Information acquisition method and device of quantum system, computer device, readable storage medium and program product

Abstract

The application relates to a quantum system information acquisition method and device, computer equipment, a readable storage medium and a program product. The method comprises the following steps: converting a preset initial quantum state of a target quantum system to obtain an initial matrix product state corresponding to the initial quantum state; optimizing the initial matrix product state to obtain a target matrix product state by taking minimizing a quantum energy value as a target; obtaining initial circuit parameters of a quantum circuit according to the target matrix product state; optimizing the initial circuit parameters of the quantum circuit to obtain target circuit parameters of the quantum circuit; and determining a ground state energy value of the target quantum system according to the target circuit parameters. The method can improve the ground state energy acquisition efficiency of a noisy quantum system.

Description

Technical Field

[0001] This application relates to the field of quantum computing technology, and in particular to a method, apparatus, computer device, computer-readable storage medium, and computer program product for acquiring information from a quantum system. Background Technology

[0002] Quantum computing has important applications in quantum chemistry, materials science, and the simulation of complex physical systems, among which the solution for the ground state energy is one of the key problems. Current mainstream quantum computing schemes mainly rely on hybrid quantum-classical optimization methods such as variational quantum eigensolvers (VQEs), which construct the target quantum state by parameterizing quantum circuits and optimize its parameters to minimize the system energy.

[0003] However, when dealing with highly entangled quantum states, existing methods typically rely on heuristically parameterized quantum circuits to construct trial states with randomly generated initial parameters. Furthermore, to accurately characterize the ground state of strongly correlated systems, complex quantum circuits with considerable depth are usually required. However, due to hardware limitations and unavoidable noise interference in current medium-noise quantum devices, the reliable execution capability of quantum circuits is significantly constrained. This causes the actual energy measurement error to accumulate and amplify with increasing circuit complexity, resulting in low efficiency in obtaining the ground state energy of the quantum system. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, computer device, computer-readable storage medium, and computer program product for acquiring information from a quantum system that can improve the efficiency of acquiring the ground state energy of a noisy quantum system, in order to address the above-mentioned technical problems.

[0005] In a first aspect, this application provides a method for acquiring information in a quantum system, including:

[0006] The preset initial quantum state of the target quantum system is transformed to obtain the initial matrix product state corresponding to the initial quantum state;

[0007] With the goal of minimizing quantum energy, the initial matrix product state is optimized to obtain the target matrix product state;

[0008] Based on the product state of the target matrix, the initial circuit parameters of the quantum circuit are obtained;

[0009] The initial circuit parameters of the quantum circuit are optimized to obtain the target circuit parameters of the quantum circuit;

[0010] Based on the target circuit parameters, determine the ground state energy value of the target quantum system.

[0011] In one embodiment, optimizing the initial matrix product state to obtain the target matrix product state with the goal of minimizing quantum energy includes:

[0012] Use the initial matrix product state as the current matrix product state;

[0013] Calculate the quantum energy value corresponding to the current matrix product state based on the Hamiltonian of the target quantum system;

[0014] The current matrix product state is normalized to a central orthogonal form to obtain the normalized current matrix product state.

[0015] Based on the normalized current matrix product state, a stochastic gradient descent optimization model is used to adjust the current matrix product state to obtain a new current matrix product state. Then, the step of calculating the quantum energy value corresponding to the current matrix product state based on the Hamiltonian of the target quantum system is returned to be executed until the quantum energy value corresponding to the current matrix product state no longer decreases. Then, the last obtained current matrix product state is taken as the target matrix product state.

[0016] In one embodiment, obtaining the initial circuit parameters of the quantum circuit based on the target matrix product state includes:

[0017] Initialize an initial mapping matrix product state that conforms to the circuit structure of the target quantum system;

[0018] The initial mapping matrix product state is iteratively adjusted until the similarity between the adjusted initial mapping matrix product state and the target matrix product state is maximized. Then, the adjusted initial mapping matrix product state is taken as the target mapping matrix product state.

[0019] The initial circuit parameters of the quantum circuit are obtained by mapping the product state of the target mapping matrix.

[0020] In one embodiment, optimizing the initial circuit parameters of the quantum circuit to obtain the target circuit parameters of the quantum circuit includes:

[0021] Use the initial circuit parameters as the current circuit parameters;

[0022] Based on the current circuit parameters, determine the quantum energy value of the quantum circuit;

[0023] Based on the stochastic gradient descent optimization model, the current circuit parameters are updated to obtain new current circuit parameters. Then, the step of determining the quantum energy value of the quantum circuit based on the current circuit parameters is returned to be executed until the quantum energy value of the quantum circuit no longer decreases. Then, the last obtained current circuit parameters are used as the target circuit parameters of the quantum circuit.

[0024] In one embodiment, determining the ground state energy value of the target quantum system based on the target circuit parameters includes:

[0025] By performing circuit folding and stretching operations on the target quantum circuit, multiple equivalent quantum circuits with different noise scaling parameters are obtained; the target quantum circuit is a quantum circuit whose corresponding circuit parameters are the target circuit parameters.

[0026] Determine the quantum energy value of each equivalent quantum circuit;

[0027] Based on the quantum energy values ​​of each equivalent quantum circuit, the noise-free quantum energy value of the target quantum circuit is obtained, which is then used as the noise-free ground state energy value of the target quantum system.

[0028] In one embodiment, obtaining the noise-free quantum energy value of the target quantum circuit based on the quantum energy values ​​of the respective equivalent quantum circuits includes:

[0029] Based on the quantum energy values ​​of each equivalent quantum circuit and the noise scaling parameters of each equivalent quantum circuit, the mapping relationship between the noise scaling parameters and the quantum energy values ​​of the target quantum circuit is obtained.

[0030] Based on the mapping relationship, the noise-free quantum energy value of the target quantum circuit is obtained.

[0031] Secondly, this application also provides an information acquisition device for a quantum system, comprising:

[0032] A quantum state conversion module is used to convert a preset initial quantum state of a target quantum system to obtain an initial matrix product state corresponding to the initial quantum state;

[0033] The product state optimization module is used to optimize the initial matrix product state with the goal of minimizing the quantum energy value, so as to obtain the target matrix product state.

[0034] An initial parameter determination module is used to obtain the initial circuit parameters of the quantum circuit based on the target matrix product state.

[0035] The target parameter determination module is used to optimize the initial circuit parameters of the quantum circuit to obtain the target circuit parameters of the quantum circuit.

[0036] The energy value determination module is used to determine the ground state energy value of the target quantum system based on the target circuit parameters.

