Systems, apparatuses, and methods for quantum artificial intelligence

Abstract

A classical computing entity of a hybrid quantum-classical computing system may generate one or more quantum circuits at least in part through operation of program code characterized by an artificial intelligence (AI) and / or machine learning architecture; provide the one or more quantum circuits (e.g., for receipt by a quantum computer); and receive measurement results generated by performance of at least one quantum circuit of the one or more quantum circuits by the quantum computer. The classical computing entity may utilize the measurement results and the program code characterized by an artificial intelligence (AI) and / or machine learning architecture to (a) update a model, (b) evaluate a loss function, (c) generate one or more updated quantum circuits for performance by the quantum computer, and / or (d) determine and / or provide a solution to a problem.

Description

SYSTEMS, APPARATUSES, AND METHODS FOR QUANTUM ARTIFICIAL INTELLIGENCECROSS-REFERENCE TO RELATED APPLICATION(S)

[0001] This application claims priority to U.S. Application No. 63 / 755,013, filed February 6, 2025, the content of which is incorporated by reference in its entirety.TECHNICAL FIELD

[0002] The present disclosure relates to the use of quantum computation in artificial intelligence (Al) and / or machine learning (ML) applications. For example, some embodiments relate to the use of machine learning models in generating quantum information and / or the use of quantum information in the training of machine learning models.BACKGROUND

[0003] Al and ML applications analyze large amounts of data to identify patterns in the data that may then be applied for various tasks such as visual perception, speech recognition, decision-making, translating between languages, and / or the like. Quantum computation is a method of performing computations that exploits quantum mechanical phenomena, such as superposition of states.

[0004] Through applied effort, ingenuity, and innovation many deficiencies of integrating quantum computations and AI / ML have been solved by developing solutions that are structured in accordance with the embodiments of the present invention, many examples of which are described in detail herein.BRIEF SUMMARY OF EXAMPLE EMBODIMENTS

[0005] Various embodiments provide methods, systems, apparatus, computer program products, and / or the like for integrating quantum computations (e.g., computations performed using quantum hardware) with AI / ML applications. Some example applications arise in the fields of chemistry, optimization, biology, finance, and various other fields.

[0006] According to a first aspect of the present disclosure, a method is provided. In an example embodiment, the method is performed by a controller of a quantum computer or -1- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vla classical computing entity. For example, the method may be performed via one or more classical processing elements of a hybrid quantum -cl as si cal AI / ML system. In an example embodiment, the method comprises causing a circuit large language model (LLM) to generate one or more quantum circuits based at least in part on request information; causing performance of the one or more quantum circuits on a quantum computer; receiving quantum information generated via performance of the one or more quantum circuits on the quantum computer; determining whether one or more stop criteria are satisfied by the quantum information; and responsive to determining that the one or more stop criteria are satisfied, providing an output corresponding to the request information. Responsive to determining that the one or more stop criteria are not satisfied, the method further includes updating at least one of the circuit LLM or a solution parameter of the one or more quantum circuits, causing one or more updated quantum circuits to be performed by the quantum computer, wherein the one or more updated quantum circuits are at least one of generated by the circuit LLM based at least in part on the solution parameter, and receiving the quantum information generated via performance of the one or more updated quantum circuits by the quantum computer.

[0007] In an example embodiment, the method further includes determining the output based at least in part on the quantum information generated via performance of at least one quantum circuit by the quantum computer.

[0008] In an example embodiment, determining the output comprises causing a semantic interpreter of a blueprint engine to generate the output based on the quantum information generated via performance of the at least one quantum circuit by the quantum computer wherein the blueprint engine acts as an interface between a user interface and the circuit LLM.

[0009] In an example embodiment, a blueprint engine is configured to process the request information responsive to receipt of a request comprising the request information to generate an input vector that is provided as input to the circuit LLM.

[0010] In an example embodiment, the request information comprises a data set, the blueprint engine extracts a sub-graph from the data set, generates the input vector representing the sub-graph, and provides the input vector representing the sub-graph to the circuit LLM.

[0011] In an example embodiment, the circuit LLM generates a quantum topological data analysis (QTDA) circuit configured to perform QTDA of the sub-graph.-2- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vl

[0012] In an example embodiment, the blueprint engine comprises a semantic interpreter configured to classify whether an anomaly is detected in the set based at least in part on the quantum information generated via performance of the QTDA circuit by the quantum computer.

[0013] In an example embodiment, the quantum information generated via performance of the QTDA circuit by the quantum computer is an approximate Betti-k number for the sub -graph.

[0014] In an example embodiment, the blueprint engine is configured to execute a classifier configured to determine whether to perform QTDA on additional sub-graphs from the data set and it is determined that the one or more stop criteria are satisfied responsive to the classifier determining that performance of QTDA on additional subgraphs is not required and it is determined that the one or more stop criteria are not satisfied responsive to the classifier determining that performance of QTDA on additional sub-graphs is required.

[0015] In an example embodiment, the request information provides a Hamiltonian description of a system and the output is at least one of an Eigen solution to the Hamiltonian description of the system or an Eigen value corresponding to the Eigen solution to the Hamiltonian description of the system.

[0016] In an example embodiment, a blueprint engine is configured to generate an ansatz based at least in part on the Hamiltonian description of the system and to cause the circuit LLM to generate the one or more quantum circuits based at least in part on the ansatz.

[0017] In an example embodiment, the at least one of the circuit LLM or the solution parameter is updated in accordance with a quantum eigensolver protocol.

[0018] In an example embodiment, it is determined that the one or more stop criteria are satisfied responsive to determining that convergence of the Eigen value corresponding to the Eigen solution to the Hamiltonian description of the system satisfies convergence criteria and it is determined that the one or more stop criteria are not satisfied responsive to determining that the convergence of the Eigen value corresponding to the Eigen solution to the Hamiltonian description of the system does not satisfy convergence criteria.

[0019] In an example embodiment, the Eigen solution to the Hamiltonian description of the system is a ground state or lowest cost state of the system and the Eigen value-3- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlcorresponding to the Eigen solution to the Hamiltonian description of the system is a ground state energy of the ground state or a cost corresponding to the lowest cost state.

[0020] In an example embodiment, the method further includes receiving a request comprising the request information, wherein the request is received responsive to user interaction with a user interface.

[0021] In an example embodiment, the method further includes, prior to causing the circuit LLM to generate the one or more quantum circuits based at least in part on the request information, executing a resource estimation oracle configured to analyze the request information to determine whether to analyze the request using classical or hybrid quantum-classical techniques, and the circuit LLM is caused to generate the one or more quantum circuits responsive to the resource estimation oracle determining to analyze the request using hybrid quantum-classical techniques.

[0022] In an example embodiment, the resource estimation oracle comprises a machine learning trained model configured to estimate the quantum complexity of a request based on the request information thereof.

[0023] According to another aspect, a classical computing entity is provided. In an example embodiment, the classical computing entity includes one or more classical processing elements; a communication interface configured for communicating with a quantum computer; and a memory storing computer executable instructions. The computer executable instructions are configured to, when executed by the one or more classical processing elements, cause the classical computing entity to at least perform causing a circuit large language model (LLM) to generate one or more quantum circuits based at least in part on request information; causing performance of the one or more quantum circuits on a quantum computer; receiving quantum information generated via performance of the one or more quantum circuits on the quantum computer; determining whether one or more stop criteria are satisfied by the quantum information generated via performance of the one or more quantum circuits on the quantum computer; responsive to determining that the one or more stop criteria are satisfied, providing an output corresponding to the request information; and responsive to determining that the one or more stop criteria are not satisfied, updating at least one of the circuit LLM or a solution parameter, causing one or more updated quantum circuits to be performed by the quantum computer, wherein the one or more updated quantum circuits are at least one of generated by the circuit LLM based at least in part on the solution parameter, and receiving-4- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlquantum information generated via performance of the one or more updated quantum circuits on the quantum computer.

[0024] In an example embodiment, the classical computing entity further comprises a user interface configured for receiving user input and for providing the output in a human perceivable manner.

[0025] According to another aspect a system, such as a hybrid quantum-classical AI / ML system is provided. In an example embodiment, the system includes a classical computing entity and a quantum computer. The classical computing entity comprises one or more classical processing elements, a communication interface configured for communicating with a quantum computer, and a memory storing classical computer executable instructions. The quantum computer includes a plurality of qubits, qubit manipulation elements, sensors, and a controller configured to control operation of the qubit manipulation elements and receive respective sensor signals generated by the sensors. The classical computer executable instructions are configured to, when executed by the one or more classical processing elements, cause the classical computing entity to at least perform causing a circuit large language model (LLM) to generate one or more quantum circuits based at least in part on request information; causing performance of the one or more quantum circuits on the quantum computer; receiving quantum information generated via performance of the one or more quantum circuits on the quantum computer; determining whether one or more stop criteria are satisfied by the quantum information generated via performance of the one or more quantum circuits on the quantum computer; responsive to determining that the one or more stop criteria are satisfied, providing an output corresponding to the request information; and responsive to determining that the one or more stop criteria are not satisfied updating at least one of the circuit LLM or a solution parameter of the one or more quantum circuits, causing one or more updated quantum circuits to be performed by the quantum computer, wherein the one or more updated quantum circuits are at least one of generated by the circuit LLM based at least in part on the solution parameter, and receiving quantum information generated via performance of the one or more quantum circuits on the quantum computer. The controller of the quantum computer is configured to perform receiving the one or more quantum circuits, executing the one or more quantum circuits to generate the quantum information, wherein executing a quantum circuit of the one or more quantum circuits comprises controlling operation of the qubit manipulation elements to cause a controlled -5- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlquantum state evolution of one or more qubits of the plurality of physical qubits in accordance with the quantum circuit, analyzing the respective sensor signals to determine resultant quantum states of the one or more qubits, and determining the quantum information based on the resultant quantum states of the one or more qubits, and providing the quantum information configured for receipt by the classical computing entity.BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWING(S)

[0026] Having thus described the invention in general terms, reference will now be made to the accompanying drawings, which are not necessarily drawn to scale, and wherein:

[0027] Figure 1 is a schematic diagram illustrating an example hybrid quantum-classical computing system, according to an example embodiment.

