Method, device, storage medium and equipment for quantizing combinatorial optimization problems

Abstract

The application discloses a quantum solution method and device for a combination optimization problem, a storage medium and computer equipment. The quantum solution method comprises the following steps: mapping a loss function corresponding to a combination optimization problem to be solved into a whole Hamiltonian; decomposing the whole Hamiltonian into a plurality of sub-Hamiltonians, and the sum of the quantum bit numbers of each sub-Hamiltonian being equal to the quantum bit number of the whole Hamiltonian; obtaining eigenvalues and corresponding eigenstates of each sub-Hamiltonian satisfying a preset combination condition based on a variational quantum algorithm, taking the product of the eigenvalues of each sub-Hamiltonian as the eigenvalue of the whole Hamiltonian, and taking the tensor product of the eigenstates of each sub-Hamiltonian as the eigenstate of the whole Hamiltonian. The method converts the measurement of the whole Hamiltonian into the measurement of the plurality of sub-Hamiltonians, so that the optimal solution of the combination optimization problem to be solved is obtained, and a larger-scale problem is calculated and a larger-scale system is simulated by using fewer quantum resources.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of quantum computing, and in particular relates to a quantumized solving method and device for a combinatorial optimization problem, a computer readable storage medium and a computer device. BACKGROUND

[0002] A combinatorial optimization problem (COP) refers to a type of optimization problem of finding an optimal solution in a set of a limited number of feasible solutions. Typical combinatorial optimization problems include the traveling salesman problem (TSP), the knapsack problem, the graph coloring problem, etc. Combinatorial optimization problems have important applications in the fields of information technology, economic management, industrial engineering, traffic operation, and communication networks.

[0003] With the continuous development of scientific information technology, when most actual problems are converted into combinatorial optimization problems, the decision variables and constraint conditions in the combinatorial optimization problems may reach millions, resulting in an exponential increase in the size of the feasible solution set. When some specific problems reach a certain scale, the computing capacity of current classical computers is already difficult to meet.

[0004] A quantum computer is a physical device that performs operations, stores and processes quantum information according to quantum mechanics. Quantum computing uses quantum superposition and quantum entanglement for computation, and quantum computing has been proven to have computing capacity that exceeds classical computing and can solve problems that are difficult for classical computing. For example, a quantum computer can be applied to the solving process of a combinatorial optimization problem. However, although quantum computing has far superior computing capacity to classical computing, current quantum computers are considered to be noisy intermediate-scale quantum computers due to the limited number of qubits and the coherence time of qubits. Correspondingly, the current development stage is also considered to be the noisy intermediate-scale quantum computing era (NISQ). When a combinatorial optimization problem is relatively complex, a quantum system with a large number of qubits is needed to solve it, which on the one hand requires a very large quantum resource to build a quantum system with a large number of qubits, and on the other hand a quantum system with a large number of qubits is also less stable.

[0005] Therefore, how to solve a combinatorial optimization problem using fewer quantum resources is a technical problem that the field desires to solve. SUMMARY

[0006] The technical problem solved by the present application is how to solve a large-scale combinatorial optimization problem using fewer quantum resources.

[0007] The application discloses a quantum solution method for a combination optimization problem, and the quantum solution method comprises the following steps:

[0008] mapping a loss function corresponding to the combination optimization problem to be solved into a whole Hamiltonian;

[0009] decomposing the whole Hamiltonian into a plurality of sub-Hamiltonians, and a sum of quantum bit numbers of each sub-Hamiltonian being equal to a quantum bit number of the whole Hamiltonian;

[0010] obtaining eigenvalues and corresponding eigenstates of each sub-Hamiltonian satisfying a preset combination condition based on a variational quantum algorithm, taking a product of the eigenvalues of each sub-Hamiltonian as an eigenvalue of the whole Hamiltonian, and taking a tensor product of the eigenstates of each sub-Hamiltonian as an eigenstate of the whole Hamiltonian.

