Material combination screening method and device, electronic equipment and storage medium

Abstract

This application relates to a material combination screening method, apparatus, electronic device, and storage medium. The method includes: constructing a material candidate pool comprising at least a candidate set of amine precursors, a candidate set of carbonyl precursors, and a candidate set of connection modes; extracting physicochemical descriptors obtained based on density functional theory calculations and / or molecular dynamics calculations; generating monomer scores for each component and synergistic scores between components; generating a rule set including at least selection constraints and prohibited pairing constraints; constructing a QUBO energy function comprising individual score terms, pairing synergistic terms, selection constraint penalty terms, and prohibited pairing penalty terms; and inputting the QUBO energy function into a dedicated quantum computer for solution, decoding to obtain candidate experimental combinations that satisfy the selection constraints and prohibited pairing constraints. This application solves the technical problems in related technologies where reliance on manual experience for screening and step-by-step verification leads to low screening efficiency and difficulty in characterizing component synergistic relationships.

Description

Technical Field

[0001] This application relates to the fields of materials computing and quantum computing, and in particular to a method, apparatus, electronic device and storage medium for screening materials combinations. Background Technology

[0002] Two-dimensional covalent organic frameworks (COFs) possess characteristics such as lightweight, high strength, designable structure, tunable pores, and tunable function, showing significant application potential in high-performance thin films, separation materials, flexible materials, and functional devices. In the development of COFs, different amine and carbonyl precursors, linkage modes, and configurations affect the bonding feasibility, interlayer interactions, mechanical properties, and functional performance of the materials. Simultaneously, the formation of two-dimensional polymers is influenced by both thermodynamic and kinetic factors. Factors such as the monomer structure of the precursor, its rotational freedom, coplanarity tendency, reaction pathway energy barriers, interlayer hydrogen bonds, molecular orientation, and amide bond tilt angle are interconnected; a favorable change in one factor may lead to an unfavorable change in another, making it difficult to judge the merits of candidate combinations using a single rule or simple weighted scoring. For example, while precursors with high structural rigidity are beneficial for maintaining a two-dimensional planar configuration, their reduced free rotational capability may lead to a mismatch between the spatial orientation of reactive functional groups and the target topology, thus affecting bonding feasibility. Similarly, while precursor combinations with a strong tendency to coplanarize are thermodynamically favorable for forming planar structures, if the reaction pathway energy barrier is high, kinetic bonding may be difficult to achieve under experimental conditions. The coupling relationships between these factors make traditional screening methods relying on empirical rules or independent single-factor scoring inaccurate in evaluating the overall performance of candidate combinations.

[0003] Currently, the development of two-dimensional covalent organic framework materials typically relies on empirical rules, literature inspiration, and experimental trial and error. Researchers first screen several candidate combinations based on monomer functional group types and known materials, then verify whether the candidate structures can form the target material through density functional theory (DFT) calculations, molecular dynamics (MD) calculations, or experimental characterization. This approach has the following drawbacks: the number of candidate combinations expands rapidly with the increase in the types of precursors, connection methods, and configurations; manual empirical screening or exhaustive screening is costly; synergistic or conflicting relationships between different components are difficult to accurately characterize through independent scoring; and when multiple candidate experiments need to be selected with limited experimental resources, traditional greedy screening tends to yield locally optimal combinations. Summary of the Invention

[0004] This application provides a material combination screening method, apparatus, electronic device, and storage medium to at least solve the technical problems in related technologies where screening efficiency is low and component synergy is difficult to characterize due to reliance on manual experience and individual verification.

[0005] According to a first aspect of the embodiments of this application, a material combination screening method is provided, comprising: constructing a material candidate pool containing at least an amino precursor candidate set, a carbonyl precursor candidate set, and a connection mode candidate set; extracting physicochemical descriptors based on density functional theory calculations and / or molecular dynamics calculations for the components and component pairings in the material candidate pool; generating monomer scores for each component and synergistic scores between components based on the physicochemical descriptors; generating a rule set including at least selection constraints and prohibited pairing constraints; wherein the selection constraints are for selecting a specified number of candidate components in each candidate set; and the prohibited pairing constraints are generated based on a comparison of the physicochemical descriptors with a preset threshold; defining binary decision variables; constructing a quadratic unconstrained binary optimization (QUBO) energy function including a single-item score term, a pairing synergistic term, a selection constraint penalty term, and a prohibited pairing penalty term; and inputting the QUBO energy function into a special-purpose quantum computer. The QuantumComputer (SPQC) is used to solve the problem, and the output solution is decoded into binary decision variables to obtain candidate experimental combinations that satisfy the selection constraints and prohibited pairing constraints.

[0006] Optionally, the material candidate pool further includes at least one of a process discrete level candidate set and a configuration or binding site candidate set; the process discrete level candidate set includes at least one of a temperature level, solvent system, time level, catalyst type or concentration level.

[0007] Optionally, the physicochemical descriptor includes component-level descriptors and pairing co-operational descriptors; wherein: the component-level descriptor includes at least one of precursor structural rigidity, number and spatial distribution of functional groups, free rotation capability, coplanarity tendency, reactivity, monomer size and symmetry, and thermodynamic tendency to form the target two-dimensional structure; the pairing co-operational descriptor includes at least one of bonding feasibility, functional group matching degree, reaction path energy barrier difference, post-bonding configuration planarity, steric hindrance, side reaction risk, interlayer interaction, and hydrogen bond retention rate.

[0008] Optionally, based on the physicochemical descriptor, monomer scores for each component and synergistic scores between components are generated, including: normalizing and weighted summing the physicochemical descriptor to obtain a comprehensive score; the synergistic score includes the synergistic score between the amino precursor and the carbonyl precursor, the synergistic score between the amino precursor and the linkage mode, and the synergistic score between the carbonyl precursor and the linkage mode.

[0009] Optionally, the generation rules for the prohibited pairing constraint include: if the reaction energy barrier of component pairing is higher than a first preset threshold, the planarity deviation after bonding is higher than a second preset threshold, or the structural fluctuation in molecular dynamics simulation exceeds a third preset threshold, then the component pairing is set as prohibited pairing.

[0010] Optionally, the formula for constructing the QUBO energy function is:

[0011] in, , , These are binary decision variable vectors representing the amino precursor, carbonyl precursor, and linkage mode, respectively. , , This is a single-unit score vector; , , For collaborative scoring matrix; , , To select the penalty coefficient for the constraint; , , These are the forbidden pairing matrices for amino-carbonyl, amino-linked, and carbonyl-linked groups, respectively. , , This represents the corresponding prohibited pairing penalty coefficient.

[0012] Optionally, the penalty coefficient is set to satisfy: , , , , and All are greater than the sum of the absolute values ​​of the individual scoring items and the paired collaboration items.

[0013] Optionally, after inputting the QUBO energy function into a dedicated quantum computer for solving, the method further includes: receiving multiple low-energy candidate binary strings output by the dedicated quantum computer; performing feasibility checks, deduplication, and energy sorting on the low-energy candidate binary strings to obtain a list of candidate experimental combinations.

[0014] Optionally, the method further includes: outputting the candidate experimental combination for synthesis and characterization to obtain experimental feedback results; updating the material candidate pool, the monomer score, the synergistic score, and / or the rule set based on the experimental feedback results; and returning to the step of constructing the material candidate pool or extracting physicochemical descriptors based on the updated information to perform the next round of screening.

[0015] According to a second aspect of the embodiments of this application, a material combination screening device is provided, comprising: a candidate pool construction module for constructing a material candidate pool containing at least an amino precursor candidate set, a carbonyl precursor candidate set, and a connection mode candidate set; a descriptor extraction module for extracting physicochemical descriptors based on density functional theory calculations and / or molecular dynamics calculations from the components and component pairings in the material candidate pool; a scoring generation module for generating monomer scores for each component and synergistic scores between components based on the physicochemical descriptors; a constraint generation module for generating a set of rules including at least selection constraints and prohibited pairing constraints; wherein the selection constraints are for selecting a specified number of candidate components in each candidate set; and the prohibited pairing constraints are generated based on a comparison of the physicochemical descriptors with a preset threshold; a QUBO modeling module for defining binary decision variables and constructing a QUBO energy function including a single-item scoring term, a pairing synergistic term, a selection constraint penalty term, and a prohibited pairing penalty term; and a solution and analysis module for inputting the QUBO energy function into a dedicated quantum computer for solution, decoding the output solution result into binary decision variables, and analyzing to obtain candidate experimental combinations that satisfy the selection constraints and prohibited pairing constraints.

[0016] According to a third aspect of the present application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described above.

[0017] According to a fourth aspect of the embodiments of this application, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the method described above.

