Optimizer design method and device using quantum tunneling effect, and medium

Abstract

The invention discloses an optimizer design method and device using a quantum tunneling effect and a medium, and relates to the technical field of computer science and technology and quantum information. The method comprises the following steps: converting an optimization problem into a quantum state through mathematical modeling; iterative optimization of the quantum state is realized through a parameterized quantum circuit, and a quantum tunneling effect mechanism is introduced, so that a local optimal solution is jumped out; after each quantum state evolution, evaluating the quality of a current solution by observing a target function value; a self-adaptive multi-layer optimization strategy is introduced, and the learning rate of parameters is adjusted to gradually converge to a global optimal quantity, so that the capabilities of global search and local optimization are balanced; after the quantum circuit is optimized, an observation value is measured and extracted, and a globally optimal solution is determined according to the observation value. The quantum circuit and the tunneling mechanism are introduced into hardware design, the fast jumping function is achieved through the tunneling phenomenon, the optimization performance can be improved, and a more efficient solution is provided for hardware acceleration.

Description

Technical Field

[0001] The present invention relates to the fields of computer science and technology and quantum information technology, and more specifically, to an optimizer design method, device and medium utilizing quantum tunneling effect. Background Art

[0002] In the field of hardware acceleration of optimization algorithms, most existing hardware devices are based on classic computing architectures, and their design ideas mainly rely on improving computing speed, optimizing algorithm execution paths, and using parallel computing to improve optimization performance. However, for complex optimization problems, especially global optimization tasks in high-dimensional solution spaces, the performance of such devices is often limited by local optimal problems, that is, they are prone to falling into local optimal solutions during the optimization process and find it difficult to jump to a better solution area.

[0003] At present, the methods for solving the local optimum problem mainly focus on the improvement of software algorithms, such as balancing global search and local search by introducing a variety of heuristic algorithms (such as genetic algorithms, particle swarm algorithms, and simulated annealing algorithms). However, these methods are still limited by the limitations of classical computing, especially when dealing with complex nonlinear problems, their efficiency and effectiveness are difficult to be satisfactory. In terms of hardware implementation, the design of existing optimization equipment does not provide effective support for jumping out of local optimality, and lacks a special mechanism for specific bottlenecks in the optimization process. For example, although the acceleration solutions of FPGA and GPU can improve the overall computing speed, their hardware design does not provide a direct solution for how to get rid of local optimality.

[0004] Some studies have tried to use random perturbations or high temperature mechanisms to simulate the "jumps" in the search process, but these methods rely on a large number of iterations and computing resources, and have low optimization efficiency. At the same time, hardware design is still based on classical architecture and fails to fully utilize the physical mechanisms in quantum mechanics (such as quantum tunneling effect) to construct a hardware solution that can more effectively jump out of the local optimum.

[0005] The quantum tunneling effect provides a new possibility for solving the local optimum problem. The tunneling effect is a typical quantum mechanical phenomenon, which manifests itself as a microscopic particle "traveling" through the potential barrier to the other side through the expansion characteristics of the quantum wave function when the energy is insufficient to cross the potential barrier. In the optimization problem, this characteristic can be analogized to the jumping mechanism in the optimization process, which allows the solution to "travel" from a local optimal solution area to other areas in the global search space, thereby improving the ability to jump out of the local optimum.

[0006] Although some studies have explored how to use quantum tunneling effects to solve local optimal problems at the software level, the implementation of these studies relies on complex simulation algorithms with high computational overhead, making it difficult to meet the efficiency requirements in practical applications. At the hardware level, the architecture design of existing optimization devices does not take into account the direct introduction of quantum tunneling effects, resulting in the performance of existing devices in optimization algorithms being limited to the classical logic framework and unable to fully tap the potential of quantum effects.

[0007] In summary, the existing optimization hardware devices have the following main deficiencies: First, the hardware design does not provide a special solution mechanism for the local optimal problem, and the jumping ability is limited; second, the device architecture lacks support for the quantum tunneling effect and cannot use this quantum mechanical property to improve the optimization performance. The root cause of these deficiencies is that the existing technology is mainly based on classical computing logic and lacks exploration of the application of quantum properties. Summary of the invention

[0008] In order to solve the above technical problems, the present invention provides an optimizer design method, device and medium using the quantum tunneling effect, introduces quantum circuits and tunneling mechanisms into hardware design, and realizes a fast jump function through the tunneling phenomenon, which can not only improve the optimization performance, but also make up for the shortcomings of the existing technology and provide a more efficient solution for hardware acceleration.

