Neutral atom quantum computing multichannel program parallel compiling optimization method

Abstract

The invention discloses a neutral atom quantum computing multichannel program parallel compiling optimization method, which comprises the following steps of: a parallel compiling step: intelligently distributing a plurality of quantum circuit tasks to a plurality of quantum processing units (QPUs) by utilizing a greedy scheduling algorithm of circuit length priority; and a single QPU multi-channel program compiling step, including processing concurrent execution in a single QPU by using a spatial deformation model. According to the method, a set of complete neutral atom multi-channel program compiling and optimizing scheme is constructed through combination of macroscopic parallel task allocation and microcosmic dynamic space modeling and constraint solution, so that efficient concurrent execution of a plurality of quantum programs on shared hardware resources is realized; therefore, global and difficult-to-manage hardware constraints are effectively converted into a structured and solvable space-time optimization problem, and the parallel computing potential of the neutral atom architecture is released.

Description

Technical Field

[0001] This invention relates to the field of quantum operating systems, and more particularly to a method for parallel compilation and optimization of multiprogramming in neutral atom quantum computing. Background Technology

[0002] Quantum computing technology, with its unique parallel processing capabilities and the principle of quantum superposition, is leading a profound technological revolution and foreshadows the potential for exponential speedups in cryptography, materials science, and complex optimization problems. As quantum hardware, particularly quantum processing units (QPUs), continues to mature, developing complex software infrastructure capable of efficiently scheduling and managing these hardware resources becomes crucial. This naturally leads to the concept of quantum operating systems, which aim to manage and coordinate quantum computing resources, much like classical operating systems, to improve overall computational efficiency.

[0003] In operating systems, multiprogramming is a core function that allows multiple programs to execute concurrently, thereby maximizing hardware utilization and computational throughput. Applying this concept to quantum computing means developing techniques capable of simultaneously compiling and executing multiple quantum circuits (programs) on the same quantum hardware. Current research has explored multiprogramming on superconducting quantum computing platforms, for example, by improving system throughput through resource partitioning or task scheduling. This work lays the foundation for the feasibility of quantum multiprogramming. In the area of ​​neutral atom quantum computing platforms, due to their inherent high parallelism, high gate fidelity, and dynamic reconfigurability, they are considered strong candidates for building scalable quantum computing architectures. Neutral atom systems achieve qubit-to-qubit interactions by moving atoms (qubits) using acousto-optic deflectors (AODs), and this flexibility provides an ideal physical basis for achieving efficient multiprogramming.

[0004] While neutral atom quantum computing platforms possess superior hardware capabilities to support multiprogramming, existing software and compilation techniques suffer from significant limitations that severely hinder the realization of their potential. First, current compilation methods are limited to single-circuit optimization and lack concurrent processing capabilities. Almost all current neutral atom compilation techniques, whether heuristic-based or exact solver-based, are designed to optimize the execution efficiency of a single quantum circuit, such as reducing execution depth or improving fidelity. These methods do not consider scenarios where multiple programs run simultaneously on the same hardware resources, and cannot handle the complex resource contention and scheduling conflicts arising from concurrent execution. This results in hardware resources remaining idle for extended periods, leading to low utilization.

[0005] Secondly, the global physical constraints of neutral atom platforms render simple resource partitioning strategies ineffective. Unlike superconducting qubits with fixed interconnect topologies, AOD movements in neutral atom systems have a global impact on the entire qubit array. For example, moving a row or column controlled by an AOD will affect all other potential movements in that row or column to avoid path crossings. This global coupling characteristic means that, as in some classical multi-core processors or partial quantum platforms, the hardware physical space cannot be simply divided into independent regions and allocated to different programs. Atom movements in any one program can constrain other concurrent programs, a complex challenge that current technologies have failed to address.

