Resource scheduling method, electronic device, storage medium and computer program product

Abstract

The invention provides a resource scheduling method, electronic equipment, a storage medium and a computer program product. The method comprises: obtaining global information of a target computing power network, the target computing power network comprising a plurality of nodes, the plurality of nodes comprising a plurality of user nodes, a plurality of network nodes and a plurality of computing nodes, the global information at least comprising state information of the plurality of nodes and state information of a plurality of transmission paths, each transmission path comprises a user node and a computing node; constructing a resource scheduling model of the target computing power network based on the global information of the target computing power network; converting the resource scheduling model into a quantum computing model; constructing a quantum circuit based on the quantum calculation model; and determining a resource scheduling result of the target computing power network according to the measurement result of the quantum circuit. Through the scheme provided by the invention, the global resource scheduling strategy can be quickly obtained.

Description

Technical Field

[0001] This application relates to the field of quantum computing technology, and in particular to a resource scheduling method, electronic device, storage medium, and computer program product. Background Technology

[0002] A computing power network is a network that integrates and coordinates computing, storage, and network resources. It enables the optimization and efficient utilization of computing resources, supporting complex computational tasks such as artificial intelligence training. The computing power network dynamically allocates computing resources within the resource environment. To reduce waste of computing resources and improve the overall efficiency of the computing power network, resource scheduling is also required.

[0003] Resource scheduling is a crucial step in achieving efficient operation of computing networks, primarily involving the coordinated allocation of multi-dimensional resources. With the development of big data and cloud computing technologies, the user base and business types of computing networks continue to grow, leading to an increasing demand for resource scheduling. However, current resource scheduling methods struggle to quickly derive a global resource scheduling strategy. Summary of the Invention

[0004] To address the related technical issues, embodiments of this application provide a resource scheduling method, an electronic device, a storage medium, and a computer program product.

[0005] The technical solution of this application embodiment is implemented as follows: This application provides a resource scheduling method, the method comprising: Obtain global information of the target computing power network, wherein the target computing power network includes multiple nodes, including multiple user nodes, multiple network nodes and multiple computing nodes, and the global information includes at least the status information of the multiple nodes and the status information of multiple transmission paths, and each transmission path includes one user node and one computing node; Based on the global information of the target computing power network, a resource scheduling model for the target computing power network is constructed; The resource scheduling model is converted into a quantum computing model; Based on the aforementioned quantum computing model, a quantum circuit is constructed; Based on the measurement results of the quantum circuit, the resource scheduling results of the target computing network are determined.

[0006] In the above scheme, constructing a resource scheduling model for the target computing power network based on the global information of the target computing power network includes: Using the task volume corresponding to the multiple transmission paths as variables, an initial cost model for the target computing power network is constructed, wherein the initial cost model is used to minimize the total energy consumption of the target computing power network, and the total energy consumption is determined based on the task volume corresponding to the multiple transmission paths; Based on the status information of the multiple nodes and the status information of the multiple transmission paths, constraints are set for the initial cost model; Based on the initial cost model and the constraints, a Lagrangian function is constructed to obtain the resource scheduling model.

[0007] In the above scheme, converting the resource scheduling model into a quantum computing model includes: The variables in the resource scheduling model are discretized to obtain the discretized resource scheduling model. The binary variables in the discretized resource scheduling model are mapped to the first quantum spin operator to obtain the quantum computing model; wherein the number of binary variables in the discretized resource scheduling model is equal to the number of qubits corresponding to the quantum computing model.

[0008] In the above scheme, the quantum circuit includes a quantum evolution module; the construction of the quantum circuit based on the quantum computing model includes: A quantum evolution operator is determined based on a first Hamiltonian, wherein the first Hamiltonian is the Hamiltonian of the quantum computing model, and the quantum evolution operator is used to represent the transformation operation performed on the quantum state of a qubit. Based on the quantum evolution operator, a quantum evolution module of the quantum circuit is constructed, wherein the quantum evolution module is used to evolve the quantum state of the qubit based on the quantum evolution operator.

[0009] The method in the above scheme further includes: Determine the second Hamiltonian, which is a mixed Hamiltonian; The step of determining the quantum evolution operator based on the first Hamiltonian includes: The quantum evolution operator is determined based on the first Hamiltonian and the second Hamiltonian.

[0010] In the above scheme, the quantum circuit further includes a quantum initial state preparation module and a final state measurement module; the construction of the quantum circuit based on the quantum computing model includes: Construct the quantum initial state preparation module, wherein the quantum initial state preparation module is used to initialize the number of qubits corresponding to the quantum computing model; Construct the final state measurement module, wherein the final state measurement module is used to measure the quantum state of the evolved qubit.

[0011] The method in the above scheme further includes: The evolution parameters of the quantum circuit are optimized once or multiple times to obtain the optimized quantum circuit; Determining the resource scheduling result of the target computing power network based on the measurement results of the quantum circuit includes: Based on the measurement results of the optimized quantum circuit, the resource scheduling results of the target computing network are determined.

[0012] In the above scheme, optimizing the evolution parameters of the quantum circuit once or multiple times includes: In each optimization, the expected value of the first Hamiltonian is determined based on the measurement results of the quantum circuit; wherein the first Hamiltonian is the Hamiltonian of the quantum computing model. The evolution parameters of the quantum circuit are optimized based on the expected value of the first Hamiltonian.

[0013] This application also provides an electronic device, including: a processor and a memory for storing a computer program that can run on the processor; wherein, when the processor runs the computer program, it executes the method provided in this application.

[0014] This application also provides a storage medium storing a computer program thereon, which, when executed by a processor, implements the method provided in this application.

[0015] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the method provided in this application.

[0016] This application provides a resource scheduling method, electronic device, storage medium, and computer program product. It constructs a resource scheduling model for a target computing network based on global information, taking into account the states of user nodes, network nodes, computing nodes, and transmission paths from a global perspective. By converting the resource scheduling model into a quantum computing model and constructing quantum circuits based on this model, the resource scheduling problem of the target computing network can be transformed into a quantum computing task. The parallel processing capabilities of quantum computing are then used to process the resource scheduling problem, and the resource scheduling result of the target computing network is obtained based on the measurement results of the quantum circuits. This approach improves the computational efficiency of resource scheduling results and allows for the rapid acquisition of a global resource scheduling strategy. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a resource scheduling method provided in an embodiment of this application; Figure 2 This is a schematic diagram of the structure of a target computing power network provided in an embodiment of this application; Figure 3 This is a schematic diagram illustrating the evolution process of a quantum circuit provided in an embodiment of this application; Figure 4 This is a schematic diagram of the circuit structure of a quantum circuit provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of a quantum circuit device provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0018] The present application will now be described in detail with reference to the accompanying drawings and embodiments.

[0019] A computing power network is a distributed computing network comprising multiple nodes. These nodes include user nodes, network nodes, and compute nodes. The computing power network allows user nodes to access compute nodes through network nodes, and compute nodes to provide computing power services to user nodes. The goal of a computing power network is to achieve efficient resource allocation and minimize energy consumption while meeting user needs.

