DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 10/10/2025 has been entered.
Response to Arguments
Applicant's arguments filed 09/09/2025 have been fully considered and they are persuasive.
Regarding applicant’s remarks directed to the rejection of claims under 35 USC § 103, the arguments are directed to newly amended limitations that were not previously examined by the examiner. Therefore, applicants arguments are rendered moot. The examiner refers to the rejection under 35 USC § 103 in the current office action for more details.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 1-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Pub. No. US20200118025A1 Romero et al. (“Romero”) in view of U.S. Pub. No. US20200342344A1 Gambetta et al. (“Gambetta”) in further view of Salm, Marie, et al. "The NISQ analyzer: automating the selection of quantum computers for quantum algorithms." (“Salm”)
In regards to claim 1,
Romero teaches A system comprising: a memory that stores computer executable components; a processor that executes at least one of the computer executable components that trains a machine learning model to perform a defined task, wherein the training comprises:
(Romero, “[0031] A variety of schemes may be used by embodiments of the present invention to use near-term quantum computers for assisting generative modeling tasks in classical machine learning. A first example of such a scheme uses quantum computers to assist in the realization of a Helmholtz machine. A second example of such a scheme uses quantum computers for generative adversarial networks (GAN).”)
Romero teaches quantum circuits for mapping a set of classical features to a quantum feature space
(Romero, “[0040] A process of encoding classical inputs in a quantum state may be interpreted as applying a nonlinear feature map that maps data to a quantum Hilbert space, a process also called quantum feature map or quantum encoding [mapping a set of classical features to a quantum feature space], as described by Schuld et al., “Quantum Machine Learning in Feature Hilbert Spaces,” arXiv:1803.07128 (2018). The quantum circuit [a quantum circuit] implementing this mapping on a digital quantum computer corresponds to the quantum feature circuit or encoding circuit 502. Classes of such quantum encoding include amplitude encoding and variational encoding.”)
Romero teaches training, using the quantum-enhanced features, the machine learning model to perform the defined task;
(Romero, “[0045] The quantum generator 400 learns how to improve the quality of subsequently generated data by its interaction with the quantum discriminator 510 (e.g., by receiving feedback in the form of output 512 indicating that the quantum discriminator 510 classified fake data 508 as real or as fake and improving subsequent data generation methods such that subsequent fake data 508 has an increased probability of being classified as real data by the quantum discriminator 510). The quantum discriminator 510, in turn, learns to distinguish between fake and real samples [training, using the quantum-enhanced features, the machine learning model to perform the defined task]. Note that the functionality of the quantum discriminator 510 could be provided by a classical neural network while the generator 400 would still be a quantum generator. The interaction between the generator 400 and discriminator 510 may be regulated via optimization of one or more GAN cost functions, which correspond to the probability of success of the discriminator 510. The cost function of the generator 40 may be the same as the cost function of the discriminator 510. The cost function of the generator 400 may be different from the cost function of the discriminator 510 and each may be optimized independently. The optimization seeks to minimize this cost function with respect to the parameters of the generator 400 and maximize it with respect to the parameters of the discriminator 510.”)
Romero teaches an output of the machine learning model in performing the defined task using the quantum-enhanced features
(Romero, “[0010] The device may further include a quantum discriminator receiving at least one quantum output state from a second quantum encoding circuit and measuring an observable in the received at least one quantum output state [an output of the machine learning model in performing the defined task using the quantum-enhanced features]. Measuring the observable may include measuring a designated qubit. The quantum discriminator may further include functionality for applying a measurement scheme to measure the designated qubit, wherein applying the measurement scheme to measure the designated qubit modifies a physical state of the measured designated qubit. The quantum discriminator may further include functionality for identifying a probability that the quantum output state belongs to the received classical distribution based on the measurement of the designated qubit. The quantum discriminator may further include functionality for identifying a probability that the received at least one quantum output state belongs to the received data xFake based on the measurement of the designated qubit. The quantum discriminator may further include functionality for evaluating a cost function representing a level of difference between the received input from the classical distribution and the received data xFake. The quantum discriminator may further include functionality for generating an optimized cost function representing a level of difference between the received input from the classical distribution and the received data xFake, wherein the received input further comprises a plurality of data points and the received data further comprises a plurality of data points. The quantum discriminator may further include functionality for applying a variational circuit to the received at least one quantum output state. The quantum discriminator may further include functionality for applying amplitude encoding to the received at least one quantum output state.”)