[0018] In this embodiment, the physicochemical descriptors obtained from density functional theory and molecular dynamics calculations are mapped to monomer scores and inter-component synergistic scores. The generation of forbidden pairing constraints is based on an objective comparison between quantitative indicators and preset thresholds. The monomer score, synergistic score, selection constraint penalty, and forbidden pairing penalty are uniformly encoded into a QUBO energy function. A dedicated quantum computer is used to perform a global search to solve the QUBO energy function, and the output solution is decoded into binary decision variables. Candidate experimental combinations that satisfy selection and forbidden pairing constraints are obtained analytically. This achieves the goal of simultaneously handling independent contributions of components, synergistic relationships between components, and multiple types of constraints within a unified optimization framework. This enables the rapid acquisition of candidate experimental combinations that meet the requirements of chemical feasibility and structural stability when the number of candidate combinations is large. This solves the technical problems in related technologies where the reliance on manual experience for screening and verification leads to low screening efficiency and difficulty in characterizing synergistic relationships between components. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other embodiments can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart illustrating the material combination screening method provided in this application embodiment; Figure 2 A flowchart illustrating the material combination screening method based on QUBO modeling and SPQC solving provided in this application embodiment; Figure 3 A schematic diagram illustrating the candidate pool and variable encoding provided in an embodiment of this application; Figure 4 A schematic diagram of QUBO modeling provided in an embodiment of this application; Figure 5 A schematic diagram of SPQC solving provided in an embodiment of this application; Figure 6 A schematic diagram illustrating the experimental execution and feedback closed loop provided in the embodiments of this application; Figure 7 A schematic diagram of the material combination screening device provided in the embodiments of this application; Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0021] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While some embodiments of this application are shown in the drawings, it should be understood that embodiments of this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the embodiments of this application. It should be understood that the accompanying drawings and embodiments are for illustrative purposes only and are not intended to limit the scope of protection of this application.

[0022] According to a first aspect of the embodiments of this application, a method for screening material combinations is provided. Figure 1 A flowchart of the material combination screening method provided in the embodiments of this application is shown below. Figure 1 As shown, the method includes the following steps: Step S102: Construct a material candidate pool that includes at least an amino precursor candidate set, a carbonyl precursor candidate set, and a connection mode candidate set.

[0023] The construction of the material candidate pool includes: determining the types of precursor functional groups participating in the condensation reaction based on the topological structure type of the target two-dimensional covalent organic framework material; selecting amine and carbonyl precursors with corresponding functional groups from the precursor database; and selecting connection methods matching the reaction type of the functional group from the connection method database. Specifically, each precursor in the amine precursor candidate set has at least two amine functional groups, each precursor in the carbonyl precursor candidate set has at least two carbonyl functional groups, and each connection method in the connection method candidate set corresponds to a topological connection relationship. Each candidate combination in the candidate pool corresponds to a ternary group consisting of an amine precursor, a carbonyl precursor, and a connection method. This ternary group represents the monomer composition and topological structure scheme of a potential two-dimensional covalent organic framework material.

[0024] Step S104: Extract physicochemical descriptors from the components and component pairs in the material candidate pool, based on density functional theory calculations and / or molecular dynamics calculations.

[0025] For each amine precursor, each carbonyl precursor, each connection mode, and each pairing between components in the candidate pool, density functional theory (DFT) calculations and molecular dynamics (MFD) calculations were performed. DFT calculations were used to obtain the molecular orbital energy levels, electrostatic potential distribution, optimized configuration, and reaction pathway energy profiles for each precursor. MFD calculations were used to obtain the conformational distribution, thermal motion characteristics, and intermolecular interaction energies of each precursor under the target reaction conditions. The numerical outputs of the above calculation results were used as physicochemical descriptors. The specific type of physicochemical descriptor was determined based on the molecular structural characteristics of the precursor and the reaction type.

[0026] Step S106: Based on the physicochemical descriptor, generate monomer scores for each component and synergistic scores between components.

[0027] The monomer score is obtained by weighted summation of multiple physicochemical descriptors of a single component after normalization, and is used to characterize the contribution of the component alone to the formation of the target two-dimensional covalent organic framework material. The synergistic score is obtained by weighted summation of the paired synergistic descriptors of two components after normalization, and is used to characterize the additional contribution due to the synergistic effect when the two components are paired.

[0028] Step S108: Generate a rule set that includes at least selection constraints and prohibition pairing constraints; the selection constraints are to select a specified number of candidate components in each candidate set; the prohibition pairing constraints are generated based on the comparison of physicochemical descriptors with preset thresholds.

[0029] The selection constraints specify the selection of a specified number of candidate components from each of the amine precursor candidate set, carbonyl precursor candidate set, and connection mode candidate set. The specified number is determined based on the stoichiometry of the target two-dimensional covalent organic framework material. The prohibited pairing constraints are generated based on a comparison between a physicochemical descriptor and a preset threshold; for example, if the reaction barrier for pairing a component is higher than a preset barrier threshold, or the planarity deviation after bonding is higher than a preset planarity threshold, or the structural fluctuation amplitude in molecular dynamics simulations exceeds a preset fluctuation threshold, then the pairing of that component is marked as prohibited.

[0030] Step S110: Define binary decision variables and construct a QUBO energy function that includes a single-item scoring term, a pairing synergy term, a selection constraint penalty term, and a prohibition pairing penalty term.

[0031] The binary decision variables are represented by binary numbers, taking values ​​of 0 or 1, representing whether the corresponding component in the candidate pool is not selected or is selected, respectively. The QUBO energy function aims to minimize the selected component. The single-item scoring term and the pairing coordination term guide the solver to prioritize the selection of components with higher scores and component pairs. The selection constraint penalty term forces the number of selected components in each candidate set to be equal to a specified number. The prohibition pairing penalty term forces that component pairs marked as prohibited pairings cannot be selected simultaneously.

[0032] Step S112: Input the QUBO energy function into a dedicated quantum computer for solving, and decode the output solution into binary decision variables to obtain candidate experimental combinations that satisfy the selection constraints and prohibited pairing constraints.

[0033] A dedicated quantum computer searches for low-energy states in the combination space of binary decision variables through the coupling evolution between optical pulses, outputting multiple low-energy candidate binary strings. The amino precursor, carbonyl precursor, and connection mode corresponding to the binary decision variables with a value of 1 in each candidate binary string are extracted and combined into complete candidate experimental combinations. Each candidate experimental combination satisfies the requirement of selecting a specified number of candidates from each class of candidate sets as specified in the selection constraints, and does not contain any component pairings marked as prohibited in the prohibited pairing constraints.

[0034] In this embodiment, by mapping the physicochemical descriptors obtained from density functional theory calculations and molecular dynamics calculations to monomer scores and inter-component synergistic scores, and by establishing the generation of forbidden pairing constraints based on an objective comparison between quantitative indicators and preset thresholds, and by uniformly encoding monomer scores, synergistic scores, selection constraint penalties, and forbidden pairing penalties into a QUBO energy function, the goal of simultaneously handling independent component contributions, inter-component synergistic relationships, and multiple types of constraints within a unified optimization framework is achieved. Since the generation of forbidden pairing constraints does not rely on human experience but is based on an objective comparison between quantitative indicators obtained from density functional theory calculations and molecular dynamics simulations and preset thresholds, the criteria for determining forbidden pairing constraints are consistent and repeatable across different precursor systems, avoiding inconsistencies in screening criteria caused by differences in the experience of different researchers. Meanwhile, because the structure of the QUBO model matches the hardware architecture of a dedicated quantum computer, the completed QUBO model can be directly input into the dedicated quantum computer for solution without model conversion or approximation. The dedicated quantum computer quickly finds low-energy solutions in the candidate combination space through quantum parallel search. Compared with the serial search method of classical computers, it can obtain candidate experimental combinations that meet the constraints in less computation time when the number of candidate combinations is large, reducing the computational resource investment and time cost in the early screening stage. By uniformly encoding individual scoring, cooperative scoring, selection constraints, and forbidden pairing constraints into the same QUBO mathematical expression, the objective optimization and constraint satisfaction in the material screening problem can be handled simultaneously, avoiding decision conflicts that may occur when scoring calculation, feasibility determination, and constraint checking are performed step by step.

[0035] In some embodiments, the material candidate pool further includes at least one of a process discrete level candidate set and a configuration or binding site candidate set; the process discrete level candidate set includes at least one of a temperature level, solvent system, time level, catalyst type or concentration level.

[0036] When extracting physicochemical descriptors and generating synergistic scores, the aforementioned process parameter levels are treated equally with amine precursors, carbonyl precursors, and linkage modes, and uniformly coded as binary decision variables. The compatibility relationships between each process level and the matching relationships between process parameters and precursors are incorporated into the synergistic scoring matrix and the rules for prohibiting pairing constraints. For temperature levels, the differences in bonding free energy and interlayer structure retention rate under different temperature conditions are calculated based on reaction thermodynamic data obtained from DFT calculations and structural stability data obtained from MD simulations. For solvent systems, the degree of matching between the solvent and precursor is determined based on solvent polarity, hydrogen bond donation and acceptance capabilities, and solubility parameters, combined with reaction energy barrier data. For time levels, the time window required to reach the target conversion rate is determined based on reaction kinetic simulation data. For catalyst type or concentration levels, the compatibility of the catalyst-precursor combination is determined based on the magnitude of the reaction pathway energy barrier reduction and the relative energy of competing side reaction pathways obtained from DFT calculations. When a certain process parameter setting does not meet the preset matching conditions with a certain precursor or precursor combination, the pairing of that process parameter setting with the precursor or precursor combination is set to prohibited pairing.

