Distributed quantum circuit state transmission evaluation method based on state transmission group score

Abstract

The invention relates to the technical field of distributed quantum computing, in particular to a distributed quantum circuit state transfer evaluation method based on state transfer group scoring. According to the method, the problems that the existing method depends on a simplified scheduling model and channel contention possibly caused by parallel communication operation is not restrained are solved, and meanwhile, the defect that a scheduling feasibility verification mechanism is lacked is overcome. According to the technical scheme, the method comprises the following steps that S1, a quantum circuit partition structure is analyzed; s2, calculating potential scores of two bits in the global gate as state transmission bits; s3, expanding a state transmission group: moving the interference gate based on a gate exchange rule to expand the state transmission group; s4, performing scheduling feasibility verification on the state transfer group; S5, performing final state transfer frequency statistics; and calculating the total communication cost according to whether the state transfer group needs to return the cost. The method has the beneficial effects that the state transfer times are accurately evaluated; identification and optimization of the combined transmission opportunity are supported; and an evaluation strategy with adjustable parameters is supported.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of distributed quantum computing, and particularly relates to a distributed quantum circuit teleportation evaluation method based on teleportation group score. BACKGROUND

[0002] With the continuous advancement of quantum computing research, quantum technology has shown unprecedented computing advantages and development potential in dealing with high complexity problems that are difficult for traditional computers to handle. Based on unique physical mechanisms such as quantum superposition and quantum entanglement, quantum computers can achieve far more speed than classical computing architecture on certain specific tasks. At present, medium-sized quantum chips represented by superconducting qubits have been introduced one after another, and are opened to scientific research users through cloud platforms and other means, which is a key step for the practical application of quantum computing. However, with the increasing number of qubits, the complexity of physical connections required inside the chip and the precision of gate operations also increase rapidly, and limited by current hardware processes and material technology, the number of qubits on a single chip cannot be expanded indefinitely. Referring to the paper: Efficient distributed quantum computing [J]. Proceedings: Mathematical, Physical and Engineering Sciences, 2013, 469(2153): 1-20.

[0003] Reference paper: Monroe C, Raussendorf R, Ruthven A, et al. Large-scale modular quantum-computer architecture with atomic memory and photonic interconnects[J]. Physical Review A, 2014, 89(2): 022317. To break through this scale bottleneck, distributed quantum computing architecture emerges. This architecture interconnects multiple physically separated quantum processing units (QPU) to make the logical quantum circuit map to multiple chips and execute in parallel, thus supporting larger scale and more complex quantum tasks. In the distributed architecture, qubits are divided into different QPU to perform local operations, while the quantum gate operations involving two partitions need to interact through cross-chip communication means. Such operations not only depend on communication channel resources, but also involve quantum entanglement establishment, classical information exchange and other steps, whose cost is much higher than that of local gate operations, and becomes an important factor affecting system performance.

[0004] Reference paper: Verma V, Prakash H. Standard Quantum Teleportation and Controlled Quantum Teleportation of an Arbitrary N-Qubit Information State[J]. International Journal of Theoretical Physics, 2016, 55(4): 2061-2070. Usually, the gate acting on the qubits of two different partitions is called global gate, and its execution often depends on the implementation of cross-zone quantum state migration by using teleportation technology. As the basic mechanism in distributed quantum communication, the core idea of teleportation is to use the pre-established entanglement resource to perform Bell measurement at the sending end, and send the measurement results to the receiving end through the classical channel. The receiving end performs appropriate unitary operation according to the measurement results, so as to realize the remote transmission of quantum state without moving physical bits. Although this process does not involve real quantum bit migration in theory, in actual system, it still needs high-precision control operation and cross-node synchronization, which is costly.

[0005] Reference paper: Ismail G, Zohreh D, Mariam Z, et al. Connectivity matrix model of quantum circuits and its application to distributed quantum circuit optimization[J]. Quantum Information Processing, 2021, 20(7): To reduce this high-cost communication overhead, it is usually attempted to combine multiple global gates that act on the same communication qubits into one continuous teleportation operation, thus sharing the transmission and return paths, and achieving the compression of communication rounds, which is called combined transmission. However, not all global gates involving the same communication bits can be successfully combined.

[0006] In addition to the interference of qubits on the operation path, at the execution level, the overlapping of time windows between teleportation groups will also hinder the feasibility of combined execution, causing communication channel conflicts. Therefore, the assessment of the combinability of teleportation paths not only needs to consider the gate sequence structure, but also must be comprehensively analyzed in combination with resource scheduling and time-dependent factors.