[0006] Finally, concurrent execution introduces complex spatiotemporal conflicts. When multiple quantum programs compete for shared hardware resources, their requirements for atomic positions, movement paths, and the timing of quantum gate execution can lead to complex spatiotemporal conflicts.

[0007] Therefore, how to establish a model that can understand and resolve these conflicts, while ensuring the correctness of each circuit and optimizing the overall system performance, is an urgent problem to be solved in this field for the implementation of neutral atomic multiprogramming. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for optimizing the parallel compilation of multiprogramming in neutral atom quantum computing.

[0009] The objective of this invention is achieved through the following technical solution: A first aspect of the present invention provides a method for parallel compilation optimization of multiprogramming for neutral atom quantum computing, comprising: The parallel compilation steps include using a circuit length-first greedy scheduling algorithm to intelligently allocate multiple quantum circuit tasks to multiple quantum processing units (QPUs). The compilation steps for a single QPU multiprogramming include using a spatial deformation model to handle concurrent execution within a single QPU.

[0010] Furthermore, the parallel compilation step includes: It receives a set of quantum circuits to be compiled and one or more available neutral atom QPU arrays as input, and sorts them in ascending order according to execution length; Initialize the allocation by assigning the shortest batch of sorted quantum circuits to the available QPUs one by one; Incremental allocation involves finding the insertion point for each remaining quantum circuit across all QPUs that satisfies the hardware resource constraints of the QPUs and has the shortest overall execution time.

[0011] Furthermore, the parallel compilation step also includes the following following the incremental allocation: In-array optimization involves fine-tuning the execution order of circuits within each QPU after all quantum circuits have been allocated, in order to further reduce execution time.

[0012] Furthermore, the in-array optimization is fine-tuned based on criteria including the earliest feasible start time.

[0013] Furthermore, the single QPU multiprogramming compilation step includes: The QPU array is dynamically divided into the Order Preservation Zone (OPZ) and the Order Free Zone (OFZ). The OPZ includes the rows and columns traversed by the currently executing AOD movement. Any new AOD movement must follow the same relative order as the existing movement to avoid physical path intersections. The OFZ is a region unaffected by any AOD movement, where new AOD movements are not subject to directional constraints. When compiling a new quantum circuit, the dynamic order-preserving region OPZ and order-free region OFZ generated by the already scheduled circuit are taken as additional physical constraints. The constraint-based scheduling technique is used to transform the compilation problem of the new circuit into a problem that satisfies all the native hardware constraints and the dynamic concurrent constraints generated by the space deformation model. The solver finds a feasible compilation scheme that satisfies all the conditions, thereby enabling compilation.

[0014] Furthermore, the order-preserving zone OPZ includes the rows and columns traversed by the currently executing AOD movement. Any new AOD movement wishing to traverse this zone must follow the same relative order as existing movements to avoid physical path intersections, including: For horizontal AOD movement, if the starting point of the new quantum bit in the quantum circuit is to the left of the existing starting point, then the ending point of the new quantum bit in the quantum circuit must also be to the left of the existing ending point; if the movement path of the new quantum bit in the quantum circuit completely coincides with the existing movement path, then its movement must be consistent and no relative displacement should occur. For vertical AOD movement, if the movement path of the qubit in the new quantum circuit completely coincides with the existing movement path, then its movement must be consistent and no relative displacement should occur.

[0015] Furthermore, the solver is a Satisfiability Modulus (SMT) solver.

