Parallel optical Isin machine chip driven by single light source

By designing a parallel optical Isin machine chip driven by a single light source, the light source module and the branch module are used to divide the optical signals evenly, and iteratively update them in combination with electro-optical modulation, photoelectric detection and data processing modules, the existing optical Isin machine solution volume is solved The problems of large, high power consumption and high cost have achieved scale expansion and cost reduction, and have wide application prospects.

CN119002623APending Publication Date: 2024-11-22CHONGQING UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411069591.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-06
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The existing optical Isin machine solutions are large in size and high in power consumption, making it difficult to expand the scale of solving problems. The single-channel chip design is limited by the FPGA processing scale, resulting in limited scale of combined optimization problems and difficult to scale. The multi-channel light source method is relatively costly. High, difficult to widely use.

Method used

Design a parallel optical Isin machine chip driven by a single light source. Through the light source module, the branch module and multiple computing modules, the electro-optical modulation module, the photoelectric detection module and the data processing module are used to update the spin configuration iteratively, and the matrix multiplication is used. The computing realizes feedback and adjustment of spin signals, avoiding the method of integrating multiple light sources and reducing costs.

Benefits of technology

While increasing the scale of optical Isin machine chip solving problems, it greatly reduces operating costs, has great application prospects, and can effectively solve the combination optimization problem.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119002623A_ABST
    Figure CN119002623A_ABST
Patent Text Reader

Abstract

The invention discloses a parallel optical Isin machine chip driven by a single light source, which relates to the field of Isin machines and comprises a light source module, a shunt module, a light modulation module, a photoelectric detection module and a data processing module, the light source module is connected with the shunt modules, and each shunt module is sequentially connected with the light modulation module, the photoelectric detection module and the data processing module; the light source module generates a first light signal. The branching module equally divides the first optical signal into n paths of second optical signals; the electro-optical modulation module adjusts the second optical signal according to the feedback signal under the DC bias voltage to obtain a third optical signal; and the photoelectric detection module converts the third optical signal into a first photovoltage signal, preprocesses the first photovoltage signal to obtain a second photovoltage signal, determines an input voltage matrix, and performs matrix multiplication on the input voltage matrix and the coupling matrix to obtain a feedback signal. According to the method, the problem solving scale is improved, the operation cost of a chip is reduced, and the method has a relatively great application prospect.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of Ising machines, and in particular to a parallel optical Ising machine chip driven by a single light source. Background Art

[0002] Many problems in real life can be attributed to combinatorial optimization problems, such as flight route optimization, pharmaceutical manufacturing, combat planning, and logistics support. With the development of the information society, the scale of these combinatorial optimization problems is growing, and the existing von Neumann architecture-based computing solutions are difficult to effectively solve these problems. By mapping the combinatorial optimization problem to the Ising model, solving the combinatorial optimization problem in the process of finding the minimum Hamiltonian of the Ising model has been proven to be an optimal solution. Taking flight route optimization as an example, the Ising model represents the airports in the flight route network as nodes, and the flights as the connections between the nodes. Each airport node can be assigned a spin signal, which usually takes a value of +1 or -1, indicating a certain state of the airport. Define an objective function (Hamiltonian) to quantify the energy of the entire system. The goal is to find the optimal flight route by adjusting the spin variables to minimize the total energy of the system.

[0003] Ising machine refers to a physical system built according to the theoretical framework of the Ising model, which is used to simulate the solution of complex problems. Existing optical Ising machine solutions are mostly implemented using discrete components, such as space optical components, fiber optic components, etc. This solution is large in size, high in power consumption, and difficult to expand the scale of problem solving. In addition, there are some that use single-channel chip design, such as Figure 1 As shown in the figure, the single-channel chip includes a laser source (LD), an electro-optic modulator (MOD), a photodetector (PD) and a field programmable gate array (FPGA). The spin state is mapped by modulating the photovoltage x, and a single FPGA is used to perform matrix multiplication to generate a feedback signal f, which is then input into the MOD for iterative update. This single-channel chip design is limited by the scale of problems that the FPGA can handle, resulting in a limited scale of combinatorial optimization problems that are difficult to expand.