[0007] Therefore, in actual distributed quantum computing systems, how to efficiently identify the combination opportunities between global gates, avoid communication conflicts, and determine whether teleportation is needed, has become a key problem in communication optimization. Ismail G, Zohreh D, Mariam Z, et al. Connectivity matrix model of quantum circuits and its application to distributed quantum circuit optimization[J]. Quantum Information Processing, 2021, 20(7): In distributed quantum circuit optimization, existing methods generally use the number of communications as an estimate of the teleportation cost, and reduce the number of global gates (i.e., gates acting on cross-partition qubits) to approximate the reduction of communication overhead. In order to further compress the number of teleportations, some research (such as the work of Ghodsollahee et al.) proposes to identify adjacent global gates as combinable transmission groups, and by exchanging the execution order of the gates, multiple gates that act on the same transmission bit in succession are completed in one transmission, thereby reducing the number of teleportations. This "combined transmission" strategy improves the communication redundancy problem to some extent and is one of the more typical optimization methods currently.

[0008] However, such methods usually rely on simplified scheduling models and have three main limitations:

[0009] Firstly, they generally assume that the state transfer resources are unlimited or conflict-free, without making constraints on the channel contention that multiple merging groups may bring about when executed simultaneously;

[0010] Secondly, the judgment criteria for merging are mostly based on the "proximity" of the gates in the topology structure or gate sequence, lacking analysis of the actual scheduling timing between the gates and the state transfer time of the qubits, making it difficult to accurately determine whether multiple gates can be truly merged and transmitted;

[0011] Thirdly, the influence of gate operation on the occupation of qubits is not considered by the system, resulting in that the merging judgment often ignores the potential interference in the middle time slice, and the difficulty of merging is underestimated or the feasibility of merging is misjudged.

[0012] With the expansion of the quantum circuit scale and the improvement of gate density, these idealized assumptions have been difficult to establish in practical applications. The contention of communication resources, the scheduling dependence of global gates, and the occupation problem of state transfer return have become key factors limiting performance.

[0013] How to solve the above problems is the subject of the present application. SUMMARY

[0014] The purpose of the present application is to provide a distributed quantum circuit state transfer evaluation method based on state transfer group scoring to solve the problem of oversimplified scheduling model in the prior art, which is difficult to accurately evaluate the communication cost. This method can accurately depict the coordination potential and mutual interference between global gates, and more realistically reflect the actual number of communication rounds required in the state transfer execution process. The present application is an improved method aimed at the above-mentioned pain points. By introducing a state transfer conflict detection mechanism in the path construction process, the present application can dynamically identify feasible merging transmission paths, reasonably allocate state transfer scheduling order, and significantly improve the accuracy of communication cost estimation, thereby providing a more solid analysis foundation for subsequent partitioning strategies and scheduling optimization.

[0015] In order to achieve the above-mentioned application purpose, the technical scheme adopted by the present application is as follows:

[0016] A distributed quantum circuit state transfer evaluation method based on state transfer group scoring, comprising the following steps:

[0017] S1: Analyze the known quantum circuit partition structure item by item; the present application analyzes the input quantum gate sequence item by item; for identifying the global gate corresponding to the cross-zone operation, clearly communicating the demand, providing basic data for subsequent state transfer path construction.

[0018] S2: For two qubits in the global gate, evaluate them according to the scoring function respectively, and calculate their potential score as a state transfer path; for intelligently selecting the optimal state transfer bit, improving the utilization efficiency of communication resources, and avoiding blind path allocation.

[0019] S3: expanding the teleportation group: if there exists other gate operations acting on the same teleportation bit between the target global gate and the end gate of the current teleportation group, i.e., there exists an interference gate, the system attempts to move the interference gate out of the teleportation interval of the teleportation group according to the gate exchange rule;

[0020] If the interference gate is successfully moved out of the teleportation interval of the teleportation group, the gate is added to the teleportation group; otherwise, the expansion is terminated.

[0021] The gate exchange rule used in the process is based on the principle of quantum gate equivalence, which attempts to move the interference gate out of the teleportation interval of the teleportation group by reasonably adjusting the position of the interference gate in the gate sequence. This can effectively release potential path conflicts, thereby achieving a larger range of merging scheduling and creating conditions for further reduction of cross-zone communication cost.

[0022] The core idea of the mechanism is that when judging whether a global gate can be merged into the current teleportation group, the method first detects whether there are other gate operations acting on the same teleportation bit between the gate and the end gate of the teleportation group, referred to as interference gates. If there are interference gates, further judge whether these interference gates can be moved out of the teleportation interval occupied by the current teleportation group (i.e., the continuous segment of the start gate to the end gate in the gate sequence) by gate sequence transformation. To achieve this operation, the method uses the existing gate exchange rule to judge whether the interference gate can be moved out of the current teleportation interval by adjusting the order, without changing the execution semantics of the circuit, thereby attempting to schedule the interference gate outside the teleportation interval and creating conditions for subsequent merging

[0023] Under the premise of meeting the commutative rule, the interference gate is preferentially moved backward, and if it cannot be successfully moved out of the conflict interval, it is attempted to move forward in reverse, until the interference gate leaves the teleportation interval of the current teleportation group. If the interference gate is successfully moved out, it means that the conflict has been eliminated, and the global gate to be merged can be merged into the teleportation group, and the termination boundary of the teleportation group is extended; if the movement fails, it means that the gate cannot be merged at present, and the expansion of the teleportation group will be terminated.

