem in quantum mechanical calculations. Further, the operational description may include a reference state that allows for the generation of a representation of the initial qubit state on the quantum computer, and then further operations are applied to the quantum computer by manipulating the qubit state. Then, in step S14, based on the operational description, control signals can be generated to control the quantum computer, for example, by providing these control signals to a manipulation unit, which can then manipulate the qubit state based on these control signals. In step S16, the manipulation unit then applies manipulation operations to one or more qubits of the quantum computer, wherein the qubits perform quantum mechanical calculations based on these manipulation operations. After this manipulation, in step S18, a measurement signal may be generated to determine the result of the quantum mechanical calculation. This step may include reading out (i.e., measuring) the qubit state after applying manipulation to the initial qubit state. Then, in step S20, the measurement signal may be converted into a measurement quantity on a classical computer, and if in the case of a subproblem, it may be fed back into the problem to be solved. Finally, in step S22, the result of the problem computation, including quantum mechanical calculations, may be provided on a classical computing device.

[0074] Figure 4 A schematic example of a hybrid system including classical and quantum computing devices is shown. (See reference...) Figure 3 The methods described herein indicate that quantum computing devices are typically used in conjunction with classical computing devices. For example... Figure 4 As shown, the problem preparation system (e.g., a control signal generation device) can be implemented to perform, for example... Figure 3The illustrated method uses a classical computing device 110 for steps S10, S12, S20, and S22. A control unit can then be provided as an interface between the classical computing device 110 and the quantum computer 100, wherein the control unit can also be, for example, a classical computing device performing step S14. The control unit can then be communicatively coupled to a manipulation section 106, which can control the manipulators of the quantum computing device. Furthermore, the manipulation section 106 can be implemented as a classical computing device, for example, classical control hardware for controlling specific hardware components in the quantum computer that perform qubit manipulation. However, the manipulation section 106 is generally considered part of the quantum computer because it directly affects the quantum registers. The quantum computing device 100 is adapted to perform quantum operation S16 specifically by manipulating the qubits of the quantum registers. The measurement section 108, which is also generally considered part of the quantum computing device, can then perform step S18 using classical hardware. The measurement section 108 can then be communicatively coupled to the preparation system 110 for further processing of the measurement signal.

[0075] Figure 5 A schematic example of a superconductor-based quantum computing device is shown. Superconducting quantum computing devices are one type of solid-state quantum computing technology. Here, the quantum register 104 may include superconducting circuits 520, 522, and 524 based on a Josephson junction. Then, depending on the number of superconducting circuits selected to represent qubits, the qubits may refer to, for example, charge qubits, flux qubits, transport qubits, or phase qubits. Figure 5 This is a simplified illustration of a superconducting quantum computer utilizing charge qubits. For a charge qubit, the different states of the qubit are represented by an integer number of Cooper pairs on a superconducting island. In the case of gate-based quantum computing, quantum manipulation can be achieved by manipulating the qubits via microwave pulses. Resonators 512, 514, and 516 can be used to manipulate the state of the qubits by applying microwaves, or to read out the state of the qubits by measuring the corresponding microwaves; typically, different resonators are used for manipulating the state of the qubits and reading out the qubits. Furthermore, resonator 518 can be used to apply microwaves that cause entanglement of the qubits. However, besides resonator 518, entanglement can also be achieved through inductive or capacitive coupling of a superconducting circuit, or even by providing another qubit (here, a superconducting circuit) between the qubits to be entangled.

[0076] At the operational level, such systems are maintained at extremely low temperatures, for example, tens of mK. This extreme cooling keeps the superconducting material below its critical temperature and helps avoid unwanted state transitions. To maintain this low temperature, quantum information processing systems can operate within cryostats (such as dilution freezers). In some implementations, control signals are generated in a higher-temperature environment and transmitted to the quantum computer using a shielded impedance-controlled GHz transmission line (such as a coaxial cable). In some implementations, a dispersion detection scheme is used to measure the state of a superconducting qubit. To read out or detect the state of any qubit, a probe signal (e.g., traveling-wave microwave) can be excited along a readout transmission line coupled to the qubit via a corresponding readout resonator. The frequency of the probe signal can be near the resonant frequency of the readout resonator. Depending on the internal quantum mechanical state of the qubit, the intensity or phase of the probe signal transmitted along the readout transmission line can be varied because the reflectivity of the readout resonator coupled to the qubit varies according to the qubit's state. This allows for state detection of the qubit, where, during the readout of the qubit's state, the qubit's state collapses, i.e., is projected onto one of its ground states with a corresponding probability. By performing multiple quantum mechanical calculations and readouts, the corresponding probabilities can be determined. Further details of the superconducting quantum device are described in, for example, documents EP 3830867 A1, EP 3449427 A1, US 2020272925 A1, CN 212061223 U, and US 2019019099 A1.

[0077] Figure 6 A schematic example of a quantum computing device based on ions in an ion trap is shown. Similar to neutral atom traps, ion traps with, for example, positively charged calcium ions can be used to implement quantum computing devices. Here, ions 626 are trapped in an oscillating electromagnetic field 624 within a high or ultra-high vacuum. Ions 626 are cooled by a laser and held in the oscillating electric field 624. For qubit manipulation (such as superposition or entanglement), lasers 628 of different frequencies can be used.

[0078] Generally, based on the aforementioned quantum computer implementation methods, gate-based computations can be performed on quantum computer hardware architectures. Gate-based computations are based on quantum gates. Compared to classical gates, there are countless possible single-qubit quantum gates that can change the state vector of a qubit. Changing the state vector of a qubit is often referred to as a single-qubit rotation, and in this paper, it can also be referred to as a state change or a single-qubit quantum gate operation. Rotations, state changes, or single-qubit quantum gate operations can be mathematically represented using a unitary 2 × 2 matrix with complex elements. A rotation corresponds to the rotation of the qubit state within its Hilbert space, which can be conceptualized as a rotation of a vector on a Bloch sphere, where the Bloch sphere is often referred to as the geometric representation of the pure state space of a qubit. Multi-qubit gates change the quantum states of a set of qubits. For example, a two-qubit gate rotates the states of two qubits into rotations of those two qubits in a four-dimensional Hilbert space, where, as is well known, Hilbert space is an abstract vector space with an inner product structure that allows the measurement of lengths and angles. Furthermore, Hilbert space is complete, meaning there exist sufficient limits in the space to allow the use of calculus techniques.

[0079] In the following text, the term "operation description" refers to a representation of a problem that includes a sequence of quantum operations to be applied during the quantum mechanical computation of the problem. In the context of this invention, the term "quantum operation" can include all types of quantum gates as described above, and more generally includes all manipulations of quantum elements known on any quantum computer hardware. Furthermore, the term can also include operations performed on components (and optionally components representing the boson field itself) in a quantum computer that represent the coupling between the quantum elements forming qubits and the boson field. These operations then involve any kind of change representing the coupling of the components or the state of the boson field, such as switching the coupling on and off, or changes in the field frequency. Further, in some applications, quantum operations can also include measurement operations. This allows the use of measurement feedback to implement algorithms. For example, in such an algorithm, the quantum computer can execute a quantum gate defined by a sequence of quantum operations, then measure only a subset (i.e., less than all) of the qubits or other computational elements (such as the boson field state) in the quantum computer, and then determine which further quantum manipulations to perform next based on the results of one or more measurements. In particular, measurement feedback can be used to perform quantum error correction, but is not limited to it.

[0080] Not all quantum computers are gate-based quantum computers. Embodiments of the present invention are not limited to utilizing gate-based quantum computers. As an alternative example, embodiments of the present invention may also utilize, in whole or in part, quantum computers implemented using the quantum annealing paradigm, an alternative to the gate-based quantum computing paradigm. More specifically, quantum annealing is a metaheuristic method that uses the process of quantum fluctuations to find the global minimum of a given objective function on a given set of candidate solutions (candidate states). In particular, quantum annealing is closely related to adiabatic quantum computing.

[0081] Generally, the quantum annealing process also begins by using a classical computer to provide or generate an initial Hamiltonian and a final Hamiltonian based on the computational problem to be solved, and provides the initial Hamiltonian, the final Hamiltonian, and the annealing schedule as inputs to the quantum computer. In the case of an annealing process for solving an optimization problem, preferably, the final Hamiltonian refers to the Ising Hamiltonian representing the optimization problem, or a good approximation of the ground state representing the Ising Hamiltonian of the optimization problem. Then, for example, by using a corresponding control unit that controls the manipulation part of the quantum computer, the quantum computer is adapted to prepare a relatively easy-to-prepare initial state based on the initial Hamiltonian, such as a quantum mechanical superposition of all possible states (e.g., candidate states) with equal weights. After the initial state is prepared on the quantum computer, the initial state then evolves according to the annealing schedule following the time-dependent Schrödinger equation, which refers to the natural quantum mechanical evolution of the quantum computer physical system. More specifically, the state of the quantum computer undergoes time evolution under the time-dependent Hamiltonian, starting from the initial Hamiltonian and terminating at the final Hamiltonian. If the evolution is slow enough, the system will remain in a ground state close to the instantaneous Hamiltonian. At the end of the time evolution, a set of qubits (i.e., quantum elements) on the quantum annealer are in a final state that is expected to approximate the ground state of the Ising Hamiltonian, corresponding to a solution to the primal problem (e.g., an optimization problem). The final state of the quantum computer can then be measured, producing a result that can be used to solve the primal problem. The measurement operation can be performed, for example, in any of the ways described above. The classical computer can then post-process the measurement result to produce an output representing a solution to the primal computational problem. For example, the quantum annealer described above can be implemented on superconducting quantum computer hardware.

[0082] Furthermore, embodiments of the present invention can also utilize, in whole or in part, quantum computers implemented using a one-way quantum computing architecture (also known as a measurement-based quantum computing architecture). More specifically, a one-way or measurement-based quantum computer refers to a quantum computing method that first prepares an entangled resource state, typically a cluster state or graph state, and then performs a single-qubit measurement on it. It is "one-way" because the resource state is destroyed by the measurement. In such an architecture, the result of each individual measurement is random, but they are correlated in a way that ensures the computation always succeeds. Generally, the basis for subsequent measurements needs to depend on the results of previous measurements, so not all measurements can be performed simultaneously.