[0009] In a first aspect, the present invention provides an optimizer design method using a quantum tunneling effect, the method comprising: Convert the optimization problem into quantum state through mathematical modeling; Iteratively optimizing the quantum state by parameterizing the quantum circuit; Introduce quantum tunneling effect mechanism to escape from the local optimal solution; After each quantum state evolution, the quality of the current solution is evaluated by observing the value of the objective function; Introducing an adaptive multi-layer optimization strategy, by adjusting the learning rate of the parameters, gradually converges to the global optimal value, thereby balancing the capabilities of global search and local optimization; After the quantum circuit is optimized, the observation values ​​are measured and extracted, and the global optimal solution is determined based on the observation values.

[0010] Furthermore, the optimization problem is converted into quantum state through mathematical modeling, including: For variables ,in represents the real number space, n Represents the dimension of the problem, the objective function of the variable To perform operations, quantum bits are introduced to represent the state of each variable; Through quantum amplitude encoding, variables x Each possible state of is mapped to a quantum state ,in: ; in, represents the index of the qubit state, Represents the components of a quantum state; Mapping the optimization problem into a quantum state construction problem, weights , represents the possibility of each solution; When initializing the quantum circuit, the Hadamard gate is applied to each bit to generate the initial uniform superposition state through the following formula: ; in, represents the initial uniform superposition state.

[0011] Furthermore, iterative optimization of the quantum state is achieved by parameterizing the quantum circuit, including: Through a series of parametric revolving doors and controlled doors CX The quantum variational circuit is formed, which is expressed as: ; in, represents a composite operation consisting of a series of parameterized quantum gates, Represents a quantum gate that rotates along the y-axis, with the rotation angle determined by the parameter Decide, represents the index of the rotation gate in the quantum circuit, represents the number of parameterized quantum gates in the quantum circuit; The parameterized circuit is used to update the quantum state, and the parameterized target state is formed by the following formula: ; in, It is composed of revolving door, controlled door CX And the composite operation composed of CZ gate, CZ gate is a control gate, represents the quantum state after optimization of the parameterized quantum circuit. The parameterized quantum circuit is optimized by adjusting the parameters accomplish.

[0012] Furthermore, the quantum tunneling effect mechanism is introduced to escape from the local optimal solution, including: The basis of the quantum tunneling mechanism comes from the Schrödinger equation as shown below: ; in, represents the imaginary unit, represents the reduced Planck constant, The wave function of a quantum system represents the probability amplitude of a particle at a certain position and time. represents the time variable, represents the potential barrier function, represents the mass of the particle; Quantum tunneling effect mechanism via nonlinear operation of controlled rotation gate And double quantum state entanglement: ; in, Indicates a controlled revolving door, represents the phase rotation of a complex number, Represents the rotation angle or phase angle.

[0013] Furthermore, after each quantum state evolution, the quality of the current solution is evaluated by observing the value of the objective function, including: Using quantum observers to map objective function values ​​to measurement results M , the goodness of the solution is given by the probability amplitude: ; in, Representing quantum states The probability of being measured, Representing quantum states The probability amplitude, argmax means returning the index corresponding to the maximum probability in a probability distribution.

[0014] Furthermore, an adaptive multi-layer optimization strategy is introduced to adjust the learning rate of the parameters through the following formula, gradually converging to the global optimal value: ; in, represents the updated optimization parameters at the t+1th iteration, represents the optimization parameter at the tth iteration, represents the estimated value of the objective function gradient, represents the learning rate, Indicated in the parameter Below, the quantum state obtained through the quantum circuit.

[0015] Furthermore, after the quantum circuit is optimized, the observed value is measured and extracted using the following formula: ; in, represents the observed value; Indicates the optimal parameters The quantum states generated by the quantum circuit.

[0016] Furthermore, the method further comprises: Based on the set optimization requirements, check whether the objective function value meets the optimization requirements. If the optimization requirements are not met, adjust the circuit parameters through the feedback mechanism until the objective function value meets the optimization requirements.

[0017] In a second aspect, the present invention provides an optimizer design device using a quantum tunneling effect, the device comprising: A quantum state conversion unit, configured to convert the optimization problem into a quantum state through mathematical modeling; An iterative optimization unit, configured to implement iterative optimization of the quantum state through a parameterized quantum circuit; A mechanism introduction unit, configured to introduce a quantum tunneling effect mechanism to escape from a local optimal solution; A function evaluation unit, configured to evaluate the quality of the current solution by observing the target function value after each quantum state evolution; The multi-layer optimization unit is configured to introduce an adaptive multi-layer optimization strategy, which gradually converges to the global optimum by adjusting the learning rate of the parameters, thereby balancing the capabilities of global search and local optimization; The optimal solution determination unit is configured to measure and extract observation values ​​after the quantum circuit is optimized, and determine the global optimal solution based on the observation values.

[0018] In a third aspect, the present invention provides a readable storage medium, wherein the readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the method as described above.