Romero teaches [executing the selected other quantum circuit, using the quantum computer or the quantum simulator based on a result of the determining associated with the selected other quantum circuit,] to map the set of classical features to the quantum feature space to produce updated quantum-enhanced features
(Romero, “[0040] A process of encoding classical inputs in a quantum state may be interpreted as applying a nonlinear feature map that maps data to a quantum Hilbert space, a process also called quantum feature map or quantum encoding [to map the set of classical features to the quantum feature space to produce updated quantum-enhanced features], as described by Schuld et al., “Quantum Machine Learning in Feature Hilbert Spaces,” arXiv:1803.07128 (2018). The quantum circuit [updated/selected quantum circuit (provided by Gambetta)] implementing this mapping on a digital quantum computer corresponds to the quantum feature circuit or encoding circuit 502. Classes of such quantum encoding include amplitude encoding and variational encoding.”)
Romero teaches and training, using the updated quantum-enhanced features, the machine learning model to perform the defined task;
(Romero, “[0045] The quantum generator 400 learns how to improve the quality of subsequently generated data by its interaction with the quantum discriminator 510 (e.g., by receiving feedback in the form of output 512 indicating that the quantum discriminator 510 classified fake data 508 as real or as fake and improving subsequent data generation methods such that subsequent fake data 508 has an increased probability of being classified as real data by the quantum discriminator 510). The quantum discriminator 510, in turn, learns to distinguish between fake and real samples [training, using the updated quantum-enhanced features, the machine learning model to perform the defined task]. Note that the functionality of the quantum discriminator 510 could be provided by a classical neural network while the generator 400 would still be a quantum generator. The interaction between the generator 400 and discriminator 510 may be regulated via optimization of one or more GAN cost functions, which correspond to the probability of success of the discriminator 510. The cost function of the generator 40 may be the same as the cost function of the discriminator 510. The cost function of the generator 400 may be different from the cost function of the discriminator 510 and each may be optimized independently. The optimization seeks to minimize this cost function with respect to the parameters of the generator 400 and maximize it with respect to the parameters of the discriminator 510.”)
Romero teaches [assessing the quantum circuit selection based on] another output of the machine learning model in performing the defined task using the updated quantum-enhanced features
(Romero, “[0010] The device may further include a quantum discriminator receiving at least one quantum output state from a second quantum encoding circuit and measuring an observable in the received at least one quantum output state [another output of the machine learning model in performing the defined task using the updated quantum-enhanced features]. Measuring the observable may include measuring a designated qubit. The quantum discriminator may further include functionality for applying a measurement scheme to measure the designated qubit, wherein applying the measurement scheme to measure the designated qubit modifies a physical state of the measured designated qubit. The quantum discriminator may further include functionality for identifying a probability that the quantum output state belongs to the received classical distribution based on the measurement of the designated qubit. The quantum discriminator may further include functionality for identifying a probability that the received at least one quantum output state belongs to the received data xFake based on the measurement of the designated qubit. The quantum discriminator may further include functionality for evaluating a cost function representing a level of difference between the received input from the classical distribution and the received data xFake. The quantum discriminator may further include functionality for generating an optimized cost function representing a level of difference between the received input from the classical distribution and the received data xFake, wherein the received input further comprises a plurality of data points and the received data further comprises a plurality of data points. The quantum discriminator may further include functionality for applying a variational circuit to the received at least one quantum output state. The quantum discriminator may further include functionality for applying amplitude encoding to the received at least one quantum output state.”)
