Notice of Pre-AIA or AIA Status
1. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
DETAILED ACTION
2. This Office Action is sent in response to Applicant’s Communication received on 02/17/2026 for application number 18/174,328.
Response to Amendments
3. The Amendment filed 02/17/2026 has been entered. Claims 1-20 remain pending in the application.
Response to Arguments
Applicant argues that Dukatz discloses only a single machine learning model rather than a claimed “first” and “second” machine learning model. However, Examiner respectfully disagrees and notes that claim 1 does not require that the first and second models have different architectures, different code, or independent frameworks. Under the BRI, the first and second machine learning models reasonably encompass different trained instances, states, versions, or logical model paths of the same predictive framework, distinguished by their training source. Dukatz teaches a predictive machine learning model trained using historical execution-based training examples and further teaches quantum simulators, quantum gate computers, and quantum annealers as candidate execution environments. Thus, Dukatz teaches the claimed first and second machine learning models.
Applicant argues that Dukatz does not disclose separate simulation-derived and hardware-derived training datasets. However, Examiner respectfully disagrees and notes that Dukatz teaches obtaining historical data for previously performed computational tasks, including the type of computing resource to which each task was routed, and expressly teaches that previously performed simulation tasks may have been routed to quantum simulators, while other previously performed tasks may have been routed to quantum gate computers or quantum annealers. Dukatz also teaches generating training samples using prior task data, routed-resource identifiers, and execution-related properties associated with those resources.
Applicant argues that Dukatz does not disclose “estimating … respective values” for the same quantum circuit using two models, and does not disclose “comparing the estimates” before orchestration. However, Examiner respectfully disagrees and notes that Dukatz teaches that routing is based on execution-related properties such as approximate solution quality, computational time, computational cost, error tolerance, and confidence level, and teaches that the machine learning model determines which resource to route the task to base on such considerations. Dukatz also explains that the system balances tradeoffs among available resources and learns optimal routings. It is noted that these teachings correspond to estimating respective execution-related values for candidate environments, comparing those values in selecting among the candidate routes, and then routing the task accordingly.
Applicant argues that identifying a quantum simulator and quantum hardware does not satisfy the claimed orchestration framework. However, Examiner respectfully disagrees and notes that Dukatz does not rely solely on the existence of two different resource types. Rather, it relies on Dukatz’s teaching of simulator and hardware resources, training from historical executions associated with those resources, execution-related performance properties used by the predictive model, and routing the received task to the determined resource. Thus, Dukatz teaches orchestrating the task to either the simulation engine or the quantum hardware for execution.
Claim Rejections - 35 USC § 102
4. The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
5. Claims 1-4, 6, 8, 10-12, 14-16, 18, and 20 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Dukatz et al. (U.S. Patent Application Pub. No. US 20180308000 A1).
Claim 1: Dukatz teaches a method, comprising:
providing quantum circuit information, concerning a quantum circuit, to a first machine learning model (i.e. the system may train the machine learning model to route received computational tasks by generating a set of training examples using the (i) first set of data as described above with reference to step 302, (ii) the input data, as described above with reference to step 304, and (iii) the second set of data, as described above with reference to step 306. Each training example in the set of training examples may include a machine learning model input paired with a known machine learning model output. In some cases the training examples may be initially generated from combining data from independent quantum computing resource execution logs and classical computing resource execution logs; para. [0112]) that has been trained with first training data (i.e. The machine learning module 132 is configured to provide data representing the computational task or computational sub tasks to the trained machine learning model 204. The machine learning model 204 is configured to process the received data and to determine which of the one or more additional computing resources 110 a-110 d to route the received data representing the computational task or sub tasks to; para. [0088]) generated as a result of execution of a group of quantum circuits (i.e. the machine learning model 204 may determine that a received simulation task should be routed to a quantum simulator, e.g., quantum simulator 110 c. In this example, during operation (B), the machine learning model 204 may provide the quantum simulator 110 c with instructions for performing the simulation task; para. [0089, 0112]) on a simulation engine (i.e. The additional computing resources 110 a-110 d may include one or more quantum simulators, e.g., quantum simulator 110 c. A quantum simulator is a quantum computer that may be programmed to simulate other quantum systems and their properties; para. [0060]);