[0020] Nodes in a computing network can be physical devices. For example, a node can be a physical device such as a server, switch, or smart terminal. In some implementations, nodes in a computing network can also be virtualized entities or programmatic nodes deployed on physical devices. For example, a node can be a virtual machine, a container instance, or a software-defined service unit.

[0021] User nodes are the requesters of computing power services in a computing power network, used to initiate computing power service requests. User nodes can be user devices or servers. User devices can be terminal devices such as mobile phones, personal computers, and tablets. In some implementations, user nodes can also represent virtualized entities or programmatic nodes that initiate computing power requests.

[0022] A network node is a node in a computing power network that enables communication between user nodes and computing power nodes. Network nodes can be routers, switches, base stations, and controllers in software-defined networks, etc.

[0023] Compute nodes are providers of computing power services in a computing network, used to execute computing tasks corresponding to computing power service requests initiated by user nodes. Compute nodes can be compute servers, processor clusters, edge computing nodes, or virtual machine instances, etc.

[0024] To achieve efficient resource allocation and minimize energy consumption in computing networks, resource scheduling is necessary. That is, the computing network configures transmission paths for computing tasks based on the needs of user nodes, the status of network nodes, and the capabilities of computing nodes, allocating the computing tasks of user nodes to nodes. For example, when the computing network receives a computing service request from a user node, it selects an appropriate transmission path for the user node's computing task based on the user node's needs, the status of network nodes, and the capabilities of computing nodes, and then transmits the user node's computing task through that path to a computing node capable of processing the task.

[0025] Currently, resource scheduling strategies for computing networks mainly include those based on preset rules, task priorities, and load balancing. While these strategies achieve resource scheduling, they also have some limitations. For example, resource scheduling mechanisms based on preset rules are simple to implement but lack adaptability to the computing network environment and diverse task requirements, resulting in poor flexibility; resource scheduling strategies based on task priorities can ensure the execution of critical tasks but may lead to excessively long waiting times for low-priority tasks; and while resource scheduling strategies based on load balancing improve local resource utilization, they may increase network communication costs by triggering cross-domain or cross-node task migrations.

[0026] In summary, the shortcomings of the current resource scheduling strategy are mainly reflected in the following aspects: First, most resource scheduling strategies are usually limited to the local resource scheduling of user nodes, network nodes, and computing nodes, lacking unified management and scheduling of computing resources (i.e., computing nodes) and network resources (i.e., network nodes) of the computing power network, thus limiting the optimal configuration of the overall performance of the computing power network.

[0027] Furthermore, when faced with complex resource scheduling tasks involving high concurrency and multiple objectives in computing networks, most resource scheduling strategies face bottlenecks in terms of solution efficiency and accuracy, making it difficult to support real-time and accurate global optimal resource scheduling decisions, thus limiting the service capabilities and response speed of computing networks.

[0028] Based on this, this application provides a resource scheduling scheme. It constructs a resource scheduling model for the target computing power network using global information, thus considering the states of user nodes, network nodes, and computing nodes from a global perspective, and building a mathematical model for resource scheduling (i.e., the resource scheduling model). By converting the resource scheduling model into a quantum computing model and constructing quantum circuits based on this model, the resource scheduling problem of the target computing power network can be transformed into a quantum computing task. The parallel processing capability of quantum computing is used to handle the resource scheduling problem, and the resource scheduling result of the target computing power network is obtained based on the measurement results of the quantum circuits. This not only improves the computational efficiency of the resource scheduling result and enables rapid matching of the global resource scheduling strategy for the target computing power network, but also achieves efficient utilization and refined management of resources in the target computing power network, reduces the operating cost of the target computing power network, and improves its service capabilities and response speed.

[0029] First, this application provides a resource scheduling method. This method can be applied to an electronic device. The electronic device can be a resource scheduling node in a target computing power network. Alternatively, the electronic device can drive or be integrated into a resource scheduling platform used to implement resource scheduling of the target computing power network. The resource scheduling method provided in this application will be described below using an electronic device as the execution subject. Figure 1 As shown, the resource scheduling method provided in this application includes the following steps: Step 101: The electronic device acquires global information of the target computing network, which includes at least the status information of multiple nodes and the status information of multiple transmission paths; Step 102: The electronic device constructs a resource scheduling model for the target computing power network based on the global information of the target computing power network; Step 103: The electronic device converts the resource scheduling model into a quantum computing model; Step 104: Electronic devices construct quantum circuits based on quantum computing models; Step 105: The electronic device acquires the measurement results of the quantum circuit operation; Step 106: The electronic device determines the resource scheduling result of the target computing network based on the measurement results of the quantum circuit operation.

[0030] In practical applications, a target computing power network is a computing power network awaiting resource scheduling. A target computing power network comprises multiple nodes. For example, it may include multiple user nodes, multiple network nodes, and multiple compute nodes. Multiple transmission paths can be formed between the multiple user nodes and multiple compute nodes in the target computing power network. Each transmission path includes one user node and one compute node. That is, each transmission path represents the path for a computing task to be transmitted from one user node to one compute node.

[0031] A transmission path may consist of only one user node and one compute node. In this case, the user node in the transmission path can communicate with the compute node to transmit computational tasks. In some implementations, in addition to the user node and compute node, a transmission path may also include one or more network nodes. In this case, the user node in the transmission path and the compute node communicate through one or more network nodes to transmit computational tasks.

[0032] For example, such as Figure 2 As shown, the target computing network includes 2 user nodes, 3 network nodes, and 2 compute nodes. User node 1 can reach compute node 1 through network node 1 and network node 2. That is, user node 1, network node 1, network node 2, and compute node 1 correspond to one transmission path. User node 1 can also directly reach compute node 1, that is, user node 1 and compute node 1 correspond to another transmission path.

[0033] In step 101, the electronic device acquires global information about the target computing power network. This global information includes the status information of multiple nodes and the status information of multiple transmission paths. For example, the electronic device communicates with each node to acquire the status information of each node in real time or periodically. Alternatively, the electronic device determines the status information of each transmission path in the target computing power network based on the connection relationships and status information of multiple nodes in the target computing power network.

[0034] The state information of each node is used to represent the node's status. For example, state information is used to represent the node's computing power requirements, data transmission capabilities, computing power, and other node statuses. The state information of multiple nodes includes the state information of user nodes, network nodes, and compute nodes.

[0035] The state information of a user node is used to represent at least its computing power requirements and communication capabilities. For example, the state information of a user node includes task requirements, such as the total scale of the computing task and the total amount of data transmission. The state information of a user node may also include data transmission bandwidth information.

[0036] The state information of a network node is used to represent its data transmission capabilities. For example, the state information of a network node includes available bandwidth, maximum bandwidth, transmission delay, connectivity, and data transmission strategy.