However, Romero does not explicitly teach performing a quantum circuit selection of a quantum circuit from a set of quantum circuits
determining whether to execute the quantum circuit on a quantum computer or a quantum simulator based on at least one characteristic of the quantum circuit to minimize usage of quantum resources of the quantum computer while meeting a defined performance metric associated with the quantum circuit selection, wherein the determining comprises: evaluating the at least one characteristic of the quantum circuit with respect to executing the quantum circuit on the quantum computer and executing the quantum circuit on the quantum simulator, and based on the evaluation associated with the quantum circuit satisfying a first criterion, selecting the quantum computer for execution of the quantum circuit, or based on the evaluation associated with the quantum circuit satisfying a second criterion, selecting the quantum simulator for execution of the quantum circuit; executing the quantum circuit, using the quantum computer or the quantum simulator based on a result of the determining associated with the quantum circuit
assessing the quantum circuit selection.. and iteratively performing until a result of the assessing of the quantum circuit selection meets a defined criterion.. based on the result of the assessing, performing the quantum circuit selection of another quantum circuit from the set of quantum circuits for mapping the set of classical features to the quantum feature space, wherein the other quantum circuit is different from quantum circuits previously selected
determining whether to execute the selected other quantum circuit on the quantum computer or the quantum simulator based on at least one characteristic of the selected other quantum circuit to minimize the usage of the quantum resources while meeting the defined performance metric associated with the quantum circuit selection, wherein the determining comprises: evaluating the at least one characteristic of the selected other quantum circuit with respect to executing the selected other quantum circuit on the quantum computer and executing the quantum circuit on the quantum simulator, and based on the evaluation associated with the selected other quantum circuit satisfying the first criterion, selecting the quantum computer for execution of the selected other quantum circuit, or based on the evaluation associated with the selected other quantum circuit satisfying the second criterion, selecting the quantum simulator for execution of the selected other quantum circuit; executing the selected other quantum circuit, using the quantum computer or the quantum simulator based on a result of the determining associated with the selected other quantum circuit…
assessing the quantum circuit selection…
Gambetta teaches performing a quantum circuit selection of a quantum circuit from a set of quantum circuits [for mapping a set of classical features to a quantum feature space (wherein this mapping is taught by Romero)];
(Gambetta, “[0053] If only one pattern matches a portion of the quantum circuit, an embodiment selects the transformation operation corresponding to that input circuit pattern. If more than one pattern matches a portion of the quantum circuit [from a set of quantum circuits; ie alternative transformations of the provided quantum circuit], one embodiment selects the transformation operation corresponding to the largest input circuit pattern. Another embodiment selects the transformation operation corresponding to the smallest input circuit pattern. Another embodiment selects the transformation [performing a quantum circuit selection of a quantum circuit; wherein the quantum circuit transformation with the most improved efficiency score is selected] that produced the most improved efficiency score when used to transform a previous quantum circuit. Another embodiment selects the transformation operation corresponding to the input circuit pattern that best matches the portion. Other methods of selecting a transformation operation are also possible and contemplated within the scope of the illustrative embodiments.”)
Gambetta teaches while meeting a defined performance metric associated with the quantum circuit selection
(Gambetta, [0025], “Optimizing refers to refining a quantum circuit so that its execution incurs lower cost, typically by taking less time to execute, while maintaining accuracy [while meeting a defined performance metric associated with the quantum circuit selection; wherein the optimized quantum circuit is provided to Salm, thus, Salm uses the optimized quantum circuit with the maintained performance metric (accuracy)]—e.g., by using a configuration of different or fewer qubits to minimize interference or decoherence, or some combination thereof. As used herein, accuracy refers to consistency with previous results of a quantum circuit. Accuracy is synonymous with correctness.”)
Gambetta teaches assessing the quantum circuit selection based on an output [of the machine learning model in performing the defined task using the quantum-enhanced features (wherein the output is taught by Romero based on the given circuit of Gambetta)],
(Gambetta, “[0109] With reference to FIG. 7, this figure depicts another example of evaluating quantum circuit optimizations using machine learning in accordance with an illustrative embodiment. The depicted transformations can be performed using quantum circuit transpiler 314 in FIG. 3. Quantum processor 340 is the same as quantum processor 340 in FIG. 3.
[0110] In the depicted example, application 314 has transformed section 720 of quantum circuit 710 into each of section 730, section 740, and section 750. Application 300 has executed circuit 710, generating original result 712. Application 300 has also executed the quantum circuits resulting from implementing sections 730, 740, and 750, generating results 732, 742, and 752 respectively [assessing the quantum circuit selection ie results from executing the quantum circuits in the GAN of Romero]. Application 314 has also determined the correctness of the transformed quantum circuit by comparing each of results 732, 742, and 752 with original result 712. Application 314 has also determined an efficiency score for each of original result 712 and results 732, 742, and 752.
[0111] Here, the application's evaluation indicates that the circuit generating result 732 is equivalent to original result 712, but includes more gates than circuit 710. If the application determines efficiency solely by circuit size, the circuit generating result 732 is less efficient than circuit 710. However, if the application determines efficiency by execution time, the circuit generating result 732 may execute faster than circuit 710 because of the difference in qubit interactions. The application's evaluation also indicates that the circuit generating result 742 is equivalent to original result 712 and includes fewer gates than circuit 710. Here, the application has also determined that the circuit generating result 742 executes faster than circuit 710. The application's evaluation also indicates that the circuit generating result 652 is incorrect.”)
PNG
media_image1.png
460
637
media_image1.png
Greyscale
Gambetta teaches and iteratively performing until a result of the assessing of the quantum circuit selection meets a defined criterion:
(Gambetta, “[0064] An embodiment continues in this fashion, performing and evaluating transformation operations, until an end criterion is reached [iteratively performing until a result of the assessing of the quantum circuit selection meets a defined criterion]. In one embodiment, an end criterion is an efficiency score above a particular threshold. For example, an initial constraint on the transpilation process may have been to reduce the depth of the quantum circuit to a specified depth. Once the depth is at or below the specified depth, the circuit has been sufficiently transformed and transformation can end.