[0004] The above problems can be solved by integrating multiple channels on a single chip. There are many ways to implement the single-chip multi-channel integration solution. One of the implementation methods is to use a multi-channel light source. However, since laser light sources are relatively expensive, the multi-channel light source method is costly and difficult to be widely used. Therefore, it is necessary to develop a low-cost multi-channel Ising machine chip to meet the needs of solving combinatorial optimization problems. Summary of the invention

[0005] The purpose of the present application is to provide a parallel optical Ising machine chip driven by a single light source, which can reduce the cost of a multi-channel Ising machine chip.

[0006] To achieve the above objectives, this application provides the following solutions:

[0007] The present application provides a single-light-source-driven parallel optical Ising machine chip, comprising: a light source module, a branch module, and n computing modules; each of the computing modules comprises an electro-optical modulation module, a photoelectric detection module, and a data processing module connected in sequence, and an output end of the data processing module in each of the computing modules is also connected to a corresponding electro-optical modulation module; n>1; the light source module is connected to the branch module, and the branch module is connected to the electro-optical modulation module in each of the computing modules;

[0008] The light source module is used to generate a first optical signal; the branching module is used to equally divide the first optical signal into n second optical signals, and decompose the N spins automatically generated in the Ising model into The calculation module is used to update the spin configuration based on the second optical signal for K iterations; the spin configuration represents A collection of spin signals;

[0009] For the kth iteration, the electro-optical modulation module is used to:

[0010] Under the set DC bias voltage, the intensity of the second optical signal is adjusted according to the feedback signal output by the data processing module at the k-1th iteration to obtain the third optical signal at the kth iteration; 1<k≤K; for the kth iteration, the photoelectric detection module is used to:

[0011] Convert the third optical signal of the kth iteration into A first photovoltage signal;

[0012] For the kth iteration, the data processing module is used to:

[0013] For the kth iteration The first photovoltage signal is preprocessed to obtain the kth iteration A second photovoltage signal;

[0014] According to the kth iteration A second photovoltage signal determines the input voltage matrix of the kth iteration; wherein one of the second photovoltage signals is mapped to a spin signal, The spin signals form an input voltage matrix;

[0015] According to the input voltage matrix and the coupling matrix of the kth iteration, a feedback signal of the kth iteration is obtained by matrix multiplication operation; wherein the coupling matrix is ​​used to characterize the coupling loudness between different spin signals; and the feedback signal is used to characterize the feedback amount of each spin signal.

[0016] Optionally, the data processing module includes:

[0017] The AD conversion submodule is used to receive the kth iteration first photovoltage signals, and converting each of the first photovoltage signals into a photovoltage digital signal;

[0018] The preprocessing submodule is used to preprocess each of the photovoltage digital signals to obtain the kth iteration The second photovoltage signal, and according to the kth iteration The second photovoltage signal determines the input voltage matrix of the kth iteration;

[0019] A matrix multiplication submodule, used for performing a matrix multiplication operation on the input voltage matrix of the kth iteration and the coupling matrix to obtain an operation result;

[0020] The DA conversion submodule is used to convert the operation result into an analog signal, and determine the analog signal as the feedback signal of the kth iteration.

[0021] Optionally, the preprocessing submodule specifically includes:

[0022] The extreme value determination unit is used to determine the kth iteration a maximum value and a minimum value of the first photovoltage signal;

[0023] an average value determining unit, configured to calculate an average value of the maximum value and the minimum value;

[0024] A judging unit, used for judging the magnitude of each of the first photovoltage signals of the kth iteration and the average value;

[0025] A first adjustment unit, configured to adjust the first photovoltage signal to a positive value to obtain the second photovoltage signal when the first photovoltage signal is greater than or equal to the average value; the spin signal mapped by the second photovoltage signal is +1, and the corresponding spin state is spin-up;

[0026] A second adjustment unit is used for adjusting the first photovoltage signal to a negative value to obtain the second photovoltage signal when the first photovoltage signal is less than the average value; the spin signal mapped by the second photovoltage signal is -1, and the corresponding spin state is spin down;

[0027] The input voltage matrix determination unit is used to determine the k-th iteration The second photovoltage signal determines the input voltage matrix of the kth iteration.

[0028] Optionally, the spin signal obtained in the k+1th iteration and the feedback signal obtained in the kth iteration satisfy a nonlinear expression; the nonlinear expression is:

[0029]

[0030] Among them, x i '[k+1] represents the i-th spin signal mapped by the i-th photovoltage signal at the k+1th iteration; f i [k] represents the feedback amount of the i-th spin signal at the k-th iteration; When k = 1, f i [k] = 0; ζ i [k] is the noise generated by the system and satisfies Gaussian distribution.