[0016] The beneficial effects of this invention are: In an exemplary embodiment of the present invention, a complete neutral atom multiprogramming and optimization scheme is constructed by combining macroscopic parallel task allocation with microscopic dynamic space modeling and constraint solving. This enables the efficient concurrent execution of multiple quantum programs on shared hardware resources, thereby effectively transforming global and difficult-to-manage hardware constraints into structured and solvable spatiotemporal optimization problems and releasing the parallel computing potential of the neutral atom architecture. Attached Figure Description

[0017] Figure 1 A flowchart of a method for parallel compilation optimization of a neutral atom quantum computing multiprogramming program, provided as an exemplary embodiment of the present invention; Figure 2 A schematic diagram comparing the efficiency and quantum resource utilization of a conventional serial execution method with an exemplary embodiment of the present invention; Figure 3 A single quantum bit provided for an exemplary embodiment of the present invention ( A diagram illustrating the global constraints generated by the movement of (). Figure 4 A schematic diagram of AOD movement steps with a period of 0 provided as an exemplary embodiment of the present invention; Figure 5 A schematic diagram of a non-movement operation step with a cycle of 0 provided as an exemplary embodiment of the present invention; Figure 6 A schematic diagram of AOD movement steps in period 1 provided as an exemplary embodiment of the present invention; Figure 7 This is a schematic diagram of a non-moving operation step in cycle 1 provided as an exemplary embodiment of the present invention. Detailed Implementation

[0018] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0019] See Figure 1 , Figure 1 A flowchart illustrating an exemplary embodiment of the present invention provides a method for parallel compilation optimization of a neutral atom quantum computing multiprogramming program, comprising: The parallel compilation steps include using a circuit length-first greedy scheduling algorithm to intelligently allocate multiple quantum circuit tasks to multiple quantum processing units (QPUs). The compilation steps for a single QPU multiprogramming include using a spatial deformation model to handle concurrent execution within a single QPU.

[0020] Specifically, in this exemplary embodiment, a complete neutral atom multiprogramming compilation and optimization scheme is constructed by combining macroscopic parallel task allocation with microscopic dynamic space modeling and constraint solving. This enables the efficient concurrent execution of multiple quantum programs on shared hardware resources, thereby effectively transforming global and difficult-to-manage hardware constraints into structured and solvable spatiotemporal optimization problems and unleashing the parallel computing potential of the neutral atom architecture.

[0021] The following content will elaborate on the specific implementation method, which involves a set of M quantum circuits to be processed. Efficiently allocate and execute computing resources in a pool consisting of N neutral atom quantum processing units (QPUs). The implementation process seamlessly integrates macro-level multi-QPU task scheduling with micro-level single-QPU concurrent execution control: More preferably, in an exemplary embodiment, the parallel compilation step includes: It receives a set of quantum circuits to be compiled and one or more available neutral atom QPU arrays as input, and sorts them in ascending order according to execution length; Initialize the allocation by assigning the shortest batch of sorted quantum circuits to the available QPUs one by one; Incremental allocation involves finding the insertion point for each remaining quantum circuit across all QPUs that satisfies the hardware resource constraints of the QPUs and has the shortest overall execution time.

[0022] Specifically, in this exemplary embodiment, the framework of the parallel compilation step is responsible for task allocation at the macroscopic level. In this step, a set of quantum circuits to be compiled and one or more available neutral atom QPU arrays are received as input. A greedy scheduling algorithm based on circuit length priority is employed. Its basic principle is to prioritize scheduling circuits with shorter execution times to minimize their spacetime and temporal occupancy and interference to subsequent longer circuits. The algorithm optimizes task allocation through a three-stage process: 1) Initial allocation, initially allocating the shortest batch of circuits to available QPUs; 2) Incremental allocation, sequentially finding the insertion point for each remaining circuit across all QPUs that allows for the earliest start and has the shortest overall execution time.

[0023] More specifically, in the parallel compilation step, firstly, all input quantum circuits are sorted in ascending order according to their execution length (e.g., the number of operation layers) to prioritize circuits with short execution times and high scheduling flexibility. Then, the N shortest circuits are initially assigned to N QPUs, with their initial times set to zero, thus establishing a balanced initial computational load. For the remaining circuits, the framework iteratively processes them in ascending order of length. For each circuit to be assigned, the system simulates and searches for an optimal insertion time across all QPUs. This time must satisfy the hardware resource constraints of the QPU (e.g., the maximum number of parallel qubits) and minimize the total execution time of the QPU after placing the new circuit.