[0037] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0038] The preset initial quantum state of the target quantum system is transformed to obtain the initial matrix product state corresponding to the initial quantum state;

[0039] With the goal of minimizing quantum energy, the initial matrix product state is optimized to obtain the target matrix product state;

[0040] Based on the product state of the target matrix, the initial circuit parameters of the quantum circuit are obtained;

[0041] The initial circuit parameters of the quantum circuit are optimized to obtain the target circuit parameters of the quantum circuit;

[0042] Based on the target circuit parameters, determine the ground state energy value of the target quantum system.

[0043] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0044] The preset initial quantum state of the target quantum system is transformed to obtain the initial matrix product state corresponding to the initial quantum state;

[0045] With the goal of minimizing quantum energy, the initial matrix product state is optimized to obtain the target matrix product state;

[0046] Based on the product state of the target matrix, the initial circuit parameters of the quantum circuit are obtained;

[0047] The initial circuit parameters of the quantum circuit are optimized to obtain the target circuit parameters of the quantum circuit;

[0048] Based on the target circuit parameters, determine the ground state energy value of the target quantum system.

[0049] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0050] The preset initial quantum state of the target quantum system is transformed to obtain the initial matrix product state corresponding to the initial quantum state;

[0051] With the goal of minimizing quantum energy, the initial matrix product state is optimized to obtain the target matrix product state;

[0052] Based on the product state of the target matrix, the initial circuit parameters of the quantum circuit are obtained;

[0053] The initial circuit parameters of the quantum circuit are optimized to obtain the target circuit parameters of the quantum circuit;

[0054] Based on the target circuit parameters, determine the ground state energy value of the target quantum system.

[0055] The aforementioned information acquisition method, apparatus, computer equipment, computer-readable storage medium, and computer program product for quantum systems firstly transform the preset initial quantum state of the target quantum system to obtain the corresponding initial matrix product state. Transforming the initial quantum state of the target quantum system using the matrix product state reduces the storage complexity of the quantum state, enabling efficient representation of high-dimensional quantum states. The matrix product state structure effectively describes the quantum system, reduces computational redundancy, and provides a mathematically more manageable representation for subsequent optimization. Next, with the goal of minimizing the quantum energy value, the initial matrix product state is optimized to obtain the target matrix product state. By optimizing the initial matrix product state to minimize its energy, the obtained target matrix product state is ensured to be as close as possible to the ground state of the target quantum system, avoiding direct solution of high-dimensional optimization problems on quantum hardware, making the computation more stable, and fully utilizing classical computing resources for pre-optimization, avoiding getting trapped in local optima, improving convergence stability, and reducing the number of optimization iterations required for subsequent quantum computation. Then, based on the target matrix product state, the initial circuit parameters of the quantum circuit are obtained. The optimized matrix product state generates the initial parameters of the quantum circuit, enabling… The constructed quantum circuit closely approximates the ground state of the target system, effectively reducing the search space during optimization, making the optimization process more physically interpretable, avoiding local extrema problems that may arise from random initialization, and improving the optimization convergence speed. Furthermore, by optimizing the initial circuit parameters of the quantum circuit, the target circuit parameters are obtained. Through an energy-minimization-based optimization process, the parameters of the quantum circuit are further adjusted to more closely approximate the target ground state. Parameters are dynamically corrected via quantum measurement feedback on quantum hardware, fine-tuning the initial parameters to compensate for modeling errors between classical pre-training and the real quantum system. While maintaining a low circuit depth, the energy of the real ground state is further approximated, suppressing error accumulation caused by noise and improving the experimental feasibility of the final result. Finally, based on the target circuit parameters, the ground state energy of the target quantum system is determined. By measuring the output state of the optimized quantum circuit, the expected value of the Hamiltonian is calculated, ultimately obtaining the ground state energy of the target quantum system. This ensures that the calculated ground state energy matches the ground state of the actual physical system and is applicable to different quantum systems, providing a reliable computational scheme for high-precision calculations in fields such as quantum chemistry and materials simulation. In the above method, by combining matrix product state pre-training with quantum parameter optimization, highly entangled state information is compressed and near-optimal initial parameters are generated during the classical computation stage. This significantly reduces the depth of the quantum circuit, overcomes the execution limitations of noisy quantum hardware, and improves the accuracy and computational efficiency of solving the ground state energy of the target quantum system. At the same time, the structural compatibility between the pre-trained parameters and the quantum circuit enables the classical-quantum optimization process to converge efficiently, achieving high-precision solution of the ground state energy of the quantum system with limited resources.Compared to traditional variational quantum algorithms, this method avoids blind optimization directly on quantum circuits. Instead, it utilizes matrix product state optimization to provide better initial conditions, thereby reducing the optimization overhead of quantum circuits. Furthermore, this method is adaptable to quantum systems of varying sizes and can be efficiently computed on current medium-noise quantum computers, providing a more stable solution method for future quantum computing applications. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 This is a flowchart illustrating an information acquisition method for a quantum system in one embodiment;

[0058] Figure 2 This is a flowchart illustrating the steps for obtaining the ground state energy value of a target quantum system in one embodiment;

[0059] Figure 3 This is a flowchart illustrating an information acquisition method for a quantum system in another embodiment;

[0060] Figure 4 This is a schematic diagram of the matrix product state and its central orthogonal form in one embodiment;

[0061] Figure 5 This is a schematic diagram illustrating the process of calculating the energy value of the Hamiltonian based on matrix product states in one embodiment.

[0062] Figure 6 This is a schematic diagram illustrating the energy value calculation process in a centrally orthogonal configuration in one embodiment.

[0063] Figure 7 This is a schematic diagram of a quantum circuit design framework and its noise scaling in one embodiment;

[0064] Figure 8 This is a structural block diagram of an information acquisition device for a quantum system in one embodiment;

[0065] Figure 9 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0066] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0067] In one embodiment, such as Figure 1 As shown, a method for acquiring information in a quantum system is provided. This embodiment illustrates the application of this method to a terminal. It is understood that this method can also be applied to a server, and to a system including both a terminal and a server, and is implemented through interaction between the terminal and the server. The terminal or server can be a classical computer or a quantum computer. In this embodiment, the method includes the following steps:

[0068] Step S101: Transform the preset initial quantum state of the target quantum system to obtain the initial matrix product state corresponding to the initial quantum state.

[0069] Here, the target quantum system refers to a computational object composed of multiple qubits, whose state can be represented by quantum states. To describe.

[0070] The initial quantum state refers to the preset quantum state of the target quantum system, which is usually derived by classical methods (such as Hartree-Fock state) or physical models, and provides a computational basis for subsequent optimization.

[0071] Matrix product states are a mathematical structure for compactly representing many-body quantum states. They decompose high-dimensional quantum states into a product of a series of low-rank tensors, which greatly reduces the storage and computational complexity of quantum systems.