[0037] The status information of a computing node represents its computing power and data transmission capabilities. For example, the status information includes resource utilization, available computing power, and maximum computing power. The status information may also include data transmission bandwidth information.

[0038] A transmission path can include multiple communication links. A communication link represents a connection between two nodes. In a transmission path that includes only one user node and one compute node, the transmission path includes communication links between that user node and that compute node. For example... Figure 2 Taking communication link e4 as an example, communication link e4 represents a transmission path. When a transmission path includes one or more network nodes in addition to user nodes and compute nodes, the transmission path comprises multiple communication links. For example... Figure 2 Taking the transmission path consisting of user node 2, network node 3 and computing node 2 as an example, the transmission path includes communication link e7 and communication link e8.

[0039] The state information of a transmission path is used to represent the state of the transmission path. For example, it indicates the communication capability, connection status, and status of each communication link. The state information includes at least path loss and bandwidth information of the communication links.

[0040] Path loss in a transmission path represents the data transmission cost or energy consumption of that path. For example, path loss includes path distance. The greater the path distance, the higher the data transmission cost or energy consumption. In some alternative implementations, the path loss of a transmission path is related to information such as the fiber optic or link length, the number of network nodes, and path delay. Electronic devices can comprehensively determine the path loss of a transmission path based on these factors. For example, the path loss might be equal to the weighted value or weighted average of the fiber optic or link length, the number of network nodes, and the path delay.

[0041] Bandwidth information in a communication link indicates its bandwidth capacity. For example, bandwidth information includes maximum bandwidth and available bandwidth. Electronic devices can determine the bandwidth of a communication link based on the status information of each node within the link.

[0042] In step 102, the electronic device constructs a mathematical model for resource scheduling of the target computing power network based on the state information of each node and each transmission path in the target computing power network, i.e., constructs a resource scheduling model. This resource scheduling model is used to minimize the total energy consumption of the target computing power network while satisfying the constraints corresponding to the state information of each node.

[0043] To facilitate the construction of a resource scheduling model, in some optional implementations, electronic devices can divide multiple nodes in the target computing power network into multiple node regions or sets before constructing the resource scheduling model. For example, Figure 2As shown, the electronic device divides multiple nodes in the target computing power network into a user node area (i.e., user node area I), a network node area (i.e., network node area U), and a computing power node area (i.e., computing power node area J). The user node area is the set of user nodes in the target computing power network. The network node area is the set of network nodes in the target computing power network. The computing power node area is the set of computing nodes in the target computing power network.

[0044] In this embodiment of the application, by dividing multiple nodes in the target computing power network, such as dividing them into user node area, network node area and computing power node area, the multiple nodes in the target computing power network can be partitioned and managed, thereby simplifying the global resource scheduling of multiple nodes and making it easier to construct the resource scheduling model of the target computing power network.

[0045] In some alternative implementations, the electronic device aims to minimize the total energy consumption of the target computing network, using the task volume corresponding to multiple transmission paths as variables, and performs mathematical modeling, i.e., establishing a mathematical model for resource scheduling of the target computing network. This data model can be called an initial cost model or an energy cost function.

[0046] The total energy consumption of the target computing network includes the energy consumption of each transmission path; that is, the total energy consumption of the target computing network equals the sum of the energy consumption of each transmission path. The energy consumption of each transmission path represents the energy or cost consumed by that transmission path in processing the corresponding amount of tasks. Electronic devices can determine the energy consumption of a transmission path based on the amount of tasks assigned to each transmission path (i.e., the amount of tasks allocated) and path losses.

[0047] For example, the initial cost model for target computing power network resource scheduling is shown in formula (1): Formula (1) Where C represents the initial cost model; This represents the workload of the m-th transmission path. Let m represent the path loss of the m-th transmission path, which runs from user node i to computing node j; m represents the m-th transmission path, and m is a positive integer less than or equal to the total number of transmission paths; i represents the i-th user node in the user node region, and i is a positive integer less than or equal to the total number of user nodes; j represents the j-th computing node in the computing node region, and j is a positive integer less than or equal to the total number of computing nodes.

[0048] In this embodiment of the application, when establishing the resource scheduling model of the target computing power network, the state information of multiple nodes in the target computing power network and the state information of the transmission paths are also considered, such as the computing power requirements of user nodes, the capacity limit of computing nodes, and the capacity limit of transmission paths. Therefore, the electronic device also sets constraints for the initial cost model based on the state information of multiple nodes and the state information of multiple transmission paths.

[0049] In some optional implementations, the initial cost model sets constraints including one or more of the following (one or more can be understood as at least one): network constraints, computing power constraints, and user demand constraints. Specifically, network constraints are constraints set based on the upper limit of the transmission path's capacity. Computing power constraints are constraints set based on the upper limit of the computing node's capacity. User demand constraints are constraints set based on the computing power requirements of user nodes (i.e., the total workload of user nodes).

[0050] For example, electronic devices can set network constraints based on the status information of the transmission path, such as the bandwidth limit of the communication link (e.g., maximum bandwidth or available bandwidth). The network constraints are shown in formula (2): Formula (2) in, This represents the workload of the m-th transmission path; Let represent the path loss of the m-th transmission path, where m is a positive integer; e represents the e-th communication link in set G, and the e-th communication link belongs to . e is a positive integer less than or equal to the total number of communication links; G represents the set of all communication links; represents the bandwidth limit of the e-th communication link; i represents the i-th user node in the user node area, where i is a positive integer less than or equal to the total number of user nodes; j represents the j-th computing node in the computing power node area, where j is a positive integer less than or equal to the total number of computing nodes.

[0051] Electronic devices can set computing power constraints based on the status information of multiple computing nodes, such as the computing power limit of each computing node (e.g., maximum computing power or available computing power). The computing power constraints are shown in formula (3): Formula (3) in, This represents the workload of the m-th transmission path; represents the upper limit of computing power of the j-th computing node; m represents the m-th transmission path in the target computing power network, where m is a positive integer less than or equal to the total number of transmission paths; i represents the i-th user node in the user node area, where i is a positive integer less than or equal to the total number of user nodes; j represents the j-th computing node in the computing power node area, where j is a positive integer less than or equal to the total number of computing nodes.

[0052] Electronic devices can set user requirement constraints based on the status information of multiple user nodes, such as the task requirements of each user node (e.g., calculating the total task size or the total data transmission volume). The user requirement constraints are shown in formula (4): Formula (4) in, This represents the workload of the m-th transmission path; denoted by ; m represents the m-th transmission path in the target computing power network, where m is a positive integer less than or equal to the total number of transmission paths; i represents the i-th user node in the user node area, where i is a positive integer less than or equal to the total number of user nodes; j represents the j-th computing node in the computing power node area, where j is a positive integer less than or equal to the total number of computing nodes.

[0053] In this embodiment, after constructing the initial cost model and constraints of the target computing power network, the electronic device generates a resource scheduling model for the target computing power network based on the constraints and the initial cost model. Thus, the electronic device can construct the resource scheduling model of the target computing power network from a global perspective using global information about the target computing power network.