[0024] Through the mechanism, the present application effectively combines teleportation bit selection and scheduling optimization, not only relying on the scoring function to make a static guiding decision in the path selection stage, but also dynamically adjusting the gate sequence structure in the merging process to actively eliminate conflict factors that hinder merging, thereby forming a collaborative optimization system of scoring decision and path scheduling. This design greatly enhances the robustness and applicability of merging transmission, improves the compression capability of the overall communication path, and can adapt to more complex gate sequence structures and circuit constraints.

[0025] S4: Verify the scheduling feasibility of the constructed transfer groups, focusing on detecting conflicts in the concurrent use of communication resources. Since any communication qubit can only be exclusively used by one transfer group at the same time, it is necessary to ensure that all transfer groups do not overlap in time or share the same transfer bits.

[0026] This mechanism, serving as a verification step in the path construction process, primarily confirms the physical feasibility of the current transfer scheduling scheme. Since the system has minimized resource conflicts as much as possible during the path selection and gate sequence scheduling phase in step S3, scheduling conflicts are uncommon. However, this mechanism still serves as a necessary verification measure, performing consistency checks on the scheduling intervals of all transfer groups to ensure that no transfer intervals overlap, thereby guaranteeing the temporal exclusivity of transfer operations and the correctness of system execution. Through this mechanism, this invention ensures that the constructed communication path meets scheduling constraints and possesses physical realizability, thus avoiding potential transfer contention issues.

[0027] S5: Performs final state transfer count statistics; calculates the total communication cost based on whether the state transfer group requires a return cost. Used to output the actual communication cost of the quantum circuit, supporting the evaluation and performance comparison of partitioning schemes.

[0028] Step S1 is as follows:

[0029] S11: Define the global gate; filter out all two-qubit gates that operate between qubits belonging to different partitions;

[0030] Since the execution of global gates relies on cross-region communication, which is achieved through stealth transfer, it constitutes the main source of communication costs.

[0031] S12: Construct a collection of structured objects; all identified global gates will be constructed into a collection of structured objects;

[0032] The set of data includes: position in the gate sequence, gate type, control and target bit number, and corresponding partition.

[0033] This structured data will serve as the basis for path construction, merging judgment, and conflict detection.

[0034] The scoring function formula for step S2 is:

[0035] Score(i) = αM * -βR i ,i∈{1,2} (1)

[0036] Where i represents two bits of the global gate; R i The ReturnPenalty flag indicates whether a return value is needed, represented by 0 or 1.

[0037] α and β are system configurable weight parameters, the higher the Score score value, the greater the potential of the transduction bit, the smaller the number of transduction times; after selecting the bit with the highest Score score, the bit is selected as the transduction bit to construct the transduction group, and the starting time of the transduction group is recorded.

[0038] M of formula (1) * The calculation formula is:

[0039]

[0040] Wherein, M * represents the sum of the future global gate scores of the bit

[0041] M of formula (2) x The calculation formula of is:

[0042]

[0043] Wherein, M x represents the score of each global gate of the bit, d represents how many interference gates before the global gate, x represents the first global gate on the bit, and m and n represent positive integers. M x The lower the score, the lower the possibility of avoiding interference; that is, the more interference gates before the global gate, the lower the possibility of moving out to avoid interference through the gate exchange rule, so the score is lower.

[0044] The scoring function considers three factors:

[0045] (1) The merging potential of future global gates on the bit (number and interference gate penalty coefficient);

[0046] (2) Whether the path needs to perform return transduction (return cost);

[0047] (3) Weight coefficient in the scoring function (adjustable parameter).

[0048] Step S4 checks whether there is a conflict, which is to check whether there is an intersection between the set composed of the index (start_idx) of the starting global gate of each transduction group in the gate sequence and the gate sequence index (end_idx) of the ending global gate. There is no intersection, which means no conflict.

[0049] In the quantum circuit, "before" refers to the position of the operation appearing earlier in the gate sequence, and "after" refers to the position of the operation appearing later in the gate sequence, that is, the arrangement direction from left to right according to the execution order of the gates to determine the relationship between the first and the second.