[0072] For example, the terminal first acquires a preset initial quantum state of the target quantum system and then uses a tensor network-based method to convert this quantum state into a matrix product state (MPS) representation. This process involves decomposing the global quantum state into a series of local tensors and trunculating them using numerical methods such as singular value decomposition to ensure that the matrix product state meets a given truncation error threshold, thereby guaranteeing the accuracy and feasibility of the computation. After the conversion is complete, the terminal stores and outputs the generated initial matrix product state, providing input data for subsequent optimization steps.

[0073] The Hartree-Fock state is an approximate quantum state calculated using the Hartree-Fock method. Specifically, it is an approximate ground state of a multi-electron system, typically represented by a Slater determinant, used to describe the electron's wavefunction. The calculation process begins with an initial guess of the single-electron orbitals, followed by iterative solving of the Hartree-Fock equations using a self-consistent field until the electron density and wavefunction converge. The specific form of the Hartree-Fock state depends on the Hamiltonian of the molecular system and the chosen computational basis set.

[0074] Step S102: Optimize the initial matrix product state with the goal of minimizing the quantum energy value to obtain the target matrix product state.

[0075] Here, quantum energy value refers to the expected energy corresponding to the Hamiltonian H of the quantum system, i.e. This value reflects the energy of the system's current quantum state. The initial matrix product state (MPS) is a tensor network structure used for compact representation of many-body quantum states. The target matrix product state refers to the optimized MPS-form quantum state, whose energy value is lower than the initial state and approximates the system's ground state (lowest energy state) as closely as possible.

[0076] For example, the terminal first constructs the Hamiltonian operator of the system, which describes the interactions and energy relationships of the target quantum system. Then, based on a variational optimization method, the terminal optimizes the initial matrix product state with the objective of minimizing the quantum energy value. In a specific implementation, the terminal uses gradient descent or conjugate gradient methods to calculate the gradient of each MPS tensor according to the local optimization update rule and updates it according to the learning rate η, thereby gradually reducing the expected energy value of the MPS. Furthermore, to improve optimization stability, the terminal maintains the orthogonal normalization of the MPS during the optimization process, ensuring the numerical stability of the computation. After multiple rounds of optimization iterations, when the convergence condition is met (e.g., the energy change is below a set threshold), the terminal outputs the optimized target matrix product state, providing a basis for subsequent quantum circuit parameter extraction.

[0077] Step S103: Obtain the initial circuit parameters of the quantum circuit based on the product state of the target matrix.

[0078] A quantum circuit is a sequence of operations used to simulate and solve quantum systems on a quantum computer. It consists of a series of parameterized quantum gates. The initial circuit parameters are adjustable parameters in the quantum circuit (such as the angle of the rotating gate), and their settings determine the quantum states generated by the quantum circuit.

[0079] For example, the terminal first parses the tensor structure of the target matrix product state and uses tensor decomposition methods (such as singular value decomposition) to map the local tensors in the target matrix product state to the form of parameterized quantum gates. Based on the local orthogonal tensor distribution of the target matrix product state, the terminal determines the corresponding single-bit rotation gate (such as Ri). y (θ), R z The system constructs the structure of (θ) and two-qubit entangled gates (such as CNOT gates), extracts the numerical information of each tensor, and converts it into the corresponding initial circuit parameters θ0. During the parameter mapping process, the terminal uses a transform learning algorithm or a least squares fitting method to ensure that the obtained initial circuit parameters make the quantum state output by the quantum circuit as close as possible to the target matrix product state. After the conversion is completed, the terminal stores the obtained initial circuit parameters to provide input for subsequent quantum circuit optimization.

[0080] Step S104: Optimize the initial circuit parameters of the quantum circuit to obtain the target circuit parameters of the quantum circuit.

[0081] For example, the terminal first optimizes the initial circuit parameters based on the energy minimization objective using a variational quantum eigenvalue solver or other quantum optimization algorithms. The terminal executes parameterized quantum circuits on quantum hardware, measures the expected value of the system's Hamiltonian, and calculates the gradient information of the parameters. During optimization, the terminal estimates the gradient using the parameter offset method or the finite difference method, and iteratively updates the circuit parameters using classical optimization methods such as gradient descent, conjugate gradient, or natural gradient optimization.

[0082] During the optimization process, the terminal continuously evaluates the optimized energy value and sets convergence criteria (such as the energy change being less than a certain threshold or reaching the upper limit of the number of optimization iterations). When the optimization converges, the terminal outputs the final optimized target circuit parameters. These parameters are used to construct the optimal quantum circuit, making it as close as possible to the ground state of the target quantum system, providing a high-precision quantum state representation for the final calculation of the ground state energy.

[0083] Step S105: Determine the ground state energy value of the target quantum system based on the target circuit parameters.

[0084] The ground state energy value is the lowest energy eigenvalue of the Hamiltonian H of a quantum system, and usually represents the most stable state of the system.

[0085] For example, the terminal first runs the optimized parameterized quantum circuit on a quantum computer to obtain the corresponding quantum state, and then calculates the ground state energy of the target quantum system on a classical computer based on the obtained quantum state. To improve computational accuracy, the terminal can sample the quantum measurement results multiple times and employ error mitigation methods (such as measurement error correction or zero-noise extrapolation) to reduce experimental errors. Once the measurement converges, the terminal outputs the ground state energy of the target quantum system and stores the measurement data, providing reliable computational results for further quantum simulation or physical system analysis.

