Optimization processing device, optimization processing method, and program

Abstract

An optimization processing device 100 according to the present disclosure comprises: a solving unit 121 that performs solving processing within a preset solving time on an optimization problem in which are set parameters to be used for solution transition processing in accordance with solution evaluation during a solution search; a determination unit 122 that determines whether or not preset information to be determined that is identified on the basis of a solution transition in the solving processing within the solving time satisfies a preset condition; and a setting unit 123 that sets the parameters in the optimization problem by changing the parameters according to whether or not the information to be determined within the solving time satisfies the condition.

Description

Optimization processing device, optimization processing method, and program

[0001] The present disclosure relates to an optimization processing device, an optimization processing method, and a program.

[0002] When solving a constrained optimization problem using a formulated model, the problem is converted into an unconstrained optimization problem. In this case, the energy value of the constrained problem can be expressed by an objective function term and a constraint term, thereby converting the problem into a formulated model. For example, Patent Literature 1 describes converting a combinatorial optimization problem into the format of an Ising model.

[0003] International Publication No. 2020 / 196866

[0004] However, when solving the above-mentioned combinatorial optimization problem, it is necessary to set parameters for the constraint terms in the formulated model, which requires multiple solving attempts, resulting in a time-consuming problem.Furthermore, not only for combinatorial optimization problems, but also for global optimization problems such as simulated annealing, which require parameter setting, the problem of time-consuming solution occurs.

[0005] Therefore, an object of the present disclosure is to solve the above-mentioned problem that it takes a long time to solve an optimization problem.

[0006] An optimization processing device according to one aspect of the present disclosure includes: a solution finding unit that performs solution finding processing within a predetermined solution time for an optimization problem, for which parameters used in solution transition processing in response to solution evaluation during solution search are set; a determination unit that determines whether predetermined judgment target information identified based on solution transitions during the solution finding processing within the solution finding time satisfies a predetermined condition; and a setting unit that changes the parameters in response to whether the judgment target information during the solution finding time satisfies the condition, and sets the parameters for the optimization problem. Also, an optimization processing method according to one aspect of the present disclosure is configured to perform solution finding processing within a predetermined solution time for an optimization problem, for which parameters used in solution transition processing in response to solution evaluation during solution search are set; determine whether predetermined judgment target information identified based on solution transitions during the solution finding processing within the solution finding time satisfies a predetermined condition; and change the parameters in response to whether the judgment target information during the solution finding time satisfies the condition, and set the parameters for the optimization problem. Furthermore, a program according to one embodiment of the present disclosure has a configuration in which a computer executes the following processes: performing a solution process within a predetermined solution time for an optimization problem in which parameters used in a solution transition process according to an evaluation of the solution during a solution search are set; determining whether predetermined information to be judged, which is identified based on the transition of the solution in the solution process within the solution time, satisfies a predetermined condition; and changing the parameters and setting them for the optimization problem according to whether the information to be judged within the solution time satisfies the condition.

[0007] With the above-described configuration, the present disclosure can reduce the time required to solve an optimization problem.

[0008] FIG. 1 is a block diagram showing a configuration of an optimization processing device according to the present disclosure; FIG. 2 is a flowchart showing processing operations of an optimization processing device according to the present disclosure; FIG. 3 is a diagram showing an example of processing by an optimization processing device according to the present disclosure; FIG. 4 is a flowchart showing processing operations of an optimization processing device according to the present disclosure; FIG. 5 is a block diagram showing a hardware configuration of an optimization processing device according to the present disclosure; FIG. 6 is a block diagram showing a configuration of an optimization processing device according to the present disclosure.

[0009] First Embodiment A first embodiment of the present disclosure will be described with reference to the drawings. Note that the drawings may be relevant to any embodiment.

[0010] The optimization processing device 10 in this embodiment performs processing to find an optimal solution to an optimization problem. In particular, in this embodiment, the optimization problem for which the optimization processing device 10 performs processing to find an optimal solution is an optimization problem with constraints. However, the optimization processing device 10 of the present disclosure is not limited to optimization problems with constraints, and can also be applied to finding solutions to other optimization problems as described in other embodiments.

[0011] The optimization processing device 10 is composed of one or more information processing devices each equipped with a calculation device and a storage device. As shown in FIG. 1, the optimization processing device 10 includes a solution-finding unit 11, a determination unit 12, and a setting unit 13. The functions of the solution-finding unit 11, the determination unit 12, and the setting unit 13 can be realized by the calculation device executing a program for realizing each function stored in the storage device. The optimization processing device 10 also includes a problem storage unit 14 configured by a storage device. Each component and operation will be described in detail below.