[0065] In another embodiment, an end criterion is an efficiency score improvement above a particular threshold. For example, an initial constraint on the transpilation process may have been to reduce the number of gates in the quantum circuit by ten percent. Once the number of gates has been reduced by ten percent or more, the circuit has been sufficiently transformed and transformation can end.
[0066] In another embodiment, an end criterion is the execution of a particular number of transformation operations. In another embodiment, an end criterion is a cumulative use of computing resources above a threshold amount. Execution limits such as these prevent a transformation process from executing infinitely, attempting to reach an unreachable state. Other end criteria, or combinations of end criteria, are also possible and contemplated within the scope of the illustrative embodiments.”)
Gambetta teaches based on the result of the assessing, performing the quantum circuit selection of another quantum circuit from the set of quantum circuits [for mapping the set of classical features to the quantum feature space (wherein the mapping is taught by Romero)], wherein the other quantum circuit is different from quantum circuits previously selected;
(Gambetta, “[0103] Transformation library manager 460 stores transformation operations in an operation library. Module 460 also removes obsolete or deprecated operations from the library, and replaces operations with more successful operations [based on the result of the assessing ie identified more successful operations, performing the quantum circuit selection of another quantum circuit from the set of quantum circuits for mapping the set of classical features to the quantum feature space, wherein the other quantum circuit is different from quantum circuits previously selected; wherein Gambetta teaches an iterative process with an end criterion and a library manager that obsoletes previous transformations]. Module 460 also maintains, for each operation in the operation library, an input circuit pattern to which the operation applies. Module 460 also maintains, for each operation in the operation library, processor configuration dependency information for the operation. When new configuration and calibration information is received, module 460 checks the new information against processor configuration dependency information for transformation operations in the library. If a transformation operation no longer applies to the new configuration, the embodiment removes that transformation operation from the library or marks that transformation operation as deprecated and not to be used.”)
Gambetta teaches assessing the quantum circuit selection based on another output [of the machine learning model in performing the defined task using the updated quantum-enhanced features]
(Gambetta, “[0109] With reference to FIG. 7, this figure depicts another example of evaluating quantum circuit optimizations using machine learning in accordance with an illustrative embodiment. The depicted transformations can be performed using quantum circuit transpiler 314 in FIG. 3. Quantum processor 340 is the same as quantum processor 340 in FIG. 3.
[0110] In the depicted example, application 314 has transformed section 720 of quantum circuit 710 into each of section 730, section 740, and section 750. Application 300 has executed circuit 710, generating original result 712. Application 300 has also executed the quantum circuits resulting from implementing sections 730, 740, and 750, generating results 732, 742, and 752 respectively [assessing the quantum circuit selection ie results from executing the quantum circuits in the GAN of Romero]. Application 314 has also determined the correctness of the transformed quantum circuit by comparing each of results 732, 742, and 752 with original result 712. Application 314 has also determined an efficiency score for each of original result 712 and results 732, 742, and 752.
[0111] Here, the application's evaluation indicates that the circuit generating result 732 is equivalent to original result 712, but includes more gates than circuit 710. If the application determines efficiency solely by circuit size, the circuit generating result 732 is less efficient than circuit 710. However, if the application determines efficiency by execution time, the circuit generating result 732 may execute faster than circuit 710 because of the difference in qubit interactions. The application's evaluation also indicates that the circuit generating result 742 is equivalent to original result 712 and includes fewer gates than circuit 710. Here, the application has also determined that the circuit generating result 742 executes faster than circuit 710. The application's evaluation also indicates that the circuit generating result 652 is incorrect.”)