[0086] In the aforementioned method for acquiring information about a quantum system, firstly, the preset initial quantum state of the target quantum system is transformed to obtain the corresponding initial matrix product state. Transforming the initial quantum state of the target quantum system using the matrix product state reduces the storage complexity of the quantum state, enabling it to efficiently represent high-dimensional quantum states. The matrix product state structure effectively describes the quantum system, reduces computational redundancy, and provides a mathematically easier-to-process representation for subsequent optimization. Next, with the goal of minimizing the quantum energy value, the initial matrix product state is optimized to obtain the target matrix product state. By optimizing the initial matrix product state to minimize its energy, the obtained target matrix product state is ensured to be as close as possible to the ground state of the target quantum system. This avoids directly solving high-dimensional optimization problems on quantum hardware, making the computation more stable and fully utilizing classical computing resources for pre-optimization, avoiding getting trapped in local extrema, improving convergence stability, and reducing the number of optimization iterations required for subsequent quantum computation. Then, based on the target matrix product state, the initial circuit parameters of the quantum circuit are obtained. The optimized matrix product state generates the initial parameters of the quantum circuit, ensuring that the constructed quantum circuit can be as efficient as possible. By approximating the ground state of the target system, the search space during quantum circuit optimization can be effectively reduced, making the optimization process more physically interpretable, avoiding local extrema problems that may be caused by random initialization, and improving the optimization convergence speed. Furthermore, by optimizing the initial circuit parameters of the quantum circuit to obtain the target circuit parameters, the parameters of the quantum circuit are further adjusted through an energy-minimization-based optimization process to make them closer to the target ground state. On quantum hardware, parameters are dynamically corrected through quantum measurement feedback to fine-tune and optimize the initial parameters, compensating for modeling errors between classical pre-training and the real quantum system. While maintaining low circuit depth, the energy of the real ground state is further approximated, suppressing error accumulation caused by noise and improving the experimental feasibility of the final result. Finally, based on the target circuit parameters, the ground state energy value of the target quantum system is determined. By measuring the output state of the optimized quantum circuit, the expected value of the Hamiltonian is calculated, ultimately obtaining the ground state energy value of the target quantum system. This ensures that the calculated ground state energy matches the ground state of the actual physical system and is applicable to different quantum systems, providing a reliable computational scheme for high-precision calculations in fields such as quantum chemistry and materials simulation. In the aforementioned method, by synergistically combining matrix product state pre-training and quantum parameter optimization, highly entangled state information is compressed and near-optimal initial parameters are generated during the classical computation stage. This significantly reduces the quantum circuit depth, overcomes the execution limitations of noisy quantum hardware, and improves the accuracy and computational efficiency of solving the ground state energy of the target quantum system. Simultaneously, the structural compatibility between the pre-trained parameters and the quantum circuit enables efficient convergence of the classical-quantum optimization process, achieving high-precision solutions for the quantum system's ground state energy with limited resources. Compared to traditional variational quantum algorithms, this method avoids blind optimization directly on the quantum circuit, instead utilizing matrix product state optimization to provide better initial conditions, thereby reducing the optimization overhead of the quantum circuit.Furthermore, this method can be adapted to quantum systems of different sizes and can be efficiently computed on current medium-noise quantum computers, providing a more stable solution method for subsequent quantum computing applications.

[0087] In an exemplary embodiment, step S102 above, which aims to minimize the quantum energy value and optimize the initial matrix product state to obtain the target matrix product state, further includes: taking the initial matrix product state as the current matrix product state; calculating the quantum energy value corresponding to the current matrix product state based on the Hamiltonian of the target quantum system; normalizing the current matrix product state to a centroorthogonal form to obtain the normalized current matrix product state; adjusting the current matrix product state based on the normalized current matrix product state using a stochastic gradient descent optimization model to obtain a new current matrix product state, and returning to execute the step of calculating the quantum energy value corresponding to the current matrix product state based on the Hamiltonian of the target quantum system, until the quantum energy value corresponding to the current matrix product state no longer decreases, then taking the last obtained current matrix product state as the target matrix product state.

[0088] Here, the current matrix product state refers to the matrix product state at the current step during the optimization iteration process, whose energy value is continuously updated to approximate the system's ground state energy. The Hamiltonian H is the energy operator of the target quantum system, describing the system's dynamic behavior and interactions.

[0089] The central orthogonal form is a standardized representation of the matrix product state (MPS), which makes the tensor decomposition process more stable and improves optimization convergence. The stochastic gradient descent optimization model is an iterative optimization method based on gradient information, which can effectively reduce system energy and progressively optimize the representation of the matrix product state.

[0090] For example, the terminal first sets the initial matrix product state as the current matrix product state and calculates its corresponding quantum energy value based on the Hamiltonian of the target quantum system. The terminal parses the tensor structure of the current matrix product state and uses Pauli expansion to calculate the expected energy value. The calculated energy value is used to judge the optimization progress, ensuring that the energy value gradually decreases after each optimization iteration. Subsequently, the terminal normalizes the current matrix product state, i.e., converts it into a centroorthogonal form. The normalization process decomposes the current matrix product state through singular value decomposition, ensuring that each tensor satisfies the orthogonality condition. This normalization operation can improve the stability of the optimization and avoid error accumulation during numerical calculation. Based on the normalized matrix product state, the terminal uses a stochastic gradient descent optimization model to adjust the tensor parameters of the current matrix product state. During the optimization process, the terminal uses the gradient calculation formula:

[0091]

[0092] Where η is the learning rate. This represents the gradient of the current matrix product state with respect to energy. The terminal updates each tensor. The value of causes the energy of the quantum system to decrease.

[0093] The terminal repeatedly executes the following steps: calculates the quantum energy value of the current matrix product state; normalizes the current matrix product state; optimizes using a stochastic gradient descent optimization model and updates the tensor parameters of the matrix product state; and checks whether the energy has converged.

[0094] When it is detected that the quantum energy value of the current matrix product state no longer decreases in continuous iterations (i.e., the convergence condition is met), the optimization process is terminated, and the last obtained current matrix product state is output as the target matrix product state for subsequent quantum circuit parameter mapping.

[0095] In this embodiment, central orthogonal normalization ensures the numerical stability of the matrix product state, reduces the accumulation of numerical errors during optimization, and improves the convergence speed. Simultaneously, based on the stochastic gradient descent optimization model, the tensor parameters of the matrix product state are iteratively updated with energy minimization as the objective, ensuring the correct optimization direction and avoiding getting trapped in local minima. Dynamic monitoring of energy changes adaptively controls the number of optimization iterations, avoiding unnecessary computational overhead and improving computational efficiency. Furthermore, the target matrix product state more accurately represents the ground state of the target quantum system, providing higher-quality input for subsequent quantum circuit parameter extraction, thereby improving the overall solution accuracy and convergence performance of quantum computing.

[0096] In an exemplary embodiment, step S103, which obtains the initial circuit parameters of the quantum circuit based on the target matrix product state, further includes: initializing an initial mapping matrix product state that conforms to the circuit structure of the target quantum system; iteratively adjusting the initial mapping matrix product state until the similarity between the adjusted initial mapping matrix product state and the target matrix product state is maximized, then using the adjusted initial mapping matrix product state as the target mapping matrix product state; and mapping the initial circuit parameters of the quantum circuit based on the target mapping matrix product state.

[0097] Similarity is used to measure how close two quantum states are to each other, and is usually calculated using the fidelity of the state.

[0098] For example, the terminal first constructs an initial mapping matrix product state that conforms to the circuit structure of the target quantum system as the starting point for optimization. The construction of this mapping state is based on a preset quantum circuit architecture, including a fixed quantum gate topology, such as a single-qubit rotation gate (e.g., R0). x (θ), R y (θ), R z(θ) and two-qubit entanglement gates (such as CNOT gates). The terminal generates an initial mapping matrix product state through random initialization or a physics-inspired method (such as the Hartree-Fock approximation) and ensures that its tensor structure matches the target quantum system.