providing the quantum circuit information (i.e. The database 206 is configured to store data representing properties associated with using the one or more additional computing resources 110 a-110 d, e.g., one or more quantum computing resources, to solve the multiple computational tasks; para. [0084]) to a second machine learning model that has been trained with second training data generated as a result of execution of the group of quantum circuits (i.e. the system may train the machine learning model to route received computational tasks by generating a set of training examples using the (i) first set of data as described above with reference to step 302, (ii) the input data, as described above with reference to step 304, and (iii) the second set of data, as described above with reference to step 306. Each training example in the set of training examples may include a machine learning model input paired with a known machine learning model output. In some cases the training examples may be initially generated from combining data from independent quantum computing resource execution logs and classical computing resource execution logs; para. [0112]) on quantum hardware (i.e. The additional computing resources 110 a-110 d may include one or more quantum gate processors, e.g., quantum gate processor 110 b. A quantum gate processor includes one or more quantum circuits, i.e., models for quantum computation in which a computation is performed using a sequence of quantum logic gates, operating on a number of qubits (quantum bits); para. [0059]);
estimating, by the first machine learning model and the second machine learning model, respective values (i.e. The machine learning model 204 is a predictive model that may be trained to perform one or more machine learning tasks, e.g., classification tasks; para. [0082]) of a quantum circuit execution parameter of the quantum circuit (i.e. During operation (C), the machine learning module 132 is configured to receive data representing a solution to computational task, e.g., data 210, and data representing properties of using the corresponding computing resource to solve the computational task, e.g., data 212. For example, data representing properties of using the corresponding computing resource to solve the computational task may include data representing an approximate quality of the generated solution, a computational time associated with the generated solution, or a computational cost associated with the generated solution; para. [0090, 0113]);
comparing the estimates of the quantum circuit execution parameter (i.e. The system processes the received data using a machine learning model to determine which of the one or more quantum computing resources or the one or more classical computing resources to route the data representing the computational task to (step 404); para. [0116]); and
based on the comparing (i.e. A quantum computing machine learning module, as described in this specification, balances this tradeoff and learns optimal routings of computational tasks to classical or quantum computing resources. By learning when and how to utilize the power of quantum computing, a system implementing the quantum computing machine learning module may perform computational tasks more efficiently and/or accurately compared to systems that do not include quantum computing resources, or to systems that do not learn optimal routings of computational tasks to classical or quantum resources; para. [0026]), orchestrating the quantum circuit to either the simulation engine, or the quantum hardware, for execution (i.e. During operation (B), the machine learning model 204 is configured to provide the determined additional computing resource or resources with instructions for performing the respective computational task or computational sub tasks, e.g., data 208. For example, the machine learning model 204 may determine that a received optimization task should be routed to a quantum annealer, e.g., quantum annealer 110 a. In this example, during operation (B), the machine learning model 204 may provide the quantum annealer 110 a with instructions for performing the optimization task. As another example, the machine learning model 204 may determine that a received simulation task should be routed to a quantum simulator, e.g., quantum simulator 110 c. In this example, during operation (B), the machine learning model 204 may provide the quantum simulator 110 c with instructions for performing the simulation task; para. [0089]).
Claim 2: Dukatz teaches the method as recited in claim 1. Dukatz further teaches wherein a choice to orchestrate the quantum circuit to either the simulation engine, or the quantum hardware, is based on a parameter of a service level agreement (i.e. An error tolerance associated with a computational task may be used to determine which computing resource to route the computational task to. For example, some computational tasks may have smaller error tolerances than others, e.g., an error tolerance of a solution to the task of optimizing a cancer radiotherapy treatment may be smaller than an error tolerance of a solution to the task of optimizing the wastage of water in a water network. Computational tasks with smaller error tolerances may therefore be routed to computing resources that are more accurate than other computing resources, e.g., to computing resources who are less likely to introduce errors when performing a computational task; para. [0102, 0104]).