[0011] Preferably, the quantum solution method further comprises constructing a loss function corresponding to the combination optimization problem to be solved:

[0012] defining an objective function f(x), a constraint condition g(x) and a variable x according to the combination optimization problem to be solved;

[0013] constructing a loss function C(x) as follows:

[0014] C(x) = f(x) + A*D(g(x))

[0015] wherein D(g(x)) is a penalty function, D(g(x)) is related to the constraint condition g(x), and A is a penalty coefficient.

[0016] Preferably, the method for decomposing the whole Hamiltonian into a plurality of sub-Hamiltonians is as follows:

[0017]

[0018] wherein H is a whole Hamiltonian, represents a Hamiltonian component acting on the wth subsystem, i.e., the wth sub-Hamiltonian, w is a positive integer and 1≤w≤n, and n is the number of sub-Hamiltonians.

[0019] Preferably, the preset combination condition is that a product of the eigenvalues of each sub-Hamiltonian converges.

[0020] Preferably, the method for obtaining the eigenvalues and corresponding eigenstates of each sub-Hamiltonian satisfying the preset combination condition based on the variational quantum algorithm comprises:

[0021] obtaining each parameterized quantum state matched with each of the sub-Hamiltonians in the current iteration round, and measuring eigenvalues of each of the sub-Hamiltonians based on each of the parameterized quantum states;

[0022] determining whether the eigenvalues of each of the sub-Hamiltonians in the current iteration round satisfy the predetermined combination condition;

[0023] if yes, taking each parameterized quantum state in the current iteration round as an eigenstate of each sub-Hamiltonian;

[0024] if no, updating each parameterized quantum state of the next iteration round according to the product of the eigenvalues of each Hamiltonian in the current iteration round, measuring the eigenvalues of each of the sub-Hamiltonians based on each parameterized quantum state of the next iteration round, repeating the iteration until the eigenvalues of each of the sub-Hamiltonians satisfy the predetermined combination condition, and taking each parameterized quantum state in the last iteration round as an eigenstate of each sub-Hamiltonian.

[0025] Preferably, the method for obtaining each parameterized quantum state matched with each of the sub-Hamiltonians in the current iteration round is as follows:

[0026] if the current iteration round is the first round, randomly initializing the parameters of each parameterized quantum circuit constructed in advance and then applying each parameterized quantum circuit to each initial quantum state to obtain each parameterized quantum state matched with each of the sub-Hamiltonians;

[0027] if the current iteration round is not the first round, updating the parameters of each parameterized quantum circuit according to the product of the eigenvalues of each of the sub-Hamiltonians measured in the last iteration round, and then applying each updated parameterized quantum circuit to each initial quantum state to obtain each parameterized quantum state matched with each of the sub-Hamiltonians in the current iteration round.

[0028] Preferably, the quantum solution method further comprises: constructing n parameterized quantum circuits and n initial quantum states in advance, wherein the number of quantum bits of each parameterized quantum circuit and initial quantum state is equal to the number of quantum bits of each sub-Hamiltonian, and each parameterized quantum circuit is constructed by a quantum rotation gate R y (θ i ), a quantum rotation gate R z (θ j ), a quantum rotation gate R y (θ k ) and a controlled non-gate CNOT gate. l Preferably, the parameters are randomly generated in a group with a size of 3*N Preferably, the quantum rotation gate R y (θ i ), the quantum rotation gate R z (θj ), R y (θ k ) in the relationship between the coefficients i, j, k and the qth qubit is:

[0029] i = q * 3 + 1,

[0030] j = q * 3 + 2,

[0031] k = q * 3 + 3,

[0032] wherein a set of parameters of size 3*N l is randomly generated q ∈ [0, N l -1], N l is the number of qubits of the lth Hamiltonian, and each two qubits are subjected to a controlled non-gate (CNOT gate), and the number of CNOT gates is equal to the number of qubits.