[0028] Figure 2 schematically illustrates a dataflow diagram, in accordance with an example embodiment.

[0029] Figure 3 provides a flowchart illustrating various processes and / or procedures performed by a classical component of a hybrid quantum-classical computing system of Figure 7, for example, to cause a hybrid quantum-classical AI / ML system to provide an output in response to a request.

[0030] Figure 4 provides a flowchart illustrating various processes and / or procedures performed by a classical component of a hybrid quantum-classical computing system of Figure 7, for example, to perform anomaly detection, in accordance with certain embodiments.

[0031] Figure 5 provides a flowchart illustrating various processes and / or procedures performed by a classical component of a hybrid quantum-classical computing system of Figure 7, for example, to perform determination of a minimum energy / cost solution, in accordance with certain embodiments.

[0032] Figure 6 provides a flowchart illustrating various processes and / or procedures performed by a quantum component of a hybrid-classical computing system of Figure 6, for example, in accordance with certain embodiments.

[0033] Figure 7 provides a schematic diagram of an example controller of a quantum component of a hybrid-classical computing system that is configured to control operation of one or more elements of the quantum component, according to various embodiments.-6- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vl

[0034] Figure 8 provides a schematic diagram of an example classical component of a hybrid quantum-classical computing system that may be used in accordance with an example embodiment.DETAILED DESCRIPTION OF SOME EXAMPLE EMBODIMENTS

[0035] The present invention now will be described more fully hereinafter with reference to the accompanying drawings, in which some, but not all embodiments of the invention are shown. Indeed, the invention may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will satisfy applicable legal requirements. The term “or” (also denoted “ / ”) is used herein in both the alternative and conjunctive sense, unless otherwise indicated. The terms “illustrative” and “exemplary” are used to be examples with no indication of quality level. The terms “generally,” “substantially,” and “approximately” refer to within engineering and / or manufacturing tolerances and / or within user measurement capabilities, unless otherwise indicated. Like numbers refer to like elements throughout.

[0036] Various embodiments provide methods, systems, apparatus, computer program products, and / or the like for integrating quantum computations (e.g., computations performed using quantum hardware) with AI / ML applications. Some example applications arise in the fields of chemistry, optimization, biology, finance, and various other fields.

[0037] In various embodiments, an AI / ML application, program, and / or the like is executed by a classical computing entity (a classical computer, server, Cloud computing resource(s), and / or the like). The classical computing entity may use various AI / ML architectures and / or models to determine and / or generate a quantum circuit based on the problem posed to the AI / ML application, program, and / or the like. The quantum circuit may be communicated to a quantum computer which performs the quantum circuit (e.g., to prepare a quantum state of a plurality of qubits of the quantum computer) and generates measurement results. The measurement results are returned to the classical computing entity. The classical computing entity may use the measurement results to determine a solution to the problem posed to the AI / ML application, program, and / or the like. For example, the classical computing entity may generate a new quantum circuit based at least in part on the returned measurement results; evaluate a loss function based -7- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlat least in part on the returned measurement results; determine an output of the AI / ML application program, and / or the like; and / or to perform another function of the AI / ML application program, and / or the like.

[0038] In various embodiments, a hybrid quantum-classical AI / ML system may be configured to perform anomaly identification. For example, an anomaly detection engine including trained machine learning model and operating on a classical component of a hybrid quantum-classical AI / ML system may be configured to process information (e.g., a data set and / or the like) and extract a sub-graph from the information. The sub-graph may then be analyzed using quantum topological data analysis (QTDA) via operation of a quantum processing unit (QPU) (e.g., a quantum component of the hybrid quantum-classical AI / ML system). Results of the QTDA may be received by the anomaly detection engine and a new sub-graph may be extracted from the information. This process may continue to be iterated until stop criteria are satisfied, at which time an output is provided. For example, such a technique may be used to identify anomalies in the information, perform network analysis, and / or to detect fraud.

[0039] In another example, a generative quantum Eigensolver (GQE) configured to determine a ground state and corresponding ground state energy and / or a minimum cost solution and a corresponding minimum cost is provided. An initial batch of quantum circuits may be generated and performed (e.g., via a QPU of a hybrid quantum-classical AI / ML system) based on an ansatz, for example. The initial batch of quantum circuits and the results of performing the respective quantum circuits of the initial batch of quantum circuits are used to train a circuit generation model via the classical component of the hybrid quantum-classical AI / ML system. The circuit generation model generates a new batch of quantum circuits which are then performed by a quantum computer. The results of performing the respective quantum circuits of the new batch of quantum circuits are used to further train the circuit generation model. Additional batches of quantum circuits are generated and performed with the results used to further refine the circuit generation model until stop criteria are met. Various GQE systems may be used to identify minimum energy / cost solutions in various fields such as chemistry, materials science, physical systems, biomedical fields, logistics, combinatorics, and / or the like.

[0040] Various embodiments provide technical solutions by using quantum information in the training of AI / ML models and via the integration of AI / ML techniques and quantum systems such that AI / ML provide efficient use of quantum resources in-8- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vldetermining outputs. For example, variation quantum eigensolvers (VQEs) are helpful in identifying ground states of atoms or molecules. However, as the size of the atom and / or molecule increases and / or as the size of the active subset of orbitals increases (e.g., possibly due to more qubits being available for modeling orbitals) the overhead of VQEs can become quite large. Tools disclosed herein, such as a circuit LLM, a blueprint engine configured to generate an input vector for the circuit LLM, and / or a semantic interpreter of a blueprint engine are used to provide a scalable implementation of a GQE. Thus various embodiments provide technical improvements in various fields such as quantum computation, integration of AI / ML with quant computing, anomaly detection, minimum cost solutions to problems that may be represented by a Hamiltonian representation, and / or the like.Example Hybrid Quantum-Classical AI / ML System

[0041] Figure 1 provides a block diagram of an example hybrid quantum-classical computing system 100, in accordance with various embodiments. In various embodiments, the hybrid quantum-classical computing system 100 comprises a classical computing entity 110 and a quantum computer 130 (also referred to herein as a quantum processing unit (QPU)).

[0042] The quantum computer 130 comprises a controller 132, qubits 134, qubit manipulation elements 136, and sensors 138. The controller 132 is configured to control operation of the qubit manipulation elements 136 to cause desired manipulations (e.g., controlled quantum state evolution) of the qubits 134. The controller 132 is further configured to control operation of the sensors 138 that are configured to monitor, measure, and / or capture measurements corresponding to the operation of the qubit manipulation elements and detect observables indicating the respective quantum states of respective qubits 134.

[0043] For example, in various embodiments, the qubit manipulation elements 136 comprise voltage / current sources, laser sources, magnetic field sources (e.g., electromagnets and / or permanent magnets) and / or other hardware components configured for use in confining the qubits and / or manipulating the quantum state of the qubits.

[0044] In various embodiments, the sensors 138 comprise photodetectors, voltage / current sensors, temperature sensors, pressure sensors, and / or other sensors that may be used to-9- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vldetermine a quantum state of a qubit and / or monitor operation of one or more of the qubit manipulation elements 136.

[0045] In various embodiments, the qubits 134 are quantum systems that may be manipulated via the qubit manipulation elements 136 such that operation of the qubit manipulation elements 136 control evolution of the quantum states of the qubits 134. Some non-limiting examples of quantum systems that may be used as qubits in various embodiments include photons, electrons, atomic nuclei, neutral atoms, ions, Josephson junctions, quantum dots, topological anyons, and / or other quantum particles and / or systems

[0046] In various embodiments, the classical computing entity 110 is in communication with the controller 132 of the quantum computer 130 via one or more wired or wireless networks 120 and / or via direct wired and / or wireless communications. The classical computing entity 110 may be a desktop computer, laptop computer, server, datacenter, Al factory, blade, switch, smartphone, Cloud-based computing resource, and / or the like. In various embodiments, the classical computing entity 110 is configured to train and / or execute an AI / ML model, provide quantum circuits to the quantum computer 130 (e.g., the controller 132 thereof), and receive measurement results corresponding to performance of measurements of the quantum states of qubits as a result of performing quantum circuits. The communications between the classical computing entity 110 and the quantum computer 130 (e.g., the controller 1320 may be transmitted via the one or more wired or wireless networks 120 and / or as direct wired and / or wireless communications between the classical computing entity 110 and the quantum computer 130.