Claim 3: Dukatz teaches the method as recited in claim 1. Dukatz further teaches wherein the group of quantum circuits comprises one or more randomly generated quantum circuits and/or one or more quantum circuits configured to solve a particular problem (i.e. This specification provides systems and methods for determining when and how to leverage quantum computing devices when solving computational tasks. A system can receive computational tasks, e.g., optimization tasks, to be performed. For example, the system may be an optimization engine that is configured to receive input data and to generate, as output, an optimal solution to an optimization task based on the received input data. The received input data can include static and real-time data; para. [0037]).
Claim 4: Dukatz teaches the method as recited in claim 1. Dukatz further teaches wherein the estimates comprise an estimate for a time of execution of the quantum circuit on the quantum hardware, and an estimate for a time of execution of the quantum circuit on the simulation engine (i.e. computational runtimes associated with solutions generated by the one or more quantum computing resources; para. [0108]).
Claim 6: Dukatz teaches the method as recited in claim 1. Dukatz further teaches wherein each of the machine learning models is operable to learn a relationship between the quantum circuit information and quality of service metrics of running the quantum circuit on either the simulation engine or the quantum hardware (i.e. During operation (C), the machine learning module 132 is configured to receive data representing a solution to computational task, e.g., data 210, and data representing properties of using the corresponding computing resource to solve the computational task, e.g., data 212. For example, data representing properties of using the corresponding computing resource to solve the computational task may include data representing an approximate quality of the generated solution, a computational time associated with the generated solution, or a computational cost associated with the generated solution; para. [0082, 0090, 0102]).
Claim 8: Dukatz teaches the method as recited in claim 1. Dukatz further teaches wherein one of the estimates comprises an estimate for a time of execution of the quantum circuit on the quantum hardware (i.e. data representing properties of using the corresponding computing resource to solve the computational task may include data representing an approximate quality of the generated solution, a computational time associated with the generated solution, or a computational cost associated with the generated solution; para. [0090]), and the estimate for a time of execution of the quantum circuit on the quantum hardware comprises an amount of time to perform a quantum circuit cutting process (i.e. the system 100 for performing computational tasks may include a subgraph module 122. The subgraph module 122 may be configured to partition a computational task into multiple sub tasks; para. [0069]), and a quantum circuit knitting process (i.e. the machine learning module 132 may be configured to first process received data representing solutions to sub tasks of the computational task in order to generate an overall solution to the computational task; para. [0091]).
Claim 10: Dukatz teaches the method as recited in claim 1. Dukatz further teaches wherein after the orchestrating (i.e. The system processes the received data using a machine learning model to determine which of the one or more quantum computing resources or the one or more classical computing resources to route the data representing the computational task to (step 404); para. [0116]), the quantum circuit is executed on the simulation engine, or on the quantum hardware (i.e. The additional computing resources 110 a-110 d may include one or more quantum gate processors, e.g., quantum gate processor 110 b. A quantum gate processor includes one or more quantum circuits, i.e., models for quantum computation in which a computation is performed using a sequence of quantum logic gates, operating on a number of qubits (quantum bits); para. [0058-0060]).
Claim 11: Dukatz teaches a non-transitory storage medium having stored therein instructions (i.e. Implementations of the digital and/or quantum subject matter described in this specification can be implemented as one or more digital and/or quantum computer programs, i.e., one or more modules of digital and/or quantum computer program instructions encoded on a tangible non-transitory storage medium for execution by, or to control the operation of, data processing apparatus; para. [0120]) that are executable by one or more hardware processors (i.e. processor; para. [0121]) to perform operations comprising:
providing quantum circuit information, concerning a quantum circuit, to a first machine learning model (i.e. the system may train the machine learning model to route received computational tasks by generating a set of training examples using the (i) first set of data as described above with reference to step 302, (ii) the input data, as described above with reference to step 304, and (iii) the second set of data, as described above with reference to step 306. Each training example in the set of training examples may include a machine learning model input paired with a known machine learning model output. In some cases the training examples may be initially generated from combining data from independent quantum computing resource execution logs and classical computing resource execution logs; para. [0112]) that has been trained with first training data (i.e. The machine learning module 132 is configured to provide data representing the computational task or computational sub tasks to the trained machine learning model 204. The machine learning model 204 is configured to process the received data and to determine which of the one or more additional computing resources 110 a-110 d to route the received data representing the computational task or sub tasks to; para. [0088]) generated as a result of execution of a group of quantum circuits (i.e. the machine learning model 204 may determine that a received simulation task should be routed to a quantum simulator, e.g., quantum simulator 110 c. In this example, during operation (B), the machine learning model 204 may provide the quantum simulator 110 c with instructions for performing the simulation task; para. [0089, 0112]) on a simulation engine (i.e. The additional computing resources 110 a-110 d may include one or more quantum simulators, e.g., quantum simulator 110 c. A quantum simulator is a quantum computer that may be programmed to simulate other quantum systems and their properties; para. [0060]);
