Simulated annealing apparatus and simulated annealing method
The simulated annealing device and method convert constraint information into CNF format and use SAT solvers to efficiently find solution candidates that satisfy multiple constraints, enhancing solution accuracy and reducing calculation time for combinatorial optimization problems.
Patent Information
- Application Number
- JP2023528924
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2021-06-18
- Publication Date
- 2024-11-13
- Estimated Expiration
- 2041-06-18
AI Technical Summary
Existing simulated annealing methods struggle to efficiently find solution candidates that satisfy multiple constraints in combinatorial optimization problems, leading to inaccurate solutions and increased calculation time.
A simulated annealing device and method that converts constraint information into Conjunctive Normal Form (CNF) format and uses a SAT solver to find solution candidates that satisfy all constraints by selecting and flipping variable candidates while solving Boolean Satisfiability Testing (SAT) problems.
The approach allows for easily obtaining solution candidates that satisfy all constraints, improving solution accuracy and reducing calculation time, especially for complex problems like Sudoku and nurse shift scheduling.
Smart Images

Figure 0007582471000006 
Figure 0007582471000007 
Figure 0007582471000008
Abstract
Description
[Technical field]
[0001] The present invention relates to a simulated annealing device, a simulated annealing method, and a simulated annealing program. [Background technology]
[0002] Research is underway on quantum computers that can quickly solve combinatorial optimization problems using the quantum annealing method. Quantum computers represent combinatorial optimization problems using an Ising model and use quantum superposition to search for a state that minimizes an energy function, which corresponds to the objective function of the combinatorial optimization problem.
[0003] The quantum computer solves a combinatorial optimization problem using the energy function in the Ising model as an input. Examples of combinatorial optimization problems include the traveling salesman problem, the knapsack problem, the graph partitioning problem, and the nurse shift scheduling problem. The combinatorial optimization problem may also be other problems.
[0004] The Ising model is a statistical mechanical model that describes the behavior of spins in magnetic materials. In the Ising model, each spin state is represented as "1" or "-1."
[0005] Fig. 9 is an explanatory diagram showing an example of a partially coupled Ising model. As shown in Fig. 9, the Ising model is composed of lattice points that indicate one of two spin states ("1" or "-1").
[0006] For example, the arrows at the lattice points shown in Figure 9 represent spins. An upward arrow represents a spin whose state is "1." A downward arrow represents a spin whose state is "-1."
[0007] The lattice points shown in Figure 9 interact with each other. When a combinatorial optimization problem is solved, the spin directions change so that the energy function in the Ising model becomes smaller.
[0008] The energy function in the Ising model is expressed as the following equation (1).
[0009]
number
[0010] In formula (1), s i ,s j is a variable that represents the state of spin. Also, J ij is a constant depending on i and j. Also, h i is a constant that depends on i.
[0011] QUBO (Quadratic Unconstrained Binary Optimization) is also known as a model that expresses the state of each spin as “1” or “0.” The energy function in the Ising model and the energy function in QUBO can be converted to each other.
[0012] The energy function in QUBO is expressed as follows:
[0013]
number
[0014] x in formula (2) i ,x j is a variable that represents the state of spin. Also, Q ij is a constant depending on i, j.
[0015] At present, quantum computers large enough to solve combinatorial optimization problems in the real world have not yet been realized. However, the advent of quantum computers has prompted research into the application of combinatorial optimization techniques.
[0016] For example, currently, in order to universally find solutions to various combinatorial optimization problems, the simulated annealing method (hereinafter also referred to as the SA method) is used, which uses the energy function in the Ising model or QUBO as input. When the SA method is used, even a classical computer can find solutions to combinatorial optimization problems.
[0017] When the combinatorial optimization problem to be solved is expressed in the Ising model format or the QUBO format, the SA method can handle a variety of combinatorial optimization problems. Note that combinatorial optimization problems formulated in the Ising model format or the QUBO format can be solved by both quantum and classical computers.
[0018] The SA method is a general-purpose method for approximately solving combinatorial optimization problems. The SA method solves combinatorial optimization problems by performing the following four processes as one cycle.
[0019] 1. Select one spin 2. Calculate the energy change Δ when the selected spin is flipped 3. Accept the spin flip with transition probability min(exp(-Δ / T),1). 4. If the flip is accepted, flip the selected spin.
[0020] Specifically, the SA method inverts the direction of the spin selected in step 2 above from “-1” to “1” or from “1” to “-1”, and calculates the amount of energy change Δ that occurs when the spin is flipped.
[0021] The SA scheme then accepts the flip of the spin in step 3 above if the calculated Δ is negative, and accepts the flip of the spin with the probability calculated above if the calculated Δ is positive.
[0022] Next, if the flip is accepted, the SA method flips the spin selected in step 4 above. In other words, the SA method searches for a state in which the energy is small by changing the direction of the spin (variable) one by one. Note that T is a temperature parameter.
[0023] Fig. 10 is an explanatory diagram showing an example of a process for solving a combinatorial optimization problem using the SA method. The curves in each rectangle shown in Fig. 10 represent energy functions. The black circles in each rectangle shown in Fig. 10 are examples of energy values when a solution obtained by the SA method is used.
[0024] Moreover, the graph shown in Fig. 10 shows the change in temperature parameter T 1 over time in the process of solving the combinatorial optimization problem. The graph shown in Fig. 10 shows that the temperature parameter T 1 decreases over time.
[0025] The SA method starts the solution process from a random combination (solution) of multiple spin directions, and transitions by evaluating nearby solutions. Because the temperature parameter T is introduced, when T is high, the SA method can transition to a solution with a poor evaluation.
[0026] For example, as shown in the early stage of the solution-finding process in Figure 10, the SA method can overcome a high peak in the energy function if T is high. By crossing a high peak and causing the evaluation of the solution to transition in a negative direction, the SA method can avoid finding a local optimum solution as the final solution to the combinatorial optimization problem.
[0027] By lowering the temperature parameter T, the peaks that can be overcome in the energy function become gradually lower, as shown in the middle of the solution process in Figure 10. In other words, the number of transitions in the direction of poor solution evaluation decreases.
[0028] When the temperature parameter T is further decreased, the SA method is unable to get over the peak in the energy function. Finally, the SA method transitions with a high probability to the deepest valley in the energy function and becomes stable, as shown in the later stage of the solution process in FIG.