[0012] The problem storage unit 14 stores information representing an optimization problem to be solved. Here, the optimization problem in this embodiment is a constrained optimization problem, for example, where an objective function f(x) and a constraint g(x) are set, and the problem is to find a solution x that minimizes the objective function f(x) while satisfying the constraint g(x) (g(x)=0). In this case, when solving the constrained optimization problem using a formulated model, the problem is converted into an unconstrained optimization problem. In this case, by expressing the energy value of the constrained optimization problem using the objective function term and the constraint term, the problem can be converted into a formulated model (for example, an Ising model or a QUBO (Quadratic Unconstrained Binary Optimization) model).

[0013] In this embodiment, the evaluation value of a solution, which is an energy value when a constrained optimization problem in which an objective function f(x) and a constraint g(x) are set is converted into an unconstrained optimization problem, is expressed as follows using the objective function term f(x) and the constraint term αP(x): f(x) + αP(x) In this case, the constraint term P(x) is "0" if the constraint is satisfied, and "a value greater than 0" if it is not satisfied.

[0014] Here, when a solution does not satisfy the constraint g(x), the constraint term P(x) becomes "greater than 0," and therefore the energy value becomes larger. In addition, a weighting coefficient α (parameter) is added to the constraint term, and the larger the value of the weighting coefficient α, the larger the energy value corresponding to the constraint term becomes when the solution does not satisfy the constraint. In other words, the weighting coefficient α is set so that the larger the value, the worse the evaluation value of the solution. Therefore, when searching for a solution to an optimization problem, if the solution does not satisfy the constraint, the larger the value of the weighting coefficient α is set, the worse the evaluation value of the solution becomes, and a solution that does not satisfy the constraint is less likely to be selected as a transition destination.

[0015] In this embodiment, the initial value of the weighting coefficient α is set to a relatively large value in advance using the characteristics of the energy equation described above. The energy equation converted into the unconstrained optimization problem described above and the initial value of the weighting coefficient α are stored in advance as an optimization problem in the problem storage unit 14. As described below, an initial value of the solution time for executing the solution process for the optimization problem in the state set with the weighting coefficient α is set, and this initial value of the solution time is also stored in the problem storage unit 14. Here, the initial value of the solution time is set to, for example, a time significantly shorter than the time that may be required to complete the solution of the optimization problem. As described below, a formulated model is calculated by the optimization processing device 10 from the energy equation for the unconstrained optimization problem stored in the problem storage unit 14.

[0016] The solution-seeking unit 11 performs a solution-seeking process for a target optimization problem stored in the problem storage unit 14. Specifically, the solution-seeking unit 11 first sets the initial value of the weighting factor α and the initial value of the solution-seeking time stored in the problem storage unit 14 for the optimization problem (step S1 in FIG. 2 ). The solution-seeking unit 11 then calculates a formulated model from an energy value equation for the unconstrained optimization problem for which the initial value of the weighting factor α has been set, and performs a solution-seeking process using the formulated model (step S2 in FIG. 2 ). At this time, the solution-seeking unit 11 repeatedly performs the solution-seeking process to search for a solution within the set solution-seeking time. That is, the solution-seeking unit 11 calculates an evaluation value, which is the energy value of the optimization problem, within the solution-seeking time, and repeatedly performs a process of transitioning the solution so that the evaluation value becomes a better value, i.e., a smaller value, until a preset termination condition is satisfied (step S3 in FIG. 2 ). For example, the solution-seeking unit 11 performs the solution-seeking process until a solution converges or until the total solution-seeking time reaches a set value.

[0017] As described above, when the solution-finding unit 11 searches for a solution within the solution-finding time, the initial value of the weighting coefficient α is set to a large value. Therefore, if the found solution does not satisfy the constraints, the evaluation value deteriorates, and the solution that does not satisfy the constraints is unlikely to be selected as the transition destination. In other words, in the solution-finding process when the initial value of the weighting coefficient α is set, it is expected that the transition to a solution that satisfies the constraints will be more likely. In this way, the weighting coefficient α can be said to be a parameter used in the solution transition process because it affects the solution transition process according to the evaluation value of the solution during the solution search.