[0050] The calculation formula of step S5 is:

[0051] TC=a+2b (6)

[0052] Wherein, TC is the total number of transmittal, a is the number of transmittal groups without return cost, and b is the number of transmittal groups with return cost.

[0053] In the present application, the transmittal groups are divided into two categories:

[0054] The first category is the transmittal group without return, that is, the quantum bits on the selected transmittal path are not used by the subsequent global gate after the end of the group, and only one transmittal operation is needed to complete the entire communication task.

[0055] The second category is the transmittal group with return, that is, the quantum bits on the communication path still need to return to the original partition to participate in the subsequent gate operation after the end of the group, so one transmittal and one return operation are needed, a total of two transmittals.

[0056] When the system constructs each transmittal group, it records the use of the transmittal bits in the subsequent gate sequence, and then judges whether there is a "return cost".

[0057] According to the judgment result, each transmittal group is given a corresponding transmittal frequency weight: the transmittal group without return is recorded as 1 transmittal, and the transmittal group with return is recorded as 2 transmittals.

[0058] Finally, the system traverses all the legal transmittal groups, adds up the corresponding transmittal frequencies of each group, and obtains the total cross-zone transmittal frequency of the entire quantum circuit under the current partition scheme. The result can be used as an important evaluation index of the communication cost of the distributed quantum circuit, and provides a key basis for partition strategy selection, communication scheduling design and subsequent system performance optimization.

[0059] Compared with the prior art, the present application has the following advantages:

[0060] (1) Accurate evaluation of cross-zone transmittal frequency:

[0061] The present application analyzes the use behavior of the communication quantum bits in the quantum circuit in depth, without changing the existing partition scheme, and comprehensively considers the execution timing of the quantum gate, the overlapping relationship between the operations, and the conflict situation between the transmittal zones, to establish a fine-grained cross-zone transmittal modeling mechanism. Unlike the existing method which only roughly estimates the transmittal cost according to the number of global gates, the present application can accurately judge whether each transmittal has a return operation, whether there is an interference block, and whether there is a merging space, so as to more truly reflect the communication load in the actual execution process of the quantum circuit. The mechanism effectively improves the precision of the communication cost perception, and can provide more reliable quantitative basis for partition quality evaluation, system architecture design and task scheduling strategy.

[0062] (2) Support for automatic identification and optimization of transmission merging opportunities:

[0063] The application designs a set of transmission group construction and merging process based on a scoring mechanism, which can score and sort according to the future use of each global gate involved in the communication bits, the number of interference gates and whether it needs to return transmission and other factors, and preferentially select the communication path with higher merging potential. In the process of constructing the transmission group, the application further introduces an interference gate scheduling mechanism based on the commutative gate rule. Without changing the semantics of the quantum circuit, the interference gate is moved out of the transmission interval by adjusting the gate sequence, thereby releasing the originally blocked merging space. This cooperative strategy not only improves the transmission compression rate, but also enhances the adaptability of the method under complex gate sequence structures, which can significantly reduce redundant communication and improve overall execution efficiency.

[0064] (3) Support parameter-adjustable evaluation strategy to adapt to different application scenarios:

[0065] Considering the differences in communication resources, topology structure and scheduling constraints of different quantum platforms, the application introduces a set of adjustable scoring parameter system. Users can flexibly configure the merging potential score weight, return transmission penalty factor, interference gate influence coefficient and other parameters according to the actual application scenario, thereby controlling the tendency of path selection and the aggressiveness of scheduling strategy. This mechanism gives the application good versatility and adaptability, which can be used to improve merging efficiency in systems with limited communication resources, and can also guarantee scheduling feasibility in scheduling conflict-sensitive scenarios, with good scalability and engineering application prospects. BRIEF DESCRIPTION OF DRAWINGS

[0066] The accompanying drawings are included to provide a further understanding of the application, and constitute a part of the specification, which together with the embodiments of the application are used to explain the application, and do not constitute a limitation of the application.

[0067] Figure 1 The distributed quantum circuit schematic diagram of embodiment 1 of the application.

[0068] Figure 2 The two equivalent exchange schematic diagrams of the application.

[0069] Figure 3 The transmission group schematic diagram of the application.

[0070] Figure 4 The distributed quantum circuit schematic diagram in embodiment 2 of the application.

[0071] Figure 5 The schematic diagram of establishing transmission group 1 in embodiment 2 of the application.

[0072] Figure 6 The circuit transformation schematic diagram after determining the transmission group in embodiment 2 of the application.

[0073] Figure 7Establishment of transmission group 2 for embodiment 2 of the present application.

[0074] Figure 8 Final line diagram after transformation for embodiment 2 of the present application.

[0075] Figure 9 Distributed quantum line diagram in embodiment 3 of the present application.