[0029] Below, we will explain the process of solving a combinatorial optimization problem using the SA method as an example of finding a solution to the traveling salesman problem. The traveling salesman problem is a problem of determining in what order to visit each city in order to minimize the travel distance. In the traveling salesman problem, for the number of cities N (N is a positive integer), 2 spins (variables) are defined, and each spin indicates that a salesman exists at point k at time t.
[0030] FIG. 11 is an explanatory diagram showing an example of cities that are the subject of the traveling salesman problem. The black circles in FIG. 11 represent cities. The lines in FIG. 11 represent routes. In the example shown in FIG. 11, the traveling salesman problem is a question of in what order to visit cities A, B, C, and D in order to minimize the travel distance.
[0031] First, an equation that represents the energy in the traveling salesman problem is generated. In this example, the traveling salesman problem is expressed in QUBO format. The energy function in the traveling salesman problem is generated by adding a constraint term with a large weight to the objective function. The reason for adding the constraint term to the objective function is to impose a constraint condition on the optimization target (energy).
[0032] As an energy function in the traveling salesman problem, for example, the following equation (3) is generated.
[0033]
number
[0034] In formula (3), j represents time. v and u represent cities. x v,j is a variable that indicates whether a salesman is present in city v at time j, and corresponds to the spin. That is, x v,j = 1 indicates that the salesman is in city v at time j. Also, x v,j= 0 indicates that the salesman does not exist in city v at time j.
[0035] In addition, in equation (3), W uv is the distance between city u and city v. uv is determined in advance as a constant for each pair of cities. In addition, α and β are determined in advance as constants.
[0036] The first term on the right-hand side of equation (3) represents the constraint that a salesman may only pass through the same city once, i.e., may not pass through it more than twice.
[0037] Moreover, the second term on the right-hand side of equation (3) represents the constraint that a salesman can only be in one city at a given time.
[0038] Additionally, the third term on the right-hand side of equation (3) represents the sum of the distances between cities that the salesman passes through. The third term on the right-hand side is the objective function to be minimized.
[0039] Once the energy function in QUBO is obtained, the SA method identifies the state of each spin when the energy is minimized. Once the state of each spin when the energy is minimized is identified, the optimal solution to the traveling salesman problem is obtained.
[0040] Figure 12 is a schematic diagram showing an example of the expression of the state of each spin in the traveling salesman problem. "1" or "0" shown in Figure 12 corresponds to a spin (variable).
[0041] The spins shown in Fig. 12 correspond to city A, city B, city C, and city D from the left. Also, the spins shown in Fig. 12 correspond to time 1, time 2, time 3, and time 4 from the top. In other words, Fig. 12 shows a route that goes around city A, city C, city B, and city D in that order.
[0042] As described above, by applying the energy function in the Ising model or QUBO to SA, solutions to various combinatorial optimization problems can be obtained in a generalized manner. However, when the original combinatorial optimization problem is converted into the Ising model format or QUBO format, there is a problem that the problem space, which is the range of solution search, increases.
[0043] For example, the original problem space of the traveling salesman problem is N! if the number of cities is N. However, when the traveling salesman problem is converted to an Ising model format, the problem space becomes 2 N 2 It increases exponentially.
[0044] The SA method also calculates the energy change Δ for each of the many spin combinations in the problem space that do not satisfy the constraints. If the energy change Δ becomes large due to the constraint terms, the SA method removes the combinations that do not satisfy the constraints from the solution candidates.
[0045] By setting the above α to a large value, the calculated energy change amount Δ becomes large even when only one spin flips in a combination that does not satisfy the constraints. In other words, the sharpness of the solution space caused by the constraint terms of the Ising model, in which the constraints are not satisfied and the energy change amount Δ increases too much when only one spin flip occurs, makes it difficult for the SA method to transition from a local solution to another solution.
[0046] For the reasons above, when the SA method solves a combinatorial optimization problem using an energy function that includes constraint terms in the Ising model or QUBO, the accuracy of the solution obtained may deteriorate, or the calculation time required to obtain an optimal or suboptimal solution may increase.
[0047] A solution-finding system for solving the above problem is described in Patent Document 1. The solution-finding system described in Patent Document 1 flips spins so as to satisfy constraint terms.
[0048] For example, in the case of the traveling salesman problem, the solution system described in Patent Document 1 uses a one-hot constraint that ensures that only one spin among the combinations of spins (variables) satisfies "1". When flipping the only "1" spin, the solution system described in Patent Document 1 that uses a one-hot constraint also flips the other "0" spins at the same time.
[0049] Furthermore, Non-Patent Document 1 describes a method for converting constraints into SAT (Boolean Satisfiability Testing). [Prior art documents] [Patent documents]
[0050] [Patent Document 1] International Publication No. 2021 / 059338 [Non-patent literature]
[0051] [Non-Patent Document 1] Niklas Een and Niklas Sorensson, “Translating Pseudo-Boolean Constraints into SAT,” Journal on Satisfiability, Boolean Modeling and Computation, 2(1-4), 1-26, 2006. Summary of the Invention [Problem to be solved by the invention]
[0052] However, the solution-finding system described in Patent Document 1 does not support flipping spins so that all constraints are satisfied when multiple constraints are imposed on a combinatorial optimization problem.
[0053] FIG. 13 is a schematic diagram showing another example of the expression of the state of each spin in the traveling salesman problem. For example, the solution-finding system described in Patent Document 1 expresses the variables x C,1 =0 is flipped.
[0054] Variable x C,1 When flips from "0" to "1", the solution system described in Patent Document 1 changes the variable x C,1 The other variables are flipped so that the one-hot constraint directly related to x satisfies the one-hot constraint. Specifically, the solution system described in Patent Document 1 flips the variables x surrounded by circles, as in the state of each spin shown on the right in Figure 13. A,1 From “1” to “0”, the circled variable x C,2 Flip each from “1” to “0”.
[0055] However, as shown in the right side of Figure 13, the variable x A,1 and the variable x C,2 When flipped, the variables for time 2 and city A enclosed by dashed lines all become "0". In other words, the one-hot constraint for time 2 and the one-hot constraint for city A are no longer satisfied.