[0083] Figure 7 A system for controlling and / or monitoring the production and / or design of chemical products using an apparatus for generating technical application characteristics is illustrated schematically and exemplary. The apparatus includes a molecular structure representation providing unit, a partitioning unit, an intermediate solution determination unit, an intermediate characteristic calculation unit, a characteristic generation unit, and a providing unit. Further, the system includes several optional features for interacting with the apparatus. For example, the system may include a user interface that allows a user to perform machine-guided interaction with the apparatus, particularly allowing the provision and partitioning of the molecular structure representation. Further, the system may include a quantum computer and / or a classical computer accessible by the apparatus; for example, the apparatus, and particularly the intermediate solution determination unit, may be configured to generate control data for controlling the quantum computer and / or the classical computer, so that the quantum computer and / or the classical computer provide solutions to the sub-representations indicated by the control data. Further, the system can be integrated into the production of chemical products, for example by providing control data for controlling and / or monitoring the corresponding production of the chemical products, or integrated into the chemical product design process, for example by providing control data that allows monitoring and / or control of the corresponding screening of the chemical products for corresponding target application values.

[0084] The following will combine Figure 7 and Figure 8 They will be described together, among which, Figure 8 This refers to a flowchart of a method for generating technically applicable properties associated with a chemical product, the method comprising, for example... Figure 7The system and apparatus shown are executed. In the first step, the molecular structure representation providing unit can provide a molecular structure representation associated with the molecular structure of the chemical product. For example, the molecular structure representation providing unit can interact with a user interface to prompt the user to provide a molecular structure representation, for example, by providing a corresponding selection menu from which the user can select a molecular structure representation. For example, the user can provide the name or ID of the chemical product, which then allows the molecular structure providing unit to access a database on which the molecular structure of the corresponding chemical product is stored. However, the user interface can also directly allow input of SMILES or SMARTS representations, or reference to a graphical geometric representation of the chemical product. However, the molecular structure representation unit can also be configured to provide a molecular structure representation by accessing a corresponding database on which one or more molecular structure representations are also stored, for example, for performing corresponding processing.

[0085] In the next step, the partitioning unit is configured to divide the molecular structure representation into two or more sub-representations. This partitioning is associated with the geometric and / or electronic partitioning of the molecular structure represented by the molecular structure representation. For example, the partitioning unit can also be configured to interact with a user interface to partition the molecular structure representation during machine user interaction. For example, the partitioning unit can be configured to control the user interface to display the geometric and / or electronic representation of the molecular structure representation and prompt the user to indicate which parts of the corresponding representation should be part of which sub-representation. For example, appropriate 2D or 3D drawing tools can be provided to the user to allow marking the corresponding sub-representations in the graphical representation of the molecular structure representation provided by the user interface. However, the partitioning unit can also be configured to automatically provide the partitioning of the molecular structure representation, for example, based on predetermined rules for the corresponding predetermined chemical product. In particular, the rules enable the determination of which parts of the molecular structure are most relevant to generating the corresponding technical application characteristics and which structures are less relevant. For example, if the reaction kinetics of a chemical reaction should be determined, a rule can be provided indicating that all molecular structures within a predetermined radius around the binding sites involved in the chemical reaction are considered relevant, while the chemical product portions outside that radius are considered less relevant. Furthermore, additional criteria such as available quantum or classical computing resources, and the accuracy of the optimal solution, can also be considered as part of the rules used by the partitioning unit to partition the molecular structure representation. The partitioning unit can then be configured to provide the user with automatic partitioning for confirmation or modification via a user interface. However, in this case, this user-supervised step can be omitted. Furthermore, partitioning can also refer to simply receiving the sub-representations of the molecular structure representation partitioned, for example, via a user interface or through a corresponding database, without any further processing by the partitioning unit.

[0086] In the next step, the intermediate solution determination unit is configured to cause the quantum computer to generate and provide a quantum computation solution for at least one of two or more sub-representations, and to cause the classical computer to generate and provide a classical computation solution for the remaining one or more sub-representations. Specifically, the intermediate solution determination unit can be configured to provide control signals to control the respective quantum computer and / or classical computer to perform the corresponding computations, thereby providing a solution for the corresponding sub-representation. For example, the control signals may include the corresponding computations and readable formulas of the sub-representation, as well as corresponding instructions indicating that the sub-representation should be computed using the corresponding computational resources and the solutions provided to the intermediate solution determination unit. The corresponding control signals can then be provided to the quantum computer and / or classical computer accordingly. In most cases, it is preferred that the provided control signals do not provide direct control over the quantum computer and / or classical computer, but rather allow the derivation of corresponding control signals for the hardware of the quantum computer and / or classical computer through the corresponding specific control units of the quantum computer and / or classical computer. This makes the method and apparatus independent of the specific control and / or hardware of a particular quantum computer or classical computer. For example, as briefly discussed above, different quantum computers can involve very different hardware configurations and are therefore controlled by correspondingly different control signals for the hardware to compute the corresponding quantum computing problems. The same applies to classical computers, particularly classical supercomputers and clusters, which typically also provide different interfaces for controlling and / or monitoring the software and hardware configurations on the supercomputer or cluster. However, of course, if the device is configured to always utilize the same quantum computer and / or classical computer to solve the corresponding subproblems, the intermediate solution determination unit can also be configured to directly control the corresponding quantum computer and / or classical computer. The corresponding quantum and classical computations of the subrepresentations are then performed, and the results of these computations (e.g., solutions) are provided to the intermediate solution determination unit, which receives the results for further processing.

[0087] Based on the solutions (e.g., results) obtained from the corresponding calculations, the intermediate property calculation unit is configured to generate one or more intermediate properties associated with the molecular structure by combining solutions from two or more sub-representations. Examples of solution combinations for different application cases will be provided further below.

[0088] The characteristic generation unit is then configured to generate technical application characteristics associated with the chemical product based on one or more intermediate characteristics. Furthermore, examples of different application cases will be given in further detail below regarding the generation of technical application characteristics. The providing unit can then be configured to provide technical application characteristics associated with the control and / or monitoring processes of the chemical production process associated with the chemical product. For example, providing may refer to providing control data for the production process or design process of the chemical product based on the technical application characteristics. For example, the thermodynamic and / or kinetic characteristics of the chemical reaction in the chemical production system can be controlled and / or monitored during the execution of the corresponding reaction. Furthermore, the chemical production process can be monitored based on the technical application characteristics of the chemical product to determine whether the correct intended chemical product has been produced, for example, by measuring the technical application characteristics of the chemical product and comparing the measured characteristics with the generated technical application characteristics. If the measured technical application characteristics meet the generated technical application characteristics within permissible ranges, the process can continue without modification, and / or the same process can be repeated for the next batch of chemical product. If the measured technical application characteristics do not refer to the generated technical application characteristics, it can be considered that the expected chemical product has not been produced, and the chemical process can be controlled to modify one or more process parameters, or it can be determined that the generated technical application characteristics have not been determined accurately enough and must be recalibrated.

[0089] Furthermore, in the chemical production design process, the corresponding target technical application characteristics that the chemical product to be produced should meet can be provided, and the generated technical application characteristics can be compared with the corresponding target technical application characteristics. If the generated technical application characteristics meet the target technical application characteristics, the corresponding chemical product can be considered the target chemical product, and control data for producing the chemical product can be provided, such as measuring the technical application characteristics, to ensure that the generated target technical application characteristics are the same as the measured target technical application characteristics. If the corresponding generated technical application characteristics do not meet the target technical application characteristics, the method can include providing new molecular structure representations of different chemical products and regenerating the technical application characteristics. For example, new molecular structure representations can be provided from a list of possible molecular structure representations, from which chemical products that meet the corresponding target technical application characteristics should be searched. However, the user can also be prompted to provide the corresponding new molecular structure representation, or a new molecular structure representation can be automatically generated by modifying one or more molecular structures according to predetermined rules. Therefore, for this application, the corresponding generated technical application characteristics can be used to screen multiple chemical products in order to define chemical products that meet the corresponding target technical application characteristics.

[0090] The following provides several more detailed embodiments and examples of the invention. To enable quantum computing to be used in real-world applications, particularly for solving real-world and / or industrially relevant problems (such as chemical and materials problems) related to electronic structure, the invention follows a general strategy: a) preprocessing real-world chemical and materials problems to reduce the quantum resource requirements (e.g., the number of qubits utilized and / or the number of quantum operations) of the quantum computer compared to brute-force processing of the entire real-world problem on a quantum computer; and / or b) post-processing electronic structure properties as intermediate characteristics (obtained via quantum mechanical calculations on a quantum computer) to obtain technically applicable characteristics that can be directly compared to experimentally available characteristics and / or directly contribute to the design of new chemical products, materials, and processes. Examples of real-world chemical and materials problems involving electronic structure and applications that are expected to strongly benefit from quantum computing are described below.

[0091] The workflow will be described in more detail below. Figure 9 An overview of an embodiment of the workflow according to the invention is shown. In a first step, a real-world chemical or materials problem to be solved is defined. For example, the problem may be related to the fields of chemical reactivity or spectroscopy, as described later. In a second step, the molecule or material of interest is digitally represented on a classical computer as, for example, a molecular structure representation, and further preprocessed, for example, by generating a corresponding sub-representation, thereby reducing the requirements for a quantum computer. This can be achieved through a multi-scale approach that can utilize the geometric partitioning of the molecule or material, and / or (as in active space techniques) the electronic partitioning of the molecule or material involving the partitioning of electron orbitals in space.

[0092] Typically, within the framework of multiscale approaches, real-world chemical or materials problems are divided into various subproblems of different length scales, time scales, resolutions, and / or complexities. These subproblems are linked together analytically or numerically. Each of these subproblems can be computed with varying degrees of accuracy using different computational methods. This multiscale approach often allows for a good balance between accuracy and computational feasibility when attempting to solve real-world chemical or materials problems, which is usually impossible when the problem is considered only in its entirety or merely as a model version.