[0037] In this embodiment, by uniformly encoding the discrete process levels, precursors, and connection methods as binary decision variables and incorporating them into the QUBO model, the material combination screening process can simultaneously consider the synergistic relationships between material components and the matching relationships between process conditions and material components. This avoids the screening bias caused by treating process conditions as fixed parameters in traditional step-by-step optimization. Since the process levels and precursor combinations are jointly searched under a unified optimization framework, the final output candidate experimental combinations simultaneously satisfy preset constraints at both the material and process levels. These combinations can be directly used to guide actual experimental operations without the need to adjust process parameters or re-verify process conditions after screening.

[0038] In some embodiments, the physicochemical descriptor includes a component-level descriptor and a pairing co-operational descriptor; wherein: the component-level descriptor includes at least one of precursor structural rigidity, number and spatial distribution of functional groups, free rotation capability, coplanarity tendency, reactivity, monomer size and symmetry, and thermodynamic tendency to form the target two-dimensional structure; the pairing co-operational descriptor includes at least one of bonding feasibility, functional group matching degree, reaction pathway energy barrier difference, post-bonding configuration planarity, steric hindrance, side reaction risk, interlayer interaction, and hydrogen bond retention rate.

[0039] When extracting the above descriptors, density functional theory is first used to calculate the electronic structure information, molecular configuration optimization results and reaction path energy profiles of each precursor. Based on the molecular configuration optimization results, the structural rigidity parameter of the precursor is extracted. This parameter is characterized by calculating the number of rotatable dihedral angles and the distribution of rotational energy barriers in the molecule. Based on electronic structure information and frontier molecular orbital distribution, the number and spatial distribution parameters of functional groups are extracted. This parameter is used to characterize the position and relative orientation of reactive functional groups such as amine or carbonyl groups in the molecule. Based on the dihedral angle distribution of adjacent aromatic rings or conjugated units in the molecular configuration optimization results, the coplanarity tendency parameter is extracted. This parameter is used to characterize the ability of the precursor to maintain a two-dimensional planar configuration after bonding. Based on the energy difference between the transition state and the reactants in the reaction path energy profile, the reactivity parameter is extracted. This parameter is used to characterize the reaction rate trend of the precursor participating in the target reaction. Based on the three-dimensional size and shape of the precursor molecule, the monomer size and symmetry parameters are extracted. This parameter is used to characterize the degree of spatial matching when the precursors are periodically arranged in a two-dimensional plane. Based on the unit cell parameters and lattice energy of the target two-dimensional structure, the thermodynamic tendency parameter for forming the target two-dimensional structure is extracted. This parameter is used to characterize the degree to which the precursor combination thermodynamically tends to form the target topology.

[0040] For the pairing cooperation descriptor, the feasibility of bonding is determined based on the chemical compatibility between the functional groups of the two precursors; the degree of functional group matching is determined based on the spatial orientation and electronic matching degree of the two precursors; the energy barrier difference of the reaction path is obtained based on the energy difference between the transition state and the reactants in the reaction path calculated by density functional theory; the planarity parameter of the post-bonding configuration is obtained based on the dihedral angle distribution of aromatic rings or conjugated units in the optimized bonding configuration; the steric hindrance parameter is obtained based on the van der Waals repulsion energy distribution between atoms during the approach of reactant molecules; the risk of side reactions is determined based on the energy barrier height of possible competing reaction paths; the interlayer interaction parameter is obtained based on the non-covalent interaction energy and contact area between two structural units in molecular dynamics simulations; and the hydrogen bond retention rate parameter is obtained based on the percentage of frames in the molecular dynamics trajectory where the distance and angle between the hydrogen bond donor and acceptor satisfy the hydrogen bond condition. All the above descriptors are normalized to ensure that their values ​​are within the same numerical range, and then a weighted sum is used to merge them into a comprehensive score. The weight coefficients corresponding to different descriptors are preset according to the performance requirements of the target application scenario. For example, when the target is a high-strength structural material, the weights of structural rigidity and interlayer interaction are increased; when the target is a flexible thin film material, the weights of free rotation capability and configuration planarity are adjusted accordingly.

[0041] In this embodiment, by extracting multi-dimensional component-level descriptors and pairing co-descriptors from density functional theory and molecular dynamics calculations, a comprehensive and quantitative characterization of each candidate component and each component pair can be achieved. This overcomes the shortcomings of empirical rules, which struggle to quantify complex physicochemical properties, or the bias in screening results caused by relying on a single indicator. Since the aforementioned descriptors correspond to different aspects of material synthesis feasibility, structural stability, and functional performance, the comprehensive score obtained by normalizing and weighting the descriptors can reflect the comprehensive performance of candidate combinations across multiple target dimensions, thereby improving the matching degree between the screening results and practical application requirements.

[0042] In some embodiments, based on physicochemical descriptors, monomer scores for each component and synergistic scores between components are generated, including: normalizing and weighting the physicochemical descriptors to obtain a comprehensive score; the synergistic scores include the synergistic scores of amino precursors and carbonyl precursors, the synergistic scores of amino precursors and linkage modes, and the synergistic scores of carbonyl precursors and linkage modes.

[0043] For the various dimensions of physicochemical descriptors, such as component-level descriptors (precursor structural rigidity, number and spatial distribution of functional groups, free rotation capability, coplanarity tendency, reactivity, etc.) and pairing-cooperative descriptors (bonding feasibility, functional group matching degree, reaction pathway energy barrier difference, post-bonding configuration planarity, steric hindrance, side reaction risk, interlayer interaction, hydrogen bond retention rate, etc.), the original values ​​of each dimension are first normalized to convert descriptors with different dimensions and value ranges into the same numerical interval, thereby eliminating scoring bias caused by differences in units or orders of magnitude. The normalization process uses extreme value normalization, that is, the original values ​​are linearly mapped to a preset target interval based on the minimum and maximum values ​​of the descriptor among all candidate samples. After normalization, weight coefficients are assigned to each descriptor dimension according to the performance requirements of the target application scenario. The magnitude of the weight coefficient reflects the importance of the corresponding descriptor in the comprehensive score. The comprehensive score is obtained by multiplying the normalized descriptor values ​​by their corresponding weight coefficients and summing the results. The weighting coefficients are allocated according to the performance orientation of the target material. For example, when the target is a high-strength structural material, descriptors related to structural stiffness and interlaminar interaction are given higher weights; when the target is a flexible thin film material, descriptors related to free rotation capability and coplanar tendency are given higher weights.

[0044] The synergy score comprises three specific pairing relationships: the synergy score between the amino precursor and the carbonyl precursor, the synergy score between the amino precursor and the linkage mode, and the synergy score between the carbonyl precursor and the linkage mode. The synergy score between the amino and carbonyl precursors is generated based on density functional theory calculations of bonding energy, reaction energy barrier, and post-bonding configuration planarity, characterizing the degree of matching of the precursor combination when forming covalent bonds. The synergy score between the amino precursor and the linkage mode is generated based on the bond angle matching degree and spatial orientation adaptability of the precursor under different linkage topologies, characterizing the degree of fit between the amino precursor and a specific linkage mode. Similarly, the synergy score between the carbonyl precursor and the linkage mode is generated based on the bond angle matching degree and spatial orientation adaptability of the carbonyl precursor under different linkage topologies, characterizing the degree of fit between the carbonyl precursor and a specific linkage mode. In the generation of the aforementioned synergistic scores, the synergistic relationship between the precursor and the linkage is determined based on the degree of matching between the spatial orientation of the functional group and the attack direction of the amino or carbonyl group in the linkage. When the spatial orientation of the linkage is consistent with the reaction direction of the functional group, the synergistic score increases accordingly; when steric hindrance makes the reaction difficult, the synergistic score decreases accordingly. The synergistic scores of each group are also normalized and weighted summed to form a comprehensive score, enabling the monomer scores and various synergistic scores to be compared and combined within the same numerical framework.

[0045] In this embodiment, by normalizing and weighting the multidimensional physicochemical descriptors, descriptors with different dimensions and physical meanings are uniformly converted into comparable comprehensive scores, enabling the material screening problem to be transformed into an optimization problem based on a unified scoring standard. Simultaneously, by separately constructing three sets of synergistic scores—amine-carbonyl, amine-linked, and carbonyl-linked—the synergistic relationships between components and the independent contributions of each component during the material screening process are separated and expressed. This allows the two types of scores with different properties to correspond to linear and quadratic terms respectively in the subsequent QUBO model, thus adapting the mathematical structure of the combinatorial optimization model while maintaining the clarity of the physical meaning of the scoring system.