[0056] As described above, it is easy for the solution-finding system described in Patent Document 1 to flip other variables so that the constraints to which the variable to be flipped is directly related are satisfied. However, it is difficult for the solution-finding system described in Patent Document 1 to flip other variables so that other constraints to which the variable to be flipped is not directly related are also satisfied. Furthermore, Non-Patent Document 1 does not describe flipping spins so that all constraints are satisfied when multiple constraints are imposed on a combinatorial optimization problem.
[0057] Therefore, an object of the present invention is to provide a simulated annealing apparatus, a simulated annealing method, and a simulated annealing program that can easily find solution candidates that satisfy all of the multiple constraints imposed on a combinatorial optimization problem. [Means for solving the problem]
[0058] The simulated annealing apparatus according to the present invention includes a solution-finding means for finding a set of values of a plurality of variables of a combinatorial optimization problem that satisfies one or more constraints that are candidates for a solution of the combinatorial optimization problem, by solving a SAT (Boolean Satisfiability Testing) expressed in the form of a CNF (Conjunctive Normal Form) into which constraint information indicating one or more constraints imposed on the combinatorial optimization problem is converted. The solution method selects one candidate variable to be flipped from among the variables included in the provisional solution of the combinatorial optimization problem solved by the simulated annealing method, changes the value of the selected candidate variable, and then solves the SAT to find a candidate solution to the combinatorial optimization problem. It is characterized by:
[0059] The simulated annealing method according to the present invention comprises the steps of: The computer By solving SAT (Boolean Satisfiability Testing), which is expressed in the form of CNF (Conjunctive Normal Form) into which constraint information indicating one or more constraints imposed on a combinatorial optimization problem is converted, a set of values of multiple variables of the combinatorial optimization problem that satisfies one or more constraints that are candidates for the solution of the combinatorial optimization problem can be obtained. To achieve this, the computer selects one candidate variable to be flipped from among the variables included in the provisional solution of the combinatorial optimization problem solved by the simulated annealing method, and the computer changes the value of the selected candidate variable before solving the SAT to obtain a candidate solution to the combinatorial optimization problem. It is characterized by:
[0060] The simulated annealing program according to the present invention causes a computer to execute a process of solving a Boolean Satisfiability Testing (SAT) expressed in a CNF (Conjunctive Normal Form) format into which constraint information indicating one or more constraints imposed on a combinatorial optimization problem is converted, thereby determining a set of values of multiple variables of the combinatorial optimization problem that satisfies one or more constraints that are candidates for a solution of the combinatorial optimization problem. In the solution process, one candidate variable to be flipped is selected from the variables included in the provisional solution of the combinatorial optimization problem solved by the simulated annealing method, and the value of the selected candidate variable is changed before solving the SAT to find a candidate solution to the combinatorial optimization problem. It is characterized by: Effect of the Invention
[0061] According to the present invention, it is possible to easily find solution candidates that satisfy all of a plurality of constraints imposed on a combinatorial optimization problem. [Brief description of the drawings]
[0062] [Figure 1] 1 is a block diagram showing an example of the configuration of a simulated annealing apparatus according to an embodiment of the present invention; [Diagram 2] FIG. 2 is an explanatory diagram showing an example of a process in which the SAT solution unit 112 solves the SAT. [Diagram 3] FIG. 11 is an explanatory diagram showing an example in which a SAT solution finding unit 112 searches for solution candidates in a process of finding a solution to a combinatorial optimization problem. [Figure 4] FIG. 11 is an explanatory diagram showing another example in which the SAT solution finding unit 112 searches for solution candidates in the process of finding a solution to a combinatorial optimization problem. [Diagram 5] 4 is a flowchart showing an operation of a process for solving a combinatorial optimization problem by the simulated annealing apparatus 100 of the present embodiment. [Figure 6] 13 is a flowchart showing another operation of the simulated annealing apparatus 100 of the present embodiment for solving a combinatorial optimization problem. [Figure 7] FIG. 1 is an explanatory diagram showing an example of a hardware configuration of a simulated annealing apparatus according to the present invention. [Figure 8] 1 is a block diagram showing an overview of a simulated annealing apparatus according to the present invention; [Figure 9] FIG. 1 is an explanatory diagram illustrating an example of a partially coupled Ising model. [Figure 10] FIG. 1 is an explanatory diagram showing an example of a process for solving a combinatorial optimization problem using the SA method. [Figure 11] FIG. 1 is an explanatory diagram showing an example of a city that is a target of the traveling salesman problem. [Figure 12] FIG. 1 is a schematic diagram showing an example of an expression of the state of each spin in the traveling salesman problem. [Figure 13] FIG. 13 is a schematic diagram showing another example of expression of the state of each spin in the traveling salesman problem. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0063] [Configuration Description] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an embodiment of the present invention will be described with reference to the drawings. Fig. 1 is a block diagram showing an example of the configuration of a simulated annealing apparatus according to an embodiment of the present invention.
[0064] As shown in FIG. 1, a simulated annealing apparatus 100 of this embodiment includes a neighborhood solution generating unit 110, an energy calculating unit 120, a transition determining unit 130, and a temperature control unit 140.
[0065] As shown in FIG. 1, the neighborhood solution generating unit 110 of this embodiment includes a constraint information converting unit 111 and a SAT solution finding unit 112 .
[0066] The simulated annealing apparatus 100 of this embodiment is characterized in that it uses a SAT solver to search for a combination of spins (variables) that satisfies all constraints imposed on a combinatorial optimization problem.
[0067] SAT is a problem of determining whether there exists a value that makes a given logical formula true, and a SAT solver is a system that solves SAT.
[0068] Moreover, SAT is given in the form of CNF (Conjunctive Normal Form). SAT given in the form of CNF is written, for example, as in the following formula (4).
[0069]
number
[0070] In addition, x in formula (4) iis a variable that represents either true or false. In addition, "∨" in formula (4) means OR, "∧" means AND, and "¬" means NOT. In addition, (x1 ∨ ¬x2) in formula (4) and so on are called clauses.
[0071] In other words, in a SAT given in CNF format, the clauses in which variables are ORed are connected with ANDs. A SAT solver finds a set of x that satisfies a SAT given in CNF format. i If there is no variable that satisfies the SAT, the SAT solver returns unsatisfiable.
[0072] The simulated annealing apparatus 100 of this embodiment performs the following operations: i Assuming that the variables in QUBO and the SAT given in the form of CNF are constraints imposed on the combinatorial optimization problem, a set of variable values that satisfy the constraints is generated by the SAT solver. The functions of each component of the simulated annealing apparatus 100 of this embodiment will be described below.