[0047] In various embodiments, a quantum circuit is a sequence of quantum gates and / or unitary transformations to be performed on the plurality of qubits 134 and measurement operations to be performed to determine respective quantum states of qubits as a result of performance of at least a portion of the sequence of quantum gates and / or unitary transformation. The quantum computer 130 is configured to receive quantum circuits provided by the classical computing entity 110 and provide measurement results generated by performing the quantum circuits for receipt by the classical computing entity 110 via the one or more wired or wireless networks 120 and / or via direct wired and / or wireless communications between the classical computing entity 110 and the quantum computer 130.-10- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vl

[0048] Figure 2 provides a schematic diagram of a hybrid quantum-classical AI / ML system 200. The classical computing entity 110 is illustrated as a distributed system such as a client and server system, for example. For example, a user may interact with a user interface 205 of the classical computing entity 110 which is in communication with a classical computing environment 220, such a high performance computing (HPC) computing environment, an Al factory, datacenter, server, cloud-based computing network / resource, and / or the like.

[0049] For example, a user may interact with a user interface 205 of the classical computing entity 110 such as keypad 818, display 816, a touchscreen, mouse, and / or the like. The user may provide user input that provides or selects various parameters corresponding to a request 232.

[0050] In various embodiments, a request includes request information. The request information defines the problem to be solved. For example, the request information may comprise a data set or an indication of a data set and an indication that it is desired to determine whether an anomaly is present in the data set with a given confidence level and / or in light of one or more solution parameters. In another example, the request information may comprise a Hamiltonian representation of a system to be simulated / modeled and any boundary conditions and / or constraints corresponding to the system to be simulated / modeled.

[0051] The request 232 is provided to the classical computing environment 220. For example, the request 232 may be provided to a blueprint engine 224 of the classical computing environment 220. In various embodiments, the blueprint engine 224 is configured to act as an intermediary between the user interface 205 and a circuit large language model (LLM) 222 configured and / or trained to generate quantum circuits responsive to receiving an input vector. For example, in various embodiments, the blueprint engine 224 is configured to process and / or analyze request information provided and / or indicated by the request 232 and generate an input vector based thereon. The blueprint engine 224 may then provide the input vector to the circuit LLM to cause the circuit LLM 222 to generate one or more quantum circuits 234. In various embodiments, the input vector is characterized by a standardized format. For example, the blueprint engine 224 is configured to generate an input vector of a standardized format based on request information corresponding to the request 232. As should be understood, the standardized format may vary between various embodiments.-11- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vl

[0052] In various embodiments, the classical computing environment 220 includes a compiler 226. In various embodiments, the compiler 226 is configured to receive a quantum circuit generated by the circuit LLM 222 and generate an implementation based thereon. For example, generating the implementation of the quantum circuit includes transforming and / or decomposing the quantum operations of the quantum circuit into a specific quantum computing architecture - a particular universal gate set, quantum error correction (QEC) architecture, and / or the like. In various embodiments, the universal gate set and / or QEC architecture corresponds to the quantum component (e.g., quantum computer 130) of the hybrid quantum-classical AI / ML system to be used to execute the implementations of the quantum circuit. In some embodiments, the compiler 226 is configured to generate intermediate-level instructions (e.g., a quantum intermediate representation (QIR) program) which will then be transformed into and / or used to generate machine level instructions by the controller 132 of the quantum computer 130. In some embodiments, the compiler 226 is configured to generate machine level instructions for the implementation of the quantum circuit that may be directly executed by the controller 132 of the quantum computer 130. In some embodiments, the compiler 226 is stored and / or executed by the controller 132 rather than by the classical computing entity 110.

[0053] In certain embodiments, when a request 232 is received, a resource estimation oracle 228 of the classical computing environment 220 analyzes the request to determine whether the request is better suited for classical techniques or hybrid quantum-classical techniques. For example, the resource estimation oracle 228 may comprise a machine learning trained model configured and / or trained to estimate the quantum complexity of a request based on the request information corresponding to the request 232. For example, the resource estimation oracle 228 may be configured and / or trained to identify problems where a hybrid quantum-classical approach provides an advantage in determining an output 238 corresponding to the request 232. For example, the resource estimation oracle 228 may be a classifier that classifies requests 232 as requests that should be responded to using classical techniques and requests that should be responded to using hybrid quantum-classical techniques. Responsive to the resource estimation oracle 228 determining that a particular request should be responded to using hybrid quantum-classical techniques, the blueprint engine 224 may generate an input vector and provide an input vector as input to the circuit LLM 222, which causes the circuit LLM 222 to -12- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlgenerate one or more quantum circuits. For example, the circuit LLM 222 may be caused to generate the one or more quantum circuits 234 responsive to the resource estimation oracle 228 determining to analyze the request 232 using hybrid quantum-classical techniques.

[0054] The quantum circuit(s) 234 are provided (e.g., transmitted) to the quantum computer 130. For example, the classical computing entity 110 may provide the quantum circuits 234 configured for receipt by the quantum computer 130 (e.g., a controller 132 of the quantum computer). The quantum computer 130 may queue the quantum circuit(s) 234 for execution responsive to receiving the quantum circuit(s) 234. The quantum computer executes and / or performs the quantum circuit(s) 234 to generate quantum information 236. For example, the quantum information 236 is information determined and / or generated through execution and / or performance of the quantum circuit(s) 234 by the quantum computer 130. For example, the quantum information 236 may comprise results of performing the quantum circuit(s) 234.

[0055] The quantum computer 130 provides (e.g., transmits) the quantum information 236 configured for receipt by the classical computing entity 110. The classical computing entity 110 receives the quantum information 236. The blueprint engine 224 may process the quantum information 236 to determine if stop criteria has been satisfied. If the stop criteria has been satisfied, an output is determined and provided. Providing the output may include storing the output in classical memory, providing the output via the user interface 205, providing the output as input to a computing program or application being executed by the classical computing entity 110 and / or another classical computer.Responsive to determining the stop criteria is not satisfied, the blueprint engine 224 may update the circuit LLM 222 (e.g., cause further training of the circuit LLM 222), update a component (e.g., machine learning model) of the blueprint engine 224, and / or update one or more solution parameter of the one or more quantum circuits.

[0056] In various embodiments, the one or more solution parameters may be parameters of an ansatz, parameters used to extract a sub-graph from a data set, parameters determined based on a loss function and / or a gradient of a loss function, and / or other parameters that affect the generation of a next one or more quantum circuits by the circuit LLM that are not weights of the circuit LLM.

[0057] The hybrid quantum-classical AI / ML system iteratively generates one or more quantum circuits 234, generates quantum information 236 via operation of the quantum -13- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlcomputer 130 based on the one or more quantum circuits 234, and updates the circuit LLM 222, component of the blueprint engine 224, and / or one or more solution parameters until the one or more stop criteria are satisfied. Responsive to determining that the stop criteria are satisfied, an output is provided. In some embodiments, the output is generated and / or determined responsive to the stop criteria being satisfied. For example, the quantum information 236 may be analyzed and / or processed by a semantic interpreter of the blueprint engine 224. In various embodiments, the semantic interpreter may be a classifier configured to determine and / or select a classification of the output based at least in part on the quantum information 236. In some embodiments, the semantic interpreter may be configured to determine respective expectation values for one or more observables based on the quantum information 236. For example, in some embodiments, the output is an eigen-solution (e.g., an eigenstate) of a Hamiltonian description of a system and / or an eigenvalue corresponding to the eigen-solution to the Hamiltonian description of the system. For example, eigen-solution to the Hamiltonian description of the system may be a ground state (e.g., when the system is an atomic or molecular system) or lowest cost state of the system (e.g., for various other types of systems such as systems adapted from fields such as logistics, biomedical, combinatorics, and / or the like. For example, the eigenvalue corresponding to the eigen-solution to the Hamiltonian description of the system may be a ground state energy of the ground state or a cost corresponding to the lowest cost state.Example Operation of Classical Computing Entity of a Hybrid Quantum-Classical AI / ML System

[0058] Figure 3 provides a flowchart of various processes and / or procedures that may be performed by a classical computing entity 110 of hybrid quantum-classical AI / ML system 200. Starting at block 302, the classical computing entity 110 obtains problem space information. For example, the classical computing entity 110 comprises means, such as classical processing elements 808, classical memory 822, 824, communications interface 825 (e.g., network interface 820, antenna 812 coupled to receiver 806 and / or transmitter 804), user interface 815 (e.g., keypad 818, display 816), and / or the like, for obtaining problem space information.

[0059] In various embodiments, the problem space information indicates a general framework that the resulting hybrid quantum-classical AI / ML system is configured to use -14- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlto solve problems. For example, the problem space information may indicate that the hybrid quantum-classical AI / ML system is to be trained to use QTDA to determine whether an anomaly is present in a data set. In another example, the problem space information may indicate that the hybrid quantum-classical AI / ML system is to be trained to use a generative quantum eigensolver (GQE), variational quantum eigensolver (VQE), quantum phase estimation (QPE), and / or the like to provide eigen-solutions and corresponding eigenstates to Hamiltonian representations of systems being simulated.