[0054] To analyze the resource scheduling problem of a target computing network using quantum computing, in some optional implementations, electronic devices transform the constrained initial cost model into an unconstrained form, i.e., into an unconstrained resource scheduling model. For example, electronic devices can use the Lagrange multiplier method to transform the constrained initial cost model into an unconstrained resource scheduling model, thereby converting the resource scheduling problem of the target computing network into a quadratic unconstrained binary optimization (QUBO) problem for solution, improving the solution efficiency.

[0055] For example, consider constraints including network constraints, computing power constraints, and user requirement constraints. First, the electronic device transforms the inequality constraints into equality constraints. Since the network constraint is an inequality constraint, the electronic device introduces a first scaling variable to transform it from an inequality to an equality. Similarly, the computing power constraint is also an inequality constraint, and the electronic device introduces a second scaling variable to transform it from an inequality to an equality.

[0056] The equation form of the network constraints can be found in formula (5): Formula (5) in, This represents the workload of the m-th transmission path; Let represent the path loss of the m-th transmission path, where m is a positive integer; e represents the e-th communication link in set G, and the e-th communication link belongs to . e is a positive integer less than or equal to the total number of communication links; G represents the set of all communication links; This represents the maximum bandwidth of the e-th communication link; represents the first scaling variable corresponding to the e-th communication link; i represents the i-th user node in the user node area, where i is a positive integer less than or equal to the total number of user nodes; j represents the j-th computing node in the computing power node area, where j is a positive integer less than or equal to the total number of computing nodes.

[0057] The equation form of the computing power constraint can be found in formula (6): Formula (6) in, This represents the workload of the m-th transmission path; This represents the maximum computing power of the j-th computing node; represents the second scaling variable corresponding to the j-th computing node; m represents the m-th transmission path in the target computing power network, where m is a positive integer less than or equal to the total number of transmission paths; i represents the i-th user node in the user node area, where i is a positive integer less than or equal to the total number of user nodes; j represents the j-th computing node in the computing power node area, where j is a positive integer less than or equal to the total number of computing nodes.

[0058] In the example, after transforming the constraints from inequalities to equations, the electronic device uses the Lagrange multiplier method to construct a Lagrange function from the constraints and the initial cost model, thus obtaining the resource scheduling model. The resource scheduling model can be found in formula (7): Formula (7) in, This represents the workload of the transmission path (S). (i.e., the first scaling variable) The Lagrangian function (i.e., the resource scheduling model) is defined with V (i.e., the upper limit of computing power of computing nodes), I (i.e., the task requirements of user nodes), and U (i.e., the upper limit of bandwidth of communication links) as input conditions. This is the initial cost model; This represents the pre-defined Lagrange multipliers; This represents the constraint terms corresponding to the computing power constraint conditions. ; This represents the constraint terms corresponding to the user requirement constraints. ; This represents the constraint terms corresponding to the network constraints. .

[0059] In this embodiment, by constructing a Lagrangian function, constrained conditions can be transformed into unconstrained forms, thereby simplifying complex constrained problems into easily solvable unconstrained forms and providing a foundation for subsequent quantum computing. Furthermore, by constructing the Lagrangian function, the initial cost model and constraints are combined, providing a reliable computational model for resource scheduling, thus improving the efficiency and accuracy of resource scheduling.

[0060] In step 103, the electronic device maps the resource scheduling model to a quantum computing model suitable for quantum computing. For example, the quantum computing model could be the quantum Ising model.

[0061] Understandably, the Ising model is an energy function used to describe the interaction between binary variables. The measurement results of a qubit also exhibit binary states (e.g., |0|). or |1 Therefore, the Ising model can serve as an intermediary for the conversion between classical computing models (i.e., resource scheduling models) and quantum computing models. The following section uses the quantum Ising model as an example to illustrate the process of converting resource scheduling into the quantum Ising model.

[0062] In some alternative implementations, the electronic device first processes the variables in the resource scheduling model (such as S, ... and The resource scheduling model is discretized, transforming its variables into multiple binary variables to obtain a discretized resource scheduling model. Then, the electronic device maps these binary variables to a first quantum spin operator, resulting in a quantum Ising model. Through discretization and the mapping using the first quantum spin operator, the transformation from the resource scheduling model to the quantum Ising model is achieved. This allows the resource scheduling problem of the target computing network to be solved using a quantum computing architecture, providing a novel solution for resource scheduling in the target computing network.

[0063] In this embodiment, the binary variable is a binary variable. The binary variable takes the value of a first value or a second value. For example, the value of a binary variable can be 0 or 1. Alternatively, the value of a binary variable can be 1 or -1. The number of binary variables (i.e., the number of binary variables) can be determined based on the total number of user nodes, the total number of computing nodes, and the total number of transmission paths. For example, taking N to represent the number of binary variables, 2... N The number of permutations and combinations of user nodes, compute nodes, and transmission paths is greater than or equal to the number of permutations and combinations of user nodes, compute nodes, and transmission paths, so that multiple binary variables can represent any combination strategy of user nodes, compute nodes, and transmission paths.

[0064] After discretization, the number of binary variables in the resource scheduling model is equal to the number of qubits in the quantum computing model. By mapping the binary variables to the first quantum spin operator acting on the qubits, the parallelism and superposition characteristics of qubits can be utilized to improve the efficiency and accuracy of resource scheduling strategy acquisition for the target computing network. This allows for the rapid determination of the globally optimal or near-optimal resource scheduling scheme for the target computing network, meeting the high-efficiency scheduling requirements of large-scale target computing networks.

[0065] In this embodiment, the first quantum spin operator is used to characterize the state of the qubit in the spin direction. The first quantum spin operator can be a Pauli operator, such as the Pauli Z operator, Pauli X operator, Pauli Y operator, etc.

[0066] In some alternative implementations, when mapping binary variables in the discretized resource scheduling model to first quantum spin operators, the electronic device can first map multiple binary variables in the discretized resource scheduling model to multiple spin variables, obtaining a classical Ising model. For example, the electronic device maps multiple binary variables with values ​​{0,1} to multiple spin variables with values ​​{+1,-1}, obtaining a classical Ising model. Then, the electronic device transforms the classical Ising model into a quantum Ising model. That is, the electronic device replaces multiple spin variables in the classical Ising model with multiple first quantum spin operators. The eigenvalues ​​of the first quantum spin operators are consistent with the values ​​of the spin variables, thus achieving the transformation from the classical Ising model to the quantum Ising model by replacing the spin variables with first quantum spin operators.

[0067] For example, taking the first quantum spin operator as the Pauli Z operator, the quantum Ising model can be expressed by formula (8): Formula (8) in, denoted as the Hamiltonian describing the quantum Ising model (i.e., the first Hamiltonian), which can also be called the problem Hamiltonian; N represents the number of qubits, which is the same as the number of binary variables and is a positive integer. Indicates the first One quantum bit, Indicates the first One quantum bit, and All are positive integers less than or equal to N; It is the mapping of the first quantum spin operator The coefficient of a term is a definite real number; It is the mapping of the first quantum spin operator The coefficient of a term is a definite real number; Indicates the action on the first Pauli Z operator for 1 qubit; Indicates the action on the first Pauli Z operator with 100 qubits.