[0046] In some embodiments, the generation rules for prohibiting pairing constraints include: if the reaction energy barrier of component pairing is higher than a first preset threshold, the planarity deviation after bonding is higher than a second preset threshold, or the structural fluctuation in molecular dynamics simulation exceeds a third preset threshold, then the component pairing is set as prohibited pairing.

[0047] The rules for generating forbidden pairing constraints are based on a comparison of density functional theory calculations and molecular dynamics simulation results with preset thresholds. For any component pairing in the candidate pool, the reaction energy barrier, post-bonding planarity deviation, and structural fluctuation amplitude in the molecular dynamics simulation are calculated. These three quantitative indicators are compared with their respective first, second, and third preset thresholds. If any indicator exceeds the corresponding preset threshold, the component pairing is marked as forbidden pairing.

[0048] The reaction barrier is calculated using density functional theory and is specifically the energy difference between the initial state and the transition state of the reactants along the reaction path. When the reaction barrier is higher than a first preset threshold, it indicates that the pairing is unlikely to overcome the activation energy required for bonding under normal experimental conditions, and the bonding reaction is not likely to occur. Setting this pairing as prohibited can avoid screening out kinetically infeasible candidate combinations. The value of the first preset threshold is set according to the average energy barrier level of the target reaction system, for example, based on the statistical distribution of reaction barriers in successful synthesis cases from existing experimental or computational data.

[0049] Planarity deviation after bonding is determined by the final geometrically optimized configuration calculated using density functional theory. Specifically, the distance of each atom in the optimized two-dimensional structure from the average plane is statistically analyzed, and the root mean square value of the deviation distance of all atoms is used as a quantitative indicator of planarity deviation. When this deviation value is higher than a second preset threshold, it indicates that the pairing is unlikely to maintain a two-dimensional planar configuration after bonding, and instead tends to form a wrinkled or twisted non-planar structure. Setting this pairing as prohibited pairing can avoid screening out non-planar candidate combinations that do not conform to the structural characteristics of two-dimensional materials. The value of the second preset threshold is determined based on the distribution of planarity deviations of reported structures in the target two-dimensional material.

[0050] Structural fluctuations in molecular dynamics simulations are characterized by calculating the root-mean-square (RMS) fluctuations of atomic positions over time in the simulated trajectory. Specifically, molecular dynamics simulations are performed on the bonded structure at a preset temperature, and the deviation of each atom from its average position during the equilibrium phase is statistically analyzed. The average value of the RMS fluctuations of all atoms is used as a quantitative indicator of the structural fluctuation amplitude. When this fluctuation amplitude exceeds a third preset threshold, it indicates that the structure formed by this pairing is not sufficiently stable under the target temperature conditions and is prone to conformational transformation or structural collapse under thermal perturbation. Setting this pairing as prohibited avoids screening out thermodynamically unstable candidate combinations. The value of the third preset threshold is determined based on the temperature range and structural stability requirements of the target application scenario.

[0051] When generating prohibited pairing constraints, the three judgment conditions mentioned above are combined using a logical OR relationship. That is, as long as any one of the indicators of reaction energy barrier, planarity deviation, and structural fluctuation exceeds the corresponding preset threshold, the pairing of the components is prohibited, without the need for all three conditions to be met simultaneously. This judgment method ensures that the three types of constraints—chemical feasibility, structural planarity, and thermodynamic stability—are all independently guaranteed, and infeasibility in any one aspect is sufficient to exclude the candidate combination.

[0052] In this embodiment, by basing the generation of forbidden pairing constraints on an objective comparison between quantitative indicators from density functional theory calculations and molecular dynamics simulations and preset thresholds, the determination process for forbidden pairing constraints no longer relies on human experience or literature inspiration, but is automatically completed based on physicochemical simulation data, avoiding the uncertainty brought about by subjective judgment. Simultaneously, since pairing feasibility is evaluated from three independent dimensions—reaction energy barrier, post-bonding planarity deviation, and structural fluctuation—pairing is only allowed to enter the candidate range when the quantitative indicators of all three dimensions meet preset conditions, reducing the possibility of infeasible combinations appearing in the screening results.

[0053] In some embodiments, the formula for constructing the QUBO energy function is:

[0054] in, , , These are binary decision variable vectors representing the amino precursor, carbonyl precursor, and linkage mode, respectively. , , This is a single-unit score vector; , , For collaborative scoring matrix; , , To select the penalty coefficient for the constraint; , , These are the forbidden pairing matrices for amino-carbonyl, amino-linked, and carbonyl-linked groups, respectively. , , This represents the corresponding prohibited pairing penalty coefficient.

[0055] In the above formula, Used to calculate the monomer score contribution of the selected component, where This represents the sum of the monomer scores of the selected amino precursors. This represents the sum of the monomer scores for the selected carbonyl precursor. This represents the sum of monomer scores for the selected carbonyl precursors. In the energy function, the higher the monomer score, the smaller the negative values ​​of these three terms, thus lowering the energy function value. In other words, combinations with high scores tend to be selected during the optimization process.

[0056] Used to calculate the synergistic score contribution among the selected components. This indicates the synergistic score between the selected amino precursor and carbonyl precursor. This indicates the synergistic score between the selected amino precursor and the linkage mode. This represents the synergy score between the selected carbonyl precursor and the connection method. The higher the synergy score, the smaller the negative values ​​of these three terms, and the lower the energy function value. In other words, combinations with high synergy scores tend to be selected during the optimization process.

[0057] , , To select a constraint penalty term, used to constrain the number of selected components in each candidate set. Wherein, This represents the sum of the amino precursor variables, i.e., the number of selected amino precursors. The value is positive when the quantity is not equal to 1. When the number of selected amine precursors is 1, this term takes a value of 0, resulting in no penalty; when the selected quantity is 0 or greater than 1, this term takes a positive value, increasing the energy function value, thus tending to avoid such combinations in optimization. The mechanism of the constraint penalty term is the same for carbonyl precursors and connection methods. , , The value of is greater than the sum of the absolute values ​​of all individual and collaborative scoring items, making the energy value of combinations that do not meet the selection constraints higher than that of any feasible combination.

[0058] , , This is a prohibited pairing penalty term, used to penalize combinations that do not satisfy the prohibited pairing constraint. It is a prohibited pairing penalty term for amino precursors and carbonyl precursors. For example, To prohibit pairing matrix, The Middle Line number When the value of a column is 1, it indicates that the first column is... The first amino precursor and the second The carbonyl precursor is prohibited from pairing; a value of 0 indicates that the pairing is not prohibited. When a selected amino precursor and a selected carbonyl precursor are forbidden to pair, this term takes the value of 1, multiplied by This results in a positive penalty, causing the energy function value to increase. The value of is greater than the sum of the absolute values ​​of all individual and collaborative scoring items, such that the energy value of any combination containing prohibited pairings is higher than that of any combination not containing prohibited pairings. and They have the same mechanism of action, respectively used to prohibit pairing between amino precursors and linkages, and between carbonyl precursors and linkages.

[0059] After constructing the QUBO energy function using the above formula, the singleton score, cooperative score, selection constraint, and prohibited pairing constraint are uniformly encoded into a single mathematical expression. Among these, , , and , , The value of is derived from the normalized score of density functional theory calculations and molecular dynamics calculations. and The value is pre-defined based on the scoring range. This QUBO energy function can be directly input into a dedicated quantum computer for solution, without additional model transformation or reduction. The dedicated quantum computer solves the problem by analyzing the binary decision variables in this energy function. , , The system searches the combination space to find candidate combinations that minimize the energy function. During this process, low-quality combinations and combinations that do not meet the constraints are excluded due to their high energy values, while combinations that meet the constraints and have high overall quality are retained as output candidates due to their low energy values.

[0060] In this embodiment, by unifying monomer scoring, synergistic scoring, selection constraints, and forbidden pairing constraints into a single mathematical expression of the QUBO energy function, the objective function and constraints in the material screening problem can be processed simultaneously within a unified optimization framework. This avoids decision conflicts that may arise when scoring calculation, feasibility determination, and constraint checking are performed step-by-step. Since the scoring term in this formula is generated at the front end based on physicochemical descriptors derived from density functional theory and molecular dynamics calculations, and the forbidden pairing matrix in the constraint term is pre-constructed through comparison of quantitative indicators with preset thresholds, this energy function accurately reflects the physicochemical implications of the material screening problem. The output low-energy candidate combinations simultaneously meet preset conditions in terms of chemical feasibility and structural stability, reducing the elimination rate in subsequent experimental verification.

[0061] In some embodiments, the penalty coefficient is set to satisfy: , , , , and All are greater than the sum of the absolute values ​​of the individual scoring items and the paired synergy items. For the optimal case that may occur among all candidate combinations, that is, all individual scores and synergy scores are taken as their maximum values ​​and accumulated, and the sum of the corresponding positive energy gains is used as the benchmark value. Each penalty coefficient is set to a value greater than this benchmark value.