providing the quantum circuit information (i.e. The database 206 is configured to store data representing properties associated with using the one or more additional computing resources 110 a-110 d, e.g., one or more quantum computing resources, to solve the multiple computational tasks; para. [0084]) to a second machine learning model that has been trained with second training data generated as a result of execution of the group of quantum circuits (i.e. the system may train the machine learning model to route received computational tasks by generating a set of training examples using the (i) first set of data as described above with reference to step 302, (ii) the input data, as described above with reference to step 304, and (iii) the second set of data, as described above with reference to step 306. Each training example in the set of training examples may include a machine learning model input paired with a known machine learning model output. In some cases the training examples may be initially generated from combining data from independent quantum computing resource execution logs and classical computing resource execution logs; para. [0112]) on quantum hardware (i.e. The additional computing resources 110 a-110 d may include one or more quantum gate processors, e.g., quantum gate processor 110 b. A quantum gate processor includes one or more quantum circuits, i.e., models for quantum computation in which a computation is performed using a sequence of quantum logic gates, operating on a number of qubits (quantum bits); para. [0059]);
estimating, by the first machine learning model and the second machine learning model, respective values (i.e. The machine learning model 204 is a predictive model that may be trained to perform one or more machine learning tasks, e.g., classification tasks; para. [0082]) of a quantum circuit execution parameter of the quantum circuit (i.e. During operation (C), the machine learning module 132 is configured to receive data representing a solution to computational task, e.g., data 210, and data representing properties of using the corresponding computing resource to solve the computational task, e.g., data 212. For example, data representing properties of using the corresponding computing resource to solve the computational task may include data representing an approximate quality of the generated solution, a computational time associated with the generated solution, or a computational cost associated with the generated solution; para. [0090, 0113]);
comparing the estimates of the quantum circuit execution parameter (i.e. The system processes the received data using a machine learning model to determine which of the one or more quantum computing resources or the one or more classical computing resources to route the data representing the computational task to (step 404); para. [0116]); and
based on the comparing (i.e. A quantum computing machine learning module, as described in this specification, balances this tradeoff and learns optimal routings of computational tasks to classical or quantum computing resources. By learning when and how to utilize the power of quantum computing, a system implementing the quantum computing machine learning module may perform computational tasks more efficiently and/or accurately compared to systems that do not include quantum computing resources, or to systems that do not learn optimal routings of computational tasks to classical or quantum resources; para. [0026]), orchestrating the quantum circuit to either the simulation engine, or the quantum hardware, for execution (i.e. During operation (B), the machine learning model 204 is configured to provide the determined additional computing resource or resources with instructions for performing the respective computational task or computational sub tasks, e.g., data 208. For example, the machine learning model 204 may determine that a received optimization task should be routed to a quantum annealer, e.g., quantum annealer 110 a. In this example, during operation (B), the machine learning model 204 may provide the quantum annealer 110 a with instructions for performing the optimization task. As another example, the machine learning model 204 may determine that a received simulation task should be routed to a quantum simulator, e.g., quantum simulator 110 c. In this example, during operation (B), the machine learning model 204 may provide the quantum simulator 110 c with instructions for performing the simulation task; para. [0089]).
Claim 12: Dukatz teaches the non-transitory storage medium as recited in claim 11. Dukatz further teaches wherein a choice to orchestrate the quantum circuit to either the simulation engine, or the quantum hardware, is based on a parameter of a service level agreement (i.e. An error tolerance associated with a computational task may be used to determine which computing resource to route the computational task to. For example, some computational tasks may have smaller error tolerances than others, e.g., an error tolerance of a solution to the task of optimizing a cancer radiotherapy treatment may be smaller than an error tolerance of a solution to the task of optimizing the wastage of water in a water network. Computational tasks with smaller error tolerances may therefore be routed to computing resources that are more accurate than other computing resources, e.g., to computing resources who are less likely to introduce errors when performing a computational task; para. [0102, 0104]).