[0060] In various embodiments, the problem space information provides background information corresponding to the problem space. For example, when the problem space corresponds to atomic and / or molecular systems, such as identifying ground states or excited states of atomic and / or molecular systems and / or expectation values corresponding to atomic and / or molecular systems, the problem space information may include a database or other data store regarding parameters of various atoms and / or molecules, interactions between various atoms and / or molecules, and / or the like. For example, the problem space information may include a periodic table of elements. When the problem space corresponds to a non-physical system (e.g., logistics), the problem space information may indicate various rules to be used to determine how different components interact, costs within the system, and / or other rules that define the structure of possible outputs.

[0061] In various embodiments, the classical computing entity 110 receives the problem space information (e.g., via communication interface 825) from another classical computing entity. In various embodiments, the classical computing entity 110 receives the problem space information via user interaction with the user interface 815. In an example embodiment, the classical computing entity 110 obtains the problem space information by receiving (e.g., via communication interface 825 and / or user interface 815) an indication of one or more databases and / or other data stores stored by the classical computing entity 110 (e.g., in classical memory 822, 824) and / or otherwise accessible to the classical computing entity 110 that should be used as problem space information.

[0062] At block 304, the classical computing entity 110 generates and / or obtains a plurality of circuit and result pairs based at least in part on the problem space information. For example, the classical computing entity 110 comprises means, such as classical processing elements 808, classical memory 822, 824, communications interface 825 (e.g.,-15- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlnetwork interface 820, antenna 812 coupled to receiver 806 and / or transmitter 804), user interface 815 (e.g., keypad 818, display 816), and / or the like, for generating and / or obtaining a plurality of circuit and result pairs based at least in part on the problem space information. For example, the classical computing entity 110 may receive and / or have access to a plurality of quantum circuits associated with the problem space and results generated by executing and / or performing the quantum circuits. In some embodiments, the quantum circuits may be generated (e.g., via user interaction with a user interface 815) and then executed by a quantum computer 130 such that the quantum circuit and result pair is obtained.

[0063] At block 306, the classical computing entity 110 trains the circuit LLM 222 using at least a portion of the plurality of circuit and result pairs. For example, the classical computing entity 110 comprises means, such as classical processing elements 808, classical memory 822, 824, and / or the like, for training the circuit LLM 222 using at least a portion of the plurality of circuit and result pairs. In some embodiments, the training of the circuit LLM is performed using a generative adversarial network (GAN). In various embodiments, the circuit LLM 222 is trained to generate and / or establish relationships between various gadgets. For example, a library of gadgets and / or fundamental quantum circuit component may be stored in memory 822, 824. The training of the circuit LLM 222 may be used to establish relationships between the various gadgets and / or fundamental quantum circuit components. For example, the circuit LLM 222 may be configured to treat the gadgets and / or fundamental quantum circuit components as tokens and the training may enable the circuit LLM 222 to generate token sets (e.g., quantum circuits 234) by combining the gadgets and / or fundamental quantum circuit components.

[0064] In certain embodiments, a circuit LLM 222 being trained to generate quantum circuits for use a GQE application, for example, may be trained with using training data comprises a plurality of circuit-result pairs where the result is an expectation value for energy, for example. The circuit LLM 222 may then be trained to generate and / or provide as output circuits which have lower energies, as reflected in the training loss function. For example, the training loss function may be selected to train the circuit LLM 222 to generate quantum circuits having desirable attributes for a given application.

[0065] Various other components of the hybrid quantum-classical AI / ML system 200 may also be trained (e.g., using respective machine learning protocols). For example, the blueprint engine 224 may include a semantic interpreter that may include a classifier that -16- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlmay be trained to determine whether an anomaly is present in a data set based at least in part on an approximate Betti-k number for a sub-graph configured to represent a portion of the data set.

[0066] At block 308, at some point in time after the initial training of the circuit LLM 222, the classical computing entity 110 receives a request. For example, the classical computing entity 110 comprises means, such as classical processing elements 808, classical memory 822, 824, communications interface 825 (e.g., network interface 820, antenna 812 coupled to receiver 806 and / or transmitter 804), user interface 815 (e.g., keypad 818, display 816), and / or the like, for receiving a request.

[0067] For example, a user may interact with a user interface 815 and, based thereon, a request is received by the classical computing entity 110. In another example, another classical computing entity may be executing a (classical) program or application and may generate and / or provide a request as part of executing the program or application. In certain embodiments, the classical computing entity 110 is executing a (classical) program or application and the program or application generates and / or provides the request.

[0068] In various embodiments, the request comprises and / or indicates request information. The request information defines the particular system to be simulated (e.g., based at least in part on a Hamiltonian representation of the system), a data set for which it is to be determined whether an anomaly is present, and / or the like.

[0069] At block 310, the circuit LLM 222 generates one or more quantum circuits 234. For example, the classical computing engine 110 executes the circuit LLM 222 to cause the circuit LLM 222 to generate one or more quantum circuits 234 based at least in part on the request information. For example, the classical computing entity 110 comprises means, such as classical processing elements 808, classical memory 822, 824, and / or the like, for causing the circuit LLM 222 to generate one or more quantum circuits 234.

[0070] For example, the blueprint engine 224 analyzes and / or processes the request information and generates and input vector based thereon. The blueprint engine 224 then provides the input vector to an input layer of the circuit LLM 222, thereby causing the circuit LLM 222 to generate the one or more quantum circuits 234. For example, the circuit LLM 222 receives the input vector via the input layer thereof and transforms the input vector into a quantum circuit provided at the output layer of the circuit LLM 222.-17- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vl

[0071] At block 312, the classical computing entity 110 causes the quantum computer 130 to perform the quantum circuit(s) 234 and receives quantum information generated by execution of the quantum circuit(s) 234 by the quantum computer 130. For example, the classical computing entity 110 comprises means, such as the classical computing entity 110 comprises means, such as classical processing elements 808, classical memory 822, 824, communications interface 825 (e.g., network interface 820, antenna 812 coupled to receiver 806 and / or transmitter 804), user interface 815 (e.g., keypad 818, display 816), and / or the like, for causing the quantum computer 130 to perform the quantum circuit(s) 234 and receiving quantum information generated via the execution of the quantum circuit(s) by the quantum computer 130.

[0072] At block 314, the classical computing entity 110 determines whether one or more stop criteria are satisfied. In various embodiments, the classical computing entity 110 comprises means, such as classical processing elements 808, classical memory 822, 824, and / or the like, for determining whether one or more stop criteria are satisfied. In various embodiments, the stop criteria corresponds to the problem space. In various embodiments, the stop criteria are configured to determine when a sufficient amount of quantum information is available to determine an output having at least a minimum confidence and / or to determine when the results of performing the quantum circuit(s) 234 have converged (e.g., performance of additional quantum circuits is not expected to lead to a substantially more accurate or more precise output).

[0073] When the classical computing entity 110 determines, at block 314, that the stop criteria are not satisfied, the process continues to block 316. At block 316, the classical computing entity 110 updates the circuit LLM 222 (e.g., causes further training of the circuit LLM 222), updates a component (e.g., machine learning model) of the blueprint engine 224, and / or updates one or more solution parameter of the one or more quantum circuits. For example, the classical computing entity 110 may comprise means, such as classical processing elements 808, classical memory 822, 824, and / or the like for causing further training of the circuit LLM 222 or another machine learning model component of the hybrid quantum-classical AI / ML system based on the quantum circuits generated by the circuit LLM and the corresponding quantum information generated by the quantum computer via execution of the quantum circuits. For example, the classical computing entity 110 may comprise means, such as classical processing elements 808, classical memory 822, 824, and / or the like, for updating one or more solution parameters. For -18- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlexample, a loss function may be computed (e.g., by blueprint engine 224 and / or the like) and the one or more solution parameters may be updated and / or modified based on the loss function or a gradient thereof. For example, a loss function may be evaluated based at least in part on the quantum information generated via executing the one or more quantum circuits by the quantum computer. In various embodiments the updates to the circuit LLM, component(s) of the blueprint engine, and / or parameter solutions are updated based at least in part on the loss function or a gradient thereof.

[0074] The process then returns to block 310 and another round of one or more quantum circuits are generated by the circuit LLM. The quantum circuits are performed by the quantum computer to generate corresponding quantum information. The corresponding quantum information is used to update the circuit LLM 222, one or more components of the blueprint engine 224, and / or one or more solution parameters. The process continues to be iterated until the one or more stop criteria are satisfied at block 314.

[0075] At block 318, responsive to the one or more stop criteria being satisfied, the classical computing entity 110 determines an output corresponding to the request 232. For example, the classical computing entity 110 comprises means, such as classical processing elements 808, classical memory 822, 824, and / or the like, for determining an output corresponding to the request 232. In various embodiments, the output is determined based at least in part on quantum information generated via performance of at least one quantum circuit generated by the circuit LLM 222. In an example embodiment, the output is determined by providing quantum information generated via performance of the at least one quantum circuit as input to a semantic interpreter of the blueprint engine. For example, the semantic interpreter may be a machine learning trained classifier configured to determine whether one or more anomalies are present in a data set based on estimations of Betti-k numbers determined for one or more sub-graphs extracted from the data set. In another example, the semantic interpreter may be configured to generate one or more expectation values (e.g., eigenvalues such as energy / cost, and / or expectations for various observables) corresponding to eigen-solutions to a Hamiltonian representation of a system. In some embodiments, the semantic interpreter may be configured to convert and / or transform the quantum information identifying the eigen-solution of the Hamiltonian representation of the system into a human interpretable representation of the eigen-solution.-19- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vl

[0076] At block 320, the classical computing entity provides the output 238 corresponding to the request 232. For example, the classical computing entity 110 comprises means, such as classical processing elements 808, classical memory 822, 824, communications interface 825 (e.g., network interface 820, antenna 812 coupled to receiver 806 and / or transmitter 804), user interface 815 (e.g., keypad 818, display 816), and / or the like, for providing the output 238 corresponding to and / or determined responsive to receiving the request 232. Providing the output may include storing the output in classical memory 822, 824, providing the output via the user interface 815, providing the output as input to a computing program or application being executed by the classical computing entity 110 and / or another classical computer.