[0093] As an example, the so-called ONIOM ("our own n-layered integrated molecular orbital and molecular mechanics") method is described below. More information can be found in the following article: "The ONIOM Method and Its Applications", Lung Wa Chung et al., Chem. Rev. 115, 5678 (2015). ONIOM is a multi-scale method that enables the application of different ab initio, semi-empirical, and molecular mechanics methods to different geometric parts of real-world chemical or material systems (such as large molecules) to reduce overall computational costs while maintaining consistency of results. The goal is to perform high-level (e.g., highly accurate) calculations only on the small geometric parts most relevant to the properties being treated in the system, and to include the effects of the rest with lower-level (e.g., less accurate) theory, while the final result has an accuracy comparable to high-level calculations on the entire system. This approach is particularly advantageous in the context of quantum computing because it allows for the preparation of real-world chemical or materials problems, enabling the use of a quantum computer to compute only a small geometric part of the system (i.e., a so-called "model" system, which may correspond to the most electronically complex and / or chemically relevant part of the system, such as the strongly correlated part of a molecule), while a classical computer is used to compute the effects of the remaining parts (which may be less complex and / or chemically less correlated) using lower-level theory. For example, this concept can be applied to practically relevant organometallic complexes, such as chelators, transition metal catalysts, and biomolecular systems that interact with transition metal ions (e.g., iron, nickel, rhodium, or palladium). Those systems typically contain strongly correlated homonuclear or oligonuclear centers in a weakly correlated environment, where the term "nucleus" refers to the number of single ions or single atoms of transition metal ions or atoms plus optionally ligands. For example, the transition metal ions and their nearest neighboring ligand atoms in an organometallic complex, or the portion of the molecule where bonds break or form during the chemical reaction under consideration, can be chosen as the "model" system for computation using a quantum computer. As another example, in biomolecular systems (such as enzymes, e.g., peptidases or esterases), the active site can be selected as a "model" system and computed using a quantum computer. In all the examples above, the remainder and its effects on the "model" system can then be computed using a classical computer with lower-level theory. The "model" system and the remainder can be represented by corresponding sub-representations of molecular structures.

[0094] Therefore, using the ONIOM method, quantum resources are spent only on the chemically most important geometric parts of the system (such as the parts where chemical bonds are broken and / or formed) and / or the most challenging parts of the system (such as the parts exhibiting strong electronic correlations), where classical computers are known to provide insufficient accuracy or infeasible computation time. Thus, using this method, the quantum resource requirements of a quantum computer (e.g., the number of qubits and the number of quantum operations, which typically increase with the size of the system to be computed on the quantum computer) can be significantly reduced compared to processing the entire system. This allows quantum computing to be used to solve real-world chemical and materials problems much earlier than previously possible. It should be noted that the ONIOM multiscale method relies on a fairly intuitive geometric partition based on the actual geometry of the molecule or material, compared to active space techniques that rely on more abstract electronic partitioning in the space of electron orbitals.

[0095] More specifically, for example, in the bilayer ONIOM method, the complete system including all atoms is referred to as the “real” system and is processed using efficient lower-level theories on classical computers, such as truncated configurational interactions (CI) or truncated coupled clusters (CC) methods, Mueller-Plessetter (MP) perturbation theory, random phase approximation (RPA) methods, density functional theory (DFT) methods, semi-empirical methods (such as PM6 or xTB), molecular mechanics (MM) methods, or potentials based on neural networks applied to molecular representations (e.g., ANI or AIMNet). MM methods rely on predefined force fields; for more details, see, for example, the following article: “Mixed Quantum Mechanical / Molecular Mechanical Molecular Dynamics Simulations of Biological Systems in Ground and Electronically Excited States”, Elizabeth Brunk et al., Chem. Rev. 115, 6217 (2015). A selected subset of atoms is called a “model” system, which, for example, comprises a transition metal ion in a large organometallic complex and its nearest neighboring ligand atom, as described above. This “model” system can also include additional atoms not present in the “real” system, such as bonds that would be dangling if saturated in the case of covalent bonds between the “model” system and the rest of the system, but not saturated in the case of unsaturated bonds. Both low-level and high-level theories are used to handle such smaller “model” systems. The high-level theory refers to obtaining highly accurate intermediate electronic structure properties of the “model” system using quantum computing methods described above combined with, for example, active space techniques. Then, in the case of the bilayer ONIOM method, the target quantity (e.g., energy) of the “real” system under the high-level theory can be estimated via the following subtraction scheme:

[0096]

[0097] in, It is obtained using a quantum computer as a sub-representation. The two-layer ONIOM method obtains the high-level solution of the "model" system. The contribution of the remaining parts under the lower-level theory and the interaction between the remaining parts and the "model" system. Adding them together effectively approximates the high-level solutions of the "real" system. .

[0098] Similarly, the three-layer ONIOM method additionally introduces an "intermediate" model system as another sub-representation of the intermediate theory, which can refer to classical or quantum computation methods. The corresponding objective quantity of the "real" system under the higher-level theory can then be estimated again via the following subtraction scheme:

[0099]

[0100] ONIOM can be extended to even more layers, and thus to other sub-representations and properties beyond the ground state energy, such as the gradients of excited state energies and ground state energies relative to the nucleus positions. It should be noted that, in addition to the subtractive approach, an additive approach also exists, which is particularly useful when the lower-level theory corresponds to MM.

[0101] Generally, a two-layer ONIOM approach is expected to be a good approximation if the "model" system includes interactions that dominate the application of the technology and these interactions are described with the required precision by a high-level theory (e.g., quantum computing methods). Furthermore, as an additional criterion, a two-layer ONIOM approach is expected to be a good approximation if the interactions between the "model" system and the rest of the system are also described with sufficient accuracy by a low-level theory (e.g., via classical computing methods).

[0102] In practical applications of the ONIOM method, the user provides a geometric partition from the "real" molecular system to the aforementioned layers, wherein at least one "model" system is selected, comprising a subset of the atoms of the "real" molecular system, and a corresponding level of theory is selected for each layer to generate a corresponding sub-representation of the molecular system, wherein high-level computations on at least one sub-representation system are performed using a quantum computer. The output is then an estimate of the high-level theoretical (e.g., quantum computation) results for the selected intermediate electronic structure properties of the "real" system.

[0103] As mentioned above, the high-level results of the "model" system involving the most relevant sub-representations It was obtained using a quantum computer. Although This can be directly obtained using a quantum computer as described above, but preferably, further partitioning of the most relevant sub-representations using the active space technique is performed on a classical computer beforehand, allowing for a further reduction in the resource requirements of quantum computing (e.g., the number of qubits utilized and / or the number of quantum operations). While the multi-scale approach described above relies on a geometric partitioning that produces sub-representations containing subsets of the atoms of the original molecule or material, the active space technique relies on further partitioning at least the most relevant sub-representations in terms of their electronic structure (e.g., their electronic orbitals). Within the active space technique, the electronic structure representation of the most relevant sub-representations is partitioned into a first part indicating the active space of the electronic structure representation and a second part indicating the inactive space of the electronic structure representation, wherein the first part is solved with high accuracy on a quantum computer and the second part is solved with low accuracy on a classical computer, and the two solutions are ultimately computationally combined through interaction representations to generate a solution for a high-level result for the most relevant sub-representations. It should be noted that, to better distinguish between partitioning based on the activity space method and partitioning based on the ONIOM method, the parts involved in the activity space partitioning are referred to as the "first part" and the "second part." However, the "first part" and the "second part" are also sub-representations of the molecular structure representation. This partitioning can then be used to derive estimates of high-level results for "real" systems. For the example of organometallic complexes (where geometric partitioning can produce the most correlated electron representation containing the transition metal ion and its nearest neighboring ligand atoms), the active space technique can further divide the electronic orbitals of the most correlated electron representation into an active space problem as a first part and an inactive space problem as a second part. The first part contains, for example, the nearly degenerate and incompletely filled 3d atomic orbitals of the first row of transition metal ions (which are typically the most correlated electron orbitals), and the second part contains all the remaining (usually weaker correlated) orbitals of the most correlated electron representation. Furthermore, the active space problem can additionally include other electronic orbitals, such as the 4d atomic orbitals of the first row of transition metal ions and the bonding orbitals between the metal and ligands.

[0104] The following section briefly introduces the use of active and inactive space problems to determine intermediate properties. The selection of the active space can be based on information about the chemical product, the application, experience with similar problems, etc., and can be performed automatically, user-guided, or even fully input-driven. For example, the active space of a first-row transition metal complex with a metal center and saturated ligands can contain a set of 3d atomic orbitals that are almost energy-simple and not fully filled with the transition metal. Additionally, in some cases, 4d atomic orbitals, as well as bonding orbitals between the metal and ligands, can be selected into the active space. Another example relates to the computation of chemical reaction results, where, in this case, the active space preferably includes electronic orbitals expected to participate in the chemical reaction (e.g., involved in the bonding of the corresponding atoms), preferably valence electron orbitals. Furthermore, the active space can be determined based on the relevant application (e.g., based on the technical application characteristics to be determined). For example, if the goal is to determine spectral properties, it is preferable that the active space includes electronic orbitals involved in electronic transitions (e.g., electronic excitation from one set of electronic orbitals to another within a molecule). Based on this, the expected electronic orbitals in the convergent wavefunction can be identified, and appropriate starting electronic orbitals for the iterative process can be determined and provided.

[0105] Preferably, to define the initial electron orbitals, efficient mean-field methods that are simple and computationally relatively inexpensive can be used, such as Hartley-Fock (HF) theory that completely ignores electron correlations (e.g., ignoring static and dynamic correlations), or density functional theory (DFT) that partially considers electron correlations (albeit in a rather unsystematic way). Initial electron orbitals can also be obtained from other types of multi-reference calculations (e.g., fully active space self-consistent field (CASSCF) or restricted active space self-consistent field (RASSCF) calculations with active spaces of different sizes). Localization processes and population analysis may help determine electron orbitals that can be selected into the active space, for example, helping to determine the partitioning of the electronic structure representation into first and second parts. However, selecting the active space for complex electronic structure representations preferably utilizes at least partially automated methods. One possibility is to use unrestricted HF methods or correlational methods such as MP2 or coupled clusters to calculate the natural occupancy number and natural orbitals. In this case, a predetermined rule can be used: if the natural occupancy number of an orbital deviates from 2 and 0 by more than a predetermined threshold (e.g., more than 0.02), the predetermined rule determines that the orbital is included in the active space. For example, examples of defining such an active space can be found in, for instance, the following article: “UHF naturalorbitals for defining and starting MC-SCF calculations”. P. Pulay, TP Hamilton, J. Chem. Phys. 88, 4926 (1988). Furthermore, natural orbitals can also be calculated using the NEVPT2 method based on CASSCF calculations with a smaller active space, as described, for example, in the following article: “Active Space Selection Based on Natural Orbital Occupation Numbers from n‐Electron Valence Perturbation Theory”. A. Khedkar, M. Römelt, J. Chem. Theory Comput. 15, 3522-3536 (2019).Another, more complex possibility is the identification of relevant electron orbitals selectable for the active space during an automated process that utilizes electron orbital entanglement information and single-orbital entropy, calculated using methods such as the density matrix renormalization group (DMRG), as described in, for example, the following article: “Automated Selection of Active Orbital Spaces”. Ch. J. Stein, M. Reiher, J. Chem. Theory Comput. 12, 1760-1771 (2016), where electron orbitals with the highest entanglement information (e.g., the highest single-orbital entropy) are identified as part of the active space. Instead of the DMRG method, other types of exact and approximate fully configured interaction processes, as well as other related methods, can be utilized. Another possibility is the Atomic Valence Active Space (AVAS) method, which identifies a set of molecular orbitals that can be selected into the active space, similar to a predefined set of atomic orbitals, as described, for example, in the following article: "Automated Construction of Molecular Active Spaces from Atomic Valence Orbitals". ER Sayfutyarova, Q. Sun, GK-L. Chan, G. Knizia, J. Chem. Theory Comput. 13, 4063-4078 (2017). Typically, combinations of several of the above possibilities can also be used to define and provide the first and second parts of the electronic structure representation.