Claim 14: Dukatz teaches the non-transitory storage medium as recited in claim 11. Dukatz further teaches wherein the estimates comprise an estimate for a time of execution of the quantum circuit on the quantum hardware, and an estimate for a time of execution of the quantum circuit on the simulation engine (i.e. computational runtimes associated with solutions generated by the one or more quantum computing resources; para. [0108]).
Claim 15: Dukatz teaches the non-transitory storage medium as recited in claim 11. Dukatz further teaches wherein the quantum circuit is arbitrarily sized (i.e. each training example can include (i) input data relating to a previous computational task, e.g., data specifying the task, size/complexity of the task, restrictions for solving the task, error tolerance, (ii) information relating to which device was used to solve the task, or (iii) metrics indicating a quality of the solution obtained using the device, e.g., a level of confidence in the solution, computational time taken to generate the solution, or computational costs incurred; para. [0039]).
Claim 16: Dukatz teaches the non-transitory storage medium as recited in claim 11. Dukatz further teaches wherein each of the machine learning models is operable to learn a relationship between the quantum circuit information and quality of service metrics of running the quantum circuit on either the simulation engine or the quantum hardware (i.e. During operation (C), the machine learning module 132 is configured to receive data representing a solution to computational task, e.g., data 210, and data representing properties of using the corresponding computing resource to solve the computational task, e.g., data 212. For example, data representing properties of using the corresponding computing resource to solve the computational task may include data representing an approximate quality of the generated solution, a computational time associated with the generated solution, or a computational cost associated with the generated solution; para. [0082, 0090, 0102]).
Claim 18: Dukatz teaches the non-transitory storage medium as recited in claim 11. Dukatz further teaches wherein one of the estimates comprises an estimate for a time of execution of the quantum circuit on the quantum hardware (i.e. data representing properties of using the corresponding computing resource to solve the computational task may include data representing an approximate quality of the generated solution, a computational time associated with the generated solution, or a computational cost associated with the generated solution; para. [0090]), and the estimate for a time of execution of the quantum circuit on the quantum hardware comprises an amount of time to perform a quantum circuit cutting process (i.e. the system 100 for performing computational tasks may include a subgraph module 122. The subgraph module 122 may be configured to partition a computational task into multiple sub tasks; para. [0069]), and a quantum circuit knitting process (i.e. the machine learning module 132 may be configured to first process received data representing solutions to sub tasks of the computational task in order to generate an overall solution to the computational task; para. [0091]).
Claim 20: Dukatz teaches the non-transitory storage medium as recited in claim 11. Dukatz further teaches wherein after the orchestrating (i.e. The system processes the received data using a machine learning model to determine which of the one or more quantum computing resources or the one or more classical computing resources to route the data representing the computational task to (step 404); para. [0116]), the quantum circuit is executed on the simulation engine, or on the quantum hardware (i.e. The additional computing resources 110 a-110 d may include one or more quantum gate processors, e.g., quantum gate processor 110 b. A quantum gate processor includes one or more quantum circuits, i.e., models for quantum computation in which a computation is performed using a sequence of quantum logic gates, operating on a number of qubits (quantum bits); para. [0058-0060]).
Claim Rejections – 35 USC § 103
6. 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 of this title, 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.
7. Claim 5 is rejected under 35 U.S.C. 103 as being unpatentable over Dukatz in view of Lucarelli (U.S. Patent Application Pub. No. US 20200119748 A1).
Claim 5: Dukatz teaches the method as recited in claim 1. Dukatz does not explicitly teach wherein a CNOT embedding process is performed in which the quantum circuit is transformed into a representation that captures occurrences of CNOT gates in the quantum circuit, and the representation is then translated into a matrix whose rows, or columns, correspond to qubits, numbered from 1 to a maximum number of qubits on the quantum circuit, and whose rows, or columns, correspond to CNOT gates, numbered in an order that the CNOT gates appear on the quantum circuit.