[0068] Understandably, the Hamiltonian is the expression for the quantum Ising model, used to represent the total energy of the quantum computing framework. The goal of the quantum Ising model is to minimize the Hamiltonian. The resource scheduling goal of the target computing network is to minimize the total energy consumption of the target computing network. Therefore, the objective of the resource scheduling problem can be correlated with the objective of the quantum Ising model, and the optimal or near-optimal solution to the resource scheduling problem of the target computing network can be obtained through the Hamiltonian.

[0069] In step 104, the electronic device constructs a quantum circuit based on the obtained quantum computing model (such as the quantum Ising model). For example, the electronic device can use the Quantum Approximate Optimization Algorithm (QAOA) to construct the quantum circuit corresponding to the quantum computing model.

[0070] In some optional implementations, the quantum circuit includes an initial state preparation module, a quantum evolution module, and a final state measurement module. The initial state preparation module initializes the qubit, setting it to a superposition state. The quantum evolution module evolves the quantum state of the qubit through a sequence of quantum operations. The final state measurement module measures the evolved quantum state of the qubit, obtaining the measurement result. Through these modules, a quantum circuit corresponding to the quantum computing model can be obtained. The coordinated operation of multiple modules within the quantum circuit can drive the quantum state of the qubit to approach a low-energy state of the Hamiltonian of the quantum computing model, providing a decision-making scheme for resource scheduling in the target computing network.

[0071] In this embodiment, the initial state preparation module is used to initialize the qubits. The ground state of each qubit is... The initial fabrication module includes Hadamard quantum gate (H-gate) operations. The initial fabrication module performs an H-gate operation on each qubit, changing the quantum state of each qubit from its ground state. Transform into superposition state For N qubits, the initial fabrication module will convert N qubits from N... State transitions to superposition state Superposition states can serve as the initial states of qubits in quantum circuits. An initial state preparation module can initialize qubits to a superposition state, thus providing a state basis for the subsequent quantum state evolution of the qubits.

[0072] In this embodiment, the quantum evolution module is used to evolve the quantum state of a qubit to obtain the evolved final state. The quantum evolution module includes a sequence of quantum gates determined based on quantum evolution operators. A quantum evolution operator is a unitary operator describing the evolution of the quantum state of a qubit with parameters. A quantum evolution algorithm is used to represent the transformation operations performed on the quantum state of the qubit. This is represented by quantum evolution operators as follows: ( For example, after the quantum evolution module evolves the quantum states of N qubits based on the quantum evolution operator, the final state obtained can be represented by formula (9): Formula (9) in, This represents the evolved quantum state, i.e., the final state; This represents the initial quantum state, i.e., the superposition state of a qubit; This represents the quantum evolution operator, also known as the unitary operator.

[0073] In some optional implementations, the electronic device determines the quantum evolution operator of the quantum circuit based on the Hamiltonian of the quantum computing model (i.e., the first Hamiltonian), thereby constructing the quantum evolution module of the quantum circuit based on the quantum evolution operator. In the embodiments of this application, the quantum evolution algorithm can be decomposed into multiple quantum gate operations of the quantum circuit. Through the quantum evolution operator determined by the first Hamiltonian, the resource scheduling problem of the target computing power network can be transformed into a quantum circuit. Thus, based on the parallelism and superposition of quantum computing, the resource scheduling problem can be solved quickly, improving the efficiency and accuracy of resource scheduling results and meeting the scheduling requirements in high-concurrency, large-scale scenarios.

[0074] To improve the efficiency of solving quantum circuits, in some optional implementations, the electronic device can introduce a hybrid Hamiltonian (i.e., a second Hamiltonian) when determining the quantum evolution operator. That is, the electronic device determines the quantum evolution operator based on the first and second Hamiltonians. The hybrid Hamiltonian is used to achieve the mixing of quantum states of multiple qubits. Introducing a hybrid Hamiltonian when determining the quantum evolution operator can reduce the probability that the measurement result of the quantum circuit is a local optimum.

[0075] Hybrid Hamiltonians can be realized using a second quantum spin operator acting on multiple qubits. This second quantum spin operator can be a Pauli operator, such as the Pauli Z, Pauli X, or Pauli Y operator. The second quantum spin operator is a different quantum spin operator from the first quantum spin operator. For example, if the first quantum spin operator is the Pauli Z operator, the second quantum spin operator could be the Pauli X operator.

[0076] For example, taking the second quantum spin operator as the Pauli X operator, the hybrid Hamiltonian (i.e., the second Hamiltonian) can be expressed by formula (10): Formula (10) in, This represents the mixed Hamiltonian, i.e., the superposition state. The corresponding Hamiltonian; denoted as Pauli X operator; i represents the i-th qubit, where i is a positive integer less than or equal to N; N represents the number of qubits.

[0077] Accordingly, the quantum evolution operator determined by the first Hamiltonian and the second Hamiltonian may include a first evolution operator corresponding to the first Hamiltonian and a second evolution operator corresponding to the second Hamiltonian. The first evolution operator and the second Hamiltonian act alternately on the qubit, enabling the quantum circuit to search for the optimal or near-optimal solution to the target computing power network resource scheduling problem globally.

[0078] For example, the quantum evolution operator can be represented by formula (11): Formula (11) in, Represents the quantum evolution operator. This represents the first evolution operator corresponding to the first Hamiltonian; The second evolution operator is represented by the second Hamiltonian; k represents the k-th layer of evolution, where k is a positive integer less than or equal to n; n is the number of evolution layers (or the depth of the quantum circuit), where n is a positive integer. n can be set according to actual application scenarios or needs, and the embodiment of this application does not limit the value of n.

[0079] Understandably, quantum evolution operators are parameterized unitary operators that can realize parameterized quantum circuits. The evolution parameters of a quantum circuit are represented. By iteratively optimizing the evolution parameters of the quantum circuit, the target computing power network resource scheduling problem can be solved. The initial values ​​of the evolution parameters of the quantum circuit can be set according to the actual application scenario or requirements, and this application embodiment does not limit this.

[0080] For example, the first evolution operator corresponding to the first Hamiltonian can be represented by formula (12): Formula (12) in, The first evolution operator represents the k-th layer evolution, that is, it represents the first Hamiltonian with the imaginary unit i as the base. and evolution parameters The product of these is the unitary operator of the exponent; N is the number of qubits. and All are positive integers less than or equal to N. Substituting and expanding the expression for the first Hamiltonian, the first evolutionary algorithm can be decomposed into quantum gate operations acting on the qubits: Indicates the first The first qubit is the control bit, and the second qubit is the control bit. A controlled NOT gate with one qubit as the target bit; the controlled NOT gate is used to establish entanglement relationships between different qubits; Indicates the action on the first A qubit-based rotational gate around the Z-axis (abbreviated as...) Door), Let be the rotation angle; where, It is a coefficient in the first Hamiltonian. For the k-th layer of evolution Evolutionary parameters.