[0077] A hybrid quantum-classical AI / ML system may continue to train a circuit LLM 222, machine learning components and / or models of a blueprint engine 224 (e.g., configured to generate an input vector for the circuit LLM 222 based on possibly a natural language inquiry of a request, a semantic interpreter configured as a classifier and / or the like), and / or a resource estimation oracle 228 via various iterations of receiving requests, determining outputs responsive to the received request, and providing the outputs.

[0078] Some example embodiments regarding the problem spaces of anomaly detection and GQE will now be described with respect to Figures 4 and 5, respectively.

[0079] Figure 4 provides a flowchart illustrating various processes and / or procedures that may be performed by a classical computing entity 110 to provide an output indicating whether an anomaly is present in a data set. Starting at block 402, the classical computing entity 110 obtains information to be analyzed. For example, the classical computing entity 110 may receive a request that includes request information comprising the data set and / or indicating a data set (e.g., comprises an indication / pathname / URL of where the data set is stored and / or how the data set may be accessed). In various embodiments, the data set may be a collection of a plurality of data records.

[0080] At block 404, the blueprint engine extracts a sub-graph from the data set. For example, the classical computing entity 110 causes execution of a blueprint engine to cause the blueprint engine to extract a sub-graph from the data set. In various embodiments, the sub-graph is a structure representing a subset of the data records of the data set. For example, the data set may comprise a plurality of instances of data. The data set may be represented as a graph where each data record is a node of the graph and -20- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlrelationships between data records are indicated as links between the respective nodes. The graph may be defined based on a plurality of axes and / or dimensions based on various fields of the plurality of instances of data. For example, an instance of data may be represented in the graph by a point or node in a multidimensional space corresponding to the values assigned to one or more fields of the instance of data. In some embodiments, relationships between two instances of data may be represented as a graph edge or link that connects the corresponding points or nodes.

[0081] The sub-graph may be extracted using various techniques. In an example embodiment, a sub-graph extraction model, which is a machine learning trained model, is configured to extract the sub-graph from the graph. In another example, the sub-graph is extracted from the graph by selecting nodes of interest and extracting the nodes of interest and any other nodes within two edges or links from the selected nodes of interest. The extracted sub-graph provides a representation of two or more (but generally not all) of the instances of data of the plurality of instances of data of the data set. For example, the blueprint engine may map a subset of a plurality of instances of data into a sub-graph and generate an input vector representing the sub-graph. The input vector is then provided to an input layer of the circuit LLM.

[0082] At block 406, the classical computing entity 110 causes the circuit LLM to generate a QTDA circuit for the extracted sub-graph. For example, the circuit LLM may generate one or more QTDA circuits configured to determine and / or compute one or more topological features or parameters of the extracted sub-graph. For example, the one or more QTDA circuits may be configured for estimating a Betti-k number for the extracted sub-graph.

[0083] At block 408, classical computing entity 110 causes the quantum computer 130 to perform the one or more quantum circuits. For example, the classical computing entity 110 may provide (e.g., transmit) the quantum circuit configured for receipt by the quantum computer 130. The quantum computer may, responsive to receiving the quantum circuit, perform the quantum circuit to generate quantum information. The quantum information is generated via the performance of the quantum circuit by the quantum computer. The quantum computer provides the results configured for receipt by the classical computing entity 110.

[0084] At block 410, the classical computing entity 110 receives the quantum information. At block 412, the classical computing entity 110 determines whether the one -21- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlor more stop criteria are satisfied. In an example embodiment, the stop criteria are determined to be satisfied responsive to a semantic interpreter (e.g., a machine learning trained classifier in some embodiments) and the semantic interpreter determines whether an anomaly is present in the data set (e.g., whether the quantum information indicates that the data set is in a class of no anomalies or in a class indicating the data set includes one or more anomalies). The semantic interpreter may also determine a confidence level with which the classification is made and / or determined. In certain embodiments, the classical computing entity 110 determines that stop criteria are satisfied when the confidence level with which the classification is made and / or determined is greater than a threshold confidence level (e.g., 67%, 70%, 75%, 80%, 85%, 90%, 95%, 99%, 99.9%, and / or the like). In some embodiments the threshold confidence level is provided and / or a selectin of the threshold confidence level from a plurality of defined threshold confidence levels is provided by the request information.

[0085] At block 414, responsive to the classical computing entity 110 determining that the stop criteria are not satisfied, the blueprint engine updates the extracted sub-graph based on the quantum information. In some embodiments, the blueprint engine selects a new subset of instances data from the plurality of instances of data of the data set and maps the new subset to a sub-graph. The new subset may have a non-zero overlap with the previous subset. In various embodiments, one or more solution parameters which are used by the blueprint engine to select the new subset of instances of data are updated based at least in part on the quantum information. For example, weights used to select the new subset of instances of data may be updated based at least in part on the quantum information. In various embodiments, the new subset may include more, fewer, or the same number members (e.g., instances of data) as the previous subset.

[0086] The process then returns to block 406 and the circuit LLM generates one or more quantum circuits corresponding to the new subset of instances of data. The quantum circuit(s) are provided to a quantum computer 130 for execution, and quantum information generated via execution of the quantum circuit(s) are received. This iterative process continues until the stop criteria are determined to be satisfied. In some embodiments, the stop criteria are determined to be satisfied responsive to a maximum number of iterations being performed, a maximum amount of time has elapsed since the process was initiated, and / or the like.-22- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vl

[0087] Responsive to determining, at block 412, that the one or more stop criteria are satisfied, the process continues to block 416. At block 416, if not previously determined, the output is determined. For example, the quantum information generated via performance of one or more QTDA circuits by the quantum computer 130 may be processed via a semantic interpreter configured to generate the output. In an example embodiment, the semantic interpreter is a classifier that indicates whether an anomaly is present in the data set or not. In some embodiments, the semantic interpreter is a multiclass classifier configured to process quantum information (e.g., approximation of Betti-k numbers and / or the like) configured to classify a data set as not including an anomaly or as including one or more a particular types of anomalies, to classify the data set based on a number of anomalies included in the data set and / or an anomaly rate of the data set, and / or the like.

[0088] At block 418, the classical computing entity 110 provides the output (e.g., generated by the semantic interpreter of the blueprint engine). In various embodiments, providing the output includes storing the output in classical memory, rendering a graphical representation of the output via the user interface, and / or providing the output as input to a computing program or application being executed by the classical computing entity 110 and / or another classical computer.

[0089] Figure 5 provides a flowchart illustrating various processes and / or procedures for a flowchart illustrating various processes and / or procedures that may be performed by a classical computing entity 110 to provide an output comprising an eigen-solution and / or eigenvalue corresponding to the eigen-solution for a system represented as a Hamiltonian. A Hamiltonian is a function that describes interactions within a system, generally in terms of the generalized coordinates q and their conjugate momentum p. In general, the result of a Hamiltonian acting on a state that is an eigenstate (also referred to herein as an eigen-solution) of the Hamiltonian is a cost or energy corresponding to the eigenstate or eigen-solution of the Hamiltonian. A ground state or lowest cost state of the Hamiltonian and the corresponding ground state energy and / or cost may be determined using a GQE or other quantum eigensolver. The Hamiltonian may represent an atom or molecule or other physical system. For example, the Hamiltonian may be used to identify and / or characterize (e.g., determine expectation values for observables of) ground states and / or excited stats of the atom, molecule, or other physical system. In some embodiments, the system may be a biomedical system, an engineered material, a logistics system (e.g., a -23- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlroute finder algorithm, allocation of resources, etc.), and / or other system that may be modeled and / or represented as a Hamiltonian.

[0090] Starting at block 502, the classical computing entity 110 obtains simulated system information to be analyzed. For example, the classical computing entity 110 may receive a request that includes request information comprising the simulated system information. For example, the simulated system information may include a Hamiltonian representation of the system to be simulated and any boundary conditions and / or constraints to be taken into consideration during simulation of the system.

[0091] In certain embodiments, the request information includes an ansatz to be used as starting point for simulating the system (e.g.., determining one or more eigenstates or eigen-solutions of the system and corresponding eigen values). In some embodiments, the blueprint engine includes an ansatz generator configured to generate an ansatz based at least in part on the Hamiltonian representation of the system, a type associated with the system, boundary conditions and / or constraints of the system, and / or the like. At block 504, blueprint engine executes the ansatz generator to cause the ansatz generator to generate an ansatz. For example, the ansatz generator may be a machine learning trained model configured and / or trained to generate efficient ansatzes based at least in part on a Hamiltonian representation of a system to be simulated.