[0018] The determination unit 12 determines whether the solution transitioned during the solution-finding process within the solution-finding time satisfies the constraint conditions (step S4 in FIG. 2). In other words, the determination unit 12 determines whether the solution transitioned during the short solution-finding time satisfies the constraint conditions and g(x) = 0. In this way, the solution transitioned during the solution-finding time can be said to be information (information to be determined) identified based on the solution transition. The determination unit 12 then notifies the setting unit 13 of the determination result.

[0019] The setting unit 13 changes the value of the weighting factor α in accordance with the determination result by the determination unit 12 and sets it as a new weighting factor α for the optimization problem. Specifically, if the solution that becomes the transition destination within the solution-finding time does not satisfy the constraints (No in step S4 of FIG. 2 ), the setting unit 13 changes the weighting factor α to a value greater than the initial value and sets it (step S5 of FIG. 2 ). Furthermore, if the solution that becomes the transition destination within the solution-finding time satisfies the constraints (Yes in step S4 of FIG. 2 ), the setting unit 13 changes the weighting factor α to a value smaller than the initial value and sets it (step S6 of FIG. 2 ). Furthermore, at this time, if the solution that becomes the transition destination within the solution-finding time satisfies the constraints, the setting unit 13 changes the solution-finding time to a time longer than the initial value and sets it (step S7 of FIG. 2 ). If the solution to be transitioned to within the solution-finding time does not satisfy the constraints, the setting unit 13 may not change the solution-finding time, or may change the solution-finding time to a time longer or shorter than the initial value, as described above (step S7 in FIG. 2).The setting unit 13 then notifies the solution-finding unit 11 of the changes to the weighting coefficient α and the solution-finding time.

[0020] As described above, upon receiving notification from the setting unit 13 of a change in the weighting factor α or the solution-finding time, the solution-finding unit 11 performs further solution processing for the optimization problem under the changed weighting factor α and solution-finding time (return to step S2 in FIG. 2 ). For example, as shown in FIG. 3 , if the solution satisfies the constraints within the solution-finding time set to the initial value, the solution-finding unit 11 subsequently sets the weighting factor α to a lower value and continues searching for a solution over a longer solution-finding time. In this way, by first setting a large weighting factor α and identifying a solution that satisfies the constraints through a short solution search, the weighting factor α is subsequently set to a smaller value to perform a long solution-finding process with high accuracy. As a result, the overall solution-finding processing time can be shortened.

[0021] In the above, we have illustrated a case where there is one constraint term in the energy value equation when a constrained optimization problem is converted into an unconstrained optimization problem. However, we will now explain an example of a method for finding a solution when there are multiple constraint terms, as in the following equation: ω objective function term + α constraint term 1 + β constraint term 2 + γ constraint term 3

[0022] Consider the case where there are three constraint terms, each with its own weighting coefficients α, β, and γ, as described above. ω is the weighting coefficient assigned to the objective function term. In this case, the weighting coefficients are determined one by one. Specifically, the following processes are performed: A1: Determine the balance between the two constraint term coefficients α and β. Search for an appropriate α by setting β = 1 and γ = ω = 0 (equivalent to an equation with only constraint term 1 and constraint term 2). A2: Add the remaining constraint terms one by one and search for the weighting coefficients of the constraint terms. For example, search for γ by setting β = 1 and ω = 0 depending on the result of α. A3: Repeat A2 above until the coefficients of all constraint terms are determined. A4: Search for the weighting coefficient ω of the objective function term. Note that in the searches A1-A3 above, if the solution satisfies the constraints in a short solution time, it is sufficient to determine the appropriate weighting coefficient; if the solution does not satisfy the constraints in a short solution time, it is sufficient to simply increase the weighting coefficient. Furthermore, in the search for weighting factors in A1 above, if the solution satisfies the constraint conditions, it is advisable to gradually increase the solution time while decreasing the weighting factors.

[0023] Second Embodiment Next, a second embodiment of the present disclosure will be described with reference to the drawings. Note that the drawings may be relevant to any of the embodiments.

[0024] The optimization processing device 10 in this embodiment is configured with one or more information processing devices each equipped with a calculation device and a storage device, as in the first embodiment, and includes a solution-finding unit 11, a determination unit 12, a setting unit 13, and a problem storage unit 14, as shown in FIG. 1 .