[0081] Indicates the action on the first 1 qubit Door, Let be the rotation angle; where, It is a coefficient in the first Hamiltonian. For the k-th layer of evolution Evolutionary parameters.

[0082] For example, the second evolution operator corresponding to the second Hamiltonian can be represented by formula (13): Formula (13) in, The second evolution operator represents the evolution at the k-th level, that is, it represents the second Hamiltonian with the imaginary unit i as the base. and evolution parameters The product of these is the unitary operator of the exponent; N is the number of qubits. In Let represent the i-th qubit, where i is a positive integer less than or equal to N. Substituting and expanding the expression for the second Hamiltonian, the second evolutionary algorithm can be decomposed into quantum gate operations acting on the qubits: This represents a rotation gate about the X-axis acting on the i-th qubit (abbreviated as...). Door), Let be the rotation angle; where, It is the evolution parameter corresponding to the i-th qubit in the k-th layer evolution.

[0083] The quantum evolution operator expressions described above can be used to construct quantum evolution modules, thereby parameterizing quantum circuits.

[0084] In this embodiment, the quantum circuit also includes a final state measurement module. The final state measurement module is used to measure the quantum state of the evolved qubit to obtain the measurement results after the quantum circuit has run. Through the final state measurement module, the final state of the evolved qubit can be converted into observable data, thereby providing a basis for the optimization of the quantum circuit or the resource scheduling results of the target computing network.

[0085] For example, taking 4 qubits as an example, the QAOA quantum circuit constructed based on the quantum Ising model is as follows: Figure 3 As shown. The QAOA quantum circuit includes an initial state preparation module (i.e., a quantum initial state preparation module) 301, a quantum evolution module 302, and a final state measurement module 303. The initial state preparation module 301 is used to move the qubit from its ground state through an H-gate. It transforms into a superposition state. Quantum evolution module 302 is used to transform the state via quantum evolution operators. (Right now The quantum state of the qubit is evolved to obtain the final state of the qubit. The final state measurement module 303 is used to measure the final state of the qubit through measurement operations to obtain the measurement result.

[0086] Accordingly, Figure 3 The specific logic circuit of the quantum circuit shown is as follows: Figure 4 As shown. In this example, the quantum evolution operator of quantum evolution module 302 is transformed into multiple quantum gates. Here, 3021 represents multiple quantum gates corresponding to one layer of evolution of the quantum evolution operator, such as representing the first layer of evolution. The multiple quantum gates in the first layer are the same as those in the second layer.

[0087] In step 105, the electronic device acquires the measurement results of the quantum circuit operation. For example, after constructing the quantum circuit, the electronic device can run the quantum circuit using a quantum simulator or a quantum computer. The electronic device acquires the measurement results obtained after the quantum simulator or quantum computer runs the quantum circuit.

[0088] To obtain accurate measurement results from quantum circuits, in some optional implementations, the electronic device can optimize the evolution parameters of the quantum circuit one or more times to obtain an optimized quantum circuit. Thus, by iteratively optimizing the evolution parameters of the quantum circuit, the accuracy of the measurement results is improved.

[0089] For example, in each optimization, the electronic device calculates the expected value of the first Hamiltonian based on the measurement results from multiple runs of the quantum circuit, i.e., calculates... In the final state Expected value on For example, in each optimization, the electronic device acquires the measurement results of multiple runs of the quantum circuit by a quantum simulator or quantum computer. Based on the measurement results of each run, the electronic device calculates the energy value corresponding to the first Hamiltonian. Then, the electronic device combines the energy values ​​calculated from the multiple runs to calculate the expected value of the first Hamiltonian.

[0090] After determining the expected value of the first Hamiltonian, the electronic device adjusts the evolution parameters of the quantum circuit based on this expected value. For example, the electronic device can use gradient descent to calculate the gradient of the expected value of the first Hamiltonian with respect to the evolution parameters, and update the evolution parameters of the quantum circuit based on this gradient to obtain the updated evolution parameters. In this way, the electronic device completes one optimization of the evolution parameters of the quantum circuit.

[0091] After each optimization, the electronic device reacquires the measurement results of the quantum circuit running multiple times with the updated evolution parameters, and then repeats the above steps of optimizing the evolution parameters of the quantum circuit until the expected value of the first Hamiltonian converges, or until the number of optimizations of the evolution parameters reaches a preset optimization number threshold.

[0092] The evolution parameters of the optimized quantum circuit are expressed as follows: For example, ,in, This represents the expected value of the first Hamiltonian. This represents minimizing the expected value. It can be seen that, after optimization, the evolution parameters of the quantum circuit minimize the expected value of the first Hamiltonian.

[0093] It is understandable that the smaller the expected value of the first Hamiltonian, the lower the total cost of the resource scheduling strategy corresponding to the measurement result, and the better the resource scheduling strategy. Therefore, in this embodiment, by adjusting the evolution parameters in the quantum circuit, the expected value of the first Hamiltonian is made to continuously approach the minimum value, thereby gradually guiding the quantum circuit to output the optimal or near-optimal measurement result. When the expected value reaches the minimum (or converges to a stable value), the corresponding evolution parameter value is the optimal or near-optimal evolution parameter. At this time, by measuring the final state of the quantum circuit, the optimal or near-optimal resource allocation result of the target computing power network can be obtained.

[0094] After optimization, the electronic device runs the optimized quantum circuit multiple times using a quantum simulator or quantum computer to obtain the measurement results of the multiple runs of the optimized quantum circuit.

[0095] In step 106, the electronic device can determine the resource scheduling result of the target computing power network based on the measurement results of multiple runs of the optimized quantum circuit. For example, the electronic device determines the measurement result with the highest probability among the multiple runs. The electronic device determines the value of the binary variable corresponding to this measurement result based on the mapping relationship between the qubit and the binary variable in the discretized resource scheduling model. Since the binary variable corresponds to... , as well as Thus, electronic devices can determine the resource scheduling results of the target computing network based on the values ​​of the binary variables corresponding to the measurement results.

[0096] The resource scheduling result is user node i, compute node j, transmission path m, and task volume. The correspondence between them is clarified. The resource scheduling results specify the amount of task to be allocated from user node i to computing node j via transmission path m. The specific values. At the same time, the resource scheduling results also clarified the communication link e in transmission path m. as well as Based on resource scheduling results, electronic devices can allocate a portion of the task load required by user node i. The portion is transmitted to computing node j for processing via transmission path m. In this way, global resource scheduling of the target computing network is achieved while minimizing or nearly minimizing the total energy consumption of the target computing network.