[0033] The application further discloses a quantumized solving device for a combinatorial optimization problem, which comprises:

[0034] a mapping module, configured to map a loss function corresponding to a combinatorial optimization problem to be solved into an overall Hamiltonian;

[0035] a decomposition module, configured to decompose the overall Hamiltonian into a plurality of Hamiltonians, and the sum of the number of qubits of each Hamiltonian is equal to the number of qubits of the overall Hamiltonian;

[0036] a solving module, configured to obtain eigenvalues and corresponding eigenstates of each Hamiltonian satisfying a preset combination condition based on a variational quantum algorithm, take the product of the eigenvalues of each Hamiltonian as the eigenvalue of the overall Hamiltonian, and take the tensor product of the eigenstates of each Hamiltonian as the eigenstate of the overall Hamiltonian.

[0037] The application further discloses a computer readable storage medium, which stores a quantumized solving program for a combinatorial optimization problem, and the quantumized solving program for the combinatorial optimization problem realizes the quantumized solving method for the combinatorial optimization problem when executed by a processor.

[0038] The application further discloses a computer device, which comprises a computer readable storage medium, a processor and a quantumized solving program for a combinatorial optimization problem stored in the computer readable storage medium, and the quantumized solving program for the combinatorial optimization problem realizes the quantumized solving method for the combinatorial optimization problem when executed by the processor.

[0039] The present invention discloses a method and device for quantizing a combinatorial optimization problem, which have the following technical effects:

[0040] After obtaining the overall Hamiltonian corresponding to the combinatorial optimization problem to be solved, the overall Hamiltonian with a larger number of quantum bits is decomposed to obtain sub-Hamiltonians with a smaller number of quantum bits. The expected value measurement of the overall Hamiltonian is converted into the measurement of each sub-Hamiltonian, thereby determining the optimal eigenvalue and optimal eigenstate of the overall Hamiltonian, and finally obtaining the optimal solution to the combinatorial optimization problem, reducing the required quantum resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a flow chart of a method for solving a combinatorial optimization problem using quantization according to a first embodiment of the present invention;

[0042] Figure 2 Schematic diagram of preparing a parameterized quantum state in Example 1 of the present invention;

[0043] Figure 3 for Figure 2 A schematic diagram of an embodiment of a parameterized quantum circuit (PQC) is shown;

[0044] Figure 4 A schematic diagram of a device for solving a combinatorial optimization problem using quantization according to a second embodiment of the present invention;

[0045] Figure 5 Schematic diagram of a computer device according to a fourth embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0047] Before describing the various embodiments of the present application in detail, the technical concept of the present application is briefly described first: In the prior art, when using quantization methods to solve combinatorial optimization problems, when the problem to be solved is relatively complex, it often requires a huge amount of quantum resources. To this end, the quantization solution method for combinatorial optimization problems provided by the present application, after obtaining the overall Hamiltonian corresponding to the combinatorial optimization problem to be solved, decomposes the overall Hamiltonian with a larger number of quantum bits to obtain sub-Hamiltonians with a smaller number of quantum bits, converts the expected value measurement of the overall Hamiltonian into the measurement of each sub-Hamiltonian, and determines the optimal eigenvalue and optimal eigenstate of the overall Hamiltonian, ultimately obtaining the optimal solution to the combinatorial optimization problem, reducing the required quantum resources.

[0048] Specifically, if Figure 1As shown, the quantum solving method of the combination optimization problem of the first embodiment includes the following steps:

[0049] Step S10, mapping the loss function corresponding to the combination optimization problem to be solved into an overall Hamiltonian;

[0050] Step S20, decomposing the overall Hamiltonian into a plurality of sub-Hamiltonians, and the sum of the quantum bits of each sub-Hamiltonian is equal to the quantum bits of the overall Hamiltonian;

[0051] Step S30, obtaining the eigenvalues and corresponding eigenstates of each sub-Hamiltonian satisfying the preset combination condition based on the variational quantum algorithm, taking the product of the eigenvalues of each sub-Hamiltonian as the eigenvalue of the overall Hamiltonian, and taking the tensor product of the eigenstates of each sub-Hamiltonian as the eigenstate of the overall Hamiltonian.