[0019] Based on the combination of QVAs and quantum tunneling effect, it aims to solve the problems of local optimal trap, low search efficiency and difficulty in analyzing complex solution space in traditional ABC in high-dimensional, nonlinear and multi-objective optimization problems. The beneficial effects of the technical solution disclosed in the present invention will be described in detail below.

[0020] 1. Overcoming the limitations of traditional ABC: The quantum optimizer proposed in this application combines QVAs with the quantum tunneling effect to address the shortcomings of traditional ABC in high-dimensional, nonlinear, and multi-objective optimization problems, especially when the objective function has multiple extreme points. The quantum tunneling effect helps the optimizer to cross the potential barriers of the solution space through the superposition of quantum states and the non-locality of quantum entanglement, thereby avoiding the local optimal trap and increasing the probability of finding the global optimal solution.

[0021] 2. Significantly improve the search efficiency of complex solution space: The quantum optimizer uses the exponential superposition capability of QVAs to make the search space grow exponentially with the increase in the number of quantum bits, and can evaluate and screen potential optimal solutions in a shorter time, thereby significantly improving computing efficiency. The parallel nature of quantum computing enables the optimization process to achieve more efficient search in multi-dimensional solution space and adapt to complex high-dimensional problems.

[0022] 3. Scalability and diversity adaptation of the optimizer: The quantum circuit structure of the present invention is highly modular and parameterizable, and can adapt to various types of optimization problems, including combinatorial optimization, continuous optimization, and multi-objective optimization problems. By modifying the encoding method of the objective function and adjusting the specific structure of the parameterized quantum circuit, the optimizer can flexibly adapt to different practical needs. This highly flexible structure makes it possible to apply the technical solution in different fields, providing a scalable solution for solving diverse optimization problems.

[0023] 4. Structural improvement and tunneling mechanism of quantum circuit: The quantum circuit designed by the present invention adopts a multi-layer parameterized structure, each layer of which is composed of a revolving gate and an entanglement gate, which can enhance the global search capability of the solution space. The revolving gate provides the ability to adjust the quantum state locally, while the entanglement gate enhances the global search capability of the solution space by introducing non-local correlations between bits. In addition, a tunneling module is added to the quantum circuit, which changes the distribution of quantum states through nonlinear phase adjustment, enhances the global search capability, and helps the system jump from the local optimal solution to a wider solution space area.

[0024] 5. The fusion design of quantum classical feedback mechanism: The present invention further improves the quantum classical feedback mechanism by measuring the quantum state in real time and dynamically adjusting the parameterized structure and learning rate of the quantum circuit according to the measurement results. This mechanism deeply integrates quantum computing with classical computing, and can adjust the optimization strategy in real time according to the quantum measurement results in the optimization process, ensuring that the optimization process always converges in the direction of reducing the objective function value. This feedback mechanism effectively improves the efficiency of the optimization process and maintains the stability and convergence of the optimization process during continuous adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A quantum circuit diagram describing the quantum tunneling effect according to an embodiment of the present invention is shown; Figure 2 A quantum circuit designed using Qiskit according to an embodiment of the present invention is shown; Figure 3 shows a statistical histogram of quantum state candidate solutions according to an embodiment of the present invention; Figure 4 A framework diagram of an optimizer based on quantum tunneling according to an embodiment of the present invention is shown; Figure 5 An overall flow chart of an optimizer design method using quantum tunneling effect according to an embodiment of the present invention is shown; Figure 6 A quantum circuit diagram adding quantum tunneling according to an embodiment of the present invention is shown; Figure 7 A structural diagram of an optimizer design device using quantum tunneling effect according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings and specific embodiments, but are not intended to limit the present invention. For the various steps described herein, if there is no necessity for a causal relationship between each other, the order in which they are described as examples herein should not be regarded as a limitation, and those skilled in the art should know that they can be adjusted in order, as long as the logic between them is not destroyed, resulting in the inability to implement the entire process.

[0027] The embodiment of the present invention provides an optimizer design method using the quantum tunneling effect, aiming to solve the technical problems that the traditional ABC is prone to fall into local optimality, insufficient global search capability and slow convergence speed in complex problems. ABC is the abbreviation of "Artificial Bee Colony Algorithm", which means artificial bee colony algorithm. For general dynamic constraint problems, the present invention introduces the quantum tunneling effect, and realizes jumping out of the local optimal solution through the quantum jump mechanism, thereby enhancing the global search capability. At the same time, combined with the ABC optimization strategy, the accuracy and stability of the algorithm in complex scenarios are improved, thereby meeting the demand for efficient and stable optimization solutions in practical applications. This method can be widely used in multi-objective optimization and dynamic constraint optimization scenarios, covering intelligent traffic scheduling, engineering design, data mining and other fields, and providing innovative solutions for complex problems.