[0099] Subsequently, the terminal iteratively adjusts the initial mapping matrix product state to maximize its similarity to the target matrix product state. The fidelity is:

[0100]

[0101] in, and These are the target matrix product state and the mapping matrix product state, respectively.

[0102] During the optimization process, the terminal uses variational optimization or stochastic gradient descent. Based on the state fidelity as the optimization objective, it calculates the gradient and updates the tensor parameters of the mapping matrix product state according to the gradient direction. The terminal repeatedly calculates the similarity of the states and continuously updates the initial mapping matrix product state until the fidelity between it and the target matrix product state reaches a set threshold, which is usually close to 1 (e.g., 0.99).

[0103] After the optimization process is complete, the terminal uses the final target mapping matrix product state for parameter mapping of the quantum circuit. The terminal parses the tensor structure of the target mapping matrix product state and extracts the initial circuit parameters required by the quantum circuit based on its local tensor values. This mapping process is based on the matrix-to-quantum gate conversion rules, for example: a single-bit tensor component is mapped to R... x (θ), R y (θ), R z The rotation angle (θ); the two-bit tensor is mapped to the parameters of a CNOT gate or a controlled rotation gate.

[0104] Finally, the terminal stores and outputs the initial circuit parameters of the quantum circuit, providing a foundation for subsequent quantum circuit optimization steps.

[0105] In this embodiment, optimization is performed by maximizing the fidelity of the states, ensuring a high degree of match between the final output state of the quantum circuit and the ground state of the target system. This reduces the optimization difficulties caused by random initialization in traditional methods. Simultaneously, a gradient-based optimization strategy effectively adjusts the tensor parameters of the mapping matrix product state, making it convertible into a parameterized expression of the quantum circuit, thus improving solution accuracy and computational efficiency. This approach provides a superior parameter initialization scheme within limited qubits and circuit depth, laying a solid initial foundation for subsequent quantum circuit optimization and improving the overall convergence speed and computational accuracy.

[0106] In an exemplary embodiment, step S104 above, which optimizes the initial circuit parameters of the quantum circuit to obtain the target circuit parameters of the quantum circuit, further includes: using the initial circuit parameters as the current circuit parameters; determining the quantum energy value of the quantum circuit based on the current circuit parameters; updating the current circuit parameters based on the stochastic gradient descent optimization model to obtain new current circuit parameters, and returning to execute the step of determining the quantum energy value of the quantum circuit based on the current circuit parameters, until the quantum energy value of the quantum circuit no longer decreases, then using the last obtained current circuit parameters as the target circuit parameters of the quantum circuit.

[0107] For example, the terminal first sets the initial circuit parameters as the current circuit parameters and constructs a parameterized quantum circuit based on these parameters. Then, the terminal executes the quantum circuit on a quantum computer or a classical simulator and obtains its quantum state. Next, it calculates the quantum energy value based on the quantum state on a classical computer. Subsequently, the terminal calculates the gradient of the circuit parameters based on a stochastic gradient descent optimization model and updates the parameters along the gradient descent direction.

[0108] After the terminal updates the circuit parameters, it repeats the following steps: calculate the quantum energy value corresponding to the current circuit parameters; update the current circuit parameters based on gradient optimization; and check whether the energy has converged.

[0109] When it is detected that the energy value of the quantum circuit no longer decreases in continuous iterations during the optimization process (i.e., the convergence condition is met, such as the energy change being lower than a preset threshold), the optimization process is terminated, and the current circuit parameters obtained from the last optimization are output as the target circuit parameters for subsequent measurement and calculation.

[0110] In this embodiment, a stochastic gradient descent optimization model is used to update the quantum circuit parameters along the optimal direction, ultimately obtaining circuit parameters that are closest to the ground state of the target quantum system. The parameter offset method is employed to calculate the gradient, avoiding the computationally complex analytical gradient calculation, while reducing the hardware requirements of the quantum computer and improving optimization stability. By dynamically monitoring changes in quantum energy values, the optimization step size can be adaptively adjusted to ensure efficient searching for the optimal parameter configuration and avoid local minima traps. Furthermore, this optimization method combines classical computation and quantum measurement, improving the convergence speed of the optimization process and reducing the consumption of quantum resources, making it suitable for efficient solutions in current medium-noise quantum computing environments.

[0111] In one exemplary embodiment, such as Figure 2 As shown, step S105 above, which determines the ground state energy value of the target quantum system based on the target circuit parameters, can also be achieved through the following steps:

[0112] Step S201: By performing circuit folding and stretching operations on the target quantum circuit, multiple equivalent quantum circuits with different noise scaling parameters are obtained.

[0113] Among them, the target quantum circuit is the quantum circuit whose corresponding circuit parameters are the target circuit parameters.

[0114] Step S202: Determine the quantum energy value of each equivalent quantum circuit.

[0115] Step S203: Based on the quantum energy values ​​of each equivalent quantum circuit, obtain the noise-free quantum energy value of the target quantum circuit, which is used as the noise-free ground state energy value of the target quantum system.

[0116] Among these, circuit folding operations refer to artificially enhancing quantum noise by adding circuit gate operations without changing the ideal calculation results. For example, replacing a quantum gate U with UU†U is theoretically equivalent to the original gate operation, but in practice, it increases noise accumulation. Circuit stretching operations refer to controlling the noise level by adjusting the execution time of quantum gates or by adding equivalent operations. For example, stretching the amplitude of a controllable rotating gate or adjusting the application of an entanglement gate can change the noise effect proportionally. The noise scaling parameter reflects the noise amplification factor of the equivalent quantum circuit relative to the original circuit and is used for subsequent zero-noise extrapolation calculations. The noise-free ground state energy value refers to the ground state energy of the quantum system under ideal noise-free conditions, which is inferred through noise scaling techniques.

[0117] For example, the terminal first constructs the target quantum circuit, where the parameters of each quantum gate are determined by the parameters of the target circuit. Subsequently, the terminal constructs multiple equivalent quantum circuits with different noise scaling parameters based on circuit folding and circuit stretching operations.

[0118] In circuit folding operations, the terminal performs equivalent transformations for single-bit gates U and two-bit gates (such as CNOT gates), for example: , .

[0119] These transformations keep the circuit's calculation results unchanged, but they can increase noise during actual execution.

[0120] In circuit stretching operations, the terminal adjusts the quantum gate execution mode, for example, by increasing the rotation gate amplitude. , where λ is the noise factor; or the physical execution time of the gate can be adjusted by repeatedly executing the equivalent gate sequence.