[0025] In this embodiment, the optimization problem solved by the optimization processing device 10 is assumed to be solved by simulated annealing. In simulated annealing, when searching for a solution, a transition is always made if the evaluation value of a neighboring solution is good (small), as indicated by arrow Y1 in Fig. 4 , but a transition can also occur probabilistically even if the evaluation value of a neighboring solution is bad (large), as indicated by arrow Y2 in Fig. 4 . The probability at this time is set by the value of the temperature parameter, and when the temperature parameter is high, the probability of transition to a solution with a bad evaluation value increases, and when the temperature parameter is low, the probability of transition to a solution with a bad evaluation value decreases.

[0026] When solving an optimization problem using the above-described algorithm, simulated annealing, in this embodiment, a high temperature parameter is first set as an initial value, and a solution is searched for within the set solution-finding time using this high temperature parameter. Furthermore, this embodiment is characterized in that the temperature parameter is changed depending on whether the transition of the solution during the solution search within the solution-finding time satisfies a predetermined condition. The configuration and operation of the optimization processing device 10 in this embodiment will be described in detail below.

[0027] As described above, the problem storage unit 14 stores information representing the optimization problem to be solved. The problem storage unit 14 also stores an initial value of a temperature parameter used in simulated annealing, which is an algorithm for solving the optimization problem, and an initial value of a solution time for executing the solution process for the optimization problem. The initial value of the temperature parameter is set to a high predetermined value that represents a high temperature, and the initial value of the solution time is set to, for example, a time that is significantly shorter than the time required to complete the solution of the optimization problem.

[0028] The solution-finding unit 11 performs a solution-finding process for a target optimization problem stored in the problem storage unit 14. Specifically, the solution-finding unit 11 first sets the initial values ​​of the temperature parameter and the initial values ​​of the solution-finding time stored in the problem storage unit 14 (step S11 in FIG. 5 ), and starts searching for a solution to the optimization problem by simulated annealing.

[0029] The solution-finding unit 11 first generates a random initial solution and calculates an energy value, which is an evaluation value of the initial solution (step S12 in FIG. 5). Next, the solution-finding unit 11 generates neighboring solutions and calculates the energy values ​​of the neighboring solutions (step S13 in FIG. 5). The solution-finding unit 11 then determines whether the neighboring solutions are acceptable, that is, whether the solution transitions to a neighboring solution (step S14 in FIG. 5). At this time, the solution-finding unit 11 performs the acceptance determination using a transition probability p calculated by the following formula: p=exp((e-e') / T), where e is the energy value before the transition, e' is the energy value after the transition, and T is a temperature parameter.

[0030] From the above equation for the transition probability p, if the post-transition energy value e' of the neighboring solution is lower, i.e., the difference between the pre-transition energy value e and the post-transition energy value e' is positive and the neighboring solution is a better solution, then the transition probability p is greater than or equal to 1, and a transition to the neighboring solution is guaranteed. On the other hand, if the post-transition energy value e' of the neighboring solution is higher, i.e., the difference between the pre-transition energy value e and the post-transition energy value e' is negative and the neighboring solution is a worse solution, then the transition probability p increases as the value of the temperature parameter T increases. In this case, since the temperature parameter set as the initial value is high, the transition probability p increases, and even if the neighboring solution is a bad solution, it is accepted as a bad solution with a high probability of transition. In this way, the temperature parameter is used when calculating the probability of determining whether or not to transition a solution depending on the solution evaluation, and affects the solution transition process depending on the solution evaluation value during solution search, so it can be said to be a parameter used in the solution transition process.

[0031] When searching for a solution within the set solution-finding time, the solution-finding unit 11 searches for a solution over a wide range because the temperature parameter is set to a high temperature and there is a high probability of transitioning to even a bad solution. If the solution-finding unit 11 determines that a nearby solution is acceptable (Yes in step S14 of FIG. 5), it transitions the solution to that nearby solution (step S15 of FIG. 5), and if it determines that a nearby solution is not acceptable (No in step S14 of FIG. 5), it does not transition the solution to that nearby solution.

[0032] The solution-finding unit 11 repeatedly performs a solution-finding process to search for a solution within a set solution-finding time. At this time, the solution-finding unit 11 ends the solution search when a preset termination condition is met (Yes in step S16 of FIG. 5). For example, the solution-finding unit 11 performs the solution-finding process until a solution converges or until the total solution-finding time reaches a set value.