However, Gambetta does not explicitly teach determining whether to execute the quantum circuit on a quantum computer or a quantum simulator based on at least one characteristic of the quantum circuit to minimize usage of quantum resources of the quantum computer [while meeting a defined performance metric associated with the quantum circuit selection], wherein the determining comprises: evaluating the at least one characteristic of the quantum circuit with respect to executing the quantum circuit on the quantum computer and executing the quantum circuit on the quantum simulator, and based on the evaluation associated with the quantum circuit satisfying a first criterion, selecting the quantum computer for execution of the quantum circuit, or based on the evaluation associated with the quantum circuit satisfying a second criterion, selecting the quantum simulator for execution of the quantum circuit; executing the quantum circuit, using the quantum computer or the quantum simulator based on a result of the determining associated with the quantum circuit; determining whether to execute the selected other quantum circuit on the quantum computer or the quantum simulator based on at least one characteristic of the selected other quantum circuit to minimize the usage of the quantum resources [while meeting the defined performance metric associated with the quantum circuit selection], wherein the determining comprises: evaluating the at least one characteristic of the selected other quantum circuit with respect to executing the selected other quantum circuit on the quantum computer and executing the quantum circuit on the quantum simulator, and based on the evaluation associated with the selected other quantum circuit satisfying the first criterion, selecting the quantum computer for execution of the selected other quantum circuit, or based on the evaluation associated with the selected other quantum circuit satisfying the second criterion, selecting the quantum simulator for execution of the selected other quantum circuit; executing the selected other quantum circuit, using the quantum computer or the quantum simulator based on a result of the determining associated with the selected other quantum circuit
Salm teaches determining whether to execute the quantum circuit on a quantum computer or a quantum simulator
(Salm, Section 3, “In this section, we introduce our concept of a NISQ Analyzer, which enables an automated analysis and selection of quantum algorithm implementations and quantum computers depending on the chosen quantum algorithm and input data. Fig. 1 depicts an overview of the approach. First, a quantum algorithm is selected. Second, implementations of the algorithm are analyzed and selected. Third, quantum computers are analyzed and selected. Finally, the selected implementation is executed on the suitable quantum computer [determining whether to execute the quantum circuit on a quantum computer or a quantum simulator].”)
Salm teaches based on at least one characteristic of the quantum circuit to minimize usage of quantum resources of the quantum computer [while meeting a defined performance metric associated with the quantum circuit selection], wherein the determining comprises: evaluating the at least one characteristic of the quantum circuit with respect to executing the quantum circuit on the quantum computer and executing the quantum circuit on the quantum simulator, and based on the evaluation associated with the quantum circuit satisfying a first criterion, selecting the quantum computer for execution of the quantum circuit, or based on the evaluation associated with the quantum circuit satisfying a second criterion, selecting the quantum simulator for execution of the quantum circuit;
(Salm, Section 3.3, “The general rule for selecting a suitable quantum computer for a particular implementation is defined as follows: [satisfying a first/second criterion]
PNG
media_image2.png
114
677
media_image2.png
Greyscale
Thereby, x is a quantum computer of the set of available quantum computers Q, e.g., IBMQ 16. y is an implementation of the set of remaining implementations R ✓ I. Qubits(q0, x) defines the provided number of qubits q0 of x. Qubits(q1, y) defines the required number of qubits, or the width, q1 of y. GreaterEquals(q0, q1) defines that ”q0 q1” to ensure that the quantum computer x does not have less qubits than required by y. Depth(d0, x) defines the maximum depth d0 executable by x. Depth(d1, y) defines the depth d1 of the transpiled circuit of y. GreaterEquals(d0, d1) defines that ”d0 d1”, such that the maximum executable depth of the quantum computer x is not smaller than required by the implementation y [based on at least one characteristic ie number of qubits (width) and depth of the quantum circuit (wherein the quantum circuit is an optimized quantum circuit provided by Gambetta) to minimize usage of quantum resources of the quantum computer]. Furthermore, the SDK s 2 S, e.g. Qiskit, used by the implementation, defined by Sdk(y, s), must also support the selected quantum computer, defined by Sdk(x, s), to ensure their compatibility. This all is true, if and only if Executable(y, x) is true [wherein the determining comprises: evaluating the at least one characteristic of the quantum circuit with respect to executing the quantum circuit on the quantum computer and executing the quantum circuit on the quantum simulator]. In the example in Fig. 1, IBMQ 16 can execute Shor15-Qiskit. If more than one executable implementation remains, the user decides which one to execute. Furthermore, the user also decides, in case, more than one quantum computer can execute the chosen implementation [selecting the quantum computer for execution of the quantum circuit].”)
PNG
media_image3.png
591
823
media_image3.png
Greyscale
Salm teaches executing the quantum circuit, using the quantum computer or the quantum simulator based on a result of the determining associated with the quantum circuit, [to map the set of classical features to the quantum feature space to produce quantum-enhanced features];
(Salm, Section 3.4, “In the (4) Execution phase, the selected implementation is executed by the selected quantum computer [executing the quantum circuit, using the quantum computer or the quantum simulator based on a result of the determining associated with the quantum circuit], as seen in Fig. 1.”)