[0028] Next, we will introduce the key technologies needed to implement the optimizer design method using the quantum tunneling effect: A core technology for implementing the optimizer design method using quantum tunneling effect is to realize the function of jumping out of the local optimal solution in ABC through hardware based on quantum tunneling effect. Quantum tunneling effect is an important physical phenomenon in quantum mechanics, which can explain how microscopic particles "pass through" insurmountable potential barriers in classical mechanics. This feature provides a new idea for solving local optimal jumps in high-dimensional optimization problems.

[0029] The quantum tunneling effect means that even if the energy of a particle is lower than the potential barrier height, there is still a certain probability that the particle can pass through the potential barrier. This phenomenon violates the view of classical mechanics, but it is widely present in the microscopic world. Its origin lies in the wave function superposition principle and probability interpretation in quantum mechanics.

[0030] According to the time-independent Schrödinger equation: ; in is the reduced Planck constant, is the mass of the particle, is the wave function of the particle, is the potential energy function, is the total energy of the particle.

[0031] Suppose the particle encounters a rectangular potential barrier, , represents the height of the potential barrier, The range is the potential barrier area, and the potential energy of the rest of the area is 0. If the total energy of the particle , according to the classical mechanics view, particles cannot pass through the potential barrier. However, according to the solution of quantum mechanics, the particle wave function exhibits exponential decay in the potential barrier region: ; Among them, A and B represent the constants of the wave function in the potential barrier region, represents the exponential decay rate of the wave function in the potential barrier region, Represents an input parameter.

[0032] The transmission coefficient of the potential barrier, that is, the probability of a particle passing through the barrier, is: ; in, represents the transmission coefficient, Represents the width of the barrier.

[0033] The probability of particle penetration varies with the barrier width and barrier height Increase and decrease rapidly.

[0034] When applied to optimization problems, the local optimal energy in the solution space is simulated as a potential barrier, and the state of the solution corresponds to the wave function of the particle. Through the tunneling effect, the local optimum can be skipped to find the global optimal solution. In the optimizer hardware, this process can be implemented through quantum circuits. The superposition state of quantum bits and quantum gate operations can reflect the evolution of particles in the tunneling barrier. In the specific implementation, the transmission probability of the barrier region can be controlled by a parameterized rotating gate, and the result of quantum measurement can be used to determine whether the tunneling is completed.

[0035] like Figure 1As shown, this is a quantum circuit diagram describing the quantum tunneling effect. Figure 1 In this example, q represents a quantum bit, H represents a Hadamard gate, represents a parameterized rotating gate, C represents a classical bit, and "0" and "1" represent the final measurement results of the quantum bit. The state transition of the quantum bit is parameterized by the gate to represent the transmission probability of the particle in the potential barrier area. Then the parameters can be designed to control the height of the potential barrier, which is expressed as an exponential relationship consistent with the actual transmission probability. The final measurement result of the quantum bit is 0 or 1, indicating whether the particle has completed tunneling.

[0036] Another core technology for implementing the optimizer design method using quantum tunneling effect is quantum variational algorithms (QVAs). Quantum variational algorithms are a hybrid quantum-classical optimization algorithm that uses classical optimizers to adjust quantum circuit parameters to minimize the expected value of the objective function. It is mainly used in quantum chemistry, combinatorial optimization problems, and machine learning.

[0037] The core of QVAs lies in the variational principle: by optimizing the parameterized quantum state The expected value of makes it close to the optimal solution of the problem. For the optimized objective function Usually written as: ; in, is the problem-relevant Hamiltonian (i.e., the operator describing the energy of the system), is a parameterized quantum state, which depends on the parameters of the quantum circuit . Then the quantum circuit will be the initial state Evolution into parameterized quantum states , expressed as: ; in It is a single-particle and multi-particle operation composed of several parameterized quantum gates. After parameterizing the quantum state, parameter optimization is performed. The goal is to optimize the parameters. , so that Minimize: ; in, represents the optimal parameter solution.

[0038] Finally, by measuring the quantum state , then update the parameters : ; in is the learning rate, is the gradient of the objective function, represents the parameter value updated in each iteration, Represents the parameters in the current iteration.

[0039] Using Qiskit from International Business Machines Corporation (IBM), you can design Figure 2 The quantum circuit shown. Figure 2 In the above example, q1, q2, and q3 all represent quantum bits, and “0”, “1”, “2”, and “3” represent the indexes of the measurement results. , , Represents a parameterized revolving door.