[0092] At block 506, the classical computing entity 110 causes the circuit LLM to generate a one or more quantum circuits based at least in part on the ansatz. In some embodiments, the circuit LLM may generate a batch of quantum circuits that explore a region of parameter space based at least in part on the ansatz. For example, the circuit LLM may generate one or more quantum circuits configured to simulate the interactions with the system as represented by the Hamiltonian representation of the system. In various embodiments, the one or more quantum circuits are generated in accordance with a GQE or other quantum eigensolver protocol.

[0093] At block 508, classical computing entity 110 causes the quantum computer 130 to perform the one or more quantum circuits. For example, the classical computing entity 110 may provide (e.g., transmit) the quantum circuit configured for receipt by the quantum computer 130. The quantum computer may, responsive to receiving the quantum circuits, perform the quantum circuits to generate respective instances quantum information (e.g., an instance of quantum information generated via the execution of a respective quantum circuit of the one or more quantum circuits). The quantum-24- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlinformation is generated via the performance of the quantum circuit by the quantum computer. The quantum computer provides the results configured for receipt by the classical computing entity 110. The classical computing entity 110 receives the quantum information.

[0094] At block 510, the classical computing entity 110 determines whether the one or more stop criteria are satisfied. In an example embodiment, the stop criteria are determined to be satisfied responsive to the eigenstate / eigen-solution and / or corresponding eigenvalue satisfying one or more convergence criteria. For example, when quantum information generated via performance of two or more quantum circuits yield eigenvalues for an eigenstate / eigen-solution of the Hamiltonian representation of the system that are within a threshold percentage / fraction of one another (e.g., the eigenvalues are equivalent to one another within significant figures), the classical computing entity 110 may determine that the one or more convergence criteria are satisfied. In another example, when quantum information generated via performance of two or more quantum circuits yield eigenvalues for an eigenstate / eigen-solution of the Hamiltonian representation of the system indicate that a (local) extremum (e.g., minimum energy / cost) has been identified, the classical computing entity 110 may determine that the one or more convergence criteria are satisfied. Various convergence criteria may be applied in various embodiments, as appropriate for the application. The stop criteria are determined to not be satisfied responsive to the eigenstate / eigen-solution and / or corresponding eigenvalue not satisfying one or more convergence criteria.

[0095] At block 512, responsive to the classical computing entity 110 determining that the stop criteria are not satisfied, the blueprint engine updates the circuit LLM, and / or one or more solution parameters (e.g., corresponding to a structure or constants of the ansatz, and / or the like) based at least in part on the quantum information. For example, the ansatz may be updated. In another example, updating the solution parameters results in the region of parameter space of the eigenstate / eigen-solution being explored being updated or modified.

[0096] The process then returns to block 506 and the circuit LLM generates one or more quantum circuits using the updated circuit LLM and / or based at least in part on the updated solution parameters. The quantum circuit(s) are provided to a quantum computer 130 for execution, and quantum information generated via execution of the quantum circuit(s) are received. This iterative process continues until the stop criteria are-25- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vldetermined to be satisfied. In some embodiments, the stop criteria are determined to be satisfied responsive to a maximum number of iterations being performed, a maximum amount of time has elapsed since the process was initiated, and / or the like.

[0097] Responsive to determining, at block 510, that the one or more stop criteria are satisfied, the process continues to block 514. At block 514, if not previously determined, the output is determined. For example, the quantum information generated via performance of one or more quantum circuits by the quantum computer 130 may be processed via a semantic interpreter configured to generate the output. In an example embodiment, the semantic interpreter is configured to generate a characterization of the eigenstate / eigen-solution identified and / or the corresponding energy / cost. For example, the semantic interpreter may generate a human language representation, a graphical representation, or a mathematical representation of the eigenstate / eigen-solution. In another example, the semantic interpreter may use the quantum information to determine an eigenvalue (e.g., energy / cost) corresponding to the eigenstate / eigen-solution and / or to determine one or more expectation values (e.g., of various observables) corresponding to the eigenstate / eigen-solution of the Hamiltonian representation of the system. The characterization of the eigenstate / eigen-solution identified may take various forms in various embodiments, as appropriate for the application.

[0098] At block 516, the classical computing entity 110 provides the output (e.g., generated by the semantic interpreter of the blueprint engine). In various embodiments, providing the output includes storing the output in classical memory, rendering a graphical representation of the output via the user interface, and / or providing the output as input to a computing program or application being executed by the classical computing entity 110 and / or another classical computer.Example Operation of Quantum Computer of a Hybrid Quantum-Classical AI / ML System

[0099] Figure 6 provides a flowchart illustrating various processes and / or procedures performed by a controller 132 of a quantum computer 130 as part of the quantum computer function within a hybrid quantum-classical AI / ML system. Starting at block 602, the controller 132 obtains a quantum circuit. For example, the controller 132 comprises means, such as classical processing elements 705, classical memory 710, communication interface 720, and / or the like, for obtaining a quantum circuit. For -26- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlexample, the classical computing entity 110 may provide (e.g., transmit) a quantum circuit and the controller 132 may receive the quantum circuit. In an example embodiment, the classical computing entity 110 provides the quantum circuit in an uncompiled format, a partially compiled format (e.g., an implementation of the quantum circuit as a QIR program, for example), or as machine level instructions. In various embodiments, the controller 132 comprises a compiler configured to compile the quantum circuit into an implementation (e.g., as a QIR program) and / or into machine level instructions.

[0100] At block 604, the controller 132 controls operation of the qubit manipulation elements and / or sensors to cause performance of the quantum circuit. For example, the controller 132 comprises means, such as processing elements 705, memory 710, driver controller elements 715, A / D converters 725, and / or the like, for controlling operation of the qubit manipulation elements 136 and / or the sensors 138 to cause performance of the quantum circuit. For example, the controller 132 may control operation of the qubit manipulation elements 136 to cause a controlled quantum state evolution of one or more qubits 134. The controller 132 may be configured to receive respective sensor signals generated by one or more sensors 138 during performance of the quantum circuit and use the respective sensor signals to determine respective quantum states of the one or more qubits 134.

[0101] At block 606, the controller 132 determines the quantum information generated via performance of the quantum circuit by the quantum computer 130. For example, the controller 132 comprises means, such as processing elements 705, memory 710, and / or the like, for determining the quantum information generated via performance of the quantum circuit by the quantum computer 130. In various embodiments, the quantum information comprises an indication of one or more quantum states determined for one or more qubits. For example, the quantum information may include a bit array that indicates the measured quantum state of the one or more qubits. In another example, the quantum states determined for one or more qubits may be processed to generate the quantum information based thereon.

[0102] At block 608, the controller 132 provides the quantum information. For example, the controller 132 comprises means, such as classical processing elements 705, classical memory 710, communication interface 720, and / or the like, for providing the quantum-27- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlinformation. For example, the controller 132 may provide (e.g., transmit) the quantum information configured for receipt by the classical computing entity 110.Exemplary Controller

[0103] In various embodiments, hybrid quantum-classical computing system 100 comprises a quantum computer 130. The quantum computer 130 is configured to perform various quantum computations and / or calculations via execution of one or more quantum circuits and / or algorithms. In various embodiments, the controller 132 is configured to control operation of one or more components of the quantum computer 130 (e.g., qubit manipulation elements 136, sensors 138), receive sensor signals indicating measurements captured by sensors 138, and / or communicate with a classical computing entity 110.

[0104] As shown in Figure 7, in various embodiments, the controller 132 may comprise various controller elements including processing elements 705, memory 710, driver controller elements 715, a communication interface 720, analog-digital converter elements 725, and / or the like. For example, the processing elements 705 may comprise one or more (classical) processing devices such as programmable logic devices (CPLDs), microprocessors, central processing units (CPUs), graphic processing units (GPUs), coprocessing entities, application-specific instruction-set processors (ASIPs), integrated circuits, application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), hardware accelerators, other processing devices and / or circuitry, and / or the like. The term circuitry may refer to an entirely hardware embodiment or a combination of hardware and computer program products. In an example embodiment, the processing element 705 of the controller 132 comprises a clock and / or is in communication with a clock.

[0105] For example, the memory 710 may comprise non-transitory memory such as volatile and / or non-volatile memory storage such as one or more of as hard disks, ROM, PROM, EPROM, EEPROM, flash memory, MMCs, SD memory cards, Memory Sticks, CBRAM, PRAM, FeRAM, RRAM, SONOS, racetrack memory, RAM, DRAM, SRAM, FPM DRAM, EDO DRAM, SDRAM, DDR SDRAM, DDR2 SDRAM, DDR3 SDRAM, RDRAM, RIMM, DIMM, SIMM, VRAM, cache memory, register memory, and / or the like. In various embodiments, the memory 710 may store a queue of commands to be executed to cause a quantum algorithm and / or circuit to be executed (e.g., an executable queue), qubit records corresponding the qubits of quantum computer (e.g., in a qubit -28- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlrecord data store, qubit record database, qubit record table, and / or the like), a calibration table, computer program code (e.g., in a one or more computer languages, specialized controller language(s), and / or the like), and / or the like. In an example embodiment, execution of at least a portion of the computer program code stored in the memory 710 (e.g., by a processing element 705) causes the controller 132 to perform one or more steps, operations, processes, procedures and / or the like described herein for controlling operation of one or more qubit manipulation elements 136, processing sensor signals indicating measurements captured by sensors 138, and / or communicating with a classical computing entity 110 of the hybrid quantum-classical computing system 100.

