Search

CN-121981292-A - Distributed quantum circuit mapping method based on block ZXZ decomposition

CN121981292ACN 121981292 ACN121981292 ACN 121981292ACN-121981292-A

Abstract

The invention discloses a distributed quantum circuit mapping method based on block ZXZ decomposition, which comprises the steps of 1 analyzing an input quantum circuit and constructing a logic quantum bit sequence, 2 obtaining distributed subsystem resource information, 3 counting CNOT gate quantity between target bits and residual sub-bits, 4 performing bit distribution by adopting a greedy strategy according to a statistical result, and 5 recursively repeating the steps 3 and 4 on residual logic quantum bits until all ZXZ decomposition and quantum proportion distribution are completed on the quantum circuit. The invention realizes the integration of quantum operator decomposition and distributed resource scheduling based on the high efficiency of block ZXZ decomposition and the dynamic optimality of greedy mapping algorithm, effectively reduces the communication overhead between systems while ensuring the decomposition precision, and provides an extensible solution for the engineering realization of distributed quantum computing.

Inventors

  • DU ZHENLONG
  • LIU ZHILONG
  • ZHOU XIANGZHEN
  • MENG FANXU

Assignees

  • 南京工业大学

Dates

Publication Date
20260505
Application Date
20260119

Claims (10)

  1. 1. The distributed quantum circuit mapping method based on block ZXZ decomposition is characterized by comprising the following steps: step 1, analyzing an input quantum circuit and constructing a logic quantum bit sequence ; Step 2, obtaining distributed subsystem resource information including subsystem number Physical bit capacity per subsystem Topology and cost models; step 3, selecting the current target bit t decomposed by the block ZXZ under the recursive frame, performing a block ZXZ decomposition on the circuit once to decouple t and count the CNOT gate interaction weight set of the rest of the quantum bits q and t ; Step 4, distributing the target bit t to the subsystem with the least current residual quantum bit quantity Then, according to the CNOT gate interaction weight set, the logic qubit with the smallest interaction weight with t is obtained Allocated to the subsystem where the remaining qubit resources are the most and different from t Thereby ensuring the least consumed cross-system resources, and maximizing the logic quantum bit of the CNOT gate interaction weight between t and t Selecting the target bit decomposed for the next block ZXZ if t is in the subsystem There are still available physical qubit resources, then Preferentially distributing to the subsystems same as t so as to reduce the number of the subsequent cross-subsystem CNOT gates; and 5, recursively repeating the step 3 and the step 4 on the residual logic qubits until all ZXZ decomposition and quantum proportion distribution are completed on the quantum circuit.
  2. 2. The distributed quantum circuit mapping method based on block ZXZ decomposition according to claim 1, wherein the step1 specifically includes: Step 1.1 input as unitary matrix OR gate sequence, analyzing to obtain logical bit number n and logical bit index set ; Step 1.2, selecting initial target bit according to the index order of logic bit Then the natural order recursion is decomposed in block ZXZ, the sequence of logical qubits to be allocated is represented as a queue Q, initialized to 。
  3. 3. The distributed quantum circuit mapping method based on block ZXZ decomposition according to claim 1, wherein the step 2 specifically includes: step 2.1 reading the subsystem set Any subsystem of (a) Physical bit capacity Reading out Number of available quantization bits in (a) And build a maintenance array Initializing And the subsystem With other subsystems Cost of quantum communication parameters between Wherein the quantum communication cost parameter For characterizing the communication overhead that needs to be paid when two qubits of a CNOT gate are located in different subsystems, respectively; Step 2.2, setting the cost of the CNOT gate in the subsystem as Let the cost of CNOT gate between systems be Maintaining an array The initialization map dictionary M is initially empty.
  4. 4. The distributed quantum circuit mapping method based on block ZXZ decomposition according to claim 1, wherein the step 3 specifically includes: step 3.1 performing a block ZXZ split decoupling operation to split the current sub-circuit in block ZXZ, separating the target bit t from the whole, resulting in a set of gates, i.e. a uniform controlled rotation Gate and n-1 bit gate, so that t and other bit interactions appear in a well-defined gate pattern, combined with Gray code will be uniformly controlled rotated If the input is a gate sequence, equivalently finding out a gate subset related to t; step 3.2, analyzing the gate sequence of the decomposition result, and counting each other logic bit CNOT gate interaction weight set with t For each remaining logical bit Statistics of And t, the number of CNOT gates that will occur in the subsequent circuit; step 3.3 pair Normalization is carried out, weighting treatment is carried out according to actual conditions, and an ordered interactive weight set array is generated 。
  5. 5. The distributed quantum circuit mapping method based on block ZXZ decomposition according to claim 1, wherein the step 4 specifically includes: Step 4.1, target t is directly put into the subsystem with the least current residual quantum bit quantity Establishing a physical mapping for the mobile terminal; step 4.2, the interaction weight is calculated Minimal logical qubits The subsystem with the most current residual qubit resources is preferentially placed, and is not in the same system with t, so that the minimum intersystem CNOT gate is generated; Step 4.3 interaction weight Maximum logical qubit As the next target bit, if Still has empty space, give preference to handle Put into So that the logic qubit with high CNOT gate interaction frequency can complete the double-bit gate operation in the same subsystem as much as possible, thereby reducing the number of CNOT gates between subsequent systems, otherwise Put into the subsystem with the least current qubit resources.
  6. 6. The distributed quantum circuit mapping method based on block ZXZ decomposition according to claim 1, wherein the step 5 specifically includes: Step 5.1, recursively carrying out the decomposition of the block ZXZ in the step 3 and the step 4, and updating t to be obtained in the last step Repeating decoupling and allocation for the new target until all logic qubits are allocated, completing all physical mapping; And 5.2, generating a final mapping M and a mapped circuit.
  7. 7. A distributed quantum circuit mapping system based on block ZXZ decomposition, comprising: the quantum circuit input module is used for inputting a quantum circuit to be decomposed and constructing a logic quantum bit sequence; A block ZXZ decomposition solution module, configured to perform a block ZXZ decomposition on the quantum circuit to decouple the target bit and count a set of CNOT interaction weights of the remaining quantum bits and the target bit; The CNOT statistical analysis module is used for counting the number of CNOT gates between the target bit and the residual sub-bit; The greedy bit mapping module is used for carrying out bit allocation by adopting greedy strategies according to the statistical results; And the recursion execution and result output module is used for repeatedly executing the recursion process of the block ZXZ decomposition and greedy mapping on all the quantum bits until the matrix decomposition is complete and the bit mapping table M is filled, so that a final distributed quantum circuit structure is formed.
  8. 8. The distributed sub-circuit mapping system based on block ZXZ decomposition of claim 7 wherein said greedy strategy performs bit allocation by allocating target bit t to a subsystem having a minimum number of currently remaining qubits Then, according to the CNOT gate interaction weight set, the logic qubit with the smallest interaction weight with t is obtained Allocated to the subsystem where the remaining qubit resources are the most and different from t Thereby ensuring the least consumed cross-system resources, and maximizing the CNOT interaction weight with t Selecting the target bit decomposed for the next block ZXZ if t is in the subsystem There are still available physical qubit resources, then Preferentially allocated to the same subsystem as t to reduce the number of subsequent cross-subsystem CNOT gates.
  9. 9. A computer device comprising a memory, a processor, and computer readable instructions stored in the memory and executable on the processor, wherein the processor, when executing the computer readable instructions, implements the distributed quantum circuit mapping method of any of claims 1 to 6.
  10. 10. A readable storage medium storing computer readable instructions that, when executed by one or more processors, cause the one or more processors to perform the distributed quantum circuit mapping method of any of claims 1 to 6.