[0031] Optionally, the matrix multiplication submodule specifically includes:

[0032] A coupling matrix generating unit, used for generating a coupling matrix according to coupling loudness of different spin signals;

[0033] Matrix multiplication unit, used to calculate Perform matrix multiplication operation on the input voltage matrix of the kth iteration and the coupling matrix to obtain the operation result; wherein, f i [k] represents the feedback amount of the i-th spin signal at the k-th iteration; x i [k] represents the spin signal of the i-th spin at the k-th iteration; x j [k] represents the jth spin signal at the kth iteration; And j≠i, α represents the feedback strength, β represents the coupling strength, and J represents the coupling matrix.

[0034] Optionally, the coupling matrix generating unit specifically includes:

[0035] The coupling matrix generating subunit is used to construct a coupling matrix in a manner that the coupling loudness of adjacent spin signals is -1 and the coupling loudness of non-adjacent spin signals is 0.

[0036] Optionally, the data processing module is a field programmable gate array.

[0037] Optionally, the branching module is an optical power splitter.

[0038] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0039] The present application provides a parallel optical Ising machine chip driven by a single light source, which adopts a light source module, divides the light source into multiple optical signals through a branching module, and sequentially connects an electro-optical modulation module, a photoelectric detection module and a data processing module with the light source module, and uses the data processing module for iterative feedback. Based on the chip structure, the method of using multiple light sources integration can be avoided, which improves the scale of solving problems of a single optical Ising machine chip and greatly reduces the operating cost of the optical Ising machine chip, and has great application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0041] Figure 1 This is the schematic diagram of the existing single-channel Ising machine;

[0042] Figure 2 A schematic diagram of the structure of a parallel optical Ising machine chip driven by a single light source provided in an embodiment of the present application;

[0043] Figure 3 A schematic diagram of a parallel optical Ising machine chip driven by a single light source provided in an embodiment of the present application;

[0044] Figure 4 A schematic diagram of matrix decomposition of a parallel optical Ising machine chip driven by a single light source is provided for one embodiment of the present application;

[0045] Figure 5 A comparison chart of Hamiltonian test results of a parallel optical Ising machine chip driven by a single light source in this application;

[0046] Figure 6 A spin amplitude chessboard diagram showing the test results of a parallel optical Ising machine chip driven by a single light source.

[0047] Reference numerals:

[0048] 1-light source module, 2-branching module, 3-electro-optical modulation module, 4-photoelectric detection module, 5-data processing module. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0050] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0051] See also Figure 2 The embodiment of the present application provides a parallel optical Ising machine chip driven by a single light source, including: a light source module 1, a branch module 2 and n computing modules; each of the computing modules includes an electro-optical modulation module 3, a photoelectric detection module 4 and a data processing module 5 connected in sequence, and the output end of the data processing module 5 in each of the computing modules is also connected to the corresponding electro-optical modulation module 3; n>1; the light source module 1 is connected to the branch module 2, and the branch module 2 is connected to the electro-optical modulation module 3 in each of the computing modules.

[0052] The light source module 1 is used to generate a first light signal; the branching module 2 is used to equally divide the first light signal into n second light signals, and decompose the N spins automatically generated in the Ising model into n spins. The calculation module is used to perform K iterations based on the second optical signal to update each spin configuration; the spin configuration represents A collection of spin signals.

[0053] For the kth iteration, the electro-optical modulation module 3 is used to: under a set DC bias voltage, adjust the intensity of the second optical signal according to the feedback signal output by the data processing module 5 in the k-1th iteration to obtain the third optical signal of the kth iteration; 1<k≤K; wherein, when k=1, the electro-optical modulation module 3 is used to adjust the intensity of the second optical signal under a set DC bias voltage to obtain the third optical signal of the 1st iteration.

[0054] For the kth iteration, the photoelectric detection module 4 is used to:

[0055] Convert the third optical signal of the kth iteration into A first photovoltage signal;

[0056] For the kth iteration, the data processing module 5 is used to: preprocessing the first photovoltage signals to obtain N second photovoltage signals of the kth iteration; and A second photovoltage signal determines the input voltage matrix of the kth iteration; wherein one of the second photovoltage signals is mapped to a spin signal, The invention relates to a method for obtaining a feedback signal of the kth iteration by using a matrix multiplication operation according to the input voltage matrix of the kth iteration and the coupling matrix; wherein the coupling matrix is ​​used to characterize the coupling loudness between different spin signals; and the feedback signal is used to characterize the feedback amount of each spin signal.