[0076] Figure 10 Establishment of transmission group 1 for embodiment 3 of the present application.

[0077] Figure 11 Line diagram after transmission group 1 is determined for embodiment 3 of the present application.

[0078] Figure 12 Establishment of transmission group 2 for embodiment 3 of the present application.

[0079] Figure 13 Final line diagram after transformation for embodiment 3 of the present application. DETAILED DESCRIPTION

[0080] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below in combination with the drawings and embodiments. Of course, the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0081] Embodiment 1

[0082] A distributed quantum line transmission evaluation method based on transmission group score, comprising the following steps:

[0083] S1: Analyze the known quantum line partition structure item by item; the present application analyzes the input quantum gate sequence item by item;

[0084] Used for identifying the global gate corresponding to the cross-zone operation, clarifying the communication demand, and providing basic data for subsequent transmission path construction.

[0085] Step S1 is specifically:

[0086] S11: Define global gate (Global Gate); filter out all double quantum bit gates acting between quantum bits belonging to different partitions;

[0087] Since the execution of the global gate depends on cross-zone communication, which is realized by teleportation, it is the main source of communication cost.

[0088] S12: Construct a structured object set; all identified global gates will be constructed into a structured object set;

[0089] The set data is: the position in the gate sequence, the gate type, the control and target bit number, and the corresponding partition.

[0090] These structured data will be the basis for path construction, merging judgment, and conflict detection.

[0091] S2: For two qubits in the global gate, evaluate them according to the scoring function respectively to calculate their potential scores as teleportation paths; used for intelligent selection of optimal teleportation bits to improve communication resource utilization efficiency and avoid blind path allocation.

[0092] The scoring function formula of step S2 is:

[0093] Score(i) = aM * - bR i , i e {1, 2} (1)

[0094] Where i represents two bits in the global gate; R i indicates whether to return to use (ReturnPenalty), represented by 0 and 1

[0095] a and b are system configurable weight parameters, the higher the Score score value, the greater the teleportation bit potential and the fewer the teleportation times; after selecting the bit with the highest Score score, the bit is used as a teleportation bit to construct a teleportation group, and the start time of the teleportation group is recorded.

[0096] M * in formula (1) is calculated as:

[0097]

[0098] Where M * represents the sum of the future global gate scores of the bit

[0099] M x in formula (2) is calculated as:

[0100]

[0101] In the formula, M x represents the score of each global gate of the bit, d represents how many interference gates there are before the global gate, x represents the first global gate on the bit, and m and n represent positive integers. M x The lower the score, the lower the possibility of avoiding interference; that is, the more interference gates before the global gate, the lower the possibility of moving out to avoid interference through gate exchange rules, so the score is lower.

[0102] The scoring function considers three factors:

[0103] (1) the merging potential of the future global gate on this bit (number and interference gate penalty coefficient);

[0104] (2) whether the path needs to perform a return transduction (return cost);

[0105] (3) the weight coefficient in the scoring function (adjustable parameter).

[0106] As shown in Figure 1 , taking global gate G1 as an example, the two quantum bits involved are q0 and q3. If the first quantum bit q0 is selected as the transduction bit, the system will evaluate its potential as a communication path. At this time, q0 is still used by global gates G3 and G4 in the subsequent gate sequence, and there are no other interference gate operations acting on q0 between these two gates, so it can form a continuous merge with the current gate G1 to constitute the same transduction group. Therefore, its M * score is 2. At the same time, the transduction bit q0 is not used by other gates after G4, meaning that it can be released directly after completing all related gate operations without performing an additional transduction return operation. Therefore, its return cost R is 0, and the values of a and b are 3 and 1 respectively. The score of q0 as a transduction bit can be calculated as 6. In contrast, if the other quantum bit q3 of G1 is selected as the transduction path, there are also two global gates that can be merged after it, with an M * score of 2. However, q3 is called again by a local gate in the subsequent gate G6, meaning that its quantum state still needs to return to the original partition to continue participating in the calculation task after transmission. Therefore, this transduction bit needs to account for a return operation, with a return cost R of 1, and a score of 5 in the scoring function. Comparing the scores of the two, the score of q0 is higher than that of q3, so the system ultimately selects q0 as the transduction bit of G1, thereby maximizing the reduction of actual transduction times.

[0107] S3: Extend the transduction group: if there are other gate operations acting on the same transduction bit between the target global gate and the last gate of the current transduction group, i.e., there are interference gates, the system will attempt to move the interference gates out of the transduction interval according to the gate exchange rule.

[0108] If the interference gate is successfully moved out of the transduction interval of the transduction group, it will be added to the transduction group; otherwise, the extension will be terminated.