[0040] The quantum circuit can transform the local search parameters of ABC into the input of the quantum circuit, and then map these parameters to the rotation angle of the quantum bit, and process the input parameters through quantum operations. Initially, it receives the local search parameters of ABC (a real number array), and then passes these parameters through the parameterized rotation gate and Mapped to quantum bits, the range of parameters can be adjusted to a suitable angle range through normalization or other methods. Then, the input parameters are quantum-state evolved, and the quantum bits are entangled using CNOT gates to generate corresponding quantum states and enhance their expressive power. Finally, the constructed quantum circuit is output and measurement is provided.

[0041] like Figure 3 As shown in the figure, the corresponding histogram of the output can be seen that the measurement result of the quantum circuit is a dictionary, and the result represents the number of occurrences of different quantum states during measurement. States with higher frequencies in the distribution may be candidates for the global solution. If the solutions corresponding to some measurement results have never been explored by the classical algorithm, these results may indicate that the algorithm has jumped out of the local optimum.

[0042] The optimizer design method using quantum tunneling effect designs a hardware optimizer that specifically enhances the global search capability of ABC through quantum tunneling effect. Traditional ABC often falls into the local optimal solution, resulting in the inability to find the global optimal solution. In order to solve this problem, the present invention introduces QVAs and quantum tunneling effect, converts the parameters of the optimization problem into quantum circuit input through quantum hardware, and uses quantum tunneling effect to jump out of the local optimal solution, thereby improving the global search capability and optimization accuracy. The framework of the optimizer based on quantum tunneling is as follows Figure 4 As shown, the main steps are: first, the optimization problem is formalized as an objective function , the input parameters are encoded by quantum states xMapping to a superposition state of quantum bits : ; in represents the discrete state of the parameter, is the corresponding superposition coefficient. Then we design a parameterized quantum circuit , the evolution of quantum state is realized through rotation gate and CNOT gate, forming: ; in, is a parameterized quantum state.

[0043] The parameters of the quantum circuit are Dynamically updated by the classical optimization algorithm. During the search process, through the quantum tunneling effect, the quantum state overcomes the energy barrier during the evolution process and jumps out of the local optimal solution. Its mathematical description is the time evolution of the quantum state: ; in H is the Hamiltonian of the system, is the quantum state after time t evolution, t is the time variable.

[0044] Finally, the expected value of the objective function is optimized by measuring: ; in, The objective function is In the quantum state The expected value under is the objective function After being converted into a quantum operator, it acts on the quantum state The above results.

[0045] Obtain feedback information of the optimal solution and adjust the quantum circuit parameters, iterating repeatedly until the global optimal solution is found, thereby achieving efficient solution to the optimization problem.

[0046] Specifically, if Figure 5 As shown, it is an overall flow chart of an optimizer design method using quantum tunneling effect provided by an embodiment of the present invention. The optimizer design method using quantum tunneling effect includes the following steps S10 to S60.

[0047] S10: Convert the optimization problem into quantum state through mathematical modeling.

[0048] In some embodiments, the optimization problem is first converted into a form suitable for quantum computing through mathematical modeling. , its objective function To perform operations, quantum bits are introduced to represent the state of each variable. Through quantum amplitude encoding, the variable x Each possible state of is mapped to a quantum state ,in: ; in, represents the index of the qubit state, Represents the components of a quantum state.

[0049] This representation maps the optimization problem to the problem of constructing a quantum state, with weights , indicating the possibility of each solution. When initializing the quantum circuit, a Hadamard gate is applied to each bit to generate an initial uniform superposition state : ; By applying the Hadamard gate to each bit, the construction of the above superposition state can be realized, which provides a basis for the subsequent parameterized evolution. Figure 6 Shown is a quantum circuit diagram with quantum tunneling added. Figure 6 In the figure, q0, q1, q2, and q3 represent the four quantum bits in the quantum circuit, H represents the Hadamard gate, and "0", "1", "2", "3", and "4" represent the indexes of the measurement results. , , and Represents the rotation angle around a specific axis revolving door; , , and Respectively represent the rotation angle around the Z axis revolving door.

[0050] S20: Iteratively optimizing the quantum state by parameterizing the quantum circuit.

[0051] In some embodiments, the quantum state is iteratively optimized through parameterized quantum circuits based on QVAs. and controlled doors CX The composition is expressed as: ; in, represents a composite operation consisting of a series of parameterized quantum gates, Represents a quantum gate that rotates along the y-axis, with the rotation angle determined by the parameter Decide, represents the index of the rotation gate in the quantum circuit, Represents the number of parameterized quantum gates in the quantum circuit.

[0052] In actual operation, a series of parameterized gate operations are added based on the initial state. The core of this step is to use parameterized circuits to update the quantum state and form a parameterized target state: ; in, It is composed of revolving door, controlled door CX And the composite operation composed of CZ gate, CZ gate is a control gate, represents the quantum state after optimization of the parameterized quantum circuit. The parameterized quantum circuit is optimized by adjusting the parameters accomplish.