[0121] The terminal performs calculations for each equivalent quantum circuit and measures its quantum energy value. The terminal collects energy value data corresponding to all noise scaling parameters; then, based on a zero-noise extrapolation method, the terminal fits the ground state energy value under the known energy data for different noise scaling parameters to the noise-free condition.

[0122] In this embodiment, multiple equivalent quantum circuits are constructed using circuit folding and stretching techniques, enabling the measurement of quantum energy values ​​under different noise levels. The ground state energy under noise-free conditions is then calculated via zero-noise extrapolation, improving computational accuracy. By controlling the noise scaling factor, the influence of noise on the measurement results can be effectively captured, and the noise effect can be eliminated through extrapolation, resulting in a more accurate ground state energy value for the quantum system. This method is applicable to medium-noise quantum computers, providing an efficient quantum noise mitigation technique in situations where noise still exists in current quantum hardware, thus enhancing the applicability of quantum computing in solving practical problems.

[0123] In an exemplary embodiment, step S203, which obtains the noise-free quantum energy value of the target quantum circuit based on the quantum energy values ​​of each equivalent quantum circuit, further includes: obtaining a mapping relationship between the noise scaling parameter and the quantum energy value of the target quantum circuit based on the quantum energy values ​​of each equivalent quantum circuit and the noise scaling parameter of each equivalent quantum circuit; and obtaining the noise-free quantum energy value of the target quantum circuit based on the mapping relationship.

[0124] For example, the terminal first collects energy measurement data under different noise scaling parameters. Then, based on the noise influence model, the terminal fits the mapping relationship between the noise scaling parameters and the quantum energy value. Common mapping models include: linear extrapolation model, exponential decay model, polynomial fitting or neural network model.

[0125] After the fitting is completed, the terminal calculates the noise-free quantum energy value of the target quantum circuit based on the fitted mapping relationship.

[0126] In this embodiment, by establishing a mapping relationship between noise scaling parameters and quantum energy values, the ground state energy under noise-free conditions can be accurately extrapolated, effectively eliminating the influence of noise on quantum measurement results. Compared to directly measuring noise-affected data, data fitting techniques can be used to reduce experimental errors and improve computational accuracy. Furthermore, a suitable extrapolation model can be selected based on the hardware characteristics of the quantum computer, adaptively optimizing the computation process to ensure efficient and more accurate acquisition of the quantum system's ground state energy value.

[0127] In another exemplary embodiment, such as Figure 3 As shown, this application provides a method for acquiring information from a quantum system, including:

[0128] Step S301: Transform the preset initial quantum state of the target quantum system to obtain the initial matrix product state corresponding to the initial quantum state.

[0129] Step S302: Use the initial matrix product state as the current matrix product state.

[0130] Step S303: Calculate the quantum energy value corresponding to the current matrix product state based on the Hamiltonian of the target quantum system.

[0131] Step S304: Normalize the current matrix product state into a central orthogonal form to obtain the normalized current matrix product state.

[0132] Step S305: Based on the normalized current matrix product state, the stochastic gradient descent optimization model is used to adjust the current matrix product state to obtain a new current matrix product state, and then return to execute step S303 until the quantum energy value corresponding to the current matrix product state no longer decreases. Then, the last obtained current matrix product state is taken as the target matrix product state.

[0133] Step S306: Initialize an initial mapping matrix product state that conforms to the circuit structure of the target quantum system.

[0134] Step S307: Iteratively adjust the initial mapping matrix product state until the similarity between the adjusted initial mapping matrix product state and the target matrix product state is maximized. Then, take the adjusted initial mapping matrix product state as the target mapping matrix product state.

[0135] Step S308: Based on the product state of the target mapping matrix, the initial circuit parameters of the quantum circuit are mapped.

[0136] Step S309: Use the initial circuit parameters as the current circuit parameters.

[0137] Step S310: Determine the quantum energy value of the quantum circuit based on the current circuit parameters.

[0138] Step S311: Based on the stochastic gradient descent optimization model, update the current circuit parameters to obtain new current circuit parameters, and return to execute step S310 until the quantum energy value of the quantum circuit no longer decreases. Then, use the last obtained current circuit parameters as the target circuit parameters of the quantum circuit.

[0139] Step S312: By performing circuit folding and stretching operations on the target quantum circuit, multiple equivalent quantum circuits with different noise scaling parameters are obtained.

[0140] Among them, the target quantum circuit is the quantum circuit whose corresponding circuit parameters are the target circuit parameters.

[0141] Step S313: Determine the quantum energy value of each equivalent quantum circuit, and based on the quantum energy value of each equivalent quantum circuit and the noise scaling parameter of each equivalent quantum circuit, obtain the mapping relationship between the noise scaling parameter and the quantum energy value of the target quantum circuit.

[0142] Step S314: Based on the mapping relationship, obtain the noise-free quantum energy value of the target quantum circuit, which is used as the noise-free ground state energy value of the target quantum system.

[0143] For example, the Hamiltonian is typically represented as a linear combination of a set of Pauli operators. For a system with n qubits, the Hamiltonian can be formally represented as:

[0144]

[0145] in These are real coefficients. Represents the tensor product of Pauli operators, for example , I is the identity operator, and X, Y, and Z are Pauli operators, which are quantum operations that act on a single bit.

[0146] Matrix product states will convert higher-order tensors Decomposed into a series of low-order tensors The product of:

[0147] ,

[0148] in,

[0149]

[0150] in, It represents a set of orthogonal and normalized computational bases.

[0151] like Figure 4 As shown, It is the local tensor corresponding to the nth qubit, where Represents a physical index with dimension d. It is a dimension of Virtual metrics. The problem description at this point is as follows: Figure 5 As shown.

[0152] For wave function Preprocessing is performed. The wave function is... Normalization to a central orthogonal form, such as Figure 4 As shown.

[0153] definition The tensors to the left of the orthogonality center satisfy the left orthogonality condition:

[0154] ,

[0155] Tensors to the right of the orthogonality center satisfy the right orthogonality condition:

[0156] .

[0157] in, For Kronecker notation, specifically a diagonal matrix.

[0158] When updating local tensors, only the tensor at the orthogonal center is updated at a time; other tensors are treated as known tensors. The orthogonal center moves from the leftmost to the rightmost and then back from the rightmost to the leftmost, thus updating each local tensor in the matrix product state twice.

[0159] like Figure 6 As shown, if the local coupling is located to the left of the orthogonal center, i.e. According to the left-right orthogonality condition, the contraction of the tensor to the left of the i-th tensor is equal to the identity operator, and the contraction of the tensor to the right of the orthogonal center is also equal to the identity operator. We only need to convert the tensor from the i-th to the... Tensor contraction occurs when the tensor and Hamiltonian coefficients contract. The calculation simplification method is exactly the same.

[0160] In the training step S305, the objective function is: .