[0109] The gate exchange rule used in this process is based on the principle of quantum gate equivalence, which attempts to move the interference gate out of the transduction interval of the transduction group by reasonably adjusting its position in the gate sequence. This effectively releases potential path conflicts, enabling greater merging and scheduling, and creating conditions for further reduction of cross-zone communication costs.

[0110] The core idea of the mechanism is that when judging whether a global gate can be added to an existing transmission group, the system first detects whether there is an interference gate between the gate and the last gate of the transmission group, and if there is, it tries to move it out of the conflict interval through equivalent gate exchange transformation. Based on the commutativity principle of quantum gate operation, the application constructs a set of gate exchange rules to determine whether two gates can be exchanged in order without affecting the execution semantics of the line. Figure 2 As shown in FIG. 8, two typical CNOT gate exchange cases are given: when the control bits of the two CNOT gates are the same and the target bits are different, the gate order can be equivalent exchanged, or when the target bits of the two gates are the same but the controls are different, the gates can also be exchanged.

[0111] Under the premise of meeting the commutativity rules, the interference gate is preferentially moved backward, and if it cannot be successfully moved out of the conflict interval, it is tried to be moved forward, until it leaves the transmission interval of the current transmission group. If the interference gate is successfully moved out, it means that the conflict has been eliminated, and the current gate to be merged can be safely included in the group, and the termination boundary of the transmission group is extended.

[0112] Through the mechanism, the application effectively combines the transmission bit selection and scheduling optimization, not only relies on the scoring function to make a static guiding decision in the path selection stage, but also dynamically adjusts the gate order structure in the merging process, actively eliminates the conflict factors that hinder the merging, and thus forms a cooperative optimization system of scoring decision and path scheduling. This design greatly enhances the robustness and applicability of the merging transmission, improves the compression capability of the overall communication path, and can adapt to more complex gate order structures and line constraints.

[0113] S4: Verify the scheduling feasibility of the constructed transmission group, which focuses on detecting the concurrent use conflicts of communication resources. Since any communication quantum bit can only be exclusively used by one transmission group in the same time period, it is necessary to ensure that all transmission groups do not overlap in time or share the same transmission bits.

[0114] As a verification link in the path construction process, the mechanism is mainly used to confirm whether the current transmission scheduling scheme is physically feasible. Generally, since the system has avoided resource conflicts as much as possible in the aforementioned path selection and gate order scheduling stages, scheduling conflicts are not common. However, this mechanism is still a necessary verification measure to check the consistency of the scheduling interval of all transmission groups to ensure that there is no overlap of transmission intervals, thereby ensuring the timing exclusivity of the transmission operation and the correctness of the system execution. Through the mechanism, the application ensures that the constructed communication path meets the scheduling constraint conditions and is physically realizable, thereby avoiding potential transmission competition problems.

[0115] Step S4 checks whether there is a conflict between the sets of indexes corresponding to the start global gate (start_idx) and the end global gate (end_idx) of each transmittal group.

[0116] In the quantum circuit, "before" refers to the operation appearing earlier in the gate sequence, and "after" refers to the operation appearing later in the gate sequence, i.e., the order of execution of the gates is determined by the arrangement direction from left to right.

[0117] As shown in Figure 3 , the communication interval of transmittal group 1 is (1, 4) and the communication interval of transmittal group 2 is (5, 8), which do not intersect, which means that there is no conflict between the two transmittal groups.

[0118] S5: Perform final transmittal count; calculate the total communication cost according to whether the transmittal group needs to return the cost.

[0119] The real communication cost of the quantum circuit is output, supporting the evaluation and performance comparison of the partition scheme.

[0120] The calculation formula of step S5 is:

[0121] TC=a+2b (6)

[0122] Where TC is the total number of transmittals, a is the number of transmittal groups that do not need to return the cost, and b is the number of transmittal groups that need to return the cost.

[0123] In the present application, transmittal groups are divided into two categories:

[0124] The first category: transmittal groups that do not need to return, i.e., the quantum bits on the selected transmittal path are not used by subsequent global gates after the end of the group, and only one transmittal operation is needed to complete the entire communication task;

[0125] The second category: transmittal groups that need to return, i.e., the quantum bits on the communication path still need to return to the original partition to participate in subsequent gate operations after the end of the group, so one transmittal and one return operation are needed, a total of two transmittals.

[0126] The system records the use of transmittal bits in subsequent gate sequences when constructing each transmittal group, and then determines whether there is a "return cost".

[0127] According to the judgment result, each transmittal group is assigned a corresponding transmittal frequency weight: transmittal groups that do not need to return are recorded as 1 transmittal, and transmittal groups that need to return are recorded as 2 transmittals.