However, Lucarelli teaches wherein a CNOT embedding process is performed in which the quantum circuit is transformed into a representation that captures occurrences of CNOT gates in the quantum circuit (i.e. For example, the logical parity encoder B (see FIG. 1) may construct the quantum circuit 130 from the resultant matrix R.sub.M by coupling a number N×n (which is the number of columns of the resultant matrix R.sub.M) of the data qubits 102 to a number M (which is the number of rows of the resultant matrix R.sub.M) of the ancilla qubits 106 as follows. A multiple-qubit gate, which may include a controlled-NOT gate (“CNOT gate”) or a controlled-PHASE gate (“CPHASE gate”) as non-limiting examples, couples data qubit “j” (illustrated as values |ψ.sub.j> below the logical parity-check matrix 120 in FIG. 4) to ancilla qubit “i” (illustrated as values |A.sub.i> to the right of the logical parity-check matrix 120 in FIG. 4) if, and only if, the binary number located in the i-th row and j-th column of the logical parity-check matrix 120 is “1.” On the other hand, if the binary number located in the i-th row and j-th column of the logical parity-check matrix 120 is “0,” the data qubit |ψ.sub.j> remains uncoupled from the ancilla qubit; para. [0059]), and the representation is then translated into a matrix whose rows, or columns, correspond to qubits, numbered from 1 to a maximum number of qubits on the quantum circuit, and whose rows, or columns, correspond to CNOT gates, numbered in an order that the CNOT gates appear on the quantum circuit (i.e. FIG. 4 illustrates an example logical parity-check matrix 120 formed by the Kronecker product of a binary matrix 122 (which is the sub-parity-check matrix P.sub.L of a Hamming [7,4,3] classical error correcting code) and the binary representation of the quantum check operator “ZIZ” 124 (which is a binary vector represented by the variable S). The quantum check operator “ZIZ” is an example “multiple-qubit Pauli Operator.” The rows of the logical parity-check matrix 120 are illustrated as values |A.sub.i> corresponding to the ancilla qubits 106 (see FIG. 1), and the columns are illustrated as values |ψ.sub.j> corresponding to the data qubits 102 (see FIG. 1) in the quantum circuit illustrated in FIG. 5. The specification includes entries each having a value selected from the set of two values (e.g., one or zero). Referring to FIG. 5, after the specification is generated, the logical parity encoder B (see FIG. 1) may construct a quantum circuit 130 in accordance with the specification. Referring to FIG. 17, as explained above, the logical parity encoder B is executed by the host processor 166. Instructions issued by the logical parity encoder B executing on the host processor 166 are passed to the control processor plane 164, which identifies and triggers quantum operations (e.g., gates) and measurements to be performed by the control and measurement plane 162 on the quantum data plane 160. In this manner, the logical parity encoder B constructs the quantum circuit 130; para. [0058, 0060]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Dukatz to include the feature of Lucarelli. One would have been motivated to make this modification because it improves computational efficiency and clarity of circuit representation.
8. Claims 7, 13, and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Dukatz in view of Wang et al. (QuEst: Graph Transformer for Quantum Circuit Reliability Estimation, arXiv, published 30 Oct 2022, pages 1-10).
Claim 7: Dukatz teaches the method as recited in claim 1. Dukatz further teaches wherein the first training data and/or the second training data comprise ground truth data (i.e. Each training example in the set of training examples may include a machine learning model input paired with a known machine learning model output; para. [0112, 0113]), and one of the quantum circuits in the group of quantum circuits, and that quantum circuit has an arbitrary size and configuration (i.e. each training example can include (i) input data relating to a previous computational task, e.g., data specifying the task, size/complexity of the task, restrictions for solving the task, error tolerance, (ii) information relating to which device was used to solve the task, or (iii) metrics indicating a quality of the solution obtained using the device, e.g., a level of confidence in the solution, computational time taken to generate the solution, or computational costs incurred; para. [0039]).
Dukatz does not explicitly teach vectorial representation of one of the quantum circuits.