[0161] The update rule for the nth local tensor is:

[0162] , where η is the learning rate.

[0163] This embodiment utilizes the structural advantages of matrix product states to design the circuit framework of the wave function, the nth local tensor. Simulate the quantum state evolution and entanglement on the nth qubit. The design framework is as follows: Figure 7 As shown, the local tensor Mapping to parameterized quantum circuits In constructing the wave function At time, operators Acting sequentially on the initial Hartley-Fock state :

[0164]

[0165] For n>0, the two-qubit unitary operator Decomposed into single-bit unitary operators And CNOT Gate:

[0166]

[0167] unitary operator of a single quantum bit when n=0 Decomposed into: .

[0168] In step S312, when performing noise scaling, the technique for increasing the circuit noise level at the gate level is to increase its depth. For example... Figure 7 As shown, perform mapping or To achieve circuit folding.

[0169] Extrapolate to the noise-free limit. Extrapolate the noise-free expected value by fitting curves to the expected values ​​measured at different noise levels. Extrapolation is performed as follows:

[0170] (1) Selecting an extrapolation model: Assuming the expected value By function Description, where f is a noise scaling parameter. and a set of real parameters The extrapolation model is used. This embodiment employs a simple neural network consisting of three fully connected layers to fit the expected value function under different noise levels.

[0171] (2) Fitting data: The extrapolation model f is used to fit the expected values ​​of the measurements under different noise scaling to obtain a set of optimal fitting parameters. }

[0172] (3) Extrapolation noise-free limit: through calculation To obtain the expected value under noise-free conditions.

[0173] In this embodiment, noise errors in quantum devices are mitigated by optimizing quantum circuit design, pre-training initialization parameters, and introducing zero-noise extrapolation techniques. Compared to previous quantum eigenvalue solvers, this method achieves significantly higher computational accuracy in noisy environments and effectively controls the increase in circuit complexity with system size. Specifically, the quantum circuit structure is designed based on the matrix product state structure, effectively controlling circuit depth and complexity. Pre-training quantum circuit parameters using matrix product states avoids the impact of initialization on the stability of quantum circuit optimization. Introducing zero-noise extrapolation techniques to mitigate quantum noise and employing neural networks to improve noise fitting accuracy significantly enhances the accuracy of noise error mitigation while increasing the applicability and flexibility of the method under different quantum circuits and noise models.

[0174] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0175] Based on the same inventive concept, this application also provides an information acquisition device for a quantum system to implement the information acquisition method for the quantum system described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the information acquisition device for a quantum system provided below can be found in the limitations of the information acquisition method for a quantum system described above, and will not be repeated here.

[0176] In one exemplary embodiment, such as Figure 8 As shown, an information acquisition device for a quantum system is provided, comprising: a quantum state transition module 801, a product state optimization module 802, an initial parameter determination module 803, a target parameter determination module 804, and an energy value determination module 805, wherein:

[0177] The quantum state conversion module 801 is used to convert the preset initial quantum state of the target quantum system to obtain the initial matrix product state corresponding to the initial quantum state;

[0178] The product state optimization module 802 is used to optimize the initial matrix product state with the goal of minimizing the quantum energy value, so as to obtain the target matrix product state.

[0179] The initial parameter determination module 803 is used to obtain the initial circuit parameters of the quantum circuit based on the product state of the target matrix.

[0180] The target parameter determination module 804 is used to optimize the initial circuit parameters of the quantum circuit and obtain the target circuit parameters of the quantum circuit.

[0181] The energy value determination module 805 is used to determine the ground state energy value of the target quantum system based on the target circuit parameters.

[0182] In one embodiment, the product state optimization module 802 is further configured to: take the initial matrix product state as the current matrix product state; calculate the quantum energy value corresponding to the current matrix product state according to the Hamiltonian of the target quantum system; normalize the current matrix product state into a centroorthogonal form to obtain the normalized current matrix product state; based on the normalized current matrix product state, use a stochastic gradient descent optimization model to adjust the current matrix product state to obtain a new current matrix product state, and return to execute the step of calculating the quantum energy value corresponding to the current matrix product state according to the Hamiltonian of the target quantum system, until the quantum energy value corresponding to the current matrix product state no longer decreases, then take the last obtained current matrix product state as the target matrix product state.

[0183] In one embodiment, the initial parameter determination module 803 is further configured to initialize an initial mapping matrix product state that conforms to the circuit structure of the target quantum system; iteratively adjust the initial mapping matrix product state until the similarity between the adjusted initial mapping matrix product state and the target matrix product state is maximized, then use the adjusted initial mapping matrix product state as the target mapping matrix product state; and map the initial circuit parameters of the quantum circuit according to the target mapping matrix product state.

[0184] In one embodiment, the target parameter determination module 804 is further configured to use the initial circuit parameters as the current circuit parameters; determine the quantum energy value of the quantum circuit based on the current circuit parameters; update the current circuit parameters based on the stochastic gradient descent optimization model to obtain new current circuit parameters, and return to execute the step of determining the quantum energy value of the quantum circuit based on the current circuit parameters until the quantum energy value of the quantum circuit no longer decreases, then the last obtained current circuit parameters are used as the target circuit parameters of the quantum circuit.

[0185] In one embodiment, the energy value determination module 805 is further configured to obtain multiple equivalent quantum circuits under different noise scaling parameters by performing circuit folding and circuit stretching operations on the target quantum circuit; the target quantum circuit is a quantum circuit whose corresponding circuit parameters are the target circuit parameters; determine the quantum energy value of each equivalent quantum circuit; and obtain the noise-free quantum energy value of the target quantum circuit based on the quantum energy value of each equivalent quantum circuit, which is used as the noise-free ground state energy value of the target quantum system.

[0186] In one embodiment, the energy value determination module 805 is further configured to obtain a mapping relationship between the noise scaling parameter and the quantum energy value of the target quantum circuit based on the quantum energy value of each equivalent quantum circuit and the noise scaling parameter of each equivalent quantum circuit; and obtain the noise-free quantum energy value of the target quantum circuit based on the mapping relationship.

[0187] Each module in the information acquisition device of the aforementioned quantum system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in hardware within or independently of the processor in a computer device, or stored in software within the memory of the computer device, so that the processor can invoke and execute the operations corresponding to each module.

[0188] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 9 As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When executed by the processor, the computer program implements a method for acquiring information in a quantum system. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0189] Those skilled in the art will understand that Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0190] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0191] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0192] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0193] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0194] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0195] The technical features of the above embodiments can be combined in any way. For the sake of brevity, 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 application.