[0024] In summary, the parallel compilation framework designed in this exemplary embodiment can effectively distribute the computational load evenly across multiple QPUs. As the number of available QPUs increases, the system can smoothly scale its parallel processing capabilities, demonstrating good scalability. This is crucial for building large-scale, multi-core quantum computing systems and is an essential capability for achieving practical quantum computing.

[0025] More preferably, in an exemplary embodiment, the parallel compilation step further includes, following the incremental allocation: In-array optimization involves fine-tuning the execution order of circuits within each QPU after all quantum circuits have been allocated, in order to further reduce execution time.

[0026] Specifically, in this exemplary embodiment, the parallel compilation step further includes: 3) In-array optimization: after all circuits are allocated, the execution order of the circuits within each QPU is locally fine-tuned to further compress the execution time. That is, after all circuits are allocated, the execution order of the circuits within each QPU is also locally optimized once, and fine-tuned according to criteria such as the earliest feasible start time (preferred exemplary embodiment) to further compress the overall computation time.

[0027] Figure 2 This intuitively demonstrates the difference in resource utilization between the parallel compilation steps of this exemplary embodiment and the traditional serial execution method. Figure 2The above describes the method proposed in this exemplary embodiment. Below is the conventional method. Different colored blocks represent different quantum circuits. It can be seen that in stage 32, all seven colored quantum circuits have completed their execution. However, in the conventional method below, the seven circuits are executed separately, only completing in stage 124. The lighter-colored portion between stages 32 and 124 indicates that, for the method of this exemplary embodiment, after completing a task in 32 stages (while the conventional method requires 124 stages), this exemplary embodiment still has the capacity to complete more circuits within the range of 32 to 124. In summary, by adopting the method of this exemplary embodiment, multiple circuits (represented by blocks of different colors) can be executed concurrently in time, significantly shortening the total number of Rydberg stages required to complete all tasks and greatly improving the utilization efficiency of quantum resources.

[0028] Experimental results show that, compared with existing technologies that combine or sequentially compile multiple circuits, this exemplary embodiment achieves a compilation speed improvement of up to 14.39 times, while reducing the total execution time (measured by the number of Rydberg stages) by an average of 50.47%. This significant improvement in spatiotemporal efficiency means that more computational tasks can be completed in the same amount of time, thereby increasing the overall throughput of the quantum computer.

[0029] More preferably, in an exemplary embodiment, the single QPU multiprogramming compilation step includes: The QPU array is dynamically divided into the Order Preservation Zone (OPZ) and the Order Free Zone (OFZ). The OPZ includes the rows and columns traversed by the currently executing AOD movement. Any new AOD movement must follow the same relative order as the existing movement to avoid physical path intersections. The OFZ is a region unaffected by any AOD movement, where new AOD movements are not subject to directional constraints. When compiling a new quantum circuit, the dynamic order-preserving region OPZ and order-free region OFZ generated by the already scheduled circuit are taken as additional physical constraints. The constraint-based scheduling technique is used to transform the compilation problem of the new circuit into a problem that satisfies all the native hardware constraints and the dynamic concurrent constraints generated by the space deformation model. The solver finds a feasible compilation scheme that satisfies all the conditions, thereby enabling compilation.

[0030] Specifically, in this exemplary embodiment, the framework of the multiprogramming compilation step of a single QPU is responsible for the microscopic execution level. Within this step, a unique multiprogramming compilation method is employed to resolve concurrency conflicts at the microscopic level. The core of this method is the Spatial Deformation Model. Its basic idea is that the movement of an already compiled and executing circuit's AOD (Active Occurrence Distance) imposes directional constraints on the physical space of the QPU within a specific time period (cycle). That is, when one or more circuits are assigned to a single QPU for concurrent execution, the execution process of the compiled circuit is first abstracted and decomposed into a series of continuous "cycles," each containing two steps: "AOD movement" and "non-movement operations" (such as quantum gate execution). Figure 3 This demonstrates the global impact of AOD movement, where the movement of a single qubit imposes global constraints on the movement of its corresponding row and column. This is the core challenge that this exemplary embodiment aims to address.