[0106] Accurate description of electronic correlations is crucial for the reliable determination of material properties, chemical reactions, and processes that define the technical applications of materials. Electronic correlations can be categorized into dynamic and static correlations. Dynamic correlations describe the correlations of electron movement and exist in all multi-electron systems (e.g., all electronic structure representations associated with chemical products such as molecules and materials). In this case, the HF wavefunction can serve as a reliable reference wavefunction for methods that incorporate dynamic correlations, such as Mueller-Plessetter (MP) perturbation theory or coupled-cluster (CC) methods.

[0107] Static correlations exist in electronic structure representations that require more than one electron orbital occupying a state to represent their wavefunction at a qualitatively correct level. This means that the wavefunction of these electronic structure representations can be constructed as a linear combination of several Slater determinants or configuration state functions. Correlating methods are often referred to as multi-configuration or multi-reference methods, such as the Multi-Configuration Self-Consistent Field (MCSCF) method. Strong static correlations often result in HF wavefunctions from HF-based dynamic correlation methods being qualitatively erroneous, exhibiting, for example, incorrect spin multiplicity of electronic states or a mixture of different spin states. A typical feature of such statically correlated electronic structure representations is a set of energy-degenerate or near-degenerate molecular orbitals. Examples include, but are not limited to, molecular orbitals that largely exhibit the characteristics of incompletely filled d-shells or f-shells found in transition metal compounds.

[0108] The following presents a possible method suitable for determining intermediate properties of molecular structures exhibiting strong static correlations. For this purpose, a problem-specific simplified representation of the electronic structure representation can be generated, for example, based on the electronic structure representation. Based on an approximate HF solution obtained, for example as described above, molecular orbitals can be partitioned into active orbitals forming the basis of a first part of the electronic structure representation, and an inactive set of double-occupied orbitals and empty virtual orbitals forming the basis of a second part of the electronic structure representation. Active orbitals define the active space. The size of the active space can be determined based on hardware constraints, particularly the number of available qubits. In other words, the number of electronic orbitals in the active space can depend on the number of qubits available to compute the energy spectrum of the corresponding Hamiltonian. On classical computers, the computational resources required to obtain an accurate solution increase exponentially with the size of the active space, thus typically limiting the active space to a relatively small size. Therefore, the method according to the invention, as described herein, is most advantageous compared to fully classical computation when the solution to the electronic structure representation would require an active space size exceeding the capabilities of classical hardware but just within the allowable range of available quantum hardware.

[0109] Preferably, for high-quality simulation results, in the case of a statically correlated chemical system, those electrons and electron orbitals primarily responsible for static correlation are included in the active space. The wavefunction within the active space method is characterized by the size of the active space (e.g., the number of active orbitals) and the number of active electrons within it. In the case of iterative optimization of orbitals during a self-consistent field (SCF) process, the energy of the electronic structure can converge to its global minimum, or, where appropriate, to one of several local minima. Therefore, the definitions of the active space and initial electron orbitals used in the iterative process can also be based on the expected results at convergence.

[0110] To solve for the molecular structure representation, single-electron and two-electron integrals can be calculated based on the molecular orbitals in the current iteration. (Arbitrary orbitals) The single-electron integral on can be defined as

[0111]

[0112] Among them, operators It includes terms that explain the kinetic energy of electrons and the Coulomb attraction between electrons and atomic nuclei; It is the "spin coordinates" that make the spin functions orthogonal:

[0113]

[0114]

[0115] Because electrons are in arbitrary orbits The two-electron integral generated by Coulomb repulsion on the electron can be defined as:

[0116]

[0117] The single-electron integral between two orbitals with different spins is always equal to zero.

[0118]

[0119] Similarly, if and or and If they have different spins, then the two-electron integral is zero:

[0120]

[0121] Preferably, during the iteration process, the two relationships described above are considered to determine which integrals to compute and store, such that the actual integrals are performed only in the spatial portion of the relevant combination of orbits in real space.

[0122] The energy contributions of inactive occupancy and inactive virtual orbitals are equivalent to expressions similar to mean-field expressions, which can be evaluated as the second part of the molecular structure representation on a classical computer. The Hamiltonian representation is constructed in active space as a mathematical representation of the first part of the molecular structure representation, which includes indexed... The labeled single-electron and two-electron integrals on the active orbitals, in which case the active orbitals are spin orbitals:

[0123]

[0124] and These are the electron creation operator and the annihilation operator, respectively. Preferably, the contribution to connecting the active orbital with the remaining orbital is entirely represented by the effective single-electron contribution to the Hamiltonian of the active space. This contribution can be used as... A portion of which is included, The inactive occupied orbitals are enumerated; in this case, these are spin orbitals:

[0125]

[0126] At this stage, well-defined approximations can also be applied. Preferably, two-electron integral terms smaller than a predetermined threshold are identified and ignored to reduce the number of quantum operations on the quantum computer. In some cases, especially if the active orbitals extend to a larger number of atoms, the number of associated two-electron integral terms can be further reduced by orbital transformations (e.g., by repositioning).

[0127] Based on the molecular structure representation provided above and the first and second parts of the separately identified electronic structure representation, the first part can be solved on a quantum computer. Preferably, to solve the active space subproblem (e.g., the first part) on a quantum computer, quantum algorithms, such as variational Hamiltonian fitting (VHA) or quantum phase estimation (QPE), are utilized. As inputs to these algorithms, the number of electronic orbitals and electrons in the active space, spin multiplicity, single-electron integrals along with effective single-electron contributions from inactive orbitals and two-electron integrals in the active space can be provided as part of the electronic structure representation. If a combination of a variational quantum eigenvalue solver (VQE) and a unit coupled cluster (UCC) fitting or a variant thereof is used as the solver for the active space subproblem (e.g., the first part), the set of active orbitals can also be further subdivided, for example, relative to a mean-field reference, into occupied and unoccupied active orbitals. Algorithms for solving the active space contribution to the solution can employ hybrid quantum-classical methods, such as VQE or VHA.

[0128] Based on the above, the first part can be implemented on a quantum computer to obtain a solution for the active space Hamiltonian and physically prepare the corresponding solution state in a qubit register. Specifically, various transformation methods can be used, allowing, for example, a transformation unit to convert the first part into an operational description, based on which the solution can be computed using a quantum computer. In one embodiment, a variational algorithm can be utilized. Typically, a variational algorithm prepares a parameter-dependent trial state on a quantum computer. In this trial state, the energy of the electronic structure can be determined based on measurements from the quantum computer, and a classical computer can be used to optimize the parameters and prepare the updated trial state on the quantum computer. Variational algorithms do not guarantee convergence to an exact ground state or other desired state for the electronic structure problem. On the other hand, compared to, for example, non-variational algorithms (such as QPE), these variational algorithms require far fewer quantum resources, such as a far fewer number of quantum operations, and can run on noisy intermediate-scale quantum (NISQ) computers. In examples where variational algorithms are available, UCC assumptions can be used within the VQE, for example, as described in more detail in the following articles: “A variational eigenvalue solver on a photonic quantum processor”, A. Peruzzo et al., Nat. Commun. 5, 4213 (2014), or “The theory of variational hybrid quantum-classical algorithms”, J. McClean et al., New J. Phys. 18, 023023 (2016). The generalization of the proposed UCC within VQE can also be used to improve the handling of static correlations, as described in the following article: "Generalized Unitary Coupled Cluster Wavefunctions for Quantum Computation" J. Lee et al., ArXiv e-Prints arXiv:1810.02327 (2019).Alternatively, VHA can be used, for example, as described in the following literature: “Progress towards practical quantum variational algorithms”, D. Wecker et al., Phys. Rev. A, 92, 042303 (2015). In another embodiment, the QPE method is used, wherein the QPE method is guaranteed to produce exact eigenstates and eigenvalues ​​of the Hamiltonian (e.g., the active-space Hamiltonian). The QPE method is not limited to the ground state. However, this method is very demanding in terms of quantum resources (e.g., the number of quantum gate operations, the number of qubits) and is therefore preferably executed on a fault-tolerant quantum computer. If a quantum computer that meets the corresponding requirements of the algorithm is used, QPE is the preferred algorithm for the first part of the computation problem on a quantum computer. Examples of how the QPE method can be used can be found in the following articles: “Quantum computational chemistry”, S. McArdle et al., Rev. Mod. Phys. 92, 015003 (2020), and “Quantum computation and quantum information: 10”. th "Anniversary Edition of Quantum Computing and Quantum Information: 10th Anniversary Edition", MANielsen, IL Chuang, ISBN: 1-107-00217-6 978-1-107-00217-3 (Cambridge University Press, New York, NY, USA, 2011).