Description

Distributed quantum circuit mapping method based on block ZXZ decomposition Technical Field The invention belongs to the technical field of quantum computing and distributed system optimization, and particularly relates to a distributed quantum circuit mapping method based on block ZXZ decomposition. Background With the rapid development of information technology, the traditional computer based on classical physical laws gradually faces the bottleneck in terms of computing power and energy consumption when dealing with problems such as large-scale optimization, complex simulation, high-dimensional data analysis and the like. Classical computers use bits as basic information units, the state of which can only be 0 or 1, and when the problem scale grows exponentially, the required computing resources tend to rise sharply, and it is difficult to complete the computing task within an acceptable time. Quantum computation is a novel computation paradigm based on the quantum mechanics principle, and a core computation unit is a quantum bit. Unlike classical bits, quantum bits can be in not only the 0 or 1 state, but also in the superposition of both states, and strong correlation between multiple bits is achieved by quantum entanglement. The characteristics enable the quantum computer to have remarkable parallel computing capability on specific problems, and potential exponential or polynomial acceleration advantages are shown in the fields of quantum chemical simulation, password analysis, combination optimization, machine learning and the like. Implementing quantum computing typically requires representing high-level quantum algorithms as a series of basic operations that can be performed on actual quantum hardware, acting on the qubits in the form of quantum gates, and constructing a quantum circuit in a predetermined order. However, limited by the current quantum hardware scale, noise level, and physical connection, quantum circuits must undergo complex decomposition, mapping, and optimization processes prior to actual execution to ensure that they can operate efficiently and reliably on a given hardware architecture. Therefore, how to efficiently construct and optimize a quantum circuit under the condition of limited hardware has become one of the key technical problems of quantum computing engineering and large-scale development. Quantum algorithms, when executed in real hardware, typically require the Gao Weiyou matrix to be decomposed into a basic set of quantum gates consisting of single-bit gates and double-bit gates (e.g., CNOT gates). This process is called quantum gate decomposition, which directly determines the depth and gate count of the quantum circuit, thereby affecting the execution time and fault tolerance requirements of the algorithm. On the existing hardware platform, the error rate of the double-bit gate is obviously higher than that of the single-bit gate, and meanwhile, the execution cost of the double-bit gate crossing physical nodes is particularly high, so that the reduction of the number of the double-bit gates and the cost of the CNOT gate crossing the system is a key for realizing large-scale quantum computation. Among the numerous decomposition methods, block ZXZ decomposition (Block-ZXZ Decomposition) is the current more efficient decomposition method. The block ZXZ decomposition has a better upper bound on the number of CNOTs than quantum shannon decomposition. For example, for a three bit unitary operator, the method requires only 19 CNOT gates to achieve decomposition. This structure breaks down an n-bit unitary matrix into 4 (n-1) bit matrices and three uniformly controlled rotations by recursively blocking the matricesThe gates, thereby effectively reducing the overall gate count. On the other hand, in practical engineering implementation, due to physical manufacturing process, noise level and error control capability, the number of qubits that can be stably controlled by a single quantum processor is still limited, and it is difficult to directly support the operation of a large-scale quantum algorithm. To break through this scale bottleneck, distributed quantum computing architectures have been developed that enable the construction of larger scale quantum systems through the synergistic work between multiple physical quantum processors. However, in a distributed quantum computing architecture, the quantum bits are distributed and deployed in different subsystems, and the execution cost of the quantum gates is not only dependent on the type of the gates, but also closely related to the physical location of the quantum bits involved in the operation. Particularly, when two qubits of a two-bit gate (such as a CNOT gate) are distributed in different subsystems, operations often need to be completed by means of mechanisms such as entanglement distribution, quantum state transmission or classical cooperative control, so that extra quantum communication overhead is introduced, an