Salm teaches determining whether to execute the selected other quantum circuit on the quantum computer or the quantum simulator
(Salm, Section 3, “In this section, we introduce our concept of a NISQ Analyzer, which enables an automated analysis and selection of quantum algorithm implementations and quantum computers depending on the chosen quantum algorithm and input data. Fig. 1 depicts an overview of the approach. First, a quantum algorithm is selected. Second, implementations of the algorithm are analyzed and selected. Third, quantum computers are analyzed and selected. Finally, the selected implementation is executed on the suitable quantum computer [determining whether to execute the selected other quantum circuit on the quantum computer or the quantum simulator].”)
Salm teaches based on at least one characteristic of the selected other quantum circuit to minimize the usage of the quantum resources [while meeting the defined performance metric associated with the quantum circuit selection], wherein the determining comprises: evaluating the at least one characteristic of the selected other quantum circuit with respect to executing the selected other quantum circuit on the quantum computer and executing the quantum circuit on the quantum simulator, and based on the evaluation associated with the selected other quantum circuit satisfying the first criterion, selecting the quantum computer for execution of the selected other quantum circuit, or based on the evaluation associated with the selected other quantum circuit satisfying the second criterion, selecting the quantum simulator for execution of the selected other quantum circuit;
(Salm, Section 3.3, “The general rule for selecting a suitable quantum computer for a particular implementation is defined as follows: [satisfying the first/second criterion]
PNG
media_image2.png
114
677
media_image2.png
Greyscale
Thereby, x is a quantum computer of the set of available quantum computers Q, e.g., IBMQ 16. y is an implementation of the set of remaining implementations R ✓ I. Qubits(q0, x) defines the provided number of qubits q0 of x. Qubits(q1, y) defines the required number of qubits, or the width, q1 of y. GreaterEquals(q0, q1) defines that ”q0 q1” to ensure that the quantum computer x does not have less qubits than required by y. Depth(d0, x) defines the maximum depth d0 executable by x. Depth(d1, y) defines the depth d1 of the transpiled circuit of y. GreaterEquals(d0, d1) defines that ”d0 d1”, such that the maximum executable depth of the quantum computer x is not smaller than required by the implementation y [based on at least one characteristic ie number of qubits (width) and depth of the selected other quantum circuit to minimize usage of quantum resources of the quantum computer]. Furthermore, the SDK s 2 S, e.g. Qiskit, used by the implementation, defined by Sdk(y, s), must also support the selected quantum computer, defined by Sdk(x, s), to ensure their compatibility. This all is true, if and only if Executable(y, x) is true [wherein the determining comprises: evaluating the at least one characteristic of the selected other quantum circuit with respect to executing the selected other quantum circuit on the quantum computer and executing the selected other quantum circuit on the quantum simulator]. In the example in Fig. 1, IBMQ 16 can execute Shor15-Qiskit. If more than one executable implementation remains, the user decides which one to execute. Furthermore, the user also decides, in case, more than one quantum computer can execute the chosen implementation [selecting the quantum computer for execution of the selected other quantum circuit].”)
Salm teaches executing the selected other quantum circuit, using the quantum computer or the quantum simulator based on a result of the determining associated with the selected other quantum circuit, [to map the set of classical features to the quantum feature space to produce updated quantum-enhanced features,]
(Salm, Section 3.4, “In the (4) Execution phase, the selected implementation is executed by the selected quantum computer [executing the selected other quantum circuit, using the quantum computer or the quantum simulator based on a result of the determining associated with the selected other quantum circuit], as seen in Fig. 1.”)
Gambetta is considered to be analogous to the claimed invention because they are in the same field of optimized quantum circuit selection. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Romero and Salm to incorporate the teachings of Gambetta in order to provide a method of optimizations utilizing transformations and a library manager to recognize patterns to improve future quantum circuits (Gambetta, “[0028] The illustrative embodiments also recognize that optimizations are often pattern-based. In other words, a portion of a quantum circuit that matches a particular pattern can often be improved by replacing that particular portion with a known replacement. For example, one particular configuration of three CNOT gates is known to be equivalent to another particular configuration of two CNOT gates. As a result, if a portion of a quantum circuit matches the particular three-CNOT configuration, the portion can be replaced with the equivalent two-CNOT configuration. Removing a gate from the circuit in this manner often improves circuit efficiency. Thus, the illustrative embodiments recognize that if a pattern-based circuit change can be found useful, the change should be learned from, to improve future quantum circuits.”)