However, Wang teaches wherein the first training data and/or the second training data comprise ground truth data (i.e. The first step of the framework is to collect a large dataset containing various randomly generated circuits and circuits from common quantum algorithms. We run the circuits on both noisy simulators and real quantum machines. On simulators, we change the properties of the qubits, such as T1 and T2, and the error rates of gates to diversify the data samples. The dataset contains over 20 thousand samples on simulators and 25 thousand samples on real quantum machines; Section 3 and pages 2-4), and a vectorial representation of one of the quantum circuits (i.e. We firstly use directed acyclic graphs (DAG) to represent the topology of quantum circuits. Each node represents one qubit, quantum gate, or measurement. Edges represent the time-dependent order of different gates. One example of extracting the graph from the circuit is presented on the left of Figure 5. The connectivity can be encoded into an adjacent matrix; Section 4.1) in the group of quantum circuits (i.e. Figure 1: The proposed fidelity prediction framework. The quantum circuit is firstly embedded into a graph in which the nodes are gates and edges are execution orders. The feature vector on each node contains the device noise information, such as gate error rates. The graph is processed by a graph transformer in TorchQuantum to estimate circuit fidelity), and that quantum circuit has an arbitrary size and configuration; page 1), and that quantum circuit has an arbitrary size and configuration (i.e. The simple NN model only takes 116 features as input, which include circuit depth, width, and counts of RZ, X, SX, and CNOT gates, single-qubit gate counts on each qubit, and two-qubit gate counts on each qubit pair; Section 5).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Dukatz to include the feature of Wang. One would have been motivated to make this modification because it enables efficient training and inference by converting circuits into vector form.
Claim 13: Dukatz teaches the non-transitory storage medium as recited in claim 11. Dukatz does not explicitly teach wherein the group of quantum circuits comprises randomly generated quantum circuits.
However, Wang teaches wherein the group of quantum circuits comprises randomly generated quantum circuits (i.e. The first step of the framework is to collect a large dataset containing various randomly generated circuits and circuits from common quantum algorithms. We run the circuits on both noisy simulators and real quantum machines. On simulators, we change the properties of the qubits, such as T1 and T2, and the error rates of gates to diversify the data samples. The dataset contains over 20 thousand samples on simulators and 25 thousand samples on real quantum machines; Section 3 and pages 2-4).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Dukatz to include the feature of Wang. One would have been motivated to make this modification because it improves model robustness.
Claim 17: Dukatz teaches the non-transitory storage medium as recited in claim 11. Dukatz further teaches wherein the first training data and/or the second training data comprise ground truth data (i.e. Each training example in the set of training examples may include a machine learning model input paired with a known machine learning model output; para. [0112, 0113]), and one of the quantum circuits in the group of quantum circuits, and that quantum circuit has an arbitrary size and configuration (i.e. each training example can include (i) input data relating to a previous computational task, e.g., data specifying the task, size/complexity of the task, restrictions for solving the task, error tolerance, (ii) information relating to which device was used to solve the task, or (iii) metrics indicating a quality of the solution obtained using the device, e.g., a level of confidence in the solution, computational time taken to generate the solution, or computational costs incurred; para. [0039]).
Dukatz does not explicitly teach vectorial representation of one of the quantum circuits.
However, Wang teaches wherein the first training data and/or the second training data comprise ground truth data (i.e. The first step of the framework is to collect a large dataset containing various randomly generated circuits and circuits from common quantum algorithms. We run the circuits on both noisy simulators and real quantum machines. On simulators, we change the properties of the qubits, such as T1 and T2, and the error rates of gates to diversify the data samples. The dataset contains over 20 thousand samples on simulators and 25 thousand samples on real quantum machines; Section 3 and pages 2-4), and a vectorial representation of one of the quantum circuits (i.e. We firstly use directed acyclic graphs (DAG) to represent the topology of quantum circuits. Each node represents one qubit, quantum gate, or measurement. Edges represent the time-dependent order of different gates. One example of extracting the graph from the circuit is presented on the left of Figure 5. The connectivity can be encoded into an adjacent matrix; Section 4.1) in the group of quantum circuits (i.e. Figure 1: The proposed fidelity prediction framework. The quantum circuit is firstly embedded into a graph in which the nodes are gates and edges are execution orders. The feature vector on each node contains the device noise information, such as gate error rates. The graph is processed by a graph transformer in TorchQuantum to estimate circuit fidelity), and that quantum circuit has an arbitrary size and configuration; page 1), and that quantum circuit has an arbitrary size and configuration (i.e. The simple NN model only takes 116 features as input, which include circuit depth, width, and counts of RZ, X, SX, and CNOT gates, single-qubit gate counts on each qubit, and two-qubit gate counts on each qubit pair; Section 5).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Dukatz to include the feature of Wang. One would have been motivated to make this modification because it enables efficient training and inference by converting circuits into vector form.