[0053] The evolution of the quantum state is the search direction and step size in the optimization problem, ensuring that the solution gradually approaches the global optimum.

[0054] S30: Introduce the quantum tunneling effect mechanism to escape from the local optimal solution.

[0055] In some embodiments, in order to escape from the local optimal solution, a quantum tunneling effect mechanism is introduced. The mathematical basis of the quantum tunneling effect comes from the Schrödinger equation: ; in, represents the imaginary unit, represents the reduced Planck constant, The wave function of a quantum system represents the probability amplitude of a particle at a certain position and time. represents the time variable, represents the potential barrier function, Represents the mass of the particle.

[0056] In quantum optimizers, this mechanism is quantized as the ability to jump in the solution space. Its realization relies on nonlinear operations and quantum entanglement. The core of the tunneling effect is controlled by nonlinear operations of rotating gates. And double quantum state entanglement: ; in, Indicates a controlled revolving door, represents the phase rotation of a complex number, Represents the rotation angle or phase angle.

[0057] The tunneling effect allows the system to jump out of the potential well where the local optimal solution is located and explore other areas in the solution space, thereby significantly improving the optimization ability.

[0058] S40: After each quantum state evolution, the quality of the current solution is evaluated by observing the objective function value.

[0059] In some embodiments, after each quantum state evolution, the quality of the current solution needs to be evaluated by observing the objective function value. A quantum observer is used to map the objective function value to the measurement result. M , the goodness of the solution is given by the probability amplitude: ; in, Representing quantum states The probability of being measured, Representing quantum states The probability amplitude, argmax means returning the index corresponding to the maximum probability in a probability distribution.

[0060] The evaluation result of the objective function provides feedback for the next step of optimization.

[0061] S50: Introduce an adaptive multi-layer optimization strategy, which gradually converges to the global optimal value by adjusting the learning rate of the parameters, thereby balancing the capabilities of global search and local optimization.

[0062] In some embodiments, in order to balance the capabilities of global search and local optimization, an adaptive multi-layer optimization strategy is introduced. The learning rate , gradually converge to the global optimum: ; in, represents the updated optimization parameters at the t+1th iteration, represents the optimization parameter at the tth iteration, represents the estimated value of the objective function gradient, represents the learning rate, Indicated in the parameter Below, the quantum state obtained through the quantum circuit.

[0063] In the initial stage, a larger learning rate helps to quickly explore the solution space; in the later stage, a smaller learning rate improves the sophistication of the solution.

[0064] S60: After the quantum circuit is optimized, the observed values ​​are measured and extracted, and the global optimal solution is determined based on the observed values.

[0065] In some embodiments, after the quantum circuit is optimized, the system extracts the final result through measurement and determines the global optimal solution based on the observed value. The observed value is calculated as follows: ; in, represents the observed value; Indicates the optimal parameters The quantum states generated by the quantum circuit.

[0066] In some embodiments, based on steps S10 to S60, the classical calculation part further verifies whether the objective function value meets the optimization requirements. If the requirements are not met, the circuit parameters are adjusted through the classical feedback mechanism to verify whether the objective function value meets the optimization requirements. If the conditions are not met, the above steps S10 to S60 are iterated.

[0067] The embodiment of the present invention also provides an optimizer design device using the quantum tunneling effect, such as Figure 7 As shown, the device comprises: A quantum state conversion unit 701 is configured to convert the optimization problem into a quantum state through mathematical modeling; An iterative optimization unit 702, configured to implement iterative optimization of the quantum state through a parameterized quantum circuit; The mechanism introduction unit 703 is configured to introduce a quantum tunneling effect mechanism to jump out of a local optimal solution; A function evaluation unit 704 is configured to evaluate the quality of the current solution by observing the target function value after each quantum state evolution; The multi-layer optimization unit 705 is configured to introduce an adaptive multi-layer optimization strategy, and gradually converge to the global optimum by adjusting the learning rate of the parameters, thereby balancing the capabilities of global search and local optimization; The optimal solution determination unit 706 is configured to measure and extract observation values ​​after the quantum circuit is optimized, and determine the global optimal solution according to the observation values.

[0068] In some embodiments, the quantum state conversion unit is further configured as follows: For variables ,in represents the real number space, n Represents the dimension of the problem, the objective function of the variable To perform operations, quantum bits are introduced to represent the state of each variable; Through quantum amplitude encoding, variables x Each possible state of is mapped to a quantum state ,in: ; in, represents the index of the qubit state, Represents the components of a quantum state; Mapping the optimization problem into a quantum state construction problem, weights , represents the possibility of each solution; When initializing the quantum circuit, the Hadamard gate is applied to each bit to generate the initial uniform superposition state through the following formula: ; in, represents the initial uniform superposition state.