[0073] The constraint information conversion unit 111 of the neighborhood solution generation unit 110 has a function of converting constraint information indicating constraints imposed on the combinatorial optimization problem to be solved into a SAT represented in the CNF format. The constraint information conversion unit 111 receives the constraint information.
[0074] A method of expressing one-hot constraints and the like in CNF is described, for example, in Non-Patent Document 1. In Non-Patent Document 1, for example, the following formula (5), which is a generalized constraint, is the target of conversion.
[0075]
number
[0076] In equation (5), “w i " is converted to "1" and "≧k" is converted to "=1", so that the constraint shown in equation (5) is converted to a one-hot constraint.
[0077] The constraint information conversion unit 111 inputs the generated SAT expressed in CNF format to the SAT solution unit 112 .
[0078] In addition, the SAT expressed in the CNF format in which constraints other than the one-hot constraint are converted may be directly input from outside to the SAT solving unit 112. When the SAT expressed in the CNF format is directly input from outside to the SAT solving unit 112, the constraint information conversion unit 111 does not need to be provided in the neighborhood solution generating unit 110.
[0079] The SAT solving unit 112 has a function of solving the SAT expressed in the input CNF format. In other words, the SAT solving unit 112 is a component equivalent to a SAT solver. The SAT solving unit 112 inputs the solution candidates of the combinatorial optimization problem obtained by solving the SAT to the energy calculation unit 120.
[0080] The energy calculation unit 120 has a function of calculating the amount of change Δ in energy based on the input solution candidates, which is the process of step 2. The energy calculation unit 120 handles, for example, an energy function as shown in equation (3).
[0081] The energy function handled by the energy calculation unit 120 does not need to include a constraint term. The reason is that a solution candidate input to the energy calculation unit 120 always satisfies the constraint, and therefore, when the solution candidate is substituted, the constraint term in the energy function becomes 0. The energy calculation unit 120 inputs the calculated change amount Δ of the energy to the transition determination unit 130.
[0082] The transition determination unit 130 has a function of determining whether or not to accept a spin flip in the solution candidate using the input energy change amount Δ, which is the processing of the above step 3. If the flip is accepted, the transition determination unit 130 flips the spin in the solution candidate as the processing of the above step 3.
[0083] The temperature control unit 140 has a function of controlling the temperature parameter T. As shown in Fig. 10, the temperature control unit 140 decreases the temperature parameter T as the time of the solution-finding process elapses.
[0084] Generally, SAT solvers stop processing when they find a single SAT solution, but SA requires that after flipping one variable from the current combination of variables, it obtains a combination that satisfies the constraints while leaving other variables unchanged as much as possible.
[0085] The specific operation of the SAT solution unit 112 of this embodiment for achieving the above object will be described below. The SAT solution unit 112 of this embodiment operates according to an algorithm called DPLL (Davis-Putnam-Logemann-Loveland).
[0086] The SAT solving unit 112 according to the DPLL finds x that is determined in the SAT expressed in the CNF format. i If x is determined, i If there is no x i Choose one variable from the list and select the selected x i Set to true or false.
[0087] Next, the SAT solution unit 112 calculates the determined x i Substitute into SAT expressed in CNF format. i When is substituted, the newly determined x i This occurs.
[0088] For example, in the case of the clause (x1 ∨ x2 ∨ x3), if x1 and x2 are false, then the clause itself must be true, so x3 is determined to be true. This operation of determining new variables is called unit propagation. The SAT solution unit 112 finds combinations of variables by repeating unit propagation.
[0089] If¬x i and xi If a contradiction occurs, such as both must be true, then the previously selected variable was incorrectly set to true or false. Therefore, when a contradiction occurs, the SAT solution unit 112 backtracks to the previously set variable and changes the value of that variable.
[0090] There are other methods known for accelerating SAT solvers. However, all of these methods are extensions of DPLL. The basic operation of these methods is the same as that of DPLL.
[0091] An example of the process in which the SAT solution unit 112 solves the SAT according to the above algorithm will be described with reference to Fig. 2. Fig. 2 is an explanatory diagram showing an example of the process in which the SAT solution unit 112 solves the SAT.
[0092] In the example shown in Fig. 2, the SAT solution unit 112 solves the SAT shown in equation (4). Note that (1), (2), ... shown in Fig. 2 represent stages of processing. Also, the circles shown in Fig. 2 represent the state of combinations of variables at each stage.
[0093] The straight arrows in Figure 2 represent the setting of variables and changes in state. i,} represents a combination of determined variables. Also, φ in Fig. 2 represents the initial state, which is an empty set. The meanings of each notation in Fig. 2 are the same in Figs. 3 and 4.
[0094] In the first step (1), the SAT solution unit 112 sets x1 to false. When x1 is set to false, the clause (x1 ∨ ¬x2) determines x2 to be false. Therefore, after step (1), the determined combination of variables is {¬x 1, It becomes ¬x2}.
[0095] In the next step (2), the SAT solution unit 112 sets x3 to false. When x3 is set to false, the clause (x1 ∨ x3 ∨ ¬ x4) determines x4 to be false. However, the clause (x3 ∨ x4) determines x4 to be true.
[0096] Since a contradiction occurs regarding x4 as described above, the SAT solution finding unit 112 determines that the setting of x3 was incorrect, and backtracks to the previous state. Next, in step (3), the SAT solution finding unit 112 resets x3 to true.
[0097] If x3 is set to true, then the clause (¬x3 ∨ ¬x5) determines that x5 is false. However, the clause (¬x3 ∨ x5) determines that x5 is true.
[0098] Since a contradiction occurs regarding x5 as described above, the SAT solution finding unit 112 determines that the setting of x1 was incorrect, and backtracks to the previous state. Next, in step (4), the SAT solution finding unit 112 resets x1 to true.
[0099] When x1 is set to true, no variables are determined. Therefore, after step (4), the determined combination of variables is {x1}.
[0100] Next, in step (5), the SAT solution unit 112 sets x3 to false. When x3 is set to false, x4 is determined to be true by the clause (x3 ∨ x4). Therefore, after step (5), the determined combination of variables is {x 1, ¬x 3, x4}.
[0101] Next, in step (6), the SAT solution unit 112 sets x2 to false. Even if x2 is set to false, there are no variables that are determined. Therefore, after step (6), the combination of variables that is determined is {x 1, ¬x 2, ¬x 3, x4}.