[0052] Specifically, before performing step S10, the first embodiment also needs to obtain the loss function corresponding to the combination optimization problem to be solved, and the specific process is as follows: according to the objective function f(x), the constraint condition g(x) and the variable x of the combination optimization problem to be solved, the loss function C(x) is constructed as follows:

[0053] C(x) = f(x) + A*D(g(x))

[0054] Where D(g(x)) is a penalty function, the penalty function is related to the constraint condition g(x), and A is a penalty coefficient.

[0055] Further, the loss function C(x) is mapped into an overall Hamiltonian. The essence of quantum solving is to convert the solving process of the loss function C(x) into an expected value measurement process of the overall Hamiltonian.

[0056] Specifically, the variable x is encoded by the bit string X = x0x1…x v , where the size of the bit string X is determined by the nature of the specific combination optimization problem to be solved. The x v in the bit string X = x0x1…x p (p∈[0,v]) is mapped into a combination of unit matrix I and Pauli matrix Z, that is, , where Z p is the Pauli matrix Z acting on the pth quantum bit, and its matrix form is: is the unit matrix, thereby mapping the loss function C(x) into the Hamiltonian H.

[0057] In step S20, the method for decomposing the overall Hamiltonian into a plurality of sub-Hamiltonians is to decompose the overall Hamiltonian as follows:

[0058]

[0059] Where H is the overall Hamiltonian, represents the Hamiltonian component acting on the w-th subsystem, that is, the w-th sub-Hamiltonian, w is a positive integer and 1≤w≤n, and n is the number of sub-Hamiltonians.

[0060] In step S30, the method for obtaining the eigenvalues ​​and corresponding eigenstates of each sub-Hamiltonian that meets the preset combination conditions based on the variational quantum algorithm includes the following steps:

[0061] Step S301: Obtain each parameterized quantum state that matches each sub-Hamiltonian in the current iteration round, and measure the eigenvalue of each sub-Hamiltonian based on each parameterized quantum state. Figure 2 As shown, n parameterized quantum circuits and n initial quantum states are pre-constructed, where the number of quantum bits of each parameterized quantum circuit and initial quantum state is equal to the number of quantum bits of each sub-Hamiltonian. Figure 3 As shown in Figure 2, a parameterized quantum circuit (PQC) with the same number of quantum bits as the sub-Hamiltonian (N=5) is constructed. The parameterized quantum circuit consists of a quantum rotation gate R y (θ i ), R z (θ j ), R y (θ k ) and control NOT gate CNOT gate to randomly generate a set of 3*N l Parameters Then the quantum revolving door R y (θ i ), R z (θ j ), R y (θ k ) The relationship between the coefficients i, j, k and the qth quantum bit is:

[0062] i=q*3+1,

[0063] j=q*3+2,

[0064] k=q*3+3,

[0065] in To randomly generate a set of size 3*N l Parameters q∈[0,N l -1],N l is the number of quantum bits of the lth sub-Hamiltonian, every two bits act as a controlled NOT gate (CNOT gate), and the number of controlled NOT gates is equal to the number of quantum bits.

[0066] If the current iteration round m is the first round, i.e., m = 1, the parameters of each parameterized quantum circuit are updated according to the product of the eigenvalues of each sub-Hamiltonian H l

[0067] If the current iteration round m is not the first round, i.e., m > 1, the parameters of each parameterized quantum circuit are updated according to the product of the eigenvalues of each sub-Hamiltonian H l

[0068] Further, the eigenvalues of each sub-Hamiltonian are measured based on each parameterized quantum state. For example, the eigenvalues of each sub-Hamiltonian in the current iteration round m are measured in a quantum computer as follows:

[0069]

[0070]

[0071] …

[0072]

[0073] Step S302, it is determined whether the eigenvalues of each sub-Hamiltonian in the current iteration round m satisfy a predetermined combination condition. Specifically, the product of the eigenvalues of each sub-Hamiltonian is calculated in a classical computer as follows:

[0074]

[0075] wherein the predetermined combination condition is that the product of the eigenvalues of each sub-Hamiltonian converges. For example, the absolute value of the difference between the product of the eigenvalues of each sub-Hamiltonian in the current iteration round m and the product of the eigenvalues of each sub-Hamiltonian in the last iteration round m-1 is less than δ, i.e., converges, wherein the value of δ is determined by the actual problem.