[0129] In all cases, the goal of converting the first part into an operational description is to prepare the active space Hamiltonian as defined above. An approximation or exact solution to the ground state (or, if necessary, the excited state), where, in this example, the active space Hamiltonian is the first part of the electronic structure representation. Preferably, the operational description utilizes unitary evolution using gate-based quantum operations. The process applied to qubit registers. Here, It is an operator of any Hermitian fermion, especially an electron operator. In some instances, The active space Hamiltonian defined above The same. Before applying this process, the quantum computer is in a state... This state represents a specific fermionic state. After applying this process, the quantum computer is in a new state. This state represents a fermionic state. Unitary evolution can be approximated by Trotter decomposition (as described, for example, in “On the product of semi-groups of operators”, HF Trotter, Proc. Am. Math. Soc. 10, 545-551 (1959)) or by similar approximations using truncated Taylor series expansions or matrix exponents. In some cases (e.g., VHA), more than one unitary evolution is applied successively.

[0130] Therefore, as described in more detail above, the active space problem, as part of the first part, particularly the Hamiltonian of the active space problem (which ultimately contains the most challenging part of the electronic structure represented by the most relevant quantum of the original molecule or material), can be solved using a quantum computer. Preferably, this is done using a gate-based quantum computer, for example, implemented in the form of a superconducting or trapped-ion quantum computer. For example, quantum phase estimation (QPE) or variational algorithms (such as variational Hamiltonian fitting (VHA) or variational quantum eigenvalue solvers (VQE) with unit-coupled cluster (UCC) fitting) combined with Jordan-Wigner or Bravyi-Kitaev transforms and CZ algorithms or similar algorithms, and FSIM network algorithms using low-rank decomposition of unitary evolution, can be used to compute the resulting active space problem on a quantum computer. Further information can be found in the following articles: “Quantum Computational Chemistry”, Sam McArdle et al., Rev. Mod. Phys. 92, 015003 (2020), and “Quantum Chemistry in the Age of Quantum Computing”, Yudong Cao et al., Chem. Rev. 119, 10856 (2019). Typically, after the quantum mechanical state has been prepared on a quantum computer, for example using the methods and algorithms described above, predetermined fundamental quantities are measured, such as the n-particle reduced density matrix (n-RDM) of the Hamiltonian of the input active space problem, which contains relevant information about the solution to the Hamiltonian of the input active space problem. It should be noted that, preferably, only those quantities required on the quantum computer to derive the intermediate electronic structure properties and the technical application properties subsequently used to solve real-world chemical or materials problems are measured.

[0131] In addition to the gate-based quantum computers that are generally available, as another example, specialized quantum computers (such as Gaussian boson sampling photonic quantum computers and quantum annealers) can also be used to determine intermediate properties related to real-world chemical or materials problems.

[0132] The selection of quantum computing methods and quantum computing hardware can be made by the user or automatically, and can depend on the availability of specific quantum hardware, details of the available quantum hardware (such as the number of available qubits, the connectivity between qubits, the decoherence time and error rate of quantum operations), and details of the active space problem Hamiltonian (such as its dimension (e.g., size), the number of targets to be measured, and the desired accuracy).

[0133] In a further step, quantities measured on a quantum computer can be processed on a classical computer to obtain intermediate properties associated with the electronic structure of a molecule. These intermediate properties describe the physical properties of the molecule or material, such as energy and electrostatic multipole moments, but in many cases, have not yet directly solved the final real-world technological application problem. Calculating intermediate electronic structure properties from the measured quantities obtained by solving the Hamiltonian of the active space problem using a quantum computer can also include combinations with results obtained using a classical computer (e.g., solutions for the inactive space portion, as well as low-level solutions for sub-representations and "real" systems). The resulting intermediate electronic structure properties are then transferred. For further processing. In some cases, quantities measured on a quantum computer can also directly refer to intermediate properties.

[0134] The following provides examples of how computable intermediate properties of an input electronic structure system can be obtained on a classical computer based on quantities previously measured from a quantum computer. The electron density (a measure of the probability of an electron being present at a given location in a molecule or material) can be calculated using a 1-RDM obtained from pre-calculated spatial components of the electron orbitals from quantum computing. The electrostatic multipole moment can be obtained on a classical computer by shrinking the RDM using pre-calculated electrostatic multipole moment integrals. For example, the dipole moment is obtained by measuring a 1-RDM using pre-calculated dipole moment integral shrinkage, where these dipole moment integrals are related to the dielectric behavior of the molecule or material or responsible for intermolecular interactions that strongly depend on the polarity of the molecule or material. The total energy of the electronic ground state can be obtained using 1-RDM and 2-RDM measured on a classical computer using pre-calculated Hamiltonian integral shrinkage. Furthermore, in some cases, a more accurate total ground state energy can be obtained by additionally post-processing the higher-order RDMs.

[0135] The technique of obtaining intermediate electronic structure properties by measuring n-RDM with pre-computed integral shrinkage on a classical computer can be used in the context of variational algorithms. As can be seen from the following further example, various intermediate electronic structure properties (e.g., total ground state energy and dipole moment) can be calculated on a classical computer based on a single determination of some fundamental quantities (e.g., 1-RDM and 2-RDM) on a quantum computer, without further calculations or measurements on a quantum computer.

[0136] Using the QPE method, the total ground-state energy of the input Hamiltonian can be directly accessed as a result of computation on a quantum computer. Other intermediate electronic structure properties can be obtained by measuring the n-RDM as described above.

[0137] It should be noted that after mapping the corresponding electronic structure Hamiltonian to the spin Hamiltonian, the quantum annealer can be used to calculate the total ground state energy of the electronic structure system. Furthermore, the quantum annealer is generally suitable for calculating other properties of spin systems that can be described using the Ising model.

[0138] Excited-state properties (such as properties of electronically excited states, such as their energies) can also be obtained as intermediate properties. For example, quantum space expansion relies on higher-order RDMs, particularly 3-RDMs and 4-RDMs, which can be measured after running variational algorithms using a quantum computer. These higher-order RDMs can then be used to solve the generalized eigenvalue problem on a classical computer. Further details can be found in the following article: “Hybrid quantum-classical hierarchy for mitigation of decoherence and determination of excited states”, Jarrod McClean et al., Phys. Rev. A, 95, 042308 (2017).

[0139] It should be noted that, in addition to electronic structure, boson degrees of freedom can also affect important properties of molecules or materials, such as excited states or spectral properties, and therefore can be taken into account. In this context, Gaussian boson sampling photonic quantum computers can be used to compute vibrational modes represented by electronic constructons, particularly Frank-Condon curves. More details of this method are described in the following article: “Boson sampling for molecular vibronic spectra”, Joonsuk Huh et al., Nature Photonics, 9, 615 (2015).

[0140] The derivatives of energies (such as the total ground-state energy and / or the excited-state energy, wrt, the nuclear position, and / or the electromagnetic field derivatives) can also be calculated as intermediate properties. In particular, the derivative of the total ground-state energy, wrt, and the nuclear position derivatives can be important for optimizing structures (e.g., the positions of atomic nuclei in molecules or materials) and for finding the most energy-stable structures. Both analytical and numerical derivatives are possible. Such energy derivatives can be obtained using algorithms based on QPE or linear response quantum space extensions. Details of these algorithms are described in the following article: “Calculating energy derivatives for quantum chemistry on a quantum computer”, Thomas E. O'Brien et al., npj Quantum Information, 5, 113 (2019). Similarly, VQE methods can be utilized using analytical gradient measurements on quantum computers, as described in the following article: “Theory of analytical energy derivatives for the variational quantum eigensolver”, Kosuke Mitarai et al., Phys. Rev. Research, 2, 013129 (2020). Another example of how to obtain the derivative of energy is the contracted VQE method, described in the following article: “Hybrid Quantum / Classical Derivative Theory: Analytical Gradients and Excited-State Dynamics for the Multistate Contracted Variational Quantum Eigensolver”, Robert M. Parrish et al., arXiv:1906.08728 (2019). It should be noted that, in the case of energy derivatives, a set of fundamental quantities measured on a quantum computer may be different from a set of fundamental quantities measured in the case of focusing only on energy itself.

[0141] In some cases, intermediate electronic structure properties may already be correlated with experimentally available properties and may support solving a portion of the real-world chemical or materials problems described below. However, in practice, most of these properties may only be correlated with technical applications utilizing complex further computational steps, or they may not be experimentally available, making them unhelpful for designing new chemical products, materials, and processes; for example, they are unsuitable for solving the original real-world chemical or materials problems. Therefore, for practical use, particularly with respect to the real-world problems described below, further processing of the quantities obtained so far is required on a classical computer. Furthermore, combining this post-processing on a classical computer with the pre-processing described above for preparing the real-world problem reduces the quantum resource requirements of a quantum computer. Therefore, combining this pre-processing and post-processing workflow is particularly advantageous.

[0142] The following describes in detail how one or more of the aforementioned intermediate characteristics (exemplarily intermediate electronic structure characteristics in this case) can be processed on a classical computer to obtain technical application characteristics, wherein at least one intermediate electronic structure characteristic is obtained based on computation using a quantum computer. In cases where more than one intermediate electronic structure characteristic is obtained based on computation using a quantum computer, it may be advantageous to utilize different quantum computer paradigms and / or hardware implementations depending on the intermediate electronic structure characteristic. It should also be noted that in some cases, a single intermediate electronic structure characteristic can be processed to produce multiple technical application characteristics. Furthermore, the different workflows described below can be combined such that the calculated technical application characteristic can be used as input in another workflow for calculating another technical application characteristic.

[0143] The following sections also provide exemplary applications of these workflows for real-world chemistry and materials problems. In particular, examples of the relationship between experimentally available technical application characteristics and intermediate properties are provided to solve corresponding real-world chemistry and materials problems, which are prerequisites for effectively and efficiently designing new chemical products and materials and improving chemical processes.

[0144] For example, it refers to in Figure 10The image schematically and exemplarily illustrates applications of spectroscopy. Spectroscopy is a non-invasive method for studying systems experimentally, comparing them with other systems or under different environments and / or different physicochemical conditions, to elucidate the molecular structure, properties, and chemical reactivity of a system. Different experimental spectroscopic techniques across different electromagnetic spectral ranges, and combinations thereof, can yield a more comprehensive understanding of the system under study, such as molecules and materials. However, the increasing complexity of these experimental techniques makes interpreting spectroscopic results increasingly difficult without the aid of computational chemistry. For example, experimental results often do not allow for the direct derivation of desired information; in some cases, molecular structures cannot be directly deduced from measured spectra. However, experimental results can be compared with computational results to obtain the desired information; for example, measured spectra can be compared with various calculated spectra assuming different molecular structures to determine the molecular structure that best matches both the measured and calculated spectra.