[0102] Then, in step (7), SAT solver 112 sets x5 to false. Once x5 is set to false, all clauses are satisfied.
[0103] Therefore, after step (7), the combination of variables is {x 1, ¬x 2, ¬x 3, x 4, ¬x5}, the SAT solving unit 112 ends the process of solving the SAT. The process of solving the SAT is a depth-first search process, as shown in FIG.
[0104] As mentioned above, SA requires that after flipping one variable from the current combination of variables, a combination (a solution candidate) that satisfies the constraints while leaving other variables unchanged as much as possible. The current combination of variables is the solution candidate {x 1, ¬x 2, ¬x 3, x 4, ¬x5} and provisional solutions to combinatorial optimization problems solved using the SA method.
[0105] Therefore, when the simulated annealing apparatus 100 of this embodiment solves a combinatorial optimization problem, the SAT solution unit 112, which sets the values of variables according to the DPLL, uses the current values of the variables as much as possible.
[0106] There are two ways to use the current variable values as much as possible: "1. Setting the candidates for the variables to be flipped first and then resolving the problem" and "2. Forcibly backtracking on the previously generated search tree." Each method is explained below.
[0107] 3 is an explanatory diagram showing an example of how the SAT solution unit 112 searches for solution candidates in a process of solving a combinatorial optimization problem. In the example shown in Fig. 3, the SAT solution unit 112 searches for solution candidates according to "1. Method of first setting variable candidates to be flipped and then resolving."
[0108] In the example shown in FIG. 3, the SAT solution finding unit 112 finds the solution candidates {x 1, ¬x 2, ¬x 3, x 4, 3, the SAT solving unit 112 solves the SAT shown in equation (4).
[0109] 3, the SAT solution finding unit 112 first sets x2 as a variable candidate to be flipped. Therefore, in step (8), the SAT solution finding unit 112 sets x2 to true.
[0110] When x2 is set to true, the clause (x1 ∨ ¬x2) determines x1 to be true. Therefore, after step (8), the determined combination of variables is {x 1, x2}.
[0111] In the next step (9), the SAT solution unit 112 sets x3, which was false in the previous solution candidate, to false. When x3 is set to false, x4 is determined to be true by the clause (x3 ∨ x4). Therefore, after step (9), the determined combination of variables is {x 1, x 2, ¬x 3, x4}.
[0112] In the next step (10), SAT solution unit 112 sets x5, which was false in the previous solution candidate, to false. When x5 is set to false, all clauses are satisfied.
[0113] Therefore, after step (10), the combination of variables is {x 1, x 2, ¬x 3, x 4, Since the answer is determined to be {¬x5}, the SAT solution unit 112 ends the process of solving the SAT.
[0114] Although not shown in the example in Figure 3, when the value of the candidate variable to be flipped is changed for the first time, in the traveling salesman problem, etc., the values of variables other than the candidate variable to be flipped may also be changed to satisfy constraints.
[0115] Also, if the constraint is not satisfied with the value of the variable after the flip, the value of the variable is changed after backtracking. In the example shown in Figure 3, even if x1 is set as a candidate variable to be flipped, the constraint is not satisfied.
[0116] Specifically, when x1 is set to a variable candidate to be flipped, that is, when x1 is set to false, no solution candidate is obtained, so SAT solution finding unit 112 sets x1 to true according to DPLL. Finally, SAT solution finding unit 112 obtains the same solution candidate as the solution candidate shown in FIG.
[0117] 4 is an explanatory diagram showing another example of how the SAT solution finding unit 112 searches for solution candidates in the process of solving a combinatorial optimization problem. In the example shown in Fig. 4, the SAT solution finding unit 112 searches for solution candidates according to "2. Method of forcibly backtracking on the previously generated search tree".
[0118] In the example shown in FIG. 4, the SAT solution finding unit 112 finds the solution candidates {x 1, ¬x 2, ¬x 3, x 4, 4, the SAT solving unit 112 solves the SAT shown in equation (4).
[0119] The search tree shown in Fig. 4 is composed of one or more nodes indicating combinations of values of multiple variables, which are generated when candidate solutions to a combinatorial optimization problem are sought. Note that a similarly generated search tree is used even when the SAT solution finding unit 112 searches for candidate solutions based on a tentative solution to a combinatorial optimization problem solved by the SA method.
[0120] 4, the SAT solution finding unit 112 backtracks from state a, where a solution candidate is first obtained, to state b, where variable x5 is set. The backtrack width when backtracking from state a to state b is "1".
[0121] Next, the SAT solution unit 112 sets x5 to true. When x5 is set to true, all clauses are satisfied. Therefore, in state c, the combination of variables is {x 1, ¬x 2, ¬x 3, x 4, x5}, the SAT solution unit 112 ends the process of solving the SAT.
[0122] To obtain other solution candidates, the SAT solution finding unit 112 further backtracks the search tree. For example, the SAT solution finding unit 112 backtracks from state a, where the first solution candidate is obtained, to state d, where the variable x2 is set. The backtrack width when backtracking from state a to state d is "2".
[0123] 4, the SAT solving unit 112 searches for nodes that have not yet been generated in the search tree by solving the SAT starting from a node that indicates a tentative solution. The SAT solving unit 112 finds a set of values of multiple variables indicated by the searched node as a candidate solution.
[0124] However, going back too far in the search tree increases the cost of solving the SAT. Therefore, if going back too far in the search tree, the SAT solution unit 112 solves the SAT again and recreates the search tree. For example, the SAT solution unit 112 changes the shape of the search tree to be generated by randomly changing the order in which variables are selected. When the shape of the search tree is changed, the variables whose values are changed after backtracking also change.
[0125] It has been confirmed that the simulated annealing apparatus 100 of this embodiment is particularly effective in solving problems in which it is difficult to satisfy all of the multiple constraints imposed, such as Sudoku (registered trademark) and nurse shift scheduling problems. It has also been confirmed that the "2. Method of forcibly backtracking on the previously generated search tree" can solve combinatorial optimization problems about 10 times faster than the "1. Method of first setting variable candidates to be flipped and then resolving the problem."