[0076] Step S303, when ​​​​​​​​​When the convergence, i.e. the eigenvalues of each sub-Hamiltonian under the current iteration round m satisfy the predetermined combination condition, each parameterized quantum state under the current iteration round m is taken as the eigenstate of each sub-Hamiltonian respectively

[0077] In step S304, when When the non-convergence, i.e. the eigenvalues of each sub-Hamiltonian under the current iteration round m do not satisfy the predetermined combination condition, according to the method of classical optimization (such as gradient descent optimization method, gradient-free optimization method), the product of the eigenvalues of each sub-Hamiltonian under the current iteration round m is updated as The parameters of each parameterized quantum circuit in the next iteration round m+1 are updated Each parameterized quantum circuit after updating the parameters is used to act on each initial quantum state respectively Each parameterized quantum state in the next iteration round m+1 is obtained Similarly, the product of the eigenvalues of each sub-Hamiltonian in the next iteration round m+1 is calculated by measurement The iteration is repeated until the eigenvalues of each sub-Hamiltonian satisfy the predetermined combination condition, i.e. satisfy And each parameterized quantum state under the last iteration round is taken as the eigenstate of each sub-Hamiltonian respectively.

[0078] Exemplarily, assuming that the eigenvalues of each sub-Hamiltonian satisfying the preset combination condition and the corresponding eigenstates are obtained, the parameters of each parameterized quantum circuit are The eigenvalues of each sub-Hamiltonian at this time can be expressed as:

[0079]

[0080]

[0081] …

[0082]

[0083] The eigenvalue of the overall Hamiltonian is:

[0084]

[0085] The eigenstate of the overall Hamiltonian is:

[0086]

[0087] At this time, since the eigenvalues of each sub-Hamiltonian and the corresponding eigenstates are the overall optimal combination, the eigenvalue of the overall Hamiltonian is the optimal eigenvalue, and the eigenstate of the overall Hamiltonian is the optimal eigenstate.

[0088] The quantum solution method of the combination optimization problem disclosed in the first embodiment decomposes the overall Hamiltonian with a large number of quantum bits to obtain a sub-Hamiltonian with a smaller number of quantum bits, converts the measurement of the expected value of the overall Hamiltonian into the measurement of each sub-Hamiltonian, determines the optimal eigenvalue and the optimal eigenstate of the overall Hamiltonian, and finally obtains the optimal solution of the combination optimization problem, thereby reducing the required quantum resources and providing more possibilities for simulating a larger quantum system by using the current NISQ quantum computer.

[0089] As shown in the second embodiment, Figure 4 The quantum solution device for the combination optimization problem disclosed in the second embodiment includes a mapping module 100, a decomposition module 200, and a solution module 300. The mapping module 100 is configured to map the loss function corresponding to the combination optimization problem to be solved into an overall Hamiltonian. The decomposition module 200 is configured to decompose the overall Hamiltonian into a plurality of sub-Hamiltonians, and the sum of the number of quantum bits of each sub-Hamiltonian is equal to the number of quantum bits of the overall Hamiltonian. The solution module 300 is configured to obtain the eigenvalue and the corresponding eigenstate of each sub-Hamiltonian that satisfies the preset combination condition based on a variational quantum algorithm, take the product of the eigenvalues of each sub-Hamiltonian as the eigenvalue of the overall Hamiltonian, and take the tensor product of the eigenstates of each sub-Hamiltonian as the eigenstate of the overall Hamiltonian.

[0090] The more detailed processing procedures of the mapping module 100, the decomposition module 200, and the solution module 300 can be referred to the description of the first embodiment, and will not be described here.