[0033] If the termination condition is not satisfied in the solution search within the set solution-finding time (No in step S16 of FIG. 5 ), the determination unit 12 checks the solution acceptance status within the solution-finding time and updates the temperature parameter according to the acceptance status (step S17 of FIG. 5 ). Specifically, the determination unit 12 calculates the solution transition determination rate within the solution-finding time, i.e., the solution acceptance rate calculated by dividing the number of acceptance attempts by the number of acceptance judgments, and updates the temperature parameter according to the acceptance rate. In this embodiment, an allowable range (predetermined range) for the acceptance rate is set in advance. The allowable range includes a first threshold value, which is the upper limit, and a second threshold value, which is the lower limit lower than the upper limit. The determination unit 12 determines whether the solution acceptance rate within the solution-finding time is greater than the upper limit of the allowable range or less than the lower limit of the allowable range. Note that, because the acceptance rate within the solution-finding time is calculated by dividing the number of acceptance attempts by the number of acceptance judgments as described above, it can be said to be information (information to be determined) identified based on the solution transitions. Then, the determination unit 12 notifies the setting unit 13 of the determination result.

[0034] The setting unit 13 changes the value of the temperature parameter in accordance with the determination result by the determination unit 12 and sets it as a new temperature parameter to be used in simulated annealing. Specifically, if the acceptance rate within the solution-finding time is lower than the lower limit of the allowable range, the setting unit 13 changes the temperature parameter to a value larger than its initial value. Also, if the acceptance rate within the solution-finding time is higher than the upper limit, the setting unit 13 changes the temperature parameter to a value smaller than its initial value. Furthermore, at this time, if the acceptance rate within the solution-finding time is within the allowable range, which is equal to or lower than the upper limit and equal to or higher than the lower limit, the setting unit 13 changes the solution-finding time to a time longer than its initial value. Then, the setting unit 13 notifies the solution-finding unit 11 of the changes to the temperature parameter and the solution-finding time.

[0035] As described above, upon receiving notification of changes to the temperature parameter or solution-seeking time from the setting unit 13, the solution-seeking unit 11 performs a simulated annealing solution process for the optimization problem using the changed temperature parameter and solution-seeking time (return to step S13 in FIG. 5 ). For example, if the temperature parameter is changed to a larger value because the acceptance rate within the solution-seeking time is lower than the lower limit of the allowable range, the solution-seeking unit 11 will search for a solution using a higher temperature parameter, resulting in a wider range of solution search. On the other hand, if the temperature parameter is changed to a smaller value because the acceptance rate within the solution-seeking time is higher than the upper limit of the allowable range, the solution-seeking unit 11 will search for a solution using a lower temperature parameter, resulting in a more accurate solution search. On the other hand, if the acceptance rate within the solution-seeking time is within the allowable range, the solution-seeking unit 11 will perform a solution-seeking process using a longer solution-seeking time.

[0036] In this way, in this embodiment, a solution is first searched for at a high temperature within a short solution-finding time, the temperature parameters are appropriately set, and then a solution-finding process is performed with high accuracy over a long period of time using the appropriately set temperature parameters, thereby shortening the overall solution-finding process time.

[0037] In the above example, the temperature parameter is changed depending on whether the acceptance rate within the solution time is within the allowable range. However, it is also possible to set only either an upper limit or a lower limit, and change the temperature parameter depending on whether the acceptance rate is greater than the upper limit or less than the lower limit.

[0038] Although the above-described embodiment illustrates the case where a temperature parameter is changed when solving an optimization problem using simulated annealing, the present invention may also be applied to the case where a parameter is changed when solving an optimization problem using other algorithms. For example, the present invention may be applied to the case where a parameter is changed when solving an optimization problem using quantum annealing. Furthermore, the present invention may be applied to the case where a mutation probability (parameter) used in a genetic algorithm is changed when solving an optimization problem using the genetic algorithm.

[0039] [Application Example] Here, as a specific application example of the first embodiment, the optimization of the allocation of medical staff such as nurses and doctors will be described. Specifically, an application example of determining work shifts when the optimization processing device 10 has the configuration shown in FIG. 1 will be described. Note that this application example may also be applied to the configuration of the first embodiment. Below, the flow of determining work shifts, including input by medical staff, will be described.

[0040] First, each medical worker who is the subject of a work shift uses a terminal device to log in to the AI ​​system (work shift management system) realized by the optimization processing device 10 and input their desired work schedule. The AI ​​system then considers the desired work schedule and other factors input by each medical worker (nurse or doctor) as constraints and generates an optimal work shift.