[0097] This application combines the resource scheduling problem of the target computing network with quantum computing, and solves the resource scheduling problem of the target computing network through a quantum computing architecture, thereby achieving more efficient and accurate dynamic matching and collaborative scheduling of resources, and thus improving the overall resource utilization efficiency and service capability of the target computing network.

[0098] To implement the resource scheduling method provided in the embodiments of this application, the embodiments of this application also provide a resource scheduling device, such as... Figure 5 As shown, the device includes: The acquisition module 51 is used to acquire global information of the target computing power network, wherein the target computing power network includes multiple nodes, including multiple user nodes, multiple network nodes and multiple computing nodes, and the global information includes at least the status information of the multiple nodes and the status information of multiple transmission paths, and each transmission path includes one user node and one computing node. Construction module 52 is used to construct a resource scheduling model for the target computing power network based on the global information of the target computing power network; convert the resource scheduling model into a quantum computing model; and construct quantum circuits based on the quantum computing model. The determination module 53 is used to determine the resource scheduling result of the target computing power network based on the measurement results of the quantum circuit.

[0099] In some optional embodiments, the building module 52 is specifically used for: Using the task volume corresponding to the multiple transmission paths as variables, an initial cost model for the target computing power network is constructed, wherein the initial cost model is used to minimize the total energy consumption of the target computing power network, and the total energy consumption is determined based on the task volume corresponding to the multiple transmission paths; Based on the status information of the multiple nodes and the status information of the multiple transmission paths, constraints are set for the initial cost model; Based on the initial cost model and the constraints, a Lagrangian function is constructed to obtain the resource scheduling model.

[0100] In some optional embodiments, the building module 52 is specifically used for: The variables in the resource scheduling model are discretized to obtain the discretized resource scheduling model. The binary variables in the discretized resource scheduling model are mapped to the first quantum spin operator to obtain the quantum computing model; wherein the number of binary variables in the discretized resource scheduling model is equal to the number of qubits corresponding to the quantum computing model.

[0101] In some optional embodiments, the quantum circuit includes a quantum evolution module; the construction module 52 is specifically used for: A quantum evolution operator is determined based on a first Hamiltonian, wherein the first Hamiltonian is the Hamiltonian of the quantum computing model, and the quantum evolution operator is used to represent the transformation operation performed on the quantum state of a qubit. Based on the quantum evolution operator, a quantum evolution module of the quantum circuit is constructed, wherein the quantum evolution module is used to evolve the quantum state of the qubit based on the quantum evolution operator.

[0102] In some optional embodiments, the building module 52 is specifically used for: Determine the second Hamiltonian, which is a mixed Hamiltonian; The step of determining the quantum evolution operator based on the first Hamiltonian includes: The quantum evolution operator is determined based on the first Hamiltonian and the second Hamiltonian.

[0103] In some optional embodiments, the quantum circuit further includes a quantum initial state preparation module and a final state measurement module; the construction module 52 is specifically used for: Construct the quantum initial state preparation module, wherein the quantum initial state preparation module is used to initialize the number of qubits corresponding to the quantum computing model; Construct the final state measurement module, wherein the final state measurement module is used to measure the quantum state of the evolved qubit.

[0104] In some optional embodiments, the determining module 53 is further configured to: The evolution parameters of the quantum circuit are optimized once or multiple times to obtain the optimized quantum circuit; Based on the measurement results of the optimized quantum circuit, the resource scheduling results of the target computing network are determined.

[0105] In some optional embodiments, the determining module 53 is specifically used for: In each optimization, the expected value of the first Hamiltonian is determined based on the measurement results of the quantum circuit; wherein the first Hamiltonian is the Hamiltonian of the quantum computing model. The evolution parameters of the quantum circuit are optimized based on the expected value of the first Hamiltonian.

[0106] In practical applications, the acquisition module 51 can be implemented by the processor in the resource scheduling device in conjunction with the communication interface, and the construction module 52 and the determination module 53 can be implemented by the processor in the resource scheduling device.

[0107] It should be noted that the resource scheduling device provided in the above embodiments is only illustrated by the division of the above-described program units when determining the resource scheduling result. In practical applications, the above processing can be assigned to different program units as needed, that is, the internal structure of the device can be divided into different program units to complete all or part of the processing described above. In addition, the resource scheduling device and the resource scheduling method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0108] Based on the hardware implementation of the above program modules, and in order to implement the method of the embodiments of this application, the embodiments of this application also provide an electronic device, such as... Figure 6 As shown, the electronic device includes: The communication interface 601 enables information exchange with other electronic devices (such as quantum computers); The processor 602 is connected to the communication interface 601 to enable information interaction with other electronic devices (such as quantum computers) and to execute the methods provided by one or more of the above-mentioned technical solutions when running computer programs; The computer program is stored in memory 603.

[0109] Specifically, the processor 602, in conjunction with the communication interface 601, obtains global information of the target computing power network. The target computing power network includes multiple nodes, which include multiple user nodes, multiple network nodes, and multiple computing nodes. The global information includes at least the status information of the multiple nodes and the status information of multiple transmission paths. Each transmission path includes one user node and one computing node. The processor 602 is also used for: Based on the global information of the target computing power network, a resource scheduling model for the target computing power network is constructed; The resource scheduling model is converted into a quantum computing model; Based on the aforementioned quantum computing model, a quantum circuit is constructed; Based on the measurement results of the quantum circuit, the resource scheduling results of the target computing network are determined.

[0110] In some alternative embodiments, the processor 602 is specifically used for: Using the task volume corresponding to the multiple transmission paths as variables, an initial cost model for the target computing power network is constructed, wherein the initial cost model is used to minimize the total energy consumption of the target computing power network, and the total energy consumption is determined based on the task volume corresponding to the multiple transmission paths; Based on the status information of the multiple nodes and the status information of the multiple transmission paths, constraints are set for the initial cost model; Based on the initial cost model and the constraints, a Lagrangian function is constructed to obtain the resource scheduling model.

[0111] In some alternative embodiments, the processor 602 is specifically used for: The variables in the resource scheduling model are discretized to obtain the discretized resource scheduling model. The binary variables in the discretized resource scheduling model are mapped to the first quantum spin operator to obtain the quantum computing model; wherein the number of binary variables in the discretized resource scheduling model is equal to the number of qubits corresponding to the quantum computing model.

[0112] In some optional embodiments, the quantum circuit includes a quantum evolution module; the processor 602 is specifically used for: A quantum evolution operator is determined based on a first Hamiltonian, wherein the first Hamiltonian is the Hamiltonian of the quantum computing model, and the quantum evolution operator is used to represent the transformation operation performed on the quantum state of a qubit. Based on the quantum evolution operator, a quantum evolution module of the quantum circuit is constructed, wherein the quantum evolution module is used to evolve the quantum state of the qubit based on the quantum evolution operator.

[0113] In some alternative embodiments, the processor 602 is specifically used for: Determine the second Hamiltonian, which is a mixed Hamiltonian; The step of determining the quantum evolution operator based on the first Hamiltonian includes: The quantum evolution operator is determined based on the first Hamiltonian and the second Hamiltonian.