[0057] In an exemplary embodiment, the data processing module 5 includes:

[0058] The AD conversion submodule is used to receive the kth iteration a first photovoltage signal, and converting each of the first photovoltage signals into a photovoltage digital signal; a preprocessing submodule, for preprocessing each of the photovoltage digital signals to obtain a k-th iteration The second photovoltage signal, and according to the kth iteration The second photovoltage signal determines the input voltage matrix of the kth iteration; the matrix multiplication submodule is used to perform a matrix multiplication operation on the input voltage matrix of the kth iteration and the coupling matrix to obtain an operation result; the DA conversion submodule is used to convert the operation result into an analog signal, and determine the analog signal as the feedback signal of the kth iteration.

[0059] Since the electro-optical modulation module 3 operates at an ideal working point, the output light intensity has a certain basic value. Therefore, the first photovoltage value output after conversion by the photoelectric detection module 4 is a positive value. In order to map the state of the spin signal to ±1, it is necessary to perform certain numerical preprocessing in the data processing module 5, specifically, calculating the average of the maximum and minimum values ​​of all spins in each spin configuration, and updating the spin signal in the spin configuration whose spin signal is greater than the average value to +1 and -1. The spin state represented by +1 is spin-up, and the spin state represented by -1 is spin-down.

[0060] In an exemplary embodiment, the preprocessing submodule specifically includes:

[0061] The extreme value determination unit is used to determine the kth iteration a maximum value and a minimum value in the first photovoltage signals; an average value determining unit, used to calculate the average value of the maximum value and the minimum value; a judging unit, used to judge the size of each first photovoltage signal of the kth iteration and the average value; a first adjusting unit, used to adjust the first photovoltage signal to a positive value when the first photovoltage signal is greater than or equal to the average value, so as to obtain the second photovoltage signal; the spin signal mapped by the second photovoltage signal is +1, and the corresponding spin direction is spin-up; a second adjusting unit, used to adjust the first photovoltage signal to a negative value when the first photovoltage signal is less than the average value, so as to obtain the second photovoltage signal; the spin signal mapped by the second photovoltage signal is -1, and the corresponding spin direction is spin-down; an input voltage matrix determining unit, used to determine the input voltage matrix of the kth iteration according to the N second photovoltage signals of the kth iteration.

[0062] In an exemplary embodiment, the matrix multiplication submodule specifically includes:

[0063] A coupling matrix generating unit is used to generate a coupling matrix according to the coupling loudness of different spin signals, and the coupling matrix is ​​a fixed value; a matrix multiplication operation unit is used to generate a coupling matrix according to the coupling loudness of different spin signals, and the coupling matrix is ​​a fixed value; Perform matrix multiplication operation on the input voltage matrix of the kth iteration and the coupling matrix to obtain the operation result; wherein, f i [k] represents the feedback amount of the i-th spin signal at the k-th iteration; x i [k] represents the spin signal of the i-th spin at the k-th iteration; x j [k] represents the jth spin signal at the kth iteration; And j≠i, α represents the feedback strength, β represents the coupling strength, and J represents the coupling matrix.

[0064] In an exemplary embodiment, the coupling matrix generating unit specifically includes:

[0065] The coupling matrix generating subunit is used to construct a coupling matrix in a manner that the coupling loudness of adjacent spin signals is -1 and the coupling loudness of non-adjacent spin signals is 0.

[0066] In an exemplary embodiment, see Figure 4 For n = 2, the Ising model randomly generates an N × N coupling matrix, which is decomposed into 2 spins after being decomposed by the optical power splitter. The spin configuration of The sub-coupling matrices are denoted by J sub1 , J sub2 , J sub3 , J sub4 , where Jsub1 and J sub2 represent the first sub-coupling matrix and the second sub-coupling matrix generated by all spins in the first spin configuration respectively; J sub3 and J sub4 They represent the third sub-coupling matrix and the fourth sub-coupling matrix generated by all spins in the second spin configuration respectively.

[0067] Furthermore, two data processing modules 5 are used to perform matrix multiplication operations respectively, and the specific operation formula is:

[0068]

[0069] Among them, f1[k] is the feedback signal of the first spin configuration at the kth iteration; f2[k] is the feedback signal of the second spin configuration at the kth iteration, x1[k] is the first group of spin configurations; x1[k] is the second group of spin configurations.