[0091] The third embodiment further discloses a computer readable storage medium storing a quantum solution program for the combination optimization problem, and the quantum solution program for the combination optimization problem is executed by a processor to implement the quantum solution method for the combination optimization problem described above.

[0092] The fourth embodiment further discloses a computer device, which, at the hardware level, includes a processor and a memory, Figure 5As shown, the computer device includes a processor 12, an internal bus 13, a network interface 14, and a computer-readable storage medium 11. The processor 12 reads the corresponding computer program from the computer-readable storage medium and then runs it, forming a request processing device at the logical level. Of course, in addition to software implementation, one or more embodiments of this specification do not exclude other implementation methods, such as logic devices or a combination of software and hardware, etc., that is, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices. The computer-readable storage medium 11 stores a quantized solution program for the combinatorial optimization problem, and when the quantized solution program for the combinatorial optimization problem is executed by the processor, it implements the above-mentioned quantized solution method for the combinatorial optimization problem.

[0093] Computer-readable storage media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer-readable storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage, quantum memory, graphene-based storage media or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0094] The above describes in detail the specific implementation methods of the present invention. Although some embodiments have been shown and described, those skilled in the art should understand that these embodiments can be modified and improved without departing from the principles and spirit of the present invention, the scope of which is defined by the claims and their equivalents. These modifications and improvements should also be within the scope of protection of the present invention.

Claims

1. A quantized solution method for combinatorial optimization problems, characterized in that: The quantization solution method includes: Map the loss function corresponding to the combinatorial optimization problem to be solved into the overall Hamiltonian; Decomposing the overall Hamiltonian into a plurality of sub-Hamiltonians, wherein the sum of the number of qubits of each sub-Hamiltonian is equal to the number of qubits of the overall Hamiltonian; Obtaining the eigenvalues ​​and corresponding eigenstates of each sub-Hamiltonian that meets a preset combination condition based on a variational quantum algorithm, taking the product of the eigenvalues ​​of each sub-Hamiltonian as the eigenvalue of the overall Hamiltonian, and taking the tensor product of the eigenstates of each sub-Hamiltonian as the eigenstate of the overall Hamiltonian; The preset combination condition is that the product of the eigenvalues ​​of each sub-Hamiltonian converges, and the method for obtaining the eigenvalues ​​and corresponding eigenstates of each sub-Hamiltonian that meets the preset combination condition based on the variational quantum algorithm includes: Obtaining each parameterized quantum state that matches each of the sub-Hamiltonians in the current iteration round, and measuring the eigenvalues ​​of each of the sub-Hamiltonians based on each of the parameterized quantum states; Determine whether the eigenvalues ​​of each of the sub-Hamiltonians in the current iteration round meet the preset combination conditions; If so, each parameterized quantum state in the current iteration round is used as the eigenstate of each sub-Hamiltonian; If not, the parameterized quantum states of the next iteration round are updated according to the product of the eigenvalues ​​of the sub-Hamiltonians in the current iteration round, and the eigenvalues ​​of the sub-Hamiltonians are respectively measured according to the parameterized quantum states of the next iteration round. The iteration is repeated until the eigenvalues ​​of the sub-Hamiltonians meet the preset combination conditions, and the parameterized quantum states of the last iteration round are respectively used as the eigenstates of the sub-Hamiltonians.

2. The quantized solution method for combinatorial optimization problems according to claim 1, characterized in that: The quantized solution method further includes constructing a loss function corresponding to the combinatorial optimization problem to be solved: Define the objective function according to the combinatorial optimization problem to be solved , Restrictions and variables ; Construct the following loss function corresponding to the combinatorial optimization problem to be solved : ； in, is the penalty function, With the restrictions A is the penalty coefficient.

3. The quantized solution method for combinatorial optimization problems according to claim 1, characterized in that: The method of decomposing the overall Hamiltonian into multiple sub-Hamiltonians is to decompose the overall Hamiltonian as follows: ； in, is the overall Hamiltonian, Indicates that it acts on The Hamiltonian component of the subsystem, that is, The Hamiltonian is a positive integer and , is the number of sub-Hamiltonians.