[0126] As described above, the SAT solving unit 112 of this embodiment solves a SAT expressed in the CNF format into which constraint information indicating one or more constraints imposed on a combinatorial optimization problem is converted, thereby finding a set of values of multiple variables of the combinatorial optimization problem that satisfies one or more constraints that are candidates for a solution of the combinatorial optimization problem. Also, the constraint information conversion unit 111 converts the constraint information into a SAT.
[0127] For example, the combinatorial optimization problem to be solved by the simulated annealing apparatus 100 is a nurse shift scheduling problem.
[0128] [Explanation of operation] Hereinafter, the operation of the simulated annealing apparatus 100 of this embodiment will be described with reference to FIGS.
[0129] Fig. 5 is a flowchart showing the operation of the combinatorial optimization problem solving process by the simulated annealing apparatus 100 of this embodiment. Fig. 5 shows the combinatorial optimization problem solving process in the case where the SAT solving unit 112 employs the method of "1. First setting variable candidates to be flipped and then resolving".
[0130] First, constraint information of a combinatorial optimization problem to be solved is input to the constraint information conversion unit 111 of the neighborhood solution generation unit 110. The constraint information conversion unit 111 converts the input constraint information into a SAT expressed in CNF format (step S101). The constraint information conversion unit 111 inputs the generated SAT to the SAT solution unit 112.
[0131] Next, the generated initial solution of the SAT is input to the SAT solution unit 112. The initial solution of the SAT is, for example, a value (true or false) specified by the user, or a set of randomly generated values.
[0132] The SAT solving unit 112 generates a set of variable values that satisfy the constraints by solving the input SAT while selecting the input initial solution value as the initial value (step S102). For example, the process shown in FIG. 2 is the process of step S102.
[0133] The SAT solution unit 112 may solve a SAT input from an external source other than the constraint information conversion unit 111. The SAT solution unit 112 inputs the generated set of variable values to the energy calculation unit 120.
[0134] Next, the energy calculation unit 120 calculates the energy based on the set of input variable values (step S103).
[0135] Next, the SAT solution finding unit 112 selects one variable candidate to be flipped from among the variables included in the tentative solution (step S104). Note that in the first iterative process, the tentative solution is the set of variable values generated in step S102.
[0136] Next, the SAT solution unit 112 initially sets the values of the selected variable candidates. Next, the SAT solution unit 112 solves the input SAT while selecting the values of the tentative solutions as initial values (step S105).
[0137] Next, the SAT solution finding unit 112 sets variables whose values are different between the tentative solution and the solution candidates found in step S105 as variable candidates to be flipped (step S106). Next, the SAT solution finding unit 112 inputs the solution candidates including the found variable candidates to be flipped to the energy calculation unit 120.
[0138] Next, the energy calculation unit 120 calculates the change in energy when the input variable candidate set is flipped (step S107). The energy calculation unit 120 inputs the calculated change in energy and the solution candidate to the transition determination unit .
[0139] Next, the transition determination unit 130 calculates the transition probability based on the input amount of change in energy and the temperature parameter (step S108).
[0140] Next, the transition determination unit 130 determines whether or not to accept the state transition based on the calculated transition probability (step S109). If it is determined that the state transition is not to be accepted ("reject" in step S109), the transition determination unit 130 proceeds to step S111.
[0141] When it is determined that the state transition is to be accepted ("accept" in step S109), the transition determination unit 130 flips the variable candidates included in the solution candidates (step S110). When the variable candidates are flipped, the solution candidate becomes the current tentative solution.
[0142] Next, the temperature control unit 140 decreases the temperature parameter T (step S111).
[0143] Next, the transition determination unit 130 determines whether or not the solution-finding process satisfies an end condition (step S112). The end condition is, for example, that the number of times that the processes of steps S104 to S111 have been repeatedly executed reaches a predetermined number.
[0144] If the solution finding process does not satisfy the end condition (No in step S112), the transition determination unit 130 inputs the current tentative solution to the SAT solution finding unit 112 (step S113). The SAT solution finding unit 112 executes the process of step S104 again.
[0145] If the solution-finding process satisfies the end condition (Yes in step S112), the transition determination unit 130 outputs the current tentative solution as the optimal solution (step S114). After outputting the tentative solution, the simulated annealing apparatus 100 ends the combinatorial optimization problem solution process.
[0146] Fig. 6 is a flowchart showing the operation of another combinatorial optimization problem solving process by the simulated annealing apparatus 100 of this embodiment. Fig. 6 shows the combinatorial optimization problem solving process in the case where the SAT solution unit 112 employs "2. Method of forcibly backtracking on the previously generated search tree".
[0147] The processes in steps S201 to S203 are similar to the processes in steps S101 to S103 shown in FIG.
[0148] Next, the SAT solution unit 112 backtracks the search tree generated when the tentative solution is obtained, the backtrack width starting from the state where the tentative solution is obtained. In the first iteration, the tentative solution is the set of variable values generated in step S202.
[0149] Next, the SAT solving unit 112 changes the values of the variables that can be changed in the backtracked state and solves the input SAT. After solving the SAT, the SAT solving unit 112 increases the backtrack width by one (step S204).
[0150] Next, the SAT solution finding unit 112 sets variables whose values are different between the tentative solution and the solution candidates found in step S204 as variable candidates to be flipped (step S205). Next, the SAT solution finding unit 112 inputs the solution candidates including the found variable candidates to be flipped to the energy calculation unit 120.
[0151] The processes in steps S206 to S211 are similar to the processes in steps S107 to S112 shown in FIG.
[0152] If the solution finding process does not satisfy the end condition (No in step S211), the SAT solution finding unit 112 determines whether the current backtrack width is greater than a threshold value (step S212).
[0153] If the current backtrack width is equal to or smaller than the threshold (No in step S212), the transition determination unit 130 inputs the current tentative solution to the SAT solution finding unit 112. The SAT solution finding unit 112 sets the input from the transition determination unit 130 as the tentative solution (step S214) and executes the process of step S204 again.
[0154] If the current backtrack width is greater than the threshold (Yes in step S212), the SAT solving unit 112 re-solves the SAT input in a random variable order while selecting the tentative solution value as the initial value. After re-solving the SAT, the SAT solving unit 112 sets the backtrack width to "1" (step S213).
[0155] Next, the SAT solution finding unit 112 sets the solution candidate found in step S213 as a tentative solution (step S214). The SAT solution finding unit 112 executes the process of step S204 again.