[0062] First, calculate the maximum absolute value of monomer scores for all amino precursors, the maximum absolute value of monomer scores for all carbonyl precursors, the maximum absolute value of monomer scores for all linkage modes, and the maximum absolute value of synergistic scores for all component pairings. Summing these maximum values ​​yields the baseline value. Then, apply the penalty coefficient for selection constraints. , , and the no-pairing penalty coefficient , , All values ​​are set to positive numbers greater than the baseline value. This setting ensures that any combination that does not satisfy the selection constraint or the prohibited pairing constraint, regardless of its individual and cooperative scores, will always have a higher energy function value than all feasible combinations that satisfy the constraints. In other words, when a dedicated quantum computer solves the problem of minimizing this QUBO energy function, the energy values ​​of all infeasible combinations will be higher than those of feasible combinations, and the optimal solution output by the solver will necessarily lie within the feasible solution space. There will be no situation where the score is improved at the expense of satisfying the constraints.

[0063] In practice, the specific values ​​of the penalty coefficients are determined based on the actual distribution of the scoring data. Provided that the values ​​are greater than the benchmark value, the penalty coefficients do not need to be equal; they can be assigned values ​​according to the importance of different constraint types. All penalty coefficients are greater than the benchmark value, which is calculated by taking the individual scoring vectors... , , The maximum absolute value of the elements in the matrix, and each collaborative scoring matrix. , , The maximum absolute value of each element is used as the base value. The sum of these six maximum values ​​is then used as the baseline value. All penalty coefficients set exceed this baseline value.

[0064] In this embodiment, by uniformly setting each penalty coefficient to a value greater than the sum of the absolute values ​​of all individual scoring items and paired synergy items, the energy cost incurred by combinations that do not satisfy selection constraints or prohibit pairing constraints exceeds any scoring gain. This prioritizes constraint feasibility over scoring optimization during the QUBO model solution process. Since the penalty coefficient values ​​are predetermined based on the absolute value range of the scoring data and do not depend on iterative adjustments during the specific solution process, this setting allows the solver to focus on searching for the optimal scoring combination within the feasible solution space while prioritizing feasible solutions. This avoids outputting infeasible solutions due to insufficient penalty coefficient values ​​or numerical accuracy issues caused by excessively large penalty coefficient values.

[0065] In some embodiments, after inputting the QUBO energy function into a dedicated quantum computer for solving, the method further includes: receiving multiple low-energy candidate binary strings output by the dedicated quantum computer; performing feasibility checks, deduplication, and energy sorting on the low-energy candidate binary strings to obtain a list of candidate experimental combinations.

[0066] After solving the QUBO energy function, the dedicated quantum computer outputs multiple candidate binary strings. These binary strings correspond to different combinations of values ​​for the binary decision variables, representing different candidate combinations in the material candidate pool. The multiple candidate binary strings output by the dedicated quantum computer are arranged from low to high according to their corresponding energy values. The low-energy candidate binary strings represent combinations with lower energy values ​​in the QUBO model, that is, candidate combinations with better overall performance after comprehensively considering individual scores, cooperative scores, and constraint penalties.

[0067] Upon receiving multiple low-energy candidate binary strings from the dedicated quantum computer, a feasibility check is first performed on each string. This check verifies whether the values ​​of the binary decision variables corresponding to each candidate string fully satisfy all conditions of the selection and prohibition pairing constraints. Candidate binary strings that satisfy both constraints are retained, while those that do not satisfy any constraint are discarded. Through this feasibility check, practically executable candidate combinations are selected from the candidate set output by the dedicated quantum computer.

[0068] Then, the candidate binary strings that pass the feasibility check are deduplicated. Deduplication involves checking whether the combinations of binary decision variable values ​​corresponding to each candidate binary string are the same, and merging the same repeated combinations into a single record. Deduplication avoids repeated evaluation and sorting of the same candidate combinations in subsequent steps, reducing unnecessary computation.

[0069] After deduplication, the remaining candidate binary strings are sorted according to their corresponding energy values, with lower energy values ​​listed first and higher energy values ​​listed last. After feasibility checks, deduplication, and energy sorting, the results are compiled into a candidate experimental combination list. Each item in the list corresponds to a candidate experimental combination that satisfies all constraints, including specific information about the corresponding amine precursor, carbonyl precursor, and connection method, as well as its predicted comprehensive score and energy value. This candidate experimental combination list is arranged in priority order; combinations at the top of the list have lower energy values, meaning they have better comprehensive screening scores and can be prioritized for experimentation. When outputting the candidate experimental combination list, based on the upper limit of actual experimental resources, several top-ranked candidate combinations can be selected as recommended experimental schemes.

[0070] In this embodiment, by performing feasibility checks, deduplication, and energy sorting on multiple low-energy candidate binary strings output by a dedicated quantum computer, the candidate solutions provided by the quantum computer are further screened and organized, ensuring that the final output candidate experimental combinations all meet the preset constraints and have no duplicates. Feasibility checks eliminate candidate combinations that may not meet the constraints during the quantum computer's solution process; deduplication eliminates the recurrence of the same combination in different samples; and energy sorting prioritizes combinations with better overall performance. The resulting list of candidate experimental combinations provides a reference for researchers to select solutions with limited experimental resources, eliminating the need for manual verification of the constraint satisfaction and priority order of each candidate combination.

[0071] In some embodiments, the above method further includes: outputting candidate experimental combinations for synthesis and characterization to obtain experimental feedback results; updating the material candidate pool, monomer score, synergistic score, and / or rule set based on the experimental feedback results; and returning to the step of constructing the material candidate pool or extracting physicochemical descriptors based on the updated information to perform the next round of screening.

[0072] After outputting candidate experimental combinations, synthesis experiments were performed according to the specific information on the amine precursor, carbonyl precursor, and linkage mode recorded in the candidate experimental combination list. The synthesized products were then characterized structurally and tested for performance. Structural characterization included one or more of powder X-ray diffraction, Fourier transform infrared spectroscopy, solid-state nuclear magnetic resonance, and transmission electron microscopy, used to verify whether the product formed the target two-dimensional covalent organic framework structure and the crystallinity and phase purity of the product. Performance testing was determined based on the target application scenario and included one or more of mechanical property testing, thermal stability analysis, porosity determination, and gas adsorption performance testing, used to evaluate the actual performance of the product.

[0073] After synthesis and characterization, experimental feedback results are obtained. These results include information on whether the product successfully formed the target structure, the crystallinity grade of the product, the percentage of phase purity of the target structure in the product, and the specific values ​​of various performance tests. The experimental feedback results are compared with the predicted results, and the relevant parameters for material screening are updated based on the comparison. The specific update methods include one or more of the following scenarios: when a precursor exhibits higher reactivity than predicted in actual experiments, the monomer score of the precursor is increased; when a component pair exhibits better bonding effects than predicted in actual experiments, the synergistic score of the pair is increased; when a component pair results in impure products due to side reactions in actual experiments, the synergistic score of the pair is decreased; when a component pair repeatedly fails to synthesize in actual experiments, the pair is added to the forbidden pairing constraint, i.e., marked as 1 at the corresponding position in the forbidden pairing matrix; when the availability of a precursor changes due to supply issues or cost changes, the precursor is removed from the candidate pool or its score is adjusted accordingly; when a new feasible component pair is tested in the experiment, the pair is added to the candidate pool and an initial synergistic score is generated for it.

[0074] Based on the updated information, the process returns to the steps of constructing a candidate pool of materials or extracting physicochemical descriptors for the next round of screening. Specifically, if the update involves adding or removing precursor types or connection methods in the candidate pool, the process returns to the steps of constructing the candidate pool, rebuilding the pool based on the updated candidate set. If the update only involves adjusting the scoring parameters or constraint rules without changing the candidate set, the process returns to the steps of extracting physicochemical descriptors, recalculating the scores based on the existing candidate pool and the updated scoring parameters, and then performing subsequent screening steps. After one or more iterations, as the scoring parameters gradually approach the actual experimental results and the constraint rules gradually eliminate practically infeasible combinations, the degree of agreement between the screening results and the actual experiments improves with the increase in the number of iterations.

[0075] In this embodiment, by feeding back the actual synthesis and characterization results of candidate experimental combinations to the candidate pool, scoring parameters, and constraint rules, the material screening model can continuously refine the screening parameters using experimental data. This allows the monomer and synergistic scores to gradually approach the true performance of each component during subsequent screening, and the prohibition of pairing constraints to gradually approach the true chemical feasibility boundary. As the number of iterations increases, the quality of the candidate experimental combinations output during the screening process gradually improves, and the actual experimental success rate of the candidate combinations also increases accordingly.