[0069] In some embodiments, the iterative optimization unit is further configured to: Through a series of parametric revolving doors and controlled doors CX The quantum variational circuit is formed, which is expressed as: ; in, represents a composite operation consisting of a series of parameterized quantum gates, Represents a quantum gate that rotates along the y-axis, with the rotation angle determined by the parameter Decide, represents the index of the rotating gate in the quantum circuit, represents the number of parameterized quantum gates in the quantum circuit; The parameterized circuit is used to update the quantum state, and the parameterized target state is formed by the following formula: ; in, It is composed of revolving door, controlled door CX And the composite operation composed of CZ gate, CZ gate is a control gate, represents the quantum state after optimization of the parameterized quantum circuit. The parameterized quantum circuit is optimized by adjusting the parameters accomplish.

[0070] In some embodiments, the mechanism introduction unit is further configured to: The basis of the quantum tunneling mechanism comes from the Schrödinger equation as shown below: ; in, represents the imaginary unit, represents the reduced Planck constant, The wave function of a quantum system represents the probability amplitude of a particle at a certain position and time. represents the time variable, represents the potential barrier function, represents the mass of the particle; Quantum tunneling effect mechanism via nonlinear operation of controlled rotation gate And double quantum state entanglement: ; in, Indicates a controlled revolving door, represents the phase rotation of a complex number, Represents the rotation angle or phase angle.

[0071] In some embodiments, the function evaluation unit is further configured to: Using quantum observers to map objective function values ​​to measurement results M , the goodness of the solution is given by the probability amplitude: ; in, Representing quantum states The probability of being measured, Representing quantum states The probability amplitude, argmax means returning the index corresponding to the maximum probability in a probability distribution.

[0072] In some embodiments, the multi-layer optimization unit is further configured to introduce an adaptive multi-layer optimization strategy to gradually converge to the global optimal value by adjusting the learning rate of the parameters through the following formula: ; in, represents the updated optimization parameters at the t+1th iteration, represents the optimization parameter at the tth iteration, represents the estimated value of the objective function gradient, represents the learning rate, Indicated in the parameter Below, the quantum state obtained through the quantum circuit.

[0073] In some embodiments, the optimal solution determination unit is further configured to measure and extract the observed value through the following formula after the quantum circuit is optimized: ; in, represents the observed value; Indicates the optimal parameters The quantum states generated by the quantum circuit.

[0074] In some embodiments, the apparatus further comprises a feedback adjustment unit, wherein the feedback adjustment unit is configured to: Based on the set optimization requirements, check whether the objective function value meets the optimization requirements. If the optimization requirements are not met, adjust the circuit parameters through the feedback mechanism until the objective function value meets the optimization requirements.

[0075] It should be noted that the structures of the optimizer design devices utilizing the quantum tunneling effect described in this embodiment belong to the same technical concept as the optimizer design method utilizing the quantum tunneling effect described previously, and achieve the same beneficial effects through the same principles, which will not be repeated here.

[0076] An embodiment of the present invention further provides a readable storage medium, which stores one or more programs. The one or more programs can be executed by one or more processors to implement the method described in any of the above embodiments.

[0077] The above description is intended to be illustrative rather than restrictive. For example, the above examples (or one or more of them) may be used in combination with each other. For example, a person of ordinary skill in the art may use other embodiments when reading the above description. In addition, in the above-mentioned specific embodiments, various features may be grouped together to simplify the present invention. This should not be interpreted as an intention that a feature of an invention that is not claimed for protection is necessary for any claim. On the contrary, the subject matter of the present invention may be less than all the features of an embodiment of a particular invention. Thus, the following claims are incorporated herein as examples or embodiments in the specific embodiments, wherein each claim is independently a separate embodiment, and it is considered that these embodiments may be combined with each other in various combinations or arrangements. The scope of the present invention should be determined with reference to the appended claims and the full scope of equivalent forms granted by these claims.

Claims

1. A method for designing an optimizer using the quantum tunneling effect, characterized in that: The method comprises: Convert the optimization problem into quantum state through mathematical modeling; Iteratively optimizing the quantum state by parameterizing the quantum circuit; Introduce quantum tunneling effect mechanism to escape from the local optimal solution; After each quantum state evolution, the quality of the current solution is evaluated by observing the value of the objective function; Introducing an adaptive multi-layer optimization strategy, by adjusting the learning rate of the parameters, gradually converges to the global optimal value, thereby balancing the capabilities of global search and local optimization; After the quantum circuit is optimized, the observation values ​​are measured and extracted, and the global optimal solution is determined based on the observation values.