[0031] The key innovation of this exemplary embodiment lies in treating the AOD movement generated by the executing circuit as a dynamic deformation of the QPU physical space. Within any given cycle, the AOD movement path of an existing circuit dynamically divides the rows and columns it traverses into "Order-Preserving Zones (OPZs)," while other unaffected regions are designated as "Order-Free Zones (OFZs)." When compiling a new quantum circuit, these dynamically generated regions serve as additional spatiotemporal constraints. To ensure the physical feasibility of concurrent execution—that is, to prevent AOD-controlled beam crossings—any newly introduced AOD movement must strictly adhere to the directional constraints defined by the OPZs.

[0032] Therefore, this exemplary embodiment dynamically divides the QPU array into two types of regions: 1) Order-Preserving Zone (OPZ): Consists of the rows and columns traversed by the currently executing AOD movement. Any new AOD movement must follow the same relative order as existing movements (e.g., it cannot cross from left to right if the existing movement is from right to left) to avoid physical path intersections. 2) Order-Free Zone (OFZ): A region unaffected by any AOD movement, where new AOD movements are not subject to directional constraints. These constraints on AOD movements, along with inherent hardware physical limitations (such as gate execution distance, parallelization requirements, etc.), constitute a constraint satisfaction problem, which is solved using solvers such as Satisfiability Modulus Theory (SMT). The solver finds a feasible compilation scheme that satisfies all conditions, thereby intelligently "embedding" the operation of the new circuit into the spatiotemporal "gaps" generated by the operation of the existing circuit.

[0033] In summary, through a single QPU multiprogramming compilation step, the operation of the new circuit is intelligently "embedded" into the spatiotemporal "gaps" left by the existing circuit, thereby achieving efficient and conflict-free concurrent execution.

[0034] More preferably, in an exemplary embodiment, the sequence-preserving area OPZ includes the rows and columns traversed by the currently executing AOD movement. Any new AOD movement, if it is to traverse this area, must follow the same relative order as existing movements to avoid physical path intersections, including: For horizontal AOD movement, if the starting point of the new quantum bit in the quantum circuit is to the left of the existing starting point, then the ending point of the new quantum bit in the quantum circuit must also be to the left of the existing ending point; if the movement path of the new quantum bit in the quantum circuit completely coincides with the existing movement path, then its movement must be consistent and no relative displacement should occur. For vertical AOD movement, if the movement path of the qubit in the new quantum circuit completely coincides with the existing movement path, then its movement must be consistent and no relative displacement should occur.

[0035] Specifically, to systematically implement this constraint, this exemplary embodiment formalizes it as a set of mathematical rules. Suppose a compiled circuit, during an AOD movement of period t, has its starting and ending coordinates as follows: and So, for a new circuit being compiled, the shift of any qubit i during period t (from coordinates...) arrive The following constraints must be met: For horizontal movement, if the starting point of the new circuit is to the left of an existing starting point, then its ending point must also be to the left of an existing ending point. This constraint can be expressed as: If the new circuit's movement path completely overlaps with an existing movement path, then its movement must remain consistent and no relative displacement should occur. Similarly, for vertical movements, the same principle of maintaining relative order must be satisfied, and its constraint can be expressed as: These constraints, along with inherent physical limitations of the hardware (such as gate execution distance and parallelization requirements), constitute a constraint satisfaction problem, which is solved using solvers such as Satisfiability Modulus Theory (SMT, the optimal solution). The solver finds a feasible compilation scheme that satisfies all conditions, thereby intelligently "embedding" the operation of the new circuit into the spatiotemporal "gaps" generated by the operation of the existing circuit.