[0076] In the above embodiments, some features can be implemented using alternative methods. For example, monomer scores and synergistic scores in the QUBO model can be obtained not only based on density functional theory calculations and molecular dynamics calculations, but also from machine learning models, experimental databases, literature rules, or expert scoring methods. These alternative methods can all provide numerical inputs for the QUBO model that reflect component quality and synergistic relationships between components. Candidate variables are not limited to amine precursors, carbonyl precursors, and linkage methods; they can also include one or more of the following: solvent type, catalyst type, reaction temperature, reaction time, reactant concentration, reaction pressure, post-treatment method, and characterization priority. All of these variables can be incorporated into the QUBO model using the same binary decision variable encoding method. The technical solution provided in this application is not only applicable to two-dimensional covalent organic framework materials, but also to two-dimensional polymer materials and other material development tasks that require selecting experimental combinations from multiple discrete components and process conditions.

[0077] The embodiments of this application will be described in detail below with reference to specific implementation diagrams.

[0078] Figure 2 A flowchart of the material combination screening method based on QUBO modeling and SPQC solution provided in the embodiments of this application is shown below. Figure 2 As shown, the method includes steps such as candidate pool construction, descriptor extraction, experimental design rule generation, QUBO modeling, SPQC solving, candidate experiment output, experimental execution feedback, and rule updating.

[0079] Figure 3 A schematic diagram of the candidate pool and variable encoding provided in the embodiments of this application, as shown below. Figure 3 As shown, the process includes steps such as material candidate pooling, candidate combination, and one-hot binary encoding of variables. The candidate pool includes at least one or more of the following candidate sets: amine precursor candidate set. Carbonyl precursor candidate set Connection method or candidate set of connection topology Candidate set of discrete process slots For example, temperature setting, solvent system, time setting, catalyst type or concentration setting; configuration or set of candidate binding sites. .

[0080] Furthermore, the candidate combinations can be expressed as: Each candidate combination corresponds to a possible material synthesis or computational verification experiment. An initial candidate pool can be generated based on literature, existing experimental records, supplier databases, molecular structure databases, or manually entered information.

[0081] Descriptors are extracted for each component and component pair in the candidate pool. Descriptors can be derived from DFT / MD calculations, existing experimental data, literature experience, or expert rules. Component-level descriptors include, but are not limited to: precursor structural rigidity; number and spatial distribution of functional groups; free rotation capability; coplanarity tendency; reactivity; monomer size and symmetry; thermodynamic tendency to form the target 2D structure; and contribution to the target mechanical properties, stability, or functional properties. Pairing or synergistic descriptors include, but are not limited to: bonding feasibility between amine and carbonyl precursors; degree of matching between amine precursors and linkage modes; degree of matching between carbonyl precursors and linkage modes; differences in planarity, interlayer interactions, and amide bond tilt angles due to different configurations or binding sites; energy barrier differences in the reaction pathway; risk of forming byproducts or uncontrollable structures; experimental availability, cost, safety, and process window.

[0082] Based on the above descriptors, construct the following scores or parameters: Indicates the first Monomer score of each amino precursor; Indicates the first Monomer score of carbonyl precursor; Indicates the first Individual score for each connection method; Indicates the first The first amino precursor and the second Synergistic scoring of individual carbonyl precursors; Indicates the first The first amino precursor and the second Collaborative scoring for each connection method; Indicates the first The carbonyl precursor and the first Collaborative scoring for each connection method.

[0083] A higher score indicates that the component or combination of components is more conducive to forming the target material or to achieving the target performance. If the score comes from multiple indicators, a comprehensive score can be obtained through normalization and weighted summation. For example:

[0084] in, , , , , These represent the synthesis feasibility score, mechanical property score, stability score, cost or availability score, and structural novelty score, respectively. These are the weighting coefficients.

[0085] Rating Vector , , It is not arbitrarily set by humans, but is calculated from the physicochemical descriptors in the COFs material formation process. For amine precursors, carbonyl precursors and linkage modes, descriptors such as the number of functional groups, spatial distribution of functional groups, molecular rigidity, coplanarity, steric hindrance, reactivity, synthetic availability and cost are extracted respectively.

[0086] DFT calculations are used to obtain the reaction free energy, reaction barrier, configurational stability, post-bonding planarity deviation, and critical bond length / bond angle for candidate combinations; MD calculations are used to obtain interlayer structural stability, hydrogen bond retention rate, configurational fluctuation amplitude, and structural retention capability after high-temperature perturbation. These indicators are then normalized to form synthesis feasibility scores, structural stability scores, mechanical property tendency scores, and experimental feasibility scores.

[0087] With amino precursor For example, its rating Represented as:

[0088] in, Indicates the reactivity score. Indicates the coplanarity score. Indicates the structural stiffness score. Indicates the availability score of the experiment. Indicates the steric hindrance penalty term. Carbonyl precursor score. Connection method rating Calculate using the same method.

[0089] Furthermore, a COF-specific cooperative matrix was constructed. Amine precursors. With carbonyl precursor Collaborative scoring The bond formation feasibility, functional group matching degree, reaction energy barrier, post-bonding configuration planarity, steric hindrance, and target structure stability are jointly determined, and expressed as:

[0090] in, Indicates the feasibility of bonding. This indicates the degree of matching between the number and position of functional groups. Indicates the reaction barrier score. Indicates coplanarity after bonding. Indicates the structural stability obtained from DFT / MD. This indicates the risk of steric hindrance or side reactions. The synergistic matrix of the amino precursor and its linkage mode. And the synergistic matrix of carbonyl precursors and linkage modes The calculations are based on connection angle matching, topology matching, configuration stability, and process feasibility.

[0091] Experimental design rules are generated based on the material design objectives and experimental conditions. The experimental design rules should include at least one or more of the following: a) Selection constraints: Select one each of amino precursors, carbonyl precursors, and connection methods, or select a specified number according to project requirements; b) Prohibited pairing constraints: If certain amino precursors cannot coexist with carbonyl precursors, amino precursors with connection methods, or carbonyl precursors with connection methods, they are added to the prohibited pairing set; c) Preference constraints: Set positive scores for preferred components, connection methods, or process levels; d) Synergy constraints: Set reward items for component pairings with synergistic gains and penalty items for conflicting component pairings; e) Experimental resource constraints: Limit the number of candidate experiments output in one iteration. f. Diversity constraint: Limit candidate experiments from being too similar to avoid top-ranking results. The combinations are concentrated on the same type of precursor or the same type of connection method; g. Legality constraints: penalties are imposed on combinations that clearly do not meet the requirements of chemical reaction, supply, process safety or equipment conditions.

[0092] The forbidden pairing set is represented as: This represents the forbidden pairing matrix between amino and carbonyl precursors; A forbidden pairing matrix representing the amino precursor and the connection mode; This represents the forbidden pairing matrix indicating the carbonyl precursor and its connection method.

[0093] like Then it means the first The first amino precursor and the second Simultaneous selection of carbonyl precursors is prohibited; if... If the pairing is not prohibited, it means that the pairing is not prohibited. and The meaning is similar.

[0094] The forbidden pairing matrix is ​​not arbitrarily set, but generated according to the specific chemical rules of COFs. For amino precursors... With carbonyl precursor If any of the following conditions are met, then set 1) The number or spatial location of functional groups cannot form the target connectivity relationship; 2) The target two-dimensional configuration cannot be formed after DFT geometric optimization; 3) The reaction energy barrier is higher than the preset threshold. ;4) Planarity deviation after bonding exceeds the preset threshold 5) Spatial steric hindrance score is higher than preset threshold 6) Structural fluctuations exceed the threshold or hydrogen bond retention rate is below the threshold in MD simulation; 7) There are obvious safety, cost, supply, or process infeasibility issues under experimental conditions.

[0095] and Generate them in the same way based on connection angle, topology matching, configuration stability, and process feasibility.

[0096] Define a binary variable: , indicating whether to select the first One amino precursor; , indicating whether to select the first One carbonyl precursor; , indicating whether to select the first There are several connection methods.

[0097] Combining all amino group variables into a vector Organize all carbonyl variables into a vector Organize all connection method variables into a vector .

[0098] Figure 4 A schematic diagram of QUBO modeling provided in the embodiments of this application, as shown below. Figure 4 As shown, the modeling process includes linear scoring terms, collaborative quadratic terms, one-hot constraints, prohibited pairing penalty terms, diversity / subset constraints, etc.

[0099] The selection constraints are as follows: .

[0100] For screening a single candidate experimental combination, the following QUBO energy function is constructed:

[0101] in, Represent the energy function to be minimized; These represent monomer score vectors for amino precursors, carbonyl precursors, and linkage modes, respectively. These represent the synergistic scoring matrices between the components; This represents the penalty coefficient for one-hot selection constraints; This represents the penalty coefficient for prohibiting pairings. The penalty coefficient should be set to a sufficiently large positive number so that combinations that do not meet the constraints are energy inferior to feasible combinations.