[0041] In this case, the AI ​​system defines each medical worker's work style for each date (e.g., whether or not they need to work) as a decision variable, sets up an optimization problem in which the desired work schedules and other factors, as well as the compatibility and interpersonal relationships between the medical workers, are defined as constraints, and generates work shifts based on a final output solution of the optimization problem. The work shifts generated by the AI ​​system are then notified to each medical worker by being displayed on a terminal device or the like. Each medical worker then checks the generated work shifts, and if there are no problems with the generated work shifts, logs in to the AI ​​system and inputs approval of the work shifts via their terminal devices. On the other hand, if each medical worker determines that there are problems with the generated work shifts, they log in to the AI ​​system and input a request for correction via their terminal devices. When the AI ​​system detects a correction request from a medical worker, it adds new constraints based on the correction request and corrects the optimization problem. The addition of the constraints described above may be performed based on user input from the AI ​​system administrator who confirmed the correction request. The AI ​​system then obtains a final output solution to the corrected optimization problem and generates work shifts indicated by the final output solution. Then, when a work shift is generated that is approved by all eligible medical personnel, the AI ​​system determines that work shift as the final work shift.

[0042] The optimization processing device of this embodiment is not limited to the allocation of medical personnel and can be applied to mathematical optimization problems in various fields. For example, this optimization processing device can be used in a wide range of fields, such as improving the efficiency of production lines in the manufacturing industry, optimizing delivery routes in the logistics industry, and optimizing inventory management in the retail industry. This can lead to efficient resource utilization, cost reduction, and improved customer satisfaction in each industry. In the field of education, applications to improving the quality of education, such as optimizing the allocation of teachers and teaching materials and class schedules, can be considered. This makes it possible to provide educational programs customized to the needs of each student. Furthermore, applications to improving the quality of public services, such as optimizing transfer times in public transportation and optimizing the placement of public facilities in urban planning, can also be considered. This can improve the convenience of citizens' lives and contribute to the realization of a more comfortable urban environment.

[0043] Third Embodiment Next, a third embodiment of the present disclosure will be described with reference to the drawings. This embodiment shows an outline of the configuration of the optimization processing device described in the above-mentioned embodiment. Note that Figures 6 and 7 are diagrams for explaining the configuration, and these drawings may be relevant to any of the embodiments.

[0044] First, the hardware configuration of the optimization processing device 100 will be described with reference to Fig. 6. The optimization processing device 100 is configured as a general information processing device, and is equipped with the following hardware configuration, for example: CPU (Central Processing Unit) 101 (arithmetic unit); ROM (Read Only Memory) 102 (storage device); RAM (Random Access Memory) 103 (storage device); programs 104 loaded into RAM 103; storage device 105 storing programs 104; drive device 106 for reading and writing data from and to a storage medium 110 external to the information processing device; communication interface 107 for connecting to a communication network 111 external to the information processing device; input / output interface 108 for inputting and outputting data; and bus 109 for connecting the various components.

[0045] 6 shows an example of the hardware configuration of the information processing device that is the optimization processing device 100, and the hardware configuration of the information processing device is not limited to the above-described case. For example, the information processing device may be configured with only a part of the above-described configuration, such as not including the drive device 106. Furthermore, instead of the above-described CPU, the information processing device may use a GPU (Graphics Processing Unit), a DSP (Digital Signal Processor), an MPU (Micro Processing Unit), an FPU (Floating Point Number Processing Unit), a PPU (Physics Processing Unit), a TPU (Tensor Processing Unit), a quantum processor, a microcontroller, or a combination thereof.

[0046] The optimization processing device 100 can be equipped with a solution-finding unit 121, a determination unit 122, and a setting unit 123 shown in FIG. 7 by having the CPU 101 acquire and execute the program group 104. The program group 104 is stored in advance in the storage device 105 or the ROM 102, for example, and is loaded into the RAM 103 and executed by the CPU 101 as needed. The program group 104 may be supplied to the CPU 101 via the communication network 111, or may be stored in advance in the storage medium 110, with the drive device 106 reading out the programs and supplying them to the CPU 101. However, the solution-finding unit 121, the determination unit 122, and the setting unit 123 described above may be constructed using dedicated electronic circuits for realizing such means.

[0047] The solution-finding unit 121 performs a solution-finding process for an optimization problem, within a preset solution-finding time, for which parameters used in a solution transition process according to an evaluation of the solution during a solution search have been set. The determination unit 122 determines whether preset determination target information identified based on the transition of the solution in the solution-finding process within the solution-finding time satisfies a preset condition. The setting unit 123 changes the parameters and sets them for the optimization problem depending on whether the determination target information within the solution-finding time satisfies the condition.

[0048] With the above