Salm is considered to be analogous to the claimed invention because they are in the same field of selecting appropriate quantum hardware for execution. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Romero and Gambetta to incorporate the teachings of Salm in order to provide a practical tool for automated analysis and selection of implementations of quantum algorithms and appropriate quantum computers that can execute a selected implementation with a certain input data (Salm, Abstract, Quantum computing can enable a variety of breakthroughs in research and industry in the future. Although some quantum algorithms already exist that show a theoretical speedup compared to the best known classical algorithms, the implementation and execution of these algorithms come with several challenges. The input data determines, for example, the required number of qubits and gates of a quantum algorithm. A quantum algorithm implementation also depends on the used Software Development Kit which restricts the set of usable quantum computers. Because of the limited capabilities of current quantum computers, choosing an appropriate one to execute a certain implementation for a given input is a dicult challenge that requires immense mathematical knowledge about the implemented quantum algorithm as well as technical knowledge about the used Software Development Kits. In this paper, we present a concept for the automated analysis and selection of implementations of quantum algorithms and appropriate quantum computers that can execute a selected implementation with a certain input data. The practical feasibility of the concept is demonstrated by the prototypical implementation of a tool that we call NISQ Analyzer.”)
In regards to claim 2,
Romero in view of Gambetta and Salm teaches The system of claim 1,
Gambetta teaches wherein the quantum circuit selection is based on respective complexities of the quantum circuits in the set.
(Gambetta, “[0038] An embodiment determines a size of the quantum circuit. One embodiment determines quantum circuit size using the total number of gates in the circuit. Another embodiment determines quantum circuit size using the depth of the circuit. Other size measurements are also possible and contemplated within the scope of the illustrative embodiments. Although circuit size is one measure of efficiency, a smaller circuit is not always a more efficient circuit. For example, due to the configuration of qubits within a quantum processor, one quantum circuit may take longer to execute on that processor than a larger quantum circuit that is better adapted to the qubit configuration of that processor. As well, one more complex gate can be replaced by a configuration of simpler, easier-to-implement gates. For example, a SWAP gate, which exchanges the states of a pair of qubits, can be deconstructed into three CNOT gates.
[0039] An embodiment converts one or more of an execution efficiency and a size of the quantum circuit to obtain an efficiency score for the quantum circuit [quantum circuit selection is based on respective complexities of the quantum circuits in the set]. One conversion technique includes normalizing each of the execution time, and the size of the quantum circuit to the same scale (e.g. 0 to 1) and computing a weighted average of the normalized values. An embodiment can be configured with different weights for each factor in the weighted average to weight one particular factor more highly than another. Other techniques of computing an efficiency score are also possible and contemplated within the scope of the illustrative embodiments.”)
In regards to claim 3,
Romero in view of Gambetta and Salm teaches The system of claim 2,
Gambetta teaches wherein the respective complexities are based at least on respective widths of the quantum circuits in the set.
(Gambetta, “[0038] An embodiment determines a size of the quantum circuit. One embodiment determines quantum circuit size using the total number of gates in the circuit [respective complexities are based at least on respective widths of the quantum circuits in the set]. Another embodiment determines quantum circuit size using the depth of the circuit. Other size measurements are also possible and contemplated within the scope of the illustrative embodiments. Although circuit size is one measure of efficiency, a smaller circuit is not always a more efficient circuit. For example, due to the configuration of qubits within a quantum processor, one quantum circuit may take longer to execute on that processor than a larger quantum circuit that is better adapted to the qubit configuration of that processor. As well, one more complex gate can be replaced by a configuration of simpler, easier-to-implement gates. For example, a SWAP gate, which exchanges the states of a pair of qubits, can be deconstructed into three CNOT gates.
[0039] An embodiment converts one or more of an execution efficiency and a size of the quantum circuit to obtain an efficiency score for the quantum circuit. One conversion technique includes normalizing each of the execution time, and the size of the quantum circuit to the same scale (e.g. 0 to 1) and computing a weighted average of the normalized values. An embodiment can be configured with different weights for each factor in the weighted average to weight one particular factor more highly than another. Other techniques of computing an efficiency score are also possible and contemplated within the scope of the illustrative embodiments.”)
In regards to claim 4,
Romero in view of Gambetta and Salm teaches The system of claim 2,
Gambetta teaches wherein the respective complexities are based at least on respective depths of the quantum circuits in the set
(Gambetta, “[0038] An embodiment determines a size of the quantum circuit. One embodiment determines quantum circuit size using the total number of gates in the circuit. Another embodiment determines quantum circuit size using the depth of the circuit [respective complexities are based at least on respective depths of the quantum circuits in the set]. Other size measurements are also possible and contemplated within the scope of the illustrative embodiments. Although circuit size is one measure of efficiency, a smaller circuit is not always a more efficient circuit. For example, due to the configuration of qubits within a quantum processor, one quantum circuit may take longer to execute on that processor than a larger quantum circuit that is better adapted to the qubit configuration of that processor. As well, one more complex gate can be replaced by a configuration of simpler, easier-to-implement gates. For example, a SWAP gate, which exchanges the states of a pair of qubits, can be deconstructed into three CNOT gates.