[0128] Finally, the system traverses all the legal teleportation groups, accumulates the corresponding teleportation times of each group, and obtains the total cross-zone teleportation times of the entire quantum circuit under the current partition scheme. This result can be used as an important evaluation index of the communication cost of the distributed quantum circuit, providing a key basis for partition strategy selection, communication scheduling design, and subsequent system performance optimization.

[0129] Embodiment 2

[0130] Based on Embodiment 1, as shown in Figure 4 , a distributed quantum circuit containing 6 qubits is set, and it is divided into two partitions. To calculate the cross-zone teleportation times of the circuit, first, according to the partition information, the quantum gate sequence is scanned to identify the two-qubit gates acting on qubits in different partitions, i.e., global gates. The results are shown in Figure 5 , a total of 6 global gates are identified: G1, G2, G3, G4, G6, and G7, where G5 is a local gate (its acting qubits are all in the same partition).

[0131] After identifying all the global gates, the system enters the teleportation group construction phase. This process starts with global gate G1 and attempts to construct the first teleportation group. G1 acts on qubits q0 and q3, so scoring is performed on these two candidate teleportation qubits. If q0 is chosen as the teleportation qubit, then among its subsequent gates, G3 and G4 also act on q0, and there are no interfering gates between them and G1, so their scores are both 1. Global gate G6 also acts on q0, but before it there is a local gate G5 that acts on q0, constituting an interference. Therefore, according to the penalty function in the scoring model, the score of this gate is In addition, since q0 is not used after G4, and G6 still needs to continue using this bit, the system determines that the teleportation path needs to perform a teleportation back operation, i.e., the return cost R = 1. Integrating the scores, the total score of using q0 as the teleportation qubit is 3(1 + 1 + 0.5) - 1 = 6.5.

[0132] Similarly, after scoring q3, the total score is 3. Therefore, the system finally chooses q0 as the teleportation qubit of teleportation group 1.

[0133] The system attempts to merge G3 and G4 into teleportation group 1, as there is no interference between them, they can be directly added. For G6, since there is an interfering gate G5 in front of it, the system determines that G5 and G6 can be equivalently exchanged according to the gate exchange rule (i.e., the exchange of gate order does not affect the circuit result), so it attempts to move G5 to the back of G6, thereby removing the interference in the teleportation path. In addition, to avoid communication resource conflicts, the system also removes other gates (such as G2) from the time range of the current teleportation group to ensure the legality of communication scheduling. The final teleportation group 1 after equivalent transformation is shown in Figure 6 .

[0134] After the construction of the first teleportation group, as shown in Figure 7 , the system finishes the marking of the processed gates and continues to process the next unprocessed global gate G2 and attempts to construct the second teleportation group. The candidate teleportation bits for G2 are q1 and q3. The scoring results are as follows:

[0135] If q1 is selected as the communication path, the subsequent global gate G7 also acts on q1, but there is an interference gate G5 in front of it, and the score is and still needs to return, R = 1, so the total score is 1.5 - 1 = 0.5;

[0136] If q3 is selected, there is no mergable gate in the subsequent, and the score is 0.

[0137] Therefore, the system still selects q1 with a higher score as the teleportation bit of the second teleportation group. Finally, the first teleportation group needs to perform 2 times of teleportation (transmission and return), and the second teleportation group only needs to transmit once, a total of 3 times of teleportation operation, to complete the cross-zone communication scheduling analysis of the entire line, and the line is finally transformed as shown in Figure 8 .

[0138] Example 3

[0139] Based on example 1, as shown in Figure 9 , a distributed quantum circuit containing 6 qubits is set, where each subzone contains 3 qubits. First, the system identifies and extracts all two-qubit gates involving qubits in different subzones, i.e., global gates, which include G1, G3, G5, G6, and G8.

[0140] Subsequently, the system begins to attempt to construct the first teleportation group, starting with global gate G1. G1 acts on qubits q0 and q3, so the two candidate teleportation bits need to be scored respectively.

[0141] If q0 is selected as the teleportation bit, among its subsequent gates, G3 and G5 also act on q0, and there is no interference gate between G3 and G1, so the score of G3 is 1, and there is an interference gate G4 between G5 and G3, so the score of G5 is In addition, since q0 is not used after G5, the system judges that this teleportation path does not need to perform a return operation, i.e., the return cost R = 0, and the total score of using q0 as the teleportation bit is 3 (1 + 0.5) - 0 = 4.5.

[0142] Next, consider the case of selecting q3 as the teleportation bit. q3 participates in two global gates G6 and G8 in the subsequent gate sequence. There is an interference gate G2 between G6 and G1, so the score of G6 is There is an interfering gate G7 between G8 and G6, meaning there are a total of two interfering gates in front of G8. Therefore, the score for G8 is... Similarly, q3 was not used after G8, so the return cost was 0. Summarizing the scores, the total score for using q3 as the transfer bit is 3(0.5+0.25)-0 = 2.25. Therefore, the system ultimately chooses q0 as the transfer bit for transfer group 1.