4. The quantized solution method for combinatorial optimization problems according to claim 1, characterized in that: The method for obtaining each parameterized quantum state that matches each sub-Hamiltonian in the current iteration round is: If the current iteration round is the first round, the parameters of each pre-constructed parameterized quantum circuit are randomly initialized, and then acted on each initial quantum state respectively to obtain each parameterized quantum state that matches each sub-Hamiltonian; If the current iteration round is not the first round, the parameters of each parameterized quantum circuit are updated according to the product of the eigenvalues ​​of each sub-Hamiltonian measured in the previous iteration round, and then the updated parameterized quantum circuits are used to act on each initial quantum state respectively to obtain each parameterized quantum state that matches each sub-Hamiltonian in the current iteration round.

5. The quantized solution method for combinatorial optimization problems according to claim 4, characterized in that: The quantization solution method further includes: pre-constructing parameterized quantum circuits and The number of quantum bits in each parameterized quantum circuit and the initial quantum state is equal to the number of quantum bits in each sub-Hamiltonian. Each parameterized quantum circuit consists of a quantum rotation gate. And control the NOT gate to randomly generate a set of Parameters ={ }, then the quantum revolving door Medium coefficient Hedi The relationship between the quantum bits is: ， ， ， in , To randomly generate a set of Parameters , , For the The number of quantum bits of the Hamiltonian is such that every two bits act as a controlled NOT gate, and the number of controlled NOT gates is equal to the number of quantum bits.

6. A quantized solution device for combinatorial optimization problems, characterized in that: The quantization solving device comprises: A mapping module is used to map the loss function corresponding to the combinatorial optimization problem to be solved into the overall Hamiltonian; A decomposition module, configured to decompose the overall Hamiltonian into a plurality of sub-Hamiltonians, wherein the sum of the number of qubits of each sub-Hamiltonian is equal to the number of qubits of the overall Hamiltonian; A solution module, configured to obtain the eigenvalues ​​and corresponding eigenstates of each sub-Hamiltonian that meets a preset combination condition based on a variational quantum algorithm, take the product of the eigenvalues ​​of each sub-Hamiltonian as the eigenvalue of the overall Hamiltonian, and take the tensor product of the eigenstates of each sub-Hamiltonian as the eigenstate of the overall Hamiltonian; The preset combination condition is that the product of the eigenvalues ​​of each sub-Hamiltonian converges, and the method for obtaining the eigenvalues ​​and corresponding eigenstates of each sub-Hamiltonian that meets the preset combination condition based on the variational quantum algorithm includes: Obtaining each parameterized quantum state that matches each of the sub-Hamiltonians in the current iteration round, and measuring the eigenvalues ​​of each of the sub-Hamiltonians based on each of the parameterized quantum states; Determine whether the eigenvalues ​​of each of the sub-Hamiltonians in the current iteration round meet the preset combination conditions; If so, each parameterized quantum state in the current iteration round is used as the eigenstate of each sub-Hamiltonian; If not, the parameterized quantum states of the next iteration round are updated according to the product of the eigenvalues ​​of the sub-Hamiltonians in the current iteration round, and the eigenvalues ​​of the sub-Hamiltonians are respectively measured according to the parameterized quantum states of the next iteration round. The iteration is repeated until the eigenvalues ​​of the sub-Hamiltonians meet the preset combination conditions, and the parameterized quantum states of the last iteration round are respectively used as the eigenstates of the sub-Hamiltonians.

7. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a quantization solution program for the combinatorial optimization problem, and when the quantization solution program for the combinatorial optimization problem is executed by a processor, the quantization solution method for the combinatorial optimization problem according to any one of claims 1 to 5 is implemented.

8. A computer device, characterized in that: The computer device includes a computer-readable storage medium, a processor, and a quantization solution program for the combinatorial optimization problem stored in the computer-readable storage medium. When the quantization solution program for the combinatorial optimization problem is executed by the processor, the quantization solution method for the combinatorial optimization problem according to any one of claims 1 to 5 is implemented.

Images