[0070] In addition, the feedback signal output by the data processing module 5 is initially set to 0 and applied to the electro-optical modulation module 3 to obtain the input voltage matrix of the first iteration. Due to the noise in the system, the spin signal in the input voltage matrix of the first iteration is a set of random numbers that satisfy the Gaussian distribution. The input voltage matrix of the first iteration is matrix multiplied by the coupling matrix through the data processing module 5 to obtain the feedback signal of the first iteration, and the feedback signal is input into the electro-optical modulation module 3 for iteration.

[0071] As the number of iterations increases, the spin signal corresponding to each spin in each spin configuration gradually bifurcates from all being near the initial value to half being positive and half being negative, and gradually fixes to two positive and negative values ​​with the same absolute value. The corresponding Hamiltonian gradually decreases, eventually decreasing to the lowest energy and remaining constant, obtaining the optimal solution to the Ising problem.

[0072] Furthermore, for the case of n>2, the above decomposition method can be generalized.

[0073] In an exemplary embodiment, the spin signal obtained at the k+1th iteration and the feedback signal obtained at the kth iteration satisfy a nonlinear expression; the nonlinear expression is:

[0074]

[0075] Among them, x i [k+1] represents the i-th spin signal mapped by the i-th photovoltage signal at the k+1th iteration; f i [k] represents the feedback amount of the i-th spin signal at the k-th iteration; when k = 1, f i [k] = 0; ζ i[k] is the noise generated by the system and satisfies Gaussian distribution; is the DC bias voltage corresponding to the electro-optic modulation module 3 operating in an ideal state.

[0076] In an exemplary embodiment, see Figure 3 The light source module 1 includes a laser source LD 1. The branching module 2 is a branching device 2. The electro-optical modulation module 3 is an electro-optical modulator 3, and the n electro-optical modulation modules 3 are respectively represented as MOD1, MOD2, MOD3, ..., and MOD n The photoelectric detection module 4 is a photoelectric detector 4, and the plurality of photoelectric detection modules 4 are respectively represented as PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9, PD10, PD11, PD22, PD33, PD n The data processing module 5 uses a field programmable gate array 5, and n field programmable gate arrays are represented as FPGA1, FPGA2, FPGA3...FPGA n .

[0077] In an exemplary embodiment, an Ising model is constructed based on entity data of a practical problem (such as a flight route optimization problem). An initial spin configuration is randomly generated based on the constructed Ising model, and the initial spin configuration is modulated to the intensity of the Gaussian beam of the Ising machine, so as to encode the spin of each point in the lattice of the Ising model according to the intensity of the beam.

[0078] This application is different from the existing Figure 1 The experimental test comparison results of the single-channel Ising machine Hamiltonian are shown in Figure 5 As shown. The experiment sets the number of spins N = 100 and the total number of iterations K = 100. It can be seen that compared with the single-channel Ising machine, the Hamiltonian of the Ising machine of the present application has decreased significantly, and can be reduced to the lowest energy -200, indicating that the Ising problem with a larger number of spins can be solved. Figure 6 The spin amplitude chessboard diagram of the test results of this application is displayed. Each grid point represents each spin, and the color represents the spin amplitude. It can be seen that the spins with positive and negative amplitudes are alternately and evenly distributed on the chessboard diagram, which also shows that the Hamiltonian of the Ising machine of this application can be reduced to the lowest energy.

[0079] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0080] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A parallel optical Ising machine chip driven by a single light source, characterized in that: The parallel optical Ising machine chip driven by a single light source comprises: a light source module, a branch module and n computing modules; each computing module comprises an electro-optical modulation module, a photoelectric detection module and a data processing module connected in sequence, and the output end of the data processing module in each computing module is also connected to the corresponding electro-optical modulation module; n>1; the light source module is connected to the branch module, and the branch module is connected to the electro-optical modulation module in each computing module; The light source module is used to generate a first light signal; the branching module is used to equally divide the first light signal into n second light signals, and decompose the N spin signals automatically generated in the Ising model into n spin signals. The calculation module is used to perform K iterations based on the second optical signal to update each spin configuration; the spin configuration represents The collection of spin states of a spin signal; For the kth iteration, the electro-optical modulation module is used to: Under the set DC bias voltage, the intensity of the second optical signal is adjusted according to the feedback signal output by the data processing module at the k-1th iteration to obtain the third optical signal at the kth iteration; 1<k≤K; For the kth iteration, the photodetection module is used to: Convert the third optical signal of the kth iteration into A first photovoltage signal; For the kth iteration, the data processing module is used to: For the kth iteration The first photovoltage signal is preprocessed to obtain the kth iteration A second photovoltage signal; According to the kth iteration The second photovoltage signal determines the input voltage matrix of the kth iteration; wherein, one of the second photovoltage signals is mapped to a spin signal, and one spin signal corresponds to one spin state, The spin signals form an input voltage matrix; According to the input voltage matrix and coupling matrix of the kth iteration, the feedback signal of the kth iteration is obtained by matrix multiplication operation; wherein the coupling matrix is ​​used to characterize the relationship between different spin signals. The feedback signal is used to characterize the feedback amount of each spin signal.