[0145] The following describes the determination of spectroscopic application characteristics based on the aforementioned intermediate electronic structure properties. In UV / Vis spectroscopy, excited-state properties obtained as intermediate electronic structure properties can be used. Specifically, the energies of electronic excited states (e.g., the energies of states with predefined spin multiplicity or electronic orbital configurations different from the ground state) and the corresponding energies of the electronic ground state (e.g., both obtained using a gate-based quantum computer) can be used. Additionally and optionally, vibrational contributions (e.g., Frank-Condon curves obtained using, for example, a boson-sampling photonic quantum computer) and other contributions (e.g., linewidths obtained via approximate calculations on a classical computer or as user input) can be used. As output, the workflow provides spectroscopic application characteristics that can be directly correlated with experimentally available properties and thus support the solution of real-world chemical or materials problems. For example, electronic absorption spectra can be calculated, directly referring to, for example, experimentally available UV / Vis spectra.

[0146] UV / Vis spectroscopy uses visible light and adjacent light ranges, where the absorption or reflection of light in the visible range by a chemical product or material directly affects its perceived color. Therefore, the calculated electronic absorption spectrum can support, for example, the design of new dyes. Another example involves photoinitiators, which are molecules that produce reactive species (such as free radicals) when exposed to radiation in the UV or visible range. These reactive species can then initiate, for example, polymerization to produce polymers. In this context, the difference in calculated energy between the lowest-energy electronically excited state and the electronically ground state is directly related to the laser wavelength required to irradiate the photoinitiator to initiate the polymerization process. Further examples of real-world chemistry and materials problems involve photovoltaics and photochemical synthesis.

[0147] Furthermore, electron emission spectra can be calculated, which directly refers to, for example, experimentally obtainable fluorescence spectra. Electron emission spectra are complementary to electron absorption spectra because the former involves the transitions of electrons from excited states to the ground state caused by photon emission, while the latter designs the transitions of electrons from the ground state to excited states caused by photon absorption. For example, as an important characteristic for technological applications, the calculated energy difference between the electronically excited states and the electronically ground state of a molecule or material is directly related to the color of the emitted light. For instance, using this workflow, the color of the emitted light of a potential organic light-emitting diode material can be calculated before the actual synthesis of that material.

[0148] In infrared (IR) spectroscopy, the derivative of the total ground-state energy with respect to the nuclear position, obtained as described above, can be used. Specifically, the second derivative is used, which can be calculated entirely analytically and numerically, or numerically in a subsequent step using the analytically calculated first derivative. The calculated spectrum (also known as the vibrational spectrum within the rigid rotor / harmonic oscillator approximation, as they refer to molecular vibrations) can be directly correlated with experimentally available IR spectra. IR spectroscopy can be used to characterize new chemicals and / or materials or to identify and verify known and unknown samples.

[0149] Furthermore, electronic vibrational spectra can be calculated, taking into account the simultaneous changes in the electronic and vibrational energy levels of chemical products or materials due to the emission of photons with appropriate energy. Electronic vibrational spectroscopy can provide information about the electronically excited states of molecules, such as bond lengths.

[0150] To facilitate comparison of the calculated spectra with the corresponding experimentally determined spectra, these calculated spectra are typically visualized by plotting the calculated transition intensities or related quantities (such as absorbance) against the corresponding calculated transition energies or related quantities (such as transition wavelengths). Further details regarding general computational spectroscopy are described in the following article, which is incorporated herein by reference: “Computational molecular spectroscopy”, Vincenzo Barone et al., Nature Reviews Methods Primers 1, 38 (2021).

[0151] A preferred example of a identifiable technical application characteristic is chemical reactivity, where a single chemical reaction is identified (such as the following chemical reaction).

[0152]

[0153] The thermodynamic and kinetic quantities depend on the calculation of the reaction free enthalpy and the activation free enthalpy, respectively. For example, the reaction free enthalpy indicates whether a chemical reaction can occur in principle, and the activation free enthalpy indicates the rate of the chemical reaction, such as its velocity.

[0154] Reaction free enthalpy It can be determined according to the following formula, by extracting from all products (in the example above). and Subtract all reactants (in the example above) from the sum of their free enthalpies. and Calculated by the sum of the free enthalpies of ).

[0155]

[0156] Among them, the free enthalpy of each individual chemical species in solution Through their corresponding stoichiometric weights Weighted. Similarly, activation free enthalpy. The free enthalpy from the transition state (in the example above) can be determined according to the following formula. The enthalpy of free energy is obtained by subtracting the stoichiometric weighted sum of the free enthalpies of all reactants from the enthalpy of free energy in the reaction. The free enthalpy of free energy in the transition state is the highest point of free enthalpy along the reaction pathway from reactants to products.

[0157]

[0158] Since chemical reactions typically occur in solution (e.g., in water) rather than in the gas phase, it is preferable to account for the effect of the solvent when calculating all free enthalpies. However, for many applications, omitting the solvent effect in the calculation can also be a reasonable approximation. The following examples consider the solvent, but an approximation can also be applied without considering the solvent by accordingly removing the solvent component. Individual chemical species (e.g., reactants) or ,product or transition state Enthalpy of free energy in solution The free enthalpy in the gas phase can be expressed by the following formula. and enthalpy of dissolution The sum is obtained as follows:

[0159]

[0160] It should be noted that the free enthalpy in a solution is sometimes also called the Gibbs free energy in the solution. According to the following formula, at a predetermined temperature... The free enthalpy obtained in the gas phase is as follows:

[0161]

[0162] in, It refers to electron energy, for example, an estimate of the higher energies of a "real" system. Higher energies can be generated using the methods described above, for example, by combining multiscale methods utilizing quantum computers with the active space techniques described above. It should be noted that, depending on the electronic state of interest, this energy can refer to both the electronic ground state and the electronic excited state. It is the zero-point energy of vibration, and These are the translational / rotational / vibrational partition functions, respectively. and Both can be calculated from vibrational spectra, which can be generated, for example, using the spectroscopic workflow described above, particularly the IR spectroscopic workflow, and utilizing a quantum computer as described above. However, these quantities can also be obtained using a combination of classical computers and methods that are generally less accurate, such as density functional theory (DFT). It is Avogadro's gas constant.

[0163] Enthalpy of dissolution of molecules in a predefined solvent This information can be obtained using classical computers via a dissolution model, such as the Real Solvent-like Conductor Shielding Model (COSMO-RS). Within the COSMO-RS method, in the first step, the electron density indicating the charge distribution in the molecule, obtained using a quantum computer as described above, can be used to calculate the shielding charge density σ on the molecular surface. In the second step, this information can be used to calculate the chemical potential µ of the molecule in a predetermined liquid solvent or mixture. The resulting chemical potential µ is then used to calculate the enthalpy of dissolution. Further information on the COSMO-RS method can be found in the book cited and incorporated herein by reference: “COSMO-RS: From Quantum Chemistry to Fluid Phase Thermodynamics and Drug Design”, Andreas Klamt, Elsevier (2005). It should be noted that computational costs can be significantly reduced by implicitly including the effects of solvent on technical application properties (such as reaction free enthalpy and activation free enthalpy) via the COSMO-RS method, rather than by directly calculating technical application properties in solution by constructing a supersystem that explicitly includes solvent molecules.

[0164] Calculating the reaction free enthalpy and activation free enthalpy in solution for all potential reaction pathways (e.g., for multiple chemical reactions taking into account possible transition states, intermediates, and products) ultimately allows for determining the outcome when predefined molecules (e.g., reactants) are mixed in a test tube or container by identifying one or more energy-optimal reaction pathways. In particular, highly accurate energy calculations for all types of chemical reactivity networks (e.g., reactants, products, transition states, and intermediates) are essential for reliable determination, and therefore essential for designing new chemical products and materials, improving industrial chemical processes, and for other technological applications, such as inhibiting the degradation of chemical products, determining the microstructure of polymeric chemicals, and thus calculating and fine-tuning the technical application properties of chemical products.

[0165] Specific industrially relevant chemical products for which chemical reactivity (e.g., reaction rate) workflows can be preferably applied are, for example, catalysts, such as... Figure 11 As shown. Catalysts play a crucial role in achieving or accelerating chemical reactions under mild conditions (e.g., mild temperatures and mild pressures) by interacting with transition states and lowering their energies, thereby reducing the activation enthalpy and increasing the rate of chemical reactions. Currently, catalysts containing 4d or 5d transition metals (such as rhodium and palladium) are frequently used; these catalysts are very expensive, such as rhodium-based Wilkinson catalysts used for hydroformylation. Many technological goals aim to replace those catalysts with those containing, for example, cheaper 3d transition metals such as cobalt or iron. Chemical reactivity workflows can be particularly advantageous in the field of homogeneous catalysis, for example, for calculating oxidation, reduction, hydrogenation, carbonylation, etc., in large-scale chemical production processes (such as polymer production processes) and in the synthesis of fine chemicals using homogeneous catalysis. For example, the activation enthalpy of a predetermined catalytic cycle (including undesirable side reactions) can be calculated using chemical reactivity workflows before synthesizing a predetermined set of catalysts in the laboratory. As mentioned above, those activation enthalpies are helpful in calculating chemical kinetics, particularly chemical reaction rates. Specifically, the chemical reaction rate of the identified potential catalyst can be compared with the target chemical reaction rate, and based on this comparison, ii) the corresponding potential catalyst can be identified as the target catalyst, or iii) a new potential catalyst can be provided, and the reaction rate determination according to the invention can be repeated. Therefore, only catalysts with chemical reaction rates greater than a predetermined threshold and without identified serious undesirable side reactions are ultimately selected for synthesis, for example, by providing control data for the corresponding synthesis of the catalyst, and for further investigation in the laboratory.