[0114] In some optional embodiments, the quantum circuit further includes a quantum initial state preparation module and a final state measurement module; the processor 602 is specifically used for: Construct the quantum initial state preparation module, wherein the quantum initial state preparation module is used to initialize the number of qubits corresponding to the quantum computing model; Construct the final state measurement module, wherein the final state measurement module is used to measure the quantum state of the evolved qubit.

[0115] In some alternative embodiments, the processor 602 is specifically used for: The evolution parameters of the quantum circuit are optimized once or multiple times to obtain the optimized quantum circuit; Determining the resource scheduling result of the target computing power network based on the measurement results of the quantum circuit includes: Based on the measurement results of the optimized quantum circuit, the resource scheduling results of the target computing network are determined.

[0116] In some alternative embodiments, the processor 602 is specifically used for: In each optimization, the expected value of the first Hamiltonian is determined based on the measurement results of the quantum circuit; wherein the first Hamiltonian is the Hamiltonian of the quantum computing model. The evolution parameters of the quantum circuit are optimized based on the expected value of the first Hamiltonian.

[0117] It should be noted that the specific processing procedures of the processor 602 and the communication interface 601 can be understood with reference to the resource scheduling method described above.

[0118] Of course, in practical applications, the various components in an electronic device are coupled together through a bus system 604. It can be understood that the bus system 604 is used to realize the connection and communication between these components. In addition to a data bus, the bus system 604 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, in... Figure 6 The general designated all buses as Bus System 604.

[0119] The memory 603 in this embodiment is used to store various types of data to support the operation of the electronic device. Examples of such data include any computer program used to operate on the electronic device.

[0120] The methods disclosed in the embodiments of this application can be applied to the processor 602, or implemented by the processor 602. The processor 602 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in the processor 602 or by instructions in the form of software. The processor 602 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 602 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods disclosed in the embodiments of this application can be directly reflected as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software modules may be located in a storage medium, which is located in the memory 603. The processor 602 reads the information in the memory 603 and combines its hardware to complete the steps of the aforementioned method.

[0121] In an exemplary embodiment, the electronic device may be implemented by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers (MCUs), microprocessors, or other electronic components to perform the aforementioned method.

[0122] It is understood that the memory 603 in this embodiment can be volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), ferromagnetic random access memory (FRAM), flash memory, magnetic surface memory, optical disc, or compact disc read-only memory (CD-ROM); magnetic surface memory can be disk storage or magnetic tape storage. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Synchronous Static Random Access Memory (SSRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), SyncLink Dynamic Random Access Memory (SLDRAM), and Direct Rambus Random Access Memory (DRRAM).The memories described in the embodiments of this application are intended to include, but are not limited to, these and any other suitable types of memories.

[0123] In an exemplary embodiment, this application also provides a storage medium, namely a computer storage medium, specifically a computer-readable storage medium, such as a memory 603 storing a computer program, which can be executed by a processor 602 of an electronic device to complete the steps described in the aforementioned method. The computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, Flash Memory, magnetic surface memory, optical disc, or CD-ROM.

[0124] In an exemplary embodiment, this application also provides a computer program product, including a computer program that can be executed by a processor 602 of an electronic device to perform the steps described in the foregoing method.

[0125] It should be noted that terms such as "first" and "second" are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0126] Furthermore, the technical solutions described in the embodiments of this application can be combined arbitrarily without conflict.

[0127] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of protection of this application.

Claims

1. A resource scheduling method, characterized in that, The method includes: Obtain global information of the target computing power network, wherein the target computing power network includes multiple nodes, including multiple user nodes, multiple network nodes and multiple computing nodes, and the global information includes at least the status information of the multiple nodes and the status information of multiple transmission paths, and each transmission path includes one user node and one computing node; Based on the global information of the target computing power network, a resource scheduling model for the target computing power network is constructed; The resource scheduling model is converted into a quantum computing model; Based on the aforementioned quantum computing model, a quantum circuit is constructed; Based on the measurement results of the quantum circuit, the resource scheduling results of the target computing network are determined.

2. The method according to claim 1, characterized in that, The step of constructing a resource scheduling model for the target computing power network based on its global information includes: Using the task volume corresponding to the multiple transmission paths as variables, an initial cost model for the target computing power network is constructed, wherein the initial cost model is used to minimize the total energy consumption of the target computing power network, and the total energy consumption is determined based on the task volume corresponding to the multiple transmission paths; Based on the status information of the multiple nodes and the status information of the multiple transmission paths, constraints are set for the initial cost model; Based on the initial cost model and the constraints, a Lagrangian function is constructed to obtain the resource scheduling model.

3. The method according to claim 1, characterized in that, The step of converting the resource scheduling model into a quantum computing model includes: The variables in the resource scheduling model are discretized to obtain the discretized resource scheduling model. The binary variables in the discretized resource scheduling model are mapped to the first quantum spin operator to obtain the quantum computing model; wherein the number of binary variables in the discretized resource scheduling model is equal to the number of qubits corresponding to the quantum computing model.

4. The method according to claim 1, characterized in that, The quantum circuit includes a quantum evolution module; the construction of the quantum circuit based on the quantum computing model includes: A quantum evolution operator is determined based on a first Hamiltonian, wherein the first Hamiltonian is the Hamiltonian of the quantum computing model, and the quantum evolution operator is used to represent the transformation operation performed on the quantum state of a qubit. Based on the quantum evolution operator, a quantum evolution module of the quantum circuit is constructed, wherein the quantum evolution module is used to evolve the quantum state of the qubit based on the quantum evolution operator.

5. The method according to claim 4, characterized in that, The quantum circuit further includes a quantum initial state preparation module and a final state measurement module; the construction of the quantum circuit based on the quantum computing model includes: Construct the quantum initial state preparation module, wherein the quantum initial state preparation module is used to initialize the number of qubits corresponding to the quantum computing model; Construct the final state measurement module, wherein the final state measurement module is used to measure the quantum state of the evolved qubit.

6. The method according to claim 1, characterized in that, The method further includes: The evolution parameters of the quantum circuit are optimized once or multiple times to obtain the optimized quantum circuit; Determining the resource scheduling result of the target computing power network based on the measurement results of the quantum circuit includes: Based on the measurement results of the optimized quantum circuit, the resource scheduling results of the target computing network are determined.

7. The method according to claim 6, characterized in that, The optimization of the evolution parameters of the quantum circuit one or more times includes: In each optimization, the expected value of the first Hamiltonian is determined based on the measurement results of the quantum circuit; wherein the first Hamiltonian is the Hamiltonian of the quantum computing model. The evolution parameters of the quantum circuit are optimized based on the expected value of the first Hamiltonian.

8. An electronic device, characterized in that, include: Processor and memory used to store computer programs that can run on the processor; When the processor is used to run a computer program, it executes the steps of the method according to any one of claims 1 to 7.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images