[0156] If the solution-finding process satisfies the end condition (Yes in step S211), the transition determination unit 130 outputs the current tentative solution as the optimal solution (step S215). After outputting the tentative solution, the simulated annealing apparatus 100 ends the combinatorial optimization problem solution process.
[0157] As described above, when "1. Method of first setting variable candidates to be flipped and then resolving" is adopted, the SAT solving unit 112 selects one variable candidate to be flipped from among the variables included in the tentative solution to the combinatorial optimization problem solved by the SA method, changes the value of the selected variable candidate, and then solves the SAT, thereby finding a candidate solution to the combinatorial optimization problem.
[0158] Furthermore, when "2. Method of forcibly backtracking on the previously generated search tree" is adopted, the SAT solving unit 112 searches for nodes that have not yet been generated in a search tree that is composed of one or more nodes indicating sets of values of multiple variables that were generated when a tentative solution to the combinatorial optimization problem being solved by the SA method was obtained, by solving the SAT starting from the node indicating the tentative solution, and obtains the set indicated by the searched node as a candidate solution.
[0159] Regardless of which method is adopted, the SAT solution finding unit 112 uses the values of the tentative solution as the initial values of the variables, and sets the variables included in the found solution candidates whose values have been changed from their initial values as variable candidates to be flipped.
[0160] In addition, the transition determination unit 130 determines whether or not to accept the flip of the variable candidate to be flipped that is included in the solution candidate. The transition determination unit 130 flips the variable candidate that is determined to accept the flip, and updates the tentative solution to the solution candidate to which the variable candidate has been flipped.
[0161] [Effect description] The simulated annealing apparatus 100 of this embodiment includes a constraint information conversion unit 111 that converts constraint information indicating constraints imposed on a combinatorial optimization problem to be solved into a SAT expressed in CNF format. The simulated annealing apparatus 100 also includes a SAT solving unit 112 that solves an input SAT expressed in CNF format.
[0162] Therefore, in the simulated annealing apparatus 100 of this embodiment, the SAT solution finding unit 112 can find solution candidates for a combinatorial optimization problem that satisfy constraints as a SAT solver, and therefore the simulated annealing apparatus 100 can proceed with the search for a solution while satisfying all constraints imposed on the combinatorial optimization problem.
[0163] By adopting the above-mentioned configuration, the simulated annealing apparatus 100 of this embodiment can easily solve a combinatorial optimization problem having multiple constraints, compared to the solution-finding system described in Patent Document 1, even when implemented by a classical computer.
[0164] A specific example of the hardware configuration of the simulated annealing apparatus 100 of this embodiment will be described below. Fig. 7 is an explanatory diagram showing an example of the hardware configuration of the simulated annealing apparatus according to the present invention.
[0165] 7 includes a CPU (Central Processing Unit) 11, a main memory unit 12, a communication unit 13, and an auxiliary memory unit 14. It also includes an input unit 15 for user operation and an output unit 16 for presenting the processing results or the progress of the processing contents to the user.
[0166] The simulated annealing apparatus 100 is realized by software, in which the CPU 11 shown in FIG. 7 executes a program that provides the functions of each component.
[0167] That is, the CPU 11 loads a program stored in the auxiliary storage unit 14 into the main storage unit 12, executes it, and controls the operation of the simulated annealing apparatus 100, whereby each function is realized by software.
[0168] 7 may include a DSP (Digital Signal Processor) instead of the CPU 11. Alternatively, the simulated annealing apparatus 100 shown in Fig. 7 may include both the CPU 11 and a DSP.
[0169] The main memory unit 12 is used as a working area for data and a temporary storage area for data. The main memory unit 12 is, for example, a RAM (Random Access Memory).
[0170] The communication unit 13 has a function of inputting and outputting data to and from peripheral devices via a wired network or a wireless network (information and communication network).
[0171] The auxiliary storage unit 14 is a non-transitory tangible storage medium. Examples of non-transitory tangible storage media include a magnetic disk, a magneto-optical disk, a CD-ROM (Compact Disk Read Only Memory), a DVD-ROM (Digital Versatile Disk Read Only Memory), and a semiconductor memory.
[0172] The input unit 15 has a function of inputting data and processing commands, and is, for example, an input device such as a keyboard or a mouse.
[0173] The output unit 16 has a function of outputting data, and is, for example, a display device such as a liquid crystal display device, or a printing device such as a printer.
[0174] As shown in FIG. 7, in the simulated annealing apparatus 100 , each component is connected to a system bus 17 .
[0175] In the simulated annealing apparatus 100, the auxiliary storage unit 14 stores programs for implementing a neighborhood solution generation unit 110, an energy calculation unit 120, a transition determination unit 130, and a temperature control unit 140.
[0176] The simulated annealing apparatus 100 may be implemented with a circuit including hardware components such as an LSI (Large Scale Integration) that realizes the functions shown in FIG.
[0177] The simulated annealing apparatus 100 may also be realized by hardware that does not include a computer function using an element such as a CPU. For example, some or all of the components may be realized by a general-purpose circuit, a dedicated circuit, a processor, or a combination of these. These may be configured by a single chip (for example, the above-mentioned LSI), or may be configured by multiple chips connected via a bus. Some or all of the components may be realized by a combination of the above-mentioned circuits, etc. and a program.
[0178] Furthermore, some or all of the components of the simulated annealing apparatus 100 may be configured with one or more information processing devices each including a calculation unit and a storage unit.
[0179] When some or all of the components are realized by a plurality of information processing devices, circuits, etc., the plurality of information processing devices, circuits, etc. may be centrally or distributed. For example, the information processing devices, circuits, etc. may be realized as a client-server system, a cloud computing system, etc., in a form in which each is connected via a communication network.
[0180] Next, an overview of the present invention will be described. Fig. 8 is a block diagram showing an overview of a simulated annealing apparatus according to the present invention. A simulated annealing apparatus 20 according to the present invention includes a solution-finding means 21 (e.g., a SAT solution unit 112) that finds a set of values of multiple variables of a combinatorial optimization problem that satisfies one or more constraints that are candidates for a solution of the combinatorial optimization problem, by solving a SAT expressed in a CNF format into which constraint information indicating one or more constraints imposed on the combinatorial optimization problem is converted.
[0181] With such a configuration, the simulated annealing apparatus can easily find solution candidates that satisfy all of the multiple constraints imposed on a combinatorial optimization problem.