[0166] Other specific industry-related chemical products that can be applied to chemically reactive workflows are chelating agents, such as... Figure 12As shown. The development of chelating agents tailored for certain metal ions can be enhanced by calculating the complex formation constant. is a thermodynamic quantity indicating the degree of thermodynamic stability of the resulting metal ion-chelate complex. The complex formation constant can be calculated using the enthalpy of reaction of the chelating agent and metal ion in solution to form the corresponding chelate complex. Reliable computational determination is preferably based on accurate calculation of the enthalpy of reaction, which is particularly challenging for reactions involving transition metal ions. Additionally, reliable determination of chelating agent selectivity is challenging because both its experimental and computational determinations are difficult. The above method allows for overcoming these challenges by providing the possibility of utilizing quantum computers in computation to improve computational accuracy and the complexity of solvable problems. Selectivity is a means of describing the degree to which a particular chelating agent is more inclined to bind with a particular metal ion compared to other metal ions, and it can be derived from the corresponding complex formation constant. A well-known example of a chelating agent is ethylenediaminetetraacetic acid (EDTA), which can be used, for example, to solubilize Fe. 3+ Ions. Many technical objectives aim to design novel chelating agents with predetermined selectivity that exhibit advantageous properties (such as biodegradability) or less harm to aquatic organisms. Chelating agents are used in a wide variety of technical applications, for example, to suppress the undesirable effects of metal ions during washing and cleaning processes. Furthermore, chelating agents are used in mining for the selective extraction of metals. For example, using the chemically reactive workflow described above, the enthalpy of reaction of a predetermined chelating agent can be calculated to obtain the corresponding complex formation constant and thus the selectivity relative to different transition metal ions. These calculations can be performed, for example, for a predetermined set of chelating agents before, for example, the synthesis of such a set in the laboratory. In particular, the selectivity of a identified potential chelating agent can be compared with the target selectivity, and based on this comparison, i) the corresponding potential chelating agent can be identified as the target chelating agent, or ii) a new potential chelating agent can be provided, and the determination of selectivity according to the invention can be repeated. Therefore, only those chelating agents that meet certain predetermined criteria / technical application characteristics (e.g., regarding selectivity) are ultimately selected for synthesis and further laboratory studies, for example, by providing control data that induces the corresponding synthesis of the corresponding catalyst.

[0167] Additionally, a slightly modified workflow based on the COSMO-RS method, as described above, can be used to calculate other technical application properties, such as activity coefficients, solubility, partition coefficients, and vapor pressures. For example, equilibrium vapor pressure is also an experimentally available quantity and is defined as the pressure exerted by vapor in thermodynamic equilibrium with its condensed phase (liquid or solid) at a given temperature in a closed environment, and is an indicator of the evaporation rate of the liquid or solid. For example, vapor pressure calculations are important for quantifying the volatility of hazardous or toxic chemicals at a given temperature, either before or in lieu of conducting real-world experiments. Additional safety measures can be established in cases where potentially hazardous or toxic chemicals are volatile (e.g., have high vapor pressures, which would result in high concentrations of the chemical in the respiration zone). Further information on the overall determination of the above-mentioned technical application characteristics can be found in the following article: "Predicting accurate absolute binding energies in aqueous solution: thermodynamic considerations for electronic structure methods", Jan H. Jensen, Phys. Chem. Chem. Phys. 17, 12441 (2015).

[0168] Another exemplary embodiment relates to quantitative structure-activity relationship (QSAR) or quantitative structure-property relationship (QSPR). QSAR or QSPR models are regression or classification models that, in the case of regression models (partial least squares regression models), correlate a set of predictor variables (also called “descriptors”) with one or more output variables (e.g., the power of the response variable), or, in the case of classification models, correlate a set of predictor variables with categorical values. Furthermore, neural networks can also be used as models.

[0169] In the context of this invention, predictor variables can refer to a set of intermediate electronic structure properties of an electronic structure system (e.g., a molecule), wherein at least one of the intermediate electronic structure properties is calculated using a quantum computer. However, predictor variables can also refer to a set of technically applicable properties of a real-world chemical or materials problem obtained as described above by processing intermediate electronic structure properties. Furthermore, combinations with predictor variables obtained from other sources (such as cheminformatics processing of structural information or experimental physicochemical properties) are also possible. Predictor variables can be electronic properties, geometric properties, structural properties, or physicochemical properties and / or molecular descriptors, and therefore refer to quantities obtainable by one of the methods described above or similar methods, while output variables refer to technically applicable properties of a real-world chemical or materials problem (such as the bioactivity or chemical properties of a molecule) that cannot be directly obtained by the methods described above or similar methods.

[0170] The QSAR / QSPR model mathematically summarizes the assumed relationships between a chosen set of predictor variables and the output variables. After model construction, it is carefully validated in terms of robustness, predictive performance, and applicability. Following successful validation, the QSAR / QSPR model can then be used to determine the output variables for new real-world chemical or materials problems, such as technological application characteristics, using a set of predictor variables obtained via quantum computing as described above. Since the quality of the predictions depends significantly on the accuracy of the provided predictor variables, the use of quantum computers is expected to be advantageous. Furthermore, the quality of the predictions also depends on other factors, such as the appropriate selection of predictor variables, the QSAR / QSPR model used, and its validation.

[0171] Specific technical application characteristics that can be calculated using the QSAR / QSPR workflow described above are, for example, the bioactivity of chemical products (such as drugs, poisons, or environmental pollutants) using the corresponding calculated intermediate electronic structure characteristics or the aforementioned technical application characteristics as predictive variables. Bioactivity can be quantitatively expressed as the concentration of the chemical product required to give a certain biological response (including desired therapeutic effects and undesirable side effects). For example, the QSAR / QSPR workflow can be used for computational toxicology assessments of novel chemical products.

[0172] Other specific technical application properties of real-world chemical or materials problems that can be calculated using the QSAR / QSPR workflow described above are, for example, reactivity, biotransformability, solubility, boiling point, pKa value, and partition coefficient.

[0173] In another embodiment, the methods and workflows described above can be applied to training alternative models, particularly for parameterizing force fields and training machine learning models. The workflows and methods described above generate data. If the amount of available data points allows, alternative models can be trained using classical computers to determine, at a much lower computational cost, either (a) the results of calculating the intermediate electronic structure properties of the most relevant sub-representations, or (b) the results of the final technical application properties of a “real” system (such as a real-world chemistry or materials problem). The range of alternative models can be from physics-based models to rather black-box models derived from machine learning. An example of a physics-based alternative model is a force field that allows the determination of the total energy of a molecule based on, for example, bond length, bond angle, and bond torsion angle. For such a model, a fairly small number of well-chosen data points are sufficient to fit a few parameters describing bond stiffness, the flexibility of bond angles, and the rotational energy barrier around the bonds.

[0174] Typically, large training datasets are required to train machine learning models based on intermediate electronic structure properties of a set of different electronic structure systems (e.g., a set of molecules) calculated using quantum computers as described above. However, machine learning models can be trained based on technical application properties and / or intermediate properties obtained as described above for a set of real-world chemical or materials problems. Generating such large training datasets becomes much easier in this case because reasonable computational resources can be utilized and training data can be obtained in a short time. For example, time- and resource-intensive experiments can be avoided. Furthermore, technical application properties and / or intermediate properties can be determined with higher accuracy using the methods and apparatus described above. This allows the application of training algorithms that require less high-quality training data, such as transfer learning algorithms. If the machine learning model is carefully validated after training, it can be used to determine, using classical computers, the corresponding intermediate electronic structure properties or technical application properties of electronic structure systems or real-world materials (e.g., molecules) not included in the training dataset at a very low computational cost. In particular, no additional computation on a quantum computer is required for this determination. Furthermore, for larger electronic structure systems or real-world materials that will therefore be more computationally demanding than the electronic structure systems or real-world materials included in the training dataset, the corresponding intermediate electronic structure properties or technical application properties can also be determined, potentially reducing the overall computational resources required.

[0175] A concrete example of this machine learning model is a widely applicable machine learning potential based on neural networks, such as the Accurate Neural Engine (ANI) for molecular energy, as described in the following article: "ANI-1: an extensible neural network potential with DFT accuracy at force field computational cost.", Justin S. Smith et al., Chem. Sci. 27, 479-496 (2017). The highly transferable ANI neural network potential is particularly well-suited for determining the total energy of the ground state of molecules at low computational cost on classical computers. The accuracy of the ANI method depends on the data used during training. Therefore, its accuracy depends both on the number, diversity, and properties of the systems (e.g., molecules) used in training, and on the accuracy of the training data, such as the accuracy of the energy used in the training steps. Therefore, it is advantageous to utilize quantum computers to generate training data for a predefined set of electronically structured systems or materials. Then, for example, the energy determined using ANI for systems not included in the training dataset can be further used as the electronic energy in the aforementioned chemical reactivity workflow. Input.

[0176] Typically, these alternative models can be used in any of the workflows described above, avoiding quantum computing steps. This avoidance either bypasses only the steps (which usually already implies a significant speedup of the corresponding workflow) or bypasses some or all of the steps executed on the classical computers by directly accessing the technical application characteristics of the entire workflow. Furthermore, alternative models providing the total molecular energy can also be advantageously used to perform molecular dynamics or molecular Monte Carlo simulations, which typically require calculating the energies of thousands to millions of molecular structures. Technical application characteristics accessible through these simulations are typically spatial distribution probabilities and free enthalpy. From this, relevant technical application characteristics can be derived, such as solubility, solubilization, adsorption on the surface, or interfacial behavior (e.g., interfacial tension, surface properties).

[0177] By studying the accompanying drawings, this disclosure, and the appended claims, those skilled in the art can understand and implement other variations of the disclosed embodiments when practicing the claimed invention.

[0178] The operations performed in the processes and methods disclosed herein may be implemented in different orders. Furthermore, the operations outlined are provided as examples only, and some of these operations may be optional, may be combined into fewer steps and operations, may be supplemented with more operations, or may be expanded into more operations without departing from the essence of the disclosed embodiments.

[0179] In the claims, the word “comprising” does not exclude other elements or steps, and the indefinite article “a / an” does not exclude multiple / types.

[0180] A single unit or device can perform the functions of several items listed in the claims. The fact that certain measures are listed in different dependent claims does not indicate that combinations of these measures cannot be used advantageously.

[0181] Processes performed by one or more units or devices (such as providing electronic structure representations, enabling quantum computers and / or classical computers to generate and provide solutions, generating solutions to electronic structure representations, etc.) can be performed by any other number of units or devices. These processes can be implemented as program code devices and / or dedicated hardware for computer programs.

[0182] Computer program products can be stored / distributed on suitable media, such as optical or solid-state storage media provided with or as part of other hardware, but can also be distributed in other forms, such as via the Internet or other wired or wireless telecommunications systems.