2. The single light source driven parallel optical Ising machine chip according to claim 1, characterized in that: The data processing module includes: The AD conversion submodule is used to receive the kth iteration first photovoltage signals, and converting each of the first photovoltage signals into a photovoltage digital signal; The preprocessing submodule is used to preprocess each of the photovoltage digital signals to obtain the kth iteration The second photovoltage signal, and according to the kth iteration The second photovoltage signal determines the input voltage matrix of the kth iteration; A matrix multiplication submodule, used for performing a matrix multiplication operation on the input voltage matrix of the kth iteration and the coupling matrix to obtain an operation result; The DA conversion submodule is used to convert the operation result into an analog signal, and determine the analog signal as the feedback signal of the kth iteration.

3. The single light source driven parallel optical Ising machine chip according to claim 1, characterized in that: The preprocessing submodule specifically includes: The extreme value determination unit is used to determine the kth iteration a maximum value and a minimum value of the first photovoltage signal; an average value determining unit, configured to calculate an average value of the maximum value and the minimum value; A judging unit, used for judging the magnitude of each of the first photovoltage signals of the kth iteration and the average value; A first adjustment unit, configured to adjust the first photovoltage signal to a positive value to obtain the second photovoltage signal when the first photovoltage signal is greater than or equal to the average value; the spin signal mapped by the second photovoltage signal is +1, and the spin state of the corresponding spin signal in the spin configuration is updated to spin up; a second adjustment unit, configured to adjust the first photovoltage signal to a negative value to obtain the second photovoltage signal when the first photovoltage signal is less than the average value; the spin signal mapped by the second photovoltage signal is -1, and the spin state of the corresponding spin signal in the spin configuration is updated to spin down; The input voltage matrix determination unit is used to determine the k-th iteration The second photovoltage signal determines the input voltage matrix of the kth iteration.

4. The single light source driven parallel optical Ising machine chip according to claim 1, characterized in that: The spin signal obtained at the k+1th iteration and the feedback signal obtained at the kth iteration satisfy a nonlinear expression; the nonlinear expression is: Among them, x i [k+1] represents the i-th spin signal mapped by the i-th photovoltage signal at the k+1th iteration; f i [k] represents the feedback amount of the i-th spin signal at the k-th iteration; When k = 1, f i [k] = 0; ζ i [k] is the noise generated by the system and satisfies Gaussian distribution.

5. The single light source driven parallel optical Ising machine chip according to claim 2, characterized in that: The matrix multiplication submodule specifically includes: A coupling matrix generating unit, used for generating a coupling matrix according to coupling loudness of different spin signals; Matrix multiplication unit, used to calculate Perform matrix multiplication operation on the input voltage matrix of the kth iteration and the coupling matrix to obtain the operation result; wherein, f i [k] represents the feedback amount of the i-th spin signal at the k-th iteration; x i [k] represents the spin signal of the i-th spin at the k-th iteration; x j [k] represents the jth spin signal at the kth iteration; And j≠i, α represents the feedback strength, β represents the coupling strength, and J represents the coupling matrix.

6. The single light source driven parallel optical Ising machine chip according to claim 5, characterized in that: The coupling matrix generating unit specifically comprises: The coupling matrix generating subunit is used to construct a coupling matrix in a manner that the coupling loudness of adjacent spin signals is -1 and the coupling loudness of non-adjacent spin signals is 0.

7. The single light source driven parallel optical Ising machine chip according to claim 1, characterized in that: The data processing module is a field programmable gate array.

8. The single light source driven parallel optical Ising machine chip according to claim 1, characterized in that: The branching module is an optical power splitter.