[0036] The following example of concurrent compilation illustrates this process.

[0037] Reference Figure 4 In the AOD movement step of period 0, an existing circuit qubit and They are moving. Their movement path (through rows) , and column , This creates gray sequence-holding regions and green sequence-free regions on the array. At this point, a new circuit... It is necessary to execute its two-bit gate. The required movement, i.e., let Move towards Because this movement requires columns Time travel has been Move the constrained column This violates the OPZ order preservation principle, therefore the move is blocked (indicated by a red cross). However, within the same cycle, Another part of the operation, namely The movement can be successfully executed because its starting and ending points are both within the OFZ and are not subject to additional constraints.

[0038] Reference Figure 5 After the AOD (Active Operation Delay) movement step is completed, the system enters the non-movement operation step of cycle 0. Based on the results of the previous step, It has been successfully moved to Next to it, a two-bit gate is satisfied. The execution distance condition was met, thus the door was able to execute successfully. The door that was blocked from moving... Because its qubits are not in place, it remains in an inactive state, waiting for an opportunity in a subsequent cycle.

[0039] Reference Figure 6 The system then enters the next execution cycle, i.e., cycle 1. During the AOD movement step of this cycle, the circuit... qubit A new move was performed, which redefined the OPZ and OFZ on the array. The key change is that the columns were previously moved in period 0. The constraints no longer exist. Therefore, moves blocked in cycle 0 no longer conflict with any OPZ constraints and can be executed successfully.

[0040] Reference Figure 7 ,exist After a successful move, the system enters the non-move operation step of cycle 1. At this time, and The circuit is now within the interaction range, satisfying the conditions for executing a two-bit gate. Therefore, the previously delayed gate gc1,1 is successfully executed. Thus, the new circuit... All operations have been with Complete concurrently and without conflict.

[0041] In summary, the neutral atom quantum computing multiprogramming parallel compilation optimization method proposed in the above exemplary embodiments has several significant advantages in improving the efficiency of quantum computing systems and promoting the development of quantum operating systems: (1) This invention significantly improves the utilization of hardware resources by executing multiple quantum programs concurrently. Experimental results show that, compared with the existing technology of merging multiple circuits or compiling them sequentially, this invention can achieve a compilation speed improvement of up to 14.39 times, while reducing the total execution time (measured by the number of Rydberg stages) by an average of 50.47%. This significant improvement in spatiotemporal efficiency means that more computational tasks can be completed in the same amount of time, thereby improving the overall throughput of the quantum computer.

[0042] (2) The parallel compilation framework designed in this invention can effectively distribute the computational load evenly across multiple QPUs. As the number of available QPUs increases, the system can smoothly expand its parallel processing capabilities, demonstrating good scalability. This is crucial for building large-scale, multi-core quantum computing systems and is an essential capability for moving towards practical quantum computing.

[0043] (3) Overcoming the key technical bottleneck of neutral atom multiprogramming: The invention’s original “spatial deformation model” cleverly solves the global constraint problem of AOD movement, which is the core obstacle preventing the neutral atom platform from realizing simple resource partitioning and multiprogramming. By dynamically dividing the hardware space into “order-preserving region” and “order-free region”, the invention transforms the complex global physical problem into a structured local constraint, providing a new and effective paradigm for realizing concurrent control on quantum hardware with global coupling characteristics.

[0044] (4) It lays the foundation for the development of quantum operating systems: Multiprogramming is the cornerstone of all modern operating systems. This invention is the first to realize efficient multiprogramming functionality on the highly promising hardware platform of neutral atoms, filling a major gap in the existing quantum compilation field. It provides the most critical underlying technical support for developing higher-level quantum operating system functions, such as resource management, task scheduling, and priority control, and is an important step in promoting quantum computing from dedicated devices to general-purpose computing systems.