2. The optimizer design method using the quantum tunneling effect according to claim 1, characterized in that: The optimization problem is converted into quantum state through mathematical modeling, including: For variables ,in represents the real number space, n Represents the dimension of the problem, the objective function of the variable To perform operations, quantum bits are introduced to represent the state of each variable; Through quantum amplitude encoding, variables x Each possible state of is mapped to a quantum state ,in: ； in, represents the index of the qubit state, Represents the components of a quantum state; Mapping the optimization problem into a quantum state construction problem, weights , represents the possibility of each solution; When initializing the quantum circuit, the Hadamard gate is applied to each bit to generate the initial uniform superposition state through the following formula: ； in, represents the initial uniform superposition state.

3. The optimizer design method using quantum tunneling effect according to claim 2, characterized in that: Iterative optimization of the quantum state is achieved through parameterized quantum circuits, including: Through a series of parametric revolving doors and controlled doors CX The quantum variational circuit is formed, which is expressed as: ； in, represents a composite operation consisting of a series of parameterized quantum gates, Represents a quantum gate that rotates along the y-axis, with the rotation angle determined by the parameter Decide, represents the index of the rotating gate in the quantum circuit, represents the number of parameterized quantum gates in the quantum circuit; The parameterized circuit is used to update the quantum state, and the parameterized target state is formed by the following formula: ； in, It is composed of revolving door, controlled door CX And the composite operation composed of CZ gate, CZ gate is a control gate, represents the quantum state after optimization of the parameterized quantum circuit. The parameterized quantum circuit is optimized by adjusting the parameters accomplish.

4. The optimizer design method using quantum tunneling effect according to claim 1, characterized in that: Introducing the quantum tunneling effect mechanism to jump out of the local optimal solution, including: The basis of the quantum tunneling mechanism comes from the Schrödinger equation as shown below: ； in, represents the imaginary unit, represents the reduced Planck constant, The wave function of a quantum system represents the probability amplitude of a particle at a certain position and time. represents the time variable, represents the potential barrier function, represents the mass of the particle; Quantum tunneling effect mechanism via nonlinear operation of controlled rotation gate And double quantum state entanglement: ； in, Indicates a controlled revolving door, represents the phase rotation of a complex number, Represents the rotation angle or phase angle.

5. The optimizer design method using quantum tunneling effect according to claim 1, characterized in that: After each quantum state evolution, the quality of the current solution is evaluated by observing the objective function value, including: Using quantum observers to map objective function values ​​to measurement results M , the goodness of the solution is given by the probability amplitude: ； in, Representing quantum states The probability of being measured, Representing quantum states The probability amplitude, argmax means returning the index corresponding to the maximum probability in a probability distribution.

6. The optimizer design method using quantum tunneling effect according to claim 1, characterized in that: An adaptive multi-layer optimization strategy is introduced to adjust the learning rate of the parameters through the following formula, gradually converging to the global optimal value: ； in, represents the updated optimization parameters at the t+1th iteration, represents the optimization parameter at the tth iteration, represents the estimated value of the objective function gradient, represents the learning rate, Indicated in the parameter Below, the quantum state obtained through the quantum circuit.

7. The optimizer design method using quantum tunneling effect according to claim 1, characterized in that: After the quantum circuit is optimized, the observed value is measured and extracted using the following formula: ； in, represents the observed value; Indicates the optimal parameters The quantum states generated by the quantum circuit.

8. The optimizer design method using quantum tunneling effect according to any one of claims 1 to 7, characterized in that: The method further comprises: Based on the set optimization requirements, check whether the objective function value meets the optimization requirements. If the optimization requirements are not met, adjust the circuit parameters through the feedback mechanism until the objective function value meets the optimization requirements.

9. An optimizer design device using quantum tunneling effect, characterized in that: The device comprises: A quantum state conversion unit, configured to convert the optimization problem into a quantum state through mathematical modeling; An iterative optimization unit, configured to implement iterative optimization of the quantum state through a parameterized quantum circuit; A mechanism introduction unit, configured to introduce a quantum tunneling effect mechanism to escape from a local optimal solution; A function evaluation unit, configured to evaluate the quality of the current solution by observing the target function value after each quantum state evolution; The multi-layer optimization unit is configured to introduce an adaptive multi-layer optimization strategy, which gradually converges to the global optimum by adjusting the learning rate of the parameters, thereby balancing the capabilities of global search and local optimization; The optimal solution determination unit is configured to measure and extract observation values ​​after the quantum circuit is optimized, and determine the global optimal solution based on the observation values. 10 . A non-transitory computer-readable storage medium storing instructions, which, when executed by a processor, perform the method according to claim 1 .

Images