[0106] In various embodiments, the driver controller elements 715 may include one or more drivers and / or controller elements each configured to control one or more drivers. In various embodiments, the driver controller elements 715 may comprise drivers and / or driver controllers. For example, the driver controllers may be configured to cause one or more corresponding drivers to be operated in accordance with executable instructions, commands, and / or the like scheduled and executed by the controller 132 (e.g., by the processing element 705). In various embodiments, the driver controller elements 715 may enable the controller 132 to operate various ones of the qubit manipulation elements 136 and / or sensors 138. In various embodiments, the drivers may comprise laser drivers configured to operate one or lasers; drivers for controlling operation of one or more voltage / current sources to cause generation and providing of one or more voltage and / or current signals; and / or various other drivers configured to control operation of respective qubit manipulation elements 136 of the quantum computer 130.

[0107] In various embodiments, the controller 132 comprises means for communicating and / or receiving signals from one or more sensors (e.g., photodetectors, voltage / current sensors, temperature sensors, pressure sensors, and / or other sensors). For example, the controller 132 may comprise one or more analog-digital converter elements 725 configured to receive signals from one or more sensors.

[0108] In various embodiments, the controller 132 comprises a communication interface 720 for interfacing and / or communicating with a classical computing entity 110 of the hybrid quantum-classical computing system 100. For example, the controller 132 may comprise a communication interface 720 for receiving quantum circuits, executable instructions, command sets, and / or the like from the classical computing entity 110 and providing output (e.g., measurement results) received from the quantum computer 130 -29- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vl(e.g., via sensors 138) to the classical computing entity 110. In various embodiments, the classical computing entity 110 and the controller 132 may communicate via a direct wired and / or wireless connection and / or via one or more wired and / or wireless networks 120.Exemplary Classical Computing Entity

[0109] Figure 8 provides an illustrative schematic representative of an example classical computing entity 10 that can be used in conjunction with embodiments of the present invention. In various embodiments, a classical computing entity 110 is configured to interface with a quantum computer 130. For example, the classical computing entity 110 is configured to interface with a quantum computer 130 so as to enable the efficient and accurate modeling of various systems using the quantum computer 130. For example, the classical computing entity 110 may be configured to communicate with the quantum computer 130 allow a user (e.g., a human user or a program operating on the classical computing entity 110) to provide input to the quantum computer 130 and receive, display, analyze, and / or the like output (e.g., measurement results) from the quantum computer 130. In various embodiments, a classical computing entity 110 may be a computer (e.g., desktop computer, laptop, tablet, smartphone, and / or the like), a server, a Cloud-based computing resource, and / or the like.

[0110] As shown in Figure 8, a classical computing entity 110 can include one or more processing elements 808. In various embodiments, the processing elements 808 may comprise one or more (classical) processing devices such as programmable logic devices (CPLDs), microprocessors, central processing units (CPUs), graphic processing units (GPUs), coprocessing entities, application-specific instruction-set processors (ASIPs), integrated circuits, application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), hardware accelerators, other processing devices and / or circuitry, and / or the like. For example, the processing elements 808 may include one or more switches and GPUs in communication with one another via an Al fabric. For example, the classical computing entity 110 may be a high performance computing environment, a datacenter or portion thereof, a server, an Al factory, and / or the like. In some embodiments, the classical computing entity is a desktop computer, a client computing entity that is in communication with a server, a laptop computer, a handheld computing device such as a tablet or smartphone, and / or the like.-30- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vl[OHl] As shown in Figure 8, a classical computing entity 110 may further include one or more communications interfaces 825. In some embodiments, the one or more communications interfaces 825 include a network interface 820. In some embodiments, the one or more communications interfaces 825 include an antenna 812, a transmitter 804 (e.g., radio), a receiver 806 (e.g., radio), and a processing element 808 that provides signals to and receives signals from the transmitter 804 and receiver 806, respectively. The signals provided to and received from the transmitter 804 and the receiver 806, respectively, may include signaling information / data in accordance with an air interface standard of applicable wireless systems to communicate with various entities, such as a controller 132, other classical computing entities 110, and / or the like. In this regard, the classical computing entity 110 may be capable of operating with one or more air interface standards, communication protocols, modulation types, and access types.

[0112] For example, the classical computing entity 110 may be configured to receive and / or provide communications using a wired data transmission protocol, such as fiber distributed data interface (FDDI), digital subscriber line (DSL), Ethernet, asynchronous transfer mode (ATM), frame relay, data over cable service interface specification (DOCSIS), or any other wired transmission protocol. Similarly, the classical computing entity 110 may be configured to communicate via wireless external communication networks using any of a variety of protocols, such as general packet radio service (GPRS), Universal Mobile Telecommunications System (UMTS), Code Division Multiple Access 2000 (CDMA2000), CDMA2000 IX (IxRTT), Wideband Code Division Multiple Access (WCDMA), Global System for Mobile Communications (GSM), Enhanced Data rates for GSM Evolution (EDGE), Time Division-Synchronous Code Division Multiple Access (TD-SCDMA), Long Term Evolution (LTE), Evolved Universal Terrestrial Radio Access Network (E-UTRAN), Evolution-Data Optimized (EVDO), High Speed Packet Access (HSPA), High-Speed Downlink Packet Access (HSDPA), IEEE 802.11 (Wi-Fi), Wi-Fi Direct, 802.16 (WiMAX), ultra wideband (UWB), infrared (IR) protocols, near field communication (NFC) protocols, Wibree, Bluetooth protocols, wireless universal serial bus (USB) protocols, and / or any other wireless protocol. The classical computing entity 110 may use such protocols and standards to communicate using Border Gateway Protocol (BGP), Dynamic Host Configuration Protocol (DHCP), Domain Name System (DNS), File Transfer Protocol (FTP), Hypertext Transfer Protocol (HTTP), HTTP over TLS / SSL / Secure, Internet -31- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlMessage Access Protocol (IMAP), Network Time Protocol (NTP), Simple Mail Transfer Protocol (SMTP), Telnet, Transport Layer Security (TLS), Secure Sockets Layer (SSL), Internet Protocol (IP), Transmission Control Protocol (TCP), User Datagram Protocol (UDP), Datagram Congestion Control Protocol (DCCP), Stream Control Transmission Protocol (SCTP), HyperText Markup Language (HTML), and / or the like.

[0113] Via these communication standards and protocols, the classical computing entity 110 can communicate with various other entities using concepts such as Unstructured Supplementary Service information / data (USSD), Short Message Service (SMS), Multimedia Messaging Service (MMS), Dual-Tone Multi-Frequency Signaling (DTMF), and / or Subscriber Identity Module Dialer (SIM dialer). The classical computing entity 110 can also download changes, add-ons, and updates, for instance, to its firmware, software (e.g., including executable instructions, applications, program modules), and operating system.

[0114] In various embodiments, the classical computing entity 110 may comprise a network interface 820 for interfacing and / or communicating with the controller 132, for example. For example, the classical computing entity 110 may comprise a network interface 820 for providing quantum circuits, executable instructions, command sets, and / or the like for receipt by the controller 132 and / or receiving output (e.g., measurement results) provided by the quantum computer 130. In various embodiments, the classical computing entity 110 and the controller 132 may communicate via a direct wired and / or wireless connection and / or via one or more wired and / or wireless networks 120.

[0115] In various embodiments, the processing elements 808 may comprise one or more processing devices such as programmable logic devices (CPLDs), microprocessors, coprocessing entities, application-specific instruction-set processors (ASIPs), integrated circuits, application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), hardware accelerators, graphics processing units (GPUs), central processing units (CPUs), other processing devices and / or circuitry, and / or the like. The term circuitry may refer to an entirely hardware embodiment or a combination of hardware and computer program products.

[0116] The classical computing entity 110 may also comprise a user interface device 815 comprising one or more user input / output interfaces (e.g., a display 816 and / or speaker / speaker driver coupled to a processing element 808 and a touch screen, keyboard,-32- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlmouse, and / or microphone coupled to a processing element 808). For instance, the user output interface may be configured to provide an application, browser, user interface, interface, dashboard, screen, webpage, page, and / or similar words used herein interchangeably executing on and / or accessible via the computing entity 10 to cause display or audible presentation of information / data and for interaction therewith via one or more user input interfaces. The user input interface can comprise any of a number of devices allowing the computing entity 10 to receive data, such as a keypad 818 (hard or soft), a touch display, mouse, voice / speech or motion interfaces, scanners, readers, or other input device. In embodiments including a keypad 818, the keypad 818 can include (or cause display of) the conventional numeric (0-9) and related keys (#, *), and other keys used for operating the classical computing entity 110 and may include a full set of alphabetic keys or set of keys that may be activated to provide a full set of alphanumeric keys. In addition to providing input, the user input interface can be used, for example, to activate or deactivate certain functions, such as screen savers and / or sleep modes.Through such inputs the classical computing entity 110 can collect information / data, user interact! on / input, and / or the like.