9. Claims 9 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Dukatz in view of Zhu et al. (Training of Quantum Circuits on a Hybrid Quantum Computer, arXiv, published 31 Oct 2019, pages 1-22).
Claim 9: Dukatz teaches the method as recited in claim 1. Dukatz further teaches wherein the first training data and/or the second training data comprise a size, depth, of one or more of the quantum circuits in the group of quantum circuits (i.e. each training example can include (i) input data relating to a previous computational task, e.g., data specifying the task, size/complexity of the task, restrictions for solving the task, error tolerance, (ii) information relating to which device was used to solve the task, or (iii) metrics indicating a quality of the solution obtained using the device, e.g., a level of confidence in the solution, computational time taken to generate the solution, or computational costs incurred; para. [0039]).
Dukatz does not explicitly teach level of entanglement.
However, Zhu teaches level of entanglement (i.e. The entanglement entropy quantifies the level of entanglement of a state, thus indicates how difficult it is to produce such state. This metric shows that the successfully trained circuits generate states that are consistent with a high level of entanglement. As a reference, the entanglement entropy of a GHZ state over any partition is S = 1; page 8).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Dukatz to include the feature of Zhu. One would have been motivated to make this modification because it provides an improvement in circuit characterization.
Claim 19: Dukatz teaches the non-transitory storage medium as recited in claim 11. Dukatz further teaches wherein the first training data and/or the second training data comprise a size, depth, of one or more of the quantum circuits in the group of quantum circuits (i.e. each training example can include (i) input data relating to a previous computational task, e.g., data specifying the task, size/complexity of the task, restrictions for solving the task, error tolerance, (ii) information relating to which device was used to solve the task, or (iii) metrics indicating a quality of the solution obtained using the device, e.g., a level of confidence in the solution, computational time taken to generate the solution, or computational costs incurred; para. [0039]).
Dukatz does not explicitly teach level of entanglement.
However, Zhu teaches level of entanglement (i.e. The entanglement entropy quantifies the level of entanglement of a state, thus indicates how difficult it is to produce such state. This metric shows that the successfully trained circuits generate states that are consistent with a high level of entanglement. As a reference, the entanglement entropy of a GHZ state over any partition is S = 1; page 8).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Dukatz to include the feature of Zhu. One would have been motivated to make this modification because it provides an improvement in circuit characterization.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant’s disclosure.
Satzinger et al. (Pub. No. US 20220358392 A1), Benchmarking techniques can be applied to determine how close a noisy quantum operation performed by quantum hardware is to an ideal unitary quantum operation, and thereby characterize quantum hardware performance. For example, benchmarking techniques can be applied to characterize the performance of implementations of two-qubit quantum gates. This can include using the quantum hardware to execute random quantum circuits that include multiple instances of the two-qubit gate, and using a classical computer to simulate the same random quantum circuits. The results of the quantum computations and classical computations can be compared to determine how noisy the quantum operations were as well as their fidelity and purity.
THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the date of this final action.
It is noted that any citation to specific pages, columns, lines, or figures in the prior art references and any interpretation of the references should not be considered to be limiting in any way. A reference is relevant for all it contains and may be relied upon for all that it would have reasonably suggested to one having ordinary skill in the art. In re Heck, 699 F.2d 1331, 1332-33, 216 U.S.P.Q. 1038, 1039 (Fed. Cir. 1983) (quoting In re Lemelson, 397 F.2d 1006, 1009, 158 U.S.P.Q. 275, 277 (C.C.P.A. 1968)).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to TAN TRAN whose telephone number is (303)297-4266. The examiner can normally be reached on Monday - Thursday - 8:00 am - 5:00 pm MT.
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, Matt Ell can be reached on 571-270-3264. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/TAN H TRAN/Primary Examiner, Art Unit 2141
Prosecution Insights