[0182] Moreover, the simulated annealing apparatus 20 may include a conversion means (for example, the constraint information conversion unit 111) that converts the constraint information into an SAT.
[0183] With such a configuration, the simulated annealing apparatus can convert the constraint information into a SAT.
[0184] In addition, the solution finding means 21 may select one variable candidate to be flipped from among variables included in a tentative solution to the combinatorial optimization problem solved by the simulated annealing method, change the value of the selected variable candidate, and then find a candidate solution to the combinatorial optimization problem by solving the SAT.
[0185] With such a configuration, the simulated annealing apparatus can easily find solution candidates that satisfy all of the multiple constraints imposed on a combinatorial optimization problem.
[0186] In addition, the solution finding means 21 may search for a node that has not yet been generated in a search tree that is composed of one or more nodes indicating sets of values of a plurality of variables that were generated when a tentative solution to the combinatorial optimization problem being solved by the simulated annealing method was obtained, by solving the SAT, and obtain the set indicated by the searched node as a candidate solution.
[0187] With such a configuration, the simulated annealing apparatus can solve combinatorial optimization problems more quickly.
[0188] Furthermore, the solution finding means 21 may use the values of the tentative solution as the initial values of the variables, and set the variables included in the found solution candidates whose values have been changed from their initial values as variable candidates to be flipped.
[0189] With such a configuration, the simulated annealing apparatus can find solution candidates close to the tentative solution.
[0190] The simulated annealing apparatus 20 may also include a determination means (e.g., transition determination unit 130) that determines whether or not to accept a flip of a variable candidate to be flipped that is included in a solution candidate. The determination means may also flip the variable candidate to be flipped that is determined to accept a flip, and update the tentative solution to the solution candidate.
[0191] With such a configuration, the simulated annealing apparatus can find a new solution close to the tentative solution.
[0192] The combinatorial optimization problem may also be a nurse shift scheduling problem.
[0193] With such a configuration, the simulated annealing apparatus can easily solve the nurse shift scheduling problem.
[0194] Although the present invention has been described above with reference to the embodiments and examples, the present invention is not limited to the above-mentioned embodiments and examples. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention. [Explanation of symbols]
[0195] 11 CPU 12 Main memory 13. Communications Department 14 Auxiliary storage 15 Input section 16 Output section 17 System Bus 20, 100 Simulated annealing device 21 Solution method 110 Neighborhood solution generator 111 Constraint information conversion unit 112 SAT Solving Section 120 Energy Calculation Department 130 Transition judgment section 140 Temperature control unit
Claims
1. a solving means for solving a Boolean Satisfiability Testing (SAT) expressed in a Conjunctive Normal Form (CNF) format into which constraint information indicating one or more constraints imposed on a combinatorial optimization problem is converted, thereby determining a set of values of a plurality of variables of the combinatorial optimization problem that satisfies the one or more constraints, which are candidates for a solution of the combinatorial optimization problem; The solution-finding means is selecting one candidate variable to be flipped from among variables included in a provisional solution of the combinatorial optimization problem solved by the simulated annealing method; Find candidate solutions to the combinatorial optimization problem by solving the SAT after changing the values of the selected candidate variables. A simulated annealing apparatus comprising:
2. Equipped with a conversion mechanism to convert constraint information into SAT 2. The simulated annealing apparatus according to claim 1.
3. A method for solving a combinatorial optimization problem, comprising: a solving means for solving a Boolean Satisfiability Testing (SAT) expressed in a form of Conjunctive Normal Form (CNF) into which constraint information indicating one or more constraints imposed on the combinatorial optimization problem is converted, thereby determining a set of values of multiple variables of the combinatorial optimization problem that satisfies the one or more constraints, which are candidates for a solution of the combinatorial optimization problem; The solution-finding means is searching for a node that has not yet been generated in a search tree that is configured with one or more nodes indicating a set of values of a plurality of variables that was generated when a provisional solution to the combinatorial optimization problem that is solved by a simulated annealing method was obtained, by solving a SAT; The pair indicated by the searched node is selected as a candidate solution. A simulated annealing apparatus comprising:
4. The solution method is Use the provisional solution values as the initial values of the variables, Among the variables included in the obtained solution candidates, the variables whose values have been changed from the initial values are set as candidates for variables to be flipped. The simulated annealing apparatus according to any one of claims 1 to 3.
5. The method includes a determination unit that determines whether or not to accept the flip of a variable candidate to be flipped that is included in a solution candidate. The simulated annealing apparatus according to any one of claims 1 to 4.
6. The determination means is Flip the flip target variable candidates that are determined to be eligible for flipping; Update the tentative solution to a candidate solution 6. The simulated annealing apparatus according to claim 5.
7. The combinatorial optimization problem is the nurse shift scheduling problem. The simulated annealing apparatus according to any one of claims 1 to 6.
8. A method for solving a Boolean Satisfiability Testing (SAT) problem represented in a Conjunctive Normal Form (CNF) format into which constraint information indicating one or more constraints imposed on a combinatorial optimization problem is converted, thereby determining a set of values of multiple variables of the combinatorial optimization problem that satisfy the one or more constraints, which are candidates for a solution of the combinatorial optimization problem; the computer selects one candidate variable to be flipped from among variables included in a tentative solution to the combinatorial optimization problem solved by a simulated annealing method; The computer obtains candidate solutions to the combinatorial optimization problem by solving the SAT after changing the values of the selected candidate variables. A simulated annealing method comprising:
9. On the computer, executing a solution process for finding a set of values of a plurality of variables of the combinatorial optimization problem that satisfies one or more constraints, which are candidates for a solution of the combinatorial optimization problem, by solving a SAT (Boolean Satisfiability Testing) expressed in the form of a CNF (Conjunctive Normal Form) into which constraint information indicating one or more constraints imposed on the combinatorial optimization problem is converted; In the solution process, selecting one candidate variable to be flipped from among variables included in a provisional solution of the combinatorial optimization problem solved by a simulated annealing method; Find candidate solutions to the combinatorial optimization problem by solving the SAT after changing the values of the selected candidate variables. A simulated annealing program for
Citation Information
Patent Citations
An Efficient Modeling Method for Embedded Memory in Finite Memory Testing
JP2007524165A
Verification program, verification method, and verification device
JP2015172873A
Information processing apparatus, PUBO solver, information processing method, and program
JP2020201598A
Solution system, solution method, and solution program
WO2021059338A1