[0117] The classical computing entity 110 can also include volatile storage or memory 822 and / or non-volatile storage or memory 824, which can be embedded and / or may be removable. For instance, the non-volatile memory may be ROM, PROM, EPROM, EEPROM, flash memory, MMCs, SD memory cards, Memory Sticks, CBRAM, PRAM, FeRAM, RRAM, SONOS, racetrack memory, and / or the like. The volatile memory may be RAM, DRAM, SRAM, FPM DRAM, EDO DRAM, SDRAM, DDR SDRAM, DDR2 SDRAM, DDR8 SDRAM, RDRAM, RIMM, DIMM, SIMM, VRAM, cache memory, register memory, and / or the like. The volatile and non-volatile storage or memory can store databases, database instances, database management system entities, data, applications, programs, program modules, scripts, source code, object code, byte code, compiled code, interpreted code, machine code, executable instructions, and / or the like to implement the functions of the classical computing entity 110.Conclusion

[0118] Many modifications and other embodiments of the invention set forth herein will come to mind to one skilled in the art to which the invention pertains having the benefit of the teachings presented in the foregoing descriptions and the associated drawings.-33- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlTherefore, it is to be understood that the invention is not to be limited to the specific embodiments disclosed and that modifications and other embodiments are intended to be included within the scope of the appended claims. Although specific terms are employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation.

[0119] In any of the above aspects, the various features may be implemented in hardware, or as software modules running on one or more processors / computers.

[0120] The invention also provides a computer program or a computer program product comprising instructions which, when executed by a computer, cause the computer to carry out any of the methods / method steps described herein, and a non-transitory computer-readable medium comprising instructions which, when executed by a computer, cause the computer to carry out any of the methods / method steps described herein. A computer program embodying the invention may be stored on a non-transitory computer-readable medium, or it could, for example, be in the form of a signal such as a downloadable data signal provided from an Internet website, or it could be in any other form.-34- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vl

Claims

CLAIMS:

1. A method comprising:causing a circuit large language model (LLM) to generate one or more quantum circuits based at least in part on request information;causing performance of the one or more quantum circuits on a quantum computer; receiving quantum information generated via performance of the one or more quantum circuits on the quantum computer;determining whether one or more stop criteria are satisfied by the quantum information;responsive to determining that the one or more stop criteria are satisfied, providing an output corresponding to the request information; andresponsive to determining that the one or more stop criteria are not satisfied:updating at least one of the circuit LLM or a solution parameter of the one or more quantum circuits,causing one or more updated quantum circuits to be performed by the quantum computer, wherein the one or more updated quantum circuits are at least one of generated by the circuit LLM based at least in part on the solution parameter, andreceiving the quantum information generated via performance of the one or more updated quantum circuits by the quantum computer.

2. The method of claim 1, further comprising determining the output based at least in part on the quantum information generated via performance of at least one quantum circuit by the quantum computer.

3. The method of claim 2, wherein determining the output comprises causing a semantic interpreter of a blueprint engine to generate the output based on the quantum information generated via performance of the at least one quantum circuit by the quantum computer wherein the blueprint engine acts as an interface between a user interface and the circuit LLM.

4. The method of any of the preceding claims, wherein a blueprint engine is configured to process the request information responsive to receipt of a request-35- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlcomprising the request information to generate an input vector that is provided as input to the circuit LLM.

5. The method of claim 4, wherein the request information comprises a data set, the blueprint engine extracts a sub-graph from the data set, generates the input vector representing the sub-graph, and provides the input vector representing the sub-graph to the circuit LLM.

6. The method of claim 5, wherein the circuit LLM generates a quantum topological data analysis (QTDA) circuit configured to perform QTDA of the sub-graph.

7. The method of claim 6, wherein the blueprint engine comprises a semantic interpreter configured to classify whether an anomaly is detected in the set based at least in part on the quantum information generated via performance of the QTDA circuit by the quantum computer.

8. The method of claim 6 or 7, wherein the quantum information generated via performance of the QTDA circuit by the quantum computer is an approximate Betti-k number for the sub-graph.

9. The method of claim 6, 7, or 8, wherein the blueprint engine is configured to execute a classifier configured to determine whether to perform QTDA on additional subgraphs from the data set and it is determined that the one or more stop criteria are satisfied responsive to the classifier determining that performance of QTDA on additional sub-graphs is not required and it is determined that the one or more stop criteria are not satisfied responsive to the classifier determining that performance of QTDA on additional sub-graphs is required.

10. The method of any of the preceding claims, wherein the request information provides a Hamiltonian description of a system and the output is at least one of an Eigen solution to the Hamiltonian description of the system or an Eigen value corresponding to the Eigen solution to the Hamiltonian description of the system.-36- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vl11. The method of claim 10, wherein a blueprint engine is configured to generate an ansatz based at least in part on the Hamiltonian description of the system and to cause the circuit LLM to generate the one or more quantum circuits based at least in part on the ansatz.

12. The method of claim 10 or 11, wherein the at least one of the circuit LLM or the solution parameter is updated in accordance with a quantum eigensolver protocol.

13. The method of claim 10, 11, or 12, wherein it is determined that the one or more stop criteria are satisfied responsive to determining that convergence of the Eigen value corresponding to the Eigen solution to the Hamiltonian description of the system satisfies convergence criteria and it is determined that the one or more stop criteria are not satisfied responsive to determining that the convergence of the Eigen value corresponding to the Eigen solution to the Hamiltonian description of the system does not satisfy convergence criteria.

14. The method of any of claims 10 to 13, wherein the Eigen solution to the Hamiltonian description of the system is a ground state or lowest cost state of the system and the Eigen value corresponding to the Eigen solution to the Hamiltonian description of the system is a ground state energy of the ground state or a cost corresponding to the lowest cost state.

15. The method of any of the preceding claims, further comprising receiving a request comprising the request information, wherein the request is received responsive to user interaction with a user interface.

16. The method of claim 15, further comprising prior to causing the circuit LLM to generate the one or more quantum circuits based at least in part on the request information, executing a resource estimation oracle configured to analyze the request information to determine whether to analyze the request using classical or hybrid quantum-classical techniques, and the circuit LLM is caused to generate the one or more quantum circuits responsive to the resource estimation oracle determining to analyze the request using hybrid quantum-classical techniques.-37- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vl17. The method of claim 16, wherein the resource estimation oracle comprises a machine learning trained model configured to estimate the quantum complexity of a request based on the request information thereof.

18. A classical computing entity comprising:one or more classical processing elements;a communication interface configured for communicating with a quantum computer; anda memory storing computer executable instructions, the computer executable instructions configured to, when executed by the one or more classical processing elements, cause the classical computing entity to at least perform:causing a circuit large language model (LLM) to generate one or more quantum circuits based at least in part on request information;causing performance of the one or more quantum circuits on a quantum computer;receiving quantum information generated via performance of the one or more quantum circuits on the quantum computer;determining whether one or more stop criteria are satisfied by the quantum information generated via performance of the one or more quantum circuits on the quantum computer;responsive to determining that the one or more stop criteria are satisfied, providing an output corresponding to the request information; andresponsive to determining that the one or more stop criteria are not satisfied: updating at least one of the circuit LLM or a solution parameter, causing one or more updated quantum circuits to be performed by the quantum computer, wherein the one or more updated quantum circuits are at least one of generated by the circuit LLM based at least in part on the solution parameter, andreceiving quantum information generated via performance of the one or more updated quantum circuits on the quantum computer.-38- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vl19. The classical computing entity of claim 18, wherein the classical computing entity further comprises a user interface configured for receiving user input and for providing the output in a human perceivable manner.

20. A system comprising:a classical computing entity comprising:one or more classical processing elements,a communication interface configured for communicating with a quantum computer, anda memory storing classical computer executable instructions, and a quantum computer comprising:a plurality of qubits,qubit manipulation elements,sensors, anda controller configured to control operation of the qubit manipulation elements and receive respective sensor signals generated by the sensors, wherein the classical computer executable instructions are configured to, when executed by the one or more classical processing elements, cause the classical computing entity to at least perform:causing a circuit large language model (LLM) to generate one or more quantum circuits based at least in part on request information;causing performance of the one or more quantum circuits on the quantum computer;receiving quantum information generated via performance of the one or more quantum circuits on the quantum computer;determining whether one or more stop criteria are satisfied by the quantum information generated via performance of the one or more quantum circuits on the quantum computer;responsive to determining that the one or more stop criteria are satisfied, providing an output corresponding to the request information; andresponsive to determining that the one or more stop criteria are not satisfied: updating at least one of the circuit LLM or a solution parameter of the one or more quantum circuits,-39- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vlcausing one or more updated quantum circuits to be performed by the quantum computer, wherein the one or more updated quantum circuits are at least one of generated by the circuit LLM based at least in part on the solution parameter, andreceiving quantum information generated via performance of the one or more quantum circuits on the quantum computer, andwherein the controller of the quantum computer is configured to perform:receiving the one or more quantum circuits,executing the one or more quantum circuits to generate the quantum information, wherein executing a quantum circuit of the one or more quantum circuits comprises controlling operation of the qubit manipulation elements to cause a controlled quantum state evolution of one or more qubits of the plurality of physical qubits in accordance with the quantum circuit, analyzing the respective sensor signals to determine resultant quantum states of the one or more qubits, and determining the quantum information based on the resultant quantum states of the one or more qubits, andproviding the quantum information configured for receipt by the classical computing entity.-40- QUK226817-WO 073374 / 642978 LEGAL02 / 48001939vl

Images