[0143] like Figure 10 As shown, the system attempts to merge G3 into transfer group 1. Since there is no interference between them, it can be added directly. For G5, since there is an interfering gate G4 preceding it, the system determines, according to the gate switching rules, that G5 and G5 can be equivalently swapped (i.e., gate order swapping does not affect the line result). Therefore, it attempts to move G4 after G5 to clear the interference in the transfer path. The final determined line of transfer group 1 after equivalent transformation is shown below. Figure 11 As shown.

[0144] After the transfer group 1 is constructed, as follows Figure 12 As shown, the system marks the processed gates and continues processing the next unprocessed global gate G2, attempting to construct transfer group 2. The candidate transfer bits for G6 are q1 and q3. The scoring results are as follows:

[0145] If q1 is chosen as the communication path, there are no subsequent global gates acting on it, so the score is 0.

[0146] If q3 is selected, a global gate G8 will follow, and there will be an interference gate G7 between G8 and G6, resulting in a score of [score missing].

[0147] Therefore, the system still selects q3, which has the higher score, as the transfer bit for transfer group 2. Ultimately, transfer group 1 and transfer group 2 each need to perform 2 transfers (transmission and return), for a total of 4 transfer operations, completing the cross-regional communication scheduling analysis of the entire line. The line is finally transformed as follows: Figure 13 As shown.

[0148] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A distributed quantum circuit teleportation evaluation method based on a teleportation group score, characterized in that, The method comprises the following steps: S1: item-by-item analysis is performed on a known quantum circuit partition structure; S2: for two qubits in a global gate, a scoring function is used to evaluate the potential score of the two qubits as a teleportation path, respectively; S3: extend the teleportation group: if there is an interfering gate between the target global gate and the last gate of the current teleportation group, i.e., there is an interfering gate, the system attempts to move the interfering gate out of the teleportation interval according to the gate exchange rule; If the interfering gate is successfully moved out of the teleportation interval of the teleportation group, the gate is added to the teleportation group; otherwise, the extension is terminated; S4: verify the scheduling feasibility of the constructed teleportation group: introduce a time conflict analysis mechanism to check whether there is a conflict between the teleportation intervals of each teleportation group; S5: perform final teleportation count statistics; calculate the total communication cost according to whether the teleportation group needs to return the cost.

2. The method of claim 1, wherein, Step S1 specifically comprises: S11: define a global gate; select all two-qubit gates acting on qubits belonging to different partitions; S12: construct a structured object set; all identified global gates will be constructed as a structured object set; The set data is: the position in the gate sequence, the gate type, the control and target qubit numbers, and the corresponding partition.

3. The method of claim 1, wherein, The scoring function formula of step S2 is: Score(i) = aM * - βR i , i e {1,2} (1) where i denotes two bits in a global gate; R i ReturnPenalty, represented by 0 and 1 α and β are configurable weight parameters of the system, and the higher the Score score value, the greater the teleportation bit potential and the smaller the teleportation count; after selecting the bit with the highest Score score, the bit is used as a teleportation bit to construct a teleportation group, and the start time of the teleportation group is recorded.

4. The method of claim 3, wherein, M of formula (1) * The calculation formula is: where M * represents the sum of the future global gate scores of the bits.

5. The method of claim 4, wherein, M of formula (2) x The calculation formula of M is: where M x represents the score of each global gate for the bit, d represents how many interfering gates are before the global gate, x represents the xth global gate on the bit, and m and n represent positive integers; M x The lower the score, the lower the likelihood of avoiding interference.

6. The method of claim 1, wherein, The step of checking whether there is a conflict in step S4 is to form a set of the index of the start global gate start_idx and the index of the end global gate end_idx corresponding to the gate sequence in the gate sequence of each teleportation group.

7. The method of claim 1, wherein, The calculation formula of step S5 is: TC=a+2b (6) Where TC is the total teleportation count, a is the number of teleportation groups without return cost, and b is the number of teleportation groups with return cost.

8. The method of claim 6, wherein, Teleportation groups are divided into two categories: The first type: teleportation group without return, which only needs one transmission operation to complete all communication tasks; The second type: teleportation group with return, which needs one transmission and one return operation, a total of two teleportations.

9. The method of claim 7, wherein, The system records the use of the teleportation bit in the subsequent gate sequence when constructing each teleportation group, and then determines whether there is a "return cost".

10. The method of claim 9, wherein, According to whether there is a return cost, determine the number of teleportation operations required by each teleportation group Teleportation group without return is recorded as 1 time teleportation; Teleportation group with return is recorded as 2 times teleportation.

Images