[0039] An embodiment converts one or more of an execution efficiency and a size of the quantum circuit to obtain an efficiency score for the quantum circuit. One conversion technique includes normalizing each of the execution time, and the size of the quantum circuit to the same scale (e.g. 0 to 1) and computing a weighted average of the normalized values. An embodiment can be configured with different weights for each factor in the weighted average to weight one particular factor more highly than another. Other techniques of computing an efficiency score are also possible and contemplated within the scope of the illustrative embodiments.”)
In regards to claim 5,
Romero in view of Gambetta and Salm teaches The system of claim 1,
Romero teaches wherein the machine learning model comprises at least one of a generative adversarial network model or a reinforcement learning model.
(Romero, “[0031] A variety of schemes may be used by embodiments of the present invention to use near-term quantum computers for assisting generative modeling tasks in classical machine learning. A first example of such a scheme uses quantum computers to assist in the realization of a Helmholtz machine. A second example of such a scheme uses quantum computers for generative adversarial networks (GAN) [a generative adversarial network model].”)
In regards to claim 6,
Romero in view of Gambetta and Salm teaches The system of claim 1,
Gambetta teaches wherein the quantum circuit selection is based on randomly introducing new quantum circuit elements.
(Gambetta, “[0055] To generate a transformation, an embodiment uses a pseudorandom number generator. One embodiment uses a pseudorandom number generator to generate variations on known transformations (e.g. by substituting a randomly-chosen gate for an existing gate in the circuit). Generating transformation operations randomly allows for the discovery of new transformation operations that may be better that already-known transformation operations [based on randomly introducing new quantum circuit elements]. A better transformation operation is one that produces a more efficient quantum circuit, as measured by the efficiency score, than a transformation operation currently in the library.”)
In regards to claim 7,
Romero in view of Gambetta and Salm teaches The system of claim 1,
Gambetta teaches wherein the quantum circuit selection comprises carrying out a selection optimization that decreases requirements placed on the quantum resources through additional penalty terms that help minimize at least one of required number of qubits, qubit connectivity, fidelity, number of quantum gates, number of quantum control gates, number of multi-qubit gates, or size of a Hilbert space.
(Gambetta, “[0040] An embodiment attempts to improve the quantum circuit, as measured by the efficiency score [additional penalty terms], by performing a transformation operation on the circuit. A transformation operation specifies one or more transformations to be performed on a quantum circuit to reconfigure the circuit into a different, but equivalent, quantum circuit. In other words, a transformation reconfigures a gate in the first quantum circuit such that a qubit used in the gate complies with a constraint on the quantum circuit design while participating in the first quantum circuit.
[0041] Reducing the number of gates in a quantum circuit improves circuit efficiency in many cases [minimize… number of quantum gates]. Thus, one example transformation operation rearranges target and control inputs of gates, according to specific transformation rules, to remove redundant gates from a circuit.”)
Claims 8 and 15 are substantially similar, and thus rejected on the same rationale under 35 U.S.C. 103, as claim 1.
Claims 9 and 16 are substantially similar, and thus rejected on the same rationale under 35 U.S.C. 103, as claim 2.
Claims 10 and 17 are substantially similar, and thus rejected on the same rationale under 35 U.S.C. 103, as claim 3.
Claims 11 and 18 are substantially similar, and thus rejected on the same rationale under 35 U.S.C. 103, as claim 4.
Claims 12 and 19 are substantially similar, and thus rejected on the same rationale under 35 U.S.C. 103, as claim 5.
Claims 13 and 20 are substantially similar, and thus rejected on the same rationale under 35 U.S.C. 103, as claim 6.
Claims 14 is substantially similar, and thus rejected on the same rationale under 35 U.S.C. 103, as claim 7.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
NPL: Weder B, Barzen J, Leymann F, Salm M. Automated quantum hardware selection for quantum workflows. Electronics. 2021 Apr 20;10(8):984
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JASMINE THAI whose telephone number is (703)756-5904. The examiner can normally be reached M-F 8-4.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Michael Huntley can be reached at (303) 297-4307. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/J.T.T./Examiner, Art Unit 2129
/MICHAEL J HUNTLEY/Supervisory Patent Examiner, Art Unit 2129