[0045] (5) The method has versatility and adaptability: Although this invention is designed for neutral atom architecture, its core idea—treating executed tasks as dynamic spatiotemporal constraints on system resources and scheduling new tasks under these constraints—has broader applicability. This dynamic resource management and constraint-based scheduling concept is expected to be referenced and applied to other quantum computing platforms or classical heterogeneous computing systems with complex resource sharing and dynamic characteristics, and has important theoretical and practical value.

[0046] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for parallel compilation and optimization of multiprogramming in neutral atom quantum computing, characterized in that: include: The parallel compilation steps include using a circuit length-first greedy scheduling algorithm to intelligently allocate multiple quantum circuit tasks to multiple quantum processing units (QPUs). The compilation steps for a single QPU multiprogramming include using a spatial deformation model to handle concurrent execution within a single QPU.

2. The method for parallel compilation and optimization of multiprogramming for neutral atom quantum computing according to claim 1, characterized in that: The parallel compilation steps include: It receives a set of quantum circuits to be compiled and one or more available neutral atom QPU arrays as input, and sorts them in ascending order according to execution length; Initialize the allocation by assigning the shortest batch of sorted quantum circuits to the available QPUs one by one; Incremental allocation involves finding the insertion point for each remaining quantum circuit across all QPUs that satisfies the hardware resource constraints of the QPUs and has the shortest overall execution time.

3. The method for parallel compilation and optimization of multiprogramming for neutral atom quantum computing according to claim 2, characterized in that: The parallel compilation step also includes a step following incremental allocation: In-array optimization involves fine-tuning the execution order of circuits within each QPU after all quantum circuits have been allocated, in order to further reduce execution time.

4. The method for parallel compilation and optimization of multiprogramming for neutral atom quantum computing according to claim 3, characterized in that: The in-array optimization is fine-tuned based on criteria including the earliest feasible start time.

5. The method for parallel compilation and optimization of multiprogramming for neutral atom quantum computing according to claim 1, characterized in that: The single QPU multiprogramming compilation steps include: The QPU array is dynamically divided into the Order Preservation Zone (OPZ) and the Order Free Zone (OFZ). The OPZ includes the rows and columns traversed by the currently executing AOD movement. Any new AOD movement must follow the same relative order as the existing movement to avoid physical path intersections. The OFZ is a region unaffected by any AOD movement, where new AOD movements are not subject to directional constraints. When compiling a new quantum circuit, the dynamic order-preserving region OPZ and order-free region OFZ generated by the already scheduled circuit are taken as additional physical constraints. The constraint-based scheduling technique is used to transform the compilation problem of the new circuit into a problem that satisfies all the native hardware constraints and the dynamic concurrent constraints generated by the space deformation model. The solver finds a feasible compilation scheme that satisfies all the conditions, thereby enabling compilation.

6. The method for parallel compilation and optimization of multiprogramming for neutral atom quantum computing according to claim 5, characterized in that: The Order Preservation Zone (OPZ) includes the rows and columns traversed by the currently executing AOD movement. Any new AOD movement wishing to pass through this zone must follow the same relative order as existing movements to avoid physical path intersections, including: For horizontal AOD movement, if the starting point of the new quantum bit in the quantum circuit is to the left of the existing starting point, then the ending point of the new quantum bit in the quantum circuit must also be to the left of the existing ending point; if the movement path of the new quantum bit in the quantum circuit completely coincides with the existing movement path, then its movement must be consistent and no relative displacement should occur. For vertical AOD movement, if the starting point of the new quantum bit in the quantum circuit is above the existing starting point, then the ending point of the new quantum bit in the quantum circuit must also be above the existing ending point; if the movement path of the new quantum bit in the quantum circuit completely coincides with the existing movement path, then its movement must be consistent and no relative displacement should occur.

7. The method for parallel compilation and optimization of multiprogramming for neutral atom quantum computing according to claim 5, characterized in that: The solver is a Satisfiability Modulus (SMT) solver.

Images