[0102] In the above model, a negative sign is used before the scoring item because SPQC or QUBO solutions usually aim to minimize energy; combinations with higher scores have lower energy and are more likely to be selected as high-quality candidates.

[0103] For a choice For each experimental combination scenario, first enumerate or generate a set of valid candidate combinations. And define binary variables Indicate whether to select the first There are 10 candidate combinations. Each candidate combination... With comprehensive score Any two candidate combinations and They have similarity or redundancy. Then construct a candidate subset selection QUBO:

[0104] in, Indicates the selection variable for candidate subsets; Indicates the diversity penalty coefficient; This represents the penalty coefficient for the constraint on the number of candidates; Indicates the first The degree to which experimental or resource rules are not met. This model allows selection from the candidate pool. A candidate experiment with high score, low redundancy, and that meets experimental constraints.

[0105] To ensure that infeasible combinations are not selected due to high scores, the penalty coefficient is set as follows: one-hot constraint penalty coefficient. , , and the no-pairing penalty coefficient , , All of these are greater than the maximum energy gain that could be generated by all reward items. For example, let: ;in, This represents the sum of the absolute values ​​of the corresponding score vector or matrix elements. With this setting, any combination that does not satisfy the one-hot selection constraint or the prohibited pairing constraint, even if it has a high individual or collaborative score, will not be the lowest energy solution.

[0106] Figure 5 This is a schematic diagram of SPQC solving provided in an embodiment of this application, as shown below. Figure 5 As shown, the solution process includes steps such as parameter input, SPQC calculation module, low-energy candidate binary string output, and candidate experimental combination analysis.

[0107] The QUBO model is represented as follows: ;in, Represents the coefficient matrix of the quadratic term. Represents the vector of coefficients of linear terms. Represents a constant term. Represents a binary variable vector.

[0108] When it is necessary to use the Ising form for solving, the binary variables... Convert to spin variable :

[0109] in, The transformed value is the Ising Hamiltonian:

[0110] in, This represents the coupling coefficient between spins. This represents the external field coefficient. Input SPQC, which searches for low-energy states through coupling between optical pulses. After solving, the spin state is output. Convert back to binary variable This yields candidate combinations or candidate experimental subsets.

[0111] Furthermore, SPQC outputs multiple low-energy candidate samples, performs feasibility checks, energy sorting, deduplication, and rule filtering on the candidate samples, and obtains the final recommended experiment list.

[0112] The candidate experimental combinations are analyzed based on the SPQC output. For example, when When, it indicates that the recommended execution is by the first The first amino group precursor, the first The first carbonyl precursor and the second Candidate experiments are constructed using various connection methods. The output information for candidate experiments includes, but is not limited to: candidate experiment number, amino precursor, carbonyl precursor, connection method or connection topology, reaction type, process level, prediction score, triggered synergists, constraint check results, and recommended priority.

[0113] Synthesized and characterized according to the recommended list, the product structure, formation rate, mechanical properties, stability, flexibility, film quality, crystallinity, and defect characteristics were obtained. The experimental results were written back to the candidate library, the component scores, synergy scores, forbidden pairing sets, and constraint weights were updated, and the product was selected for the next round of candidate screening.

[0114] Figure 6 A schematic diagram of the experimental execution and feedback closed loop provided in the embodiments of this application, as shown below. Figure 6 As shown, the closed-loop process includes: candidate pool construction, descriptor computation, QUBO modeling, SPQC solving, candidate experiment output, synthesis and characterization, rule updating, and the next round of screening.

[0115] Taking the screening of a single COF candidate combination as an example. Assume the candidate pool contains... Amine precursor, carbonyl precursors and Each connection method was analyzed. Scores for each amino precursor were calculated separately. Scoring of each carbonyl precursor Scoring of each connection method And calculate three types of collaborative scores. .

[0116] Create binary variables And add the following constraints: If the first The first amino precursor and the second If a carbonyl precursor is not suitable for use together, then And add a penalty item to QUBO. Perform the same process for all prohibited pairings.

[0117] The scoring items, synergistic items, and constraint items are combined into a QUBO matrix. The QUBO string is then converted into an Ising parameter input to the SPQC. After the SPQC outputs multiple candidate binary strings, candidates that satisfy the one-hot and prohibited pairing constraints are selected and sorted by energy from low to high. The lowest energy feasible binary string corresponds to the preferred experimental combination.

[0118] Compared to the general approach that simply standardizes the material selection problem into a QUBO and inputs it into SPQC for solving, the improvement of this application lies not in the conventional conversion process from QUBO to Ising, but in explicitly mapping the physicochemical factors obtained from DFT / MD, such as reaction barriers, configurational stability, coplanarity, interlayer interactions, hydrogen bond retention, steric hindrance, and process feasibility, into scoring vectors, synergistic matrices, forbidden pairing matrices, and penalty coefficients, targeting the COF material formation mechanism. The resulting QUBO model can eliminate chemically infeasible combinations and prioritize experimental candidates that meet the requirements for COF synthesis and structural stability.

[0119] Compared with existing technologies, the embodiments of this application have at least the following beneficial effects: The material combination screening problem is transformed into a QUBO model, enabling precursor selection, connection method selection, process grade selection, synergistic relationships, and prohibited pairing constraints to be expressed within a unified mathematical framework. Using DFT / MD results for front-end descriptor and scoring construction, instead of performing costly and detailed calculations on all candidates, reduces the proportion of low-potential candidates entering subsequent calculations or experimental stages. Searching for low-energy solutions in the combination space using SPQC allows for faster acquisition of high-quality candidate combinations that satisfy constraints within a larger candidate space. Representing the pairing relationships between components using quadratic synergistic terms avoids local optima or unreasonable combinations caused by independent sorting based solely on individual component scores. Prohibited pairing penalty terms and one-hot constraints ensure the legality of candidate combinations, reducing obviously infeasible experiments. A candidate subset selection model and diversity penalty terms can output a more representative and exploratory set of candidate experiments when experimental resources are limited. Updating the candidate library, scoring parameters, and constraint rules through experimental feedback forms a continuously iterative material development closed loop, improving the quality of subsequent screening.

[0120] According to a second aspect of the embodiments of this application, a material combination screening device is provided. Figure 7 This is a schematic diagram of the material combination screening device provided in the embodiments of this application, as shown below. Figure 7 As shown, the material combination screening device includes: a candidate pool construction module 702, a descriptor extraction module 704, a score generation module 706, a constraint generation module 708, a QUBO modeling module 710, and a solution and analysis module 712. The material combination screening device will be described in detail below.

[0121] The candidate pool construction module 702 is used to construct a material candidate pool that includes at least an amino precursor candidate set, a carbonyl precursor candidate set, and a connection mode candidate set.

[0122] The descriptor extraction module 704 is connected to the candidate pool construction module 702 and is used to extract physicochemical descriptors based on density functional theory calculations and / or molecular dynamics calculations from the components and component pairings in the material candidate pool.

[0123] The rating generation module 706 is connected to the descriptor extraction module 704 mentioned above, and is used to generate monomer scores for each component and synergistic scores between components based on physicochemical descriptors.

[0124] The constraint generation module 708, connected to the scoring generation module 706, is used to generate a set of rules that includes at least selection constraints and prohibition pairing constraints; the selection constraints are to select a specified number of candidate components in each candidate set; the prohibition pairing constraints are generated based on the comparison of physicochemical descriptors with preset thresholds.

[0125] The QUBO modeling module 710, connected to the constraint generation module 708, is used to define binary decision variables and construct a QUBO energy function that includes single-item scoring terms, pairing synergy terms, selection constraint penalty terms, and prohibition pairing penalty terms.

[0126] The solution and analysis module 712, connected to the QUBO modeling module 710, is used to input the QUBO energy function into a dedicated quantum computer for solution, and decode the output solution into binary decision variables to obtain candidate experimental combinations that satisfy the selection constraints and prohibited pairing constraints.

[0127] It should be noted that the candidate pool construction module 702, descriptor extraction module 704, score generation module 706, constraint generation module 708, QUBO modeling module 710 and solution and analysis module 712 mentioned above correspond to steps S102 to S112 in the method embodiment. The examples and application scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the content disclosed in the above method embodiment.

[0128] According to a third aspect of the embodiments of this application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described above.

[0129] refer to Figure 8 The present invention describes a structural block diagram of an electronic device that can serve as a server or client in embodiments of this application, which is an example of a hardware device that can be applied to various aspects of this application. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the application described and / or claimed herein.

[0130] like Figure 8 As shown, the electronic device includes a computing unit 801, which can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. The RAM 803 may also store various programs and data required for the operation of the electronic device. The computing unit 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0131] Multiple components in the electronic device are connected to I / O interface 805, including: input unit 806, output unit 807, storage unit 808, and communication unit 809. Input unit 806 can be any type of device capable of inputting information into the electronic device. Input unit 806 can receive input digital or character information and generate key signal inputs related to user settings and / or function control of the electronic device. Output unit 807 can be any type of device capable of presenting information and may include, but is not limited to, a display, speaker, video / audio output terminal, vibrator, and / or printer. Storage unit 808 may include, but is not limited to, disks and optical discs. Communication unit 809 allows the electronic device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, and wireless communication transceivers, such as Bluetooth devices, WiFi devices, WiMax devices, cellular communication devices, and / or the like.