[0183] Any unit described herein can be a processing unit as part of a classical computing system. Processing units can include general-purpose processors and can also include field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), or any other special-purpose circuitry. Any memory can be physical system memory, which can be volatile, non-volatile, or some combination of both. The term "memory" can include any computer-readable storage medium, such as a non-volatile mass storage device. If the computing system is distributed, the processing and / or storage capabilities can also be distributed. A computing system can include multiple structures as "executable components." The term "executable component" is a structure that is well understood in the computing field to be software, hardware, or a combination thereof. For example, when implemented as software, those skilled in the art will understand that the structure of an executable component can include software objects, routines, methods, etc., that can be executed on the computing system. This can include executable components in the computing system heap or on a computer-readable storage medium. The structure of an executable component can exist on a computer-readable medium such that, when interpreted by one or more processors of the computing system (e.g., by processor threads), it causes the computing system to perform functions. This structure can be directly read by a processor, for example, if the executable is binary, or it can be constructed to be interpretable and / or compileable, for example, whether in a single stage or multiple stages, thereby generating such binary that can be directly interpreted by the processor. In other cases, the structure can be hard-coded or hard-wired logic gates, implemented specifically or almost specifically in hardware, such as within a field-programmable gate array (FPGA), application-specific integrated circuit (ASIC), or any other dedicated circuit. Thus, the term "executable" is a term for a structure well-known to those skilled in the art of computing, whether implemented in software, hardware, or a combination thereof. Any embodiments herein are described with reference to actions performed by one or more processing units of a computing system. If such actions are implemented in software, one or more processors direct the operation of the computing system in response to the execution of computer-executable instructions constituting the executable. The computing system may also include communication channels that allow the computing system to communicate with other computing systems via, for example, a network. A "network" is defined as one or more data links that enable the transfer of electronic data between computing systems and / or modules and / or other electronic devices. When information is transmitted or provided to a computing system via a network or another communication connection (e.g., hardwired, wireless, or a combination of hardwired and wireless), the computing system correctly treats that connection as a transmission medium. The transmission medium may include a network and / or a data link, which may be used to carry desired program code in the form of computer-executable instructions or data structures, and may be accessed by a general-purpose computing system or a special-purpose computing system or a combination thereof.While not all computing systems require a user interface, in some embodiments, the computing system includes a user interface system for interacting with a user. The user interface, for example, acts as an input or output mechanism for the user via a display.

[0184] Those skilled in the art will understand that at least a portion of the present invention can be practiced in network computing environments with a variety of computing system configurations, including personal computers, desktop computers, laptop computers, message processors, handheld devices, multiprocessor systems, microprocessor-based or programmable consumer electronics, network PCs, minicomputers, mainframe computers, mobile phones, PDAs, pagers, routers, switches, data centers, wearable devices (such as glasses), etc. The present invention can also be practiced in distributed system environments, where, for example, local and remote computing systems linked by a network via hardwired data links, wireless data links, or a combination of hardwired and wireless data links jointly perform tasks. In a distributed system environment, program modules can reside on both local and remote memory storage devices.

[0185] Those skilled in the art will also understand that at least a portion of the present invention can be practiced in a cloud computing environment. A cloud computing environment can be distributed, but this is not required. When a cloud computing environment is distributed, it can be spread across multiple countries within an organization and / or have components across multiple organizations. In this specification and the appended claims, “cloud computing” is defined as a model for enabling on-demand network access to a shared pool of configurable computing resources, such as networks, servers, storage devices, applications, and services. The definition of “cloud computing” is not limited to any of the many other advantages that can be obtained from such a model when deployed. The computing system of the accompanying drawings includes various components or functional blocks that can implement the various embodiments disclosed herein as explained. These various components or functional blocks can be implemented on a local computing system or on a distributed computing system that includes elements residing in the cloud or aspects implementing cloud computing. These various components or functional blocks can be implemented as software, hardware, or a combination of software and hardware. The computing system shown in the figures may include more or fewer components than those shown in the figures, and some of these components may be combined as needed.

[0186] Any reference numerals in the claims should not be construed as limiting the scope.

[0187] This invention relates to an apparatus for generating properties associated with chemical products. A representation providing unit provides a molecular structure representation, and a partitioning unit divides this molecular structure representation into two or more sub-representations based on geometric and / or electronic partitioning. A solution determining unit enables a quantum computer to generate solutions for the sub-representations and enables a classical computer to generate solutions for the remaining sub-representations. An intermediate property calculation unit generates intermediate properties by combining the solutions of these sub-representations, wherein the intermediate properties are electronic properties of the electronic structure. A generation unit generates properties based on these intermediate properties.

Claims

1. An apparatus for generating technically applicable properties associated with a chemical product, wherein, The chemical product is associated with one or more molecular structures, wherein the device includes: - A molecular structure representation providing unit, which is used to provide a molecular structure representation associated with the molecular structure of the chemical product. - A partitioning unit, which is used to divide the molecular structure representation into two or more sub-representations, wherein the partitioning is associated with the geometric and / or electronic partitioning of the molecular structure represented by the molecular structure representation. - An intermediate solution determination unit, which enables a quantum computer to generate and provide a quantum computing solution for at least one of the two or more sub-representations, and enables a classical computer to generate and provide a classical computing solution for the remaining one or more sub-representations. - An intermediate property calculation unit, which is used to generate one or more intermediate properties associated with the molecular structure by combining the solutions of two or more sub-representations, wherein the intermediate property is an electronic property of the electronic structure of the molecular structure. - A characteristic generation unit, configured to generate technical application characteristics associated with the chemical product based on one or more intermediate characteristics, wherein the technical application characteristics refer to measurable chemical and / or physical quantities associated with the molecular structure, and - A providing unit, which provides the application characteristics of the technology.

2. The apparatus according to claim 1, wherein, The partitioning includes: identifying the most relevant molecular structure portion regarding the application characteristics of the technology based on the molecular structure representation, and partitioning the molecular structure representation such that a sub-representation is associated with the most relevant molecular structure portion and can be solved in quantum computing, wherein the intermediate characteristic computing unit is configured to enable the quantum computer to generate and provide a quantum computing solution for the sub-representation associated with the most relevant molecular structure portion.

3. The apparatus according to claim 2, wherein, The most relevant molecular structural part is associated with the most relevant geometric part of the molecular structure, and the division refers to the geometric division.

4. The apparatus according to claim 2, wherein, The most relevant molecular structural part is associated with the most relevant electronic structural part of the molecular structure, and the division refers to the electronic division.

5. The apparatus according to any one of the preceding claims, wherein, The partitioning includes determining the most relevant geometric part of the molecular structure, generating an electronic structure representation of the most relevant geometric part, and determining the most relevant electronic structure part in the electronic structure representation of the most relevant geometric part as a sub-representation, such that the sub-representation can be solved in quantum computing, wherein the intermediate characteristic computing unit is configured to enable the quantum computer to generate and provide a quantum computing solution for the sub-representation associated with the most relevant electronic structure part of the most relevant geometric part.

6. The apparatus according to any one of the preceding claims, wherein, This one or more intermediate properties are associated with one or more of the following: electron density, electrostatic multipole moment, electronic state energy, excited state properties, and the derivative of the ground state or excited state energy.

7. The apparatus according to any one of the preceding claims, wherein, The application characteristics of this technology are associated with one or more spectral properties of the molecular structure, wherein the generated intermediate properties are associated with the electronic state energies of the molecular structure.

8. The apparatus according to any one of the preceding claims, wherein, The application characteristics of this technology are associated with the thermodynamic and / or kinetic reaction characteristics of chemical reactions involving one or more chemical products, wherein one or more intermediate properties are generated for the molecular structure of the chemical product involved, wherein the one or more intermediate properties are associated with one or more electronic state energies.

9. The apparatus according to claim 8, wherein, Determining the thermodynamic and / or kinetic properties of the chemical reaction includes determining the vibrational zero-point energy and vibrational partition function of the corresponding molecular structure based on the energy of one or more electronic states as intermediate properties, wherein these thermodynamic and / or kinetic properties are further determined based on the vibrational zero-point energy and the vibrational partition function.

10. The apparatus according to any one of claims 8 and 9, wherein, At least one of the chemical products involved is in solution, and one or more intermediate properties generated for the at least one chemical product in solution are further determined based on electron density.

11. The apparatus according to any one of the preceding claims, wherein, The device further includes an alternative feature model providing unit, wherein the alternative feature model providing unit is configured to provide a machine learning-based alternative feature model, wherein the machine learning-based alternative feature model is trained based on a historical training dataset to generate the technology application feature based on one or more intermediate features, wherein the feature generation unit is configured to determine the technology application feature by providing one or more intermediate features as input to the alternative feature model and receiving the technology application feature as output to the alternative feature model.

12. The apparatus according to any one of the preceding claims, wherein, The device further includes a control unit for providing control signals based on the generated technical application characteristics to monitor, control, produce and / or process chemical products.

13. A system for generating technical application characteristics associated with a chemical product, wherein, The system includes: -The apparatus according to any one of the preceding claims - A quantum computer, which is communicatively coupled to the device, and - A classic computer, coupled to the device. The device is configured to control the quantum computer and the classical computer to generate quantum computing solutions and classical computing solutions, respectively.

14. A method, particularly a computer-implemented method, for generating technically applicable characteristics associated with a chemical product, wherein, The chemical product is associated with one or more molecular structures, wherein the method includes: - Provide a molecular structure representation associated with the molecular structure of the chemical product. - Divide the molecular structure representation into two or more sub-representations, wherein the division is associated with the geometric and / or electronic division of the molecular structure represented by the molecular structure representation. - Enables a quantum computer to generate and provide a quantum computation solution for at least one of the two or more sub-representations, and enables a classical computer to generate and provide a classical computation solution for the remaining one or more sub-representations. - By combining solutions from two or more sub-representations, one or more intermediate properties associated with the molecular structure can be generated, wherein the intermediate properties are electronic properties of the electronic structure of the molecular structure. - Based on one or more intermediate characteristics, technical application characteristics associated with the chemical product are generated, wherein the technical application characteristics refer to measurable chemical and / or physical quantities associated with the molecular structure, and - Provides the application characteristics of this technology in relation to the control and / or monitoring processes of the chemical production process associated with the chemical product.

15. A computer program product for generating technical application characteristics associated with a chemical product, wherein, The computer program product is configured to cause the apparatus according to any one of claims 1 to 12 to perform the method according to claim 14.

Images