[0196] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for acquiring information from a quantum system, characterized in that, The method includes: The preset initial quantum state of the target quantum system is transformed to obtain the initial matrix product state corresponding to the initial quantum state; With the goal of minimizing quantum energy, the initial matrix product state is optimized to obtain the target matrix product state; Based on the product state of the target matrix, the initial circuit parameters of the quantum circuit are obtained; The initial circuit parameters of the quantum circuit are optimized to obtain the target circuit parameters of the quantum circuit; Based on the target circuit parameters, determine the ground state energy value of the target quantum system; The step of obtaining the initial circuit parameters of the quantum circuit based on the product state of the target matrix includes: Initialize an initial mapping matrix product state that conforms to the circuit structure of the target quantum system; the circuit structure of the target quantum system includes a fixed quantum gate topology. The initial mapping matrix product state is iteratively adjusted until the similarity between the adjusted initial mapping matrix product state and the target matrix product state is maximized. Then, the adjusted initial mapping matrix product state is taken as the target mapping matrix product state. The similarity is the fidelity of the state. Based on the matrix-to-quantum gate conversion rules, the initial circuit parameters of the quantum circuit are obtained by mapping according to the target mapping matrix product state. Specifically, this includes: analyzing the tensor structure of the target mapping matrix product state; using tensor decomposition to map the local tensors in the target mapping matrix product state to the form of parameterized quantum gates; determining the structures of the corresponding single-qubit rotation gate and two-qubit entanglement gate according to the local orthogonal tensor distribution of the target mapping matrix product state; extracting the numerical information of each tensor; and converting it into the corresponding initial circuit parameters. During the parameter mapping process, a transformation learning algorithm or least squares fitting method is used to ensure that the obtained initial circuit parameters make the quantum state output by the quantum circuit close to the target matrix product state.

2. The method according to claim 1, characterized in that, The process of optimizing the initial matrix product state to obtain the target matrix product state with the goal of minimizing quantum energy includes: Use the initial matrix product state as the current matrix product state; Calculate the quantum energy value corresponding to the current matrix product state based on the Hamiltonian of the target quantum system; The current matrix product state is normalized to a central orthogonal form to obtain the normalized current matrix product state. Based on the normalized current matrix product state, a stochastic gradient descent optimization model is used to adjust the current matrix product state to obtain a new current matrix product state. Then, the step of calculating the quantum energy value corresponding to the current matrix product state based on the Hamiltonian of the target quantum system is returned to be executed until the quantum energy value corresponding to the current matrix product state no longer decreases. Then, the last obtained current matrix product state is taken as the target matrix product state.

3. The method according to claim 1, characterized in that, The optimization of the initial circuit parameters of the quantum circuit to obtain the target circuit parameters of the quantum circuit includes: Use the initial circuit parameters as the current circuit parameters; Based on the current circuit parameters, determine the quantum energy value of the quantum circuit; Based on the stochastic gradient descent optimization model, the current circuit parameters are updated to obtain new current circuit parameters. Then, the step of determining the quantum energy value of the quantum circuit based on the current circuit parameters is returned to be executed until the quantum energy value of the quantum circuit no longer decreases. Then, the last obtained current circuit parameters are used as the target circuit parameters of the quantum circuit.

4. The method according to claim 1, characterized in that, Determining the ground state energy value of the target quantum system based on the target circuit parameters includes: By performing circuit folding and stretching operations on the target quantum circuit, multiple equivalent quantum circuits with different noise scaling parameters are obtained; the target quantum circuit is a quantum circuit whose corresponding circuit parameters are the target circuit parameters. Determine the quantum energy value of each equivalent quantum circuit; Based on the quantum energy values ​​of each equivalent quantum circuit, the noise-free quantum energy value of the target quantum circuit is obtained, which is then used as the noise-free ground state energy value of the target quantum system.

5. The method according to claim 4, characterized in that, The step of obtaining the noise-free quantum energy value of the target quantum circuit based on the quantum energy values ​​of each equivalent quantum circuit includes: Based on the quantum energy values ​​of each equivalent quantum circuit and the noise scaling parameters of each equivalent quantum circuit, the mapping relationship between the noise scaling parameters and the quantum energy values ​​of the target quantum circuit is obtained. Based on the mapping relationship, the noise-free quantum energy value of the target quantum circuit is obtained.

6. An information acquisition device for a quantum system, characterized in that, The device includes: A quantum state conversion module is used to convert a preset initial quantum state of a target quantum system to obtain an initial matrix product state corresponding to the initial quantum state; The product state optimization module is used to optimize the initial matrix product state with the goal of minimizing the quantum energy value, so as to obtain the target matrix product state. An initial parameter determination module is used to obtain the initial circuit parameters of the quantum circuit based on the target matrix product state. The target parameter determination module is used to optimize the initial circuit parameters of the quantum circuit to obtain the target circuit parameters of the quantum circuit. An energy value determination module is used to determine the ground state energy value of the target quantum system based on the target circuit parameters. The initial parameter determination module is further configured to initialize an initial mapping matrix product state that conforms to the circuit structure of the target quantum system; the circuit structure of the target quantum system includes a fixed quantum gate topology; the initial mapping matrix product state is iteratively adjusted until the similarity between the adjusted initial mapping matrix product state and the target matrix product state is maximized, then the adjusted initial mapping matrix product state is taken as the target mapping matrix product state; the similarity is the fidelity of the state; based on the matrix-to-quantum gate conversion rule, the initial circuit parameters of the quantum circuit are mapped according to the target mapping matrix product state; The initial parameter determination module is further used to analyze the tensor structure of the target mapping matrix product state, and to map the local tensors in the target mapping matrix product state to the form of parameterized quantum gates using tensor decomposition. Based on the local orthogonal tensor distribution of the target mapping matrix product state, the structure of the corresponding single-bit rotation gate and two-bit entanglement gate is determined, and the numerical information of each tensor is extracted and converted into the corresponding initial circuit parameters. In the parameter mapping process, a transformation learning algorithm or a least squares fitting method is used to perform the mapping to ensure that the obtained initial circuit parameters can make the quantum state output by the quantum circuit close to the target matrix product state.

7. The apparatus according to claim 6, characterized in that, The product state optimization module is further configured to: use the initial matrix product state as the current matrix product state; calculate the quantum energy value corresponding to the current matrix product state based on the Hamiltonian of the target quantum system; normalize the current matrix product state into a centroorthogonal form to obtain the normalized current matrix product state; based on the normalized current matrix product state, use a stochastic gradient descent optimization model to adjust the current matrix product state to obtain a new current matrix product state, and return to execute the step of calculating the quantum energy value corresponding to the current matrix product state based on the Hamiltonian of the target quantum system, until the quantum energy value corresponding to the current matrix product state no longer decreases, then use the last obtained current matrix product state as the target matrix product state.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

Images