[0132] The computing unit 801 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, CPUs, graphics processing units (GPUs), various special-purpose artificial intelligence (AI) computing units, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any suitable processors, controllers, microcontrollers, etc. The computing unit 801 performs the various methods and processes described above. For example, in some embodiments, the method embodiments of this application can be implemented as computer programs tangibly contained in a machine-readable medium, such as storage unit 808. In some embodiments, part or all of the computer program can be loaded and / or installed on an electronic device via ROM 802 and / or communication unit 809. In some embodiments, the computing unit 801 can be configured to perform the methods described above by any other suitable means (e.g., by means of firmware).

[0133] According to a fourth aspect of the embodiments of this application, a computer-readable storage medium is provided. The computer-readable storage medium may be an internal storage unit of the aforementioned electronic device, such as a hard disk or memory of the electronic device. The computer-readable storage medium may also be an external storage device of the electronic device, such as a plug-in hard disk, smart memory card, secure digital card, flash memory card, etc., equipped on the electronic device. The computer-readable storage medium may also include both internal storage units and external storage devices of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0134] The computer-readable storage medium stores a computer program that, when executed by a processor, implements the material combination screening method described in any of the above method embodiments. When the processor executes the computer program, it constructs a material candidate pool containing at least a candidate set of amine precursors, a candidate set of carbonyl precursors, and a candidate set of connection modes. It extracts physicochemical descriptors from the components and component pairings in the candidate pool, obtained based on density functional theory calculations and / or molecular dynamics calculations. Based on these descriptors, it generates monomer scores for each component and synergistic scores between components, generates a rule set including at least selection constraints and prohibited pairing constraints, defines binary decision variables, and constructs a QUBO energy function containing individual score terms, pairing synergistic terms, selection constraint penalty terms, and prohibited pairing penalty terms. It inputs the QUBO energy function into a dedicated quantum computer for solving, and decodes the output solution to obtain candidate experimental combinations that satisfy the selection constraints and prohibited pairing constraints. By running this computer program on the processor, the screening process in the aforementioned method embodiments can be automatically executed on a computing device, completing all steps from candidate pool construction to candidate experimental combination output without manual intervention.

[0135] According to a fifth aspect of the embodiments of this application, a computer program product is provided, including a computer program that, when executed by a processor, implements the material combination screening method described in any of the above method embodiments. The computer program product may employ any combination of one or more readable media, which may be a readable signal medium or a readable storage medium. A readable storage medium may, for example, include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. A readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including, but not limited to, electromagnetic signals, optical signals, or any suitable combination thereof.

[0136] It should be noted that the term "comprising" and its variations used in the embodiments of this application are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; and the term "some embodiments" means "at least some embodiments". The modifications of "one" and "multiple" mentioned in the embodiments of this application are illustrative and not restrictive. Those skilled in the art should understand that, unless explicitly indicated otherwise in the context, they should be understood as "one or more".

[0137] The steps described in the method embodiments provided in this application can be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of protection of this application is not limited in this respect.

[0138] The term "embodiment" in this specification refers to a specific feature, structure, or characteristic described in connection with an embodiment that may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily imply the same embodiment, nor does it imply independence from or alternative to other embodiments. The various embodiments in this specification are described in a related manner, with reference to each other for similar or identical parts. In particular, for apparatus, device, and system embodiments, since they are substantially similar to method embodiments, the description is relatively simple, and relevant details are referred to in the description of the method embodiments.

[0139] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of patent protection. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the appended claims.

Claims

1. A method for screening material combinations, characterized in that, include: Construct a material candidate pool that includes at least a candidate set of amine precursors, a candidate set of carbonyl precursors, and a candidate set of connection modes; For the components and component pairs in the material candidate pool, extract physicochemical descriptors obtained from density functional theory calculations and / or molecular dynamics calculations; Based on the physicochemical descriptor, monomer scores for each component and synergistic scores between components are generated; Generate a rule set that includes at least selection constraints and prohibition pairing constraints; the selection constraints are to select a specified number of candidate components in each candidate set; the prohibition pairing constraints are generated based on a comparison of the physicochemical descriptor with a preset threshold. Define binary decision variables and construct a QUBO energy function that includes a single-item scoring term, a pairing synergy term, a selection constraint penalty term, and a prohibition pairing penalty term; The QUBO energy function is input into a dedicated quantum computer for solution, and the output solution is decoded into binary decision variables to obtain candidate experimental combinations that satisfy the selection constraints and prohibited pairing constraints.

2. The method according to claim 1, characterized in that, The material candidate pool also includes at least one of the process discrete level candidate set and the configuration or binding site candidate set; the process discrete level candidate set includes at least one of temperature level, solvent system, time level, catalyst type or concentration level.

3. The method according to claim 1, characterized in that, The physicochemical descriptors include component-level descriptors and paired co-operating descriptors; wherein: The component-level descriptor includes at least one of the following: precursor structural rigidity, number and spatial distribution of functional groups, free rotation capability, coplanarity tendency, reactivity, monomer size and symmetry, and thermodynamic tendency to form the target two-dimensional structure; The pairing co-operation descriptor includes at least one of the following: bonding feasibility, functional group matching degree, reaction pathway energy barrier difference, post-bonding configuration planarity, steric hindrance, side reaction risk, interlayer interaction, and hydrogen bond retention rate.

4. The method according to claim 1, characterized in that, Based on the physicochemical descriptor, monomer scores for each component and synergistic scores between components are generated, including: The physicochemical descriptors are normalized and weighted summed to obtain a comprehensive score; The synergistic score includes the synergistic score between the amino precursor and the carbonyl precursor, the synergistic score between the amino precursor and the linkage mode, and the synergistic score between the carbonyl precursor and the linkage mode.

5. The method according to claim 1, characterized in that, The rules for generating the prohibited pairing constraints include: If the reaction energy barrier for component pairing is higher than the first preset threshold, the planarity deviation after bonding is higher than the second preset threshold, or the structural fluctuation in molecular dynamics simulation exceeds the third preset threshold, then the pairing of that component is set to be prohibited.

6. The method according to claim 1, characterized in that, The formula for constructing the QUBO energy function is as follows: in, , , These are binary decision variable vectors representing the amino precursor, carbonyl precursor, and linkage mode, respectively. , , This is a single-unit score vector; , , For collaborative scoring matrix; , , To select the penalty coefficient for the constraint; , , These are the forbidden pairing matrices for amino-carbonyl, amino-linked, and carbonyl-linked groups, respectively. , , This represents the corresponding prohibited pairing penalty coefficient.

7. The method according to claim 1, characterized in that, The penalty coefficient is set to satisfy the following: , , , , and All are greater than the sum of the absolute values ​​of the individual scoring items and the paired collaboration items.

8. The method according to claim 1, characterized in that, After the QUBO energy function is input into a dedicated quantum computer for solving, the following steps are also included: Receive multiple low-energy candidate binary strings output by the dedicated quantum computer; The low-energy candidate binary strings are subjected to feasibility checks, deduplication, and energy sorting to obtain a list of candidate experimental combinations.

9. The method according to claim 1, characterized in that, The method further includes: The candidate experimental combinations are output for synthesis and characterization to obtain experimental feedback results; Based on the experimental feedback results, update the material candidate pool, the monomer score, the synergistic score, and / or the rule set; Based on the updated information, return to the steps of constructing a candidate pool of materials or extracting physicochemical descriptors for the next round of screening.

10. A material combination screening device, characterized in that, include: The candidate pool construction module is used to construct a material candidate pool that includes at least an amino precursor candidate set, a carbonyl precursor candidate set, and a connection mode candidate set; The descriptor extraction module is used to extract physicochemical descriptors based on density functional theory calculations and / or molecular dynamics calculations from the components and component pairs in the material candidate pool. The scoring generation module is used to generate monomer scores for each component and synergistic scores between components based on the physicochemical descriptor. A constraint generation module is used to generate a rule set that includes at least selection constraints and prohibition pairing constraints; the selection constraints are to select a specified number of candidate components in each candidate set; the prohibition pairing constraints are generated based on a comparison between the physicochemical descriptor and a preset threshold. The QUBO modeling module is used to define binary decision variables and construct a QUBO energy function that includes single-item scoring terms, paired synergistic terms, selection constraint penalty terms, and prohibited pairing penalty terms. The solution and analysis module is used to input the QUBO energy function into a dedicated quantum computer for solution, and decode the output solution into binary decision variables to obtain candidate experimental combinations that satisfy the selection constraints and prohibited pairing constraints.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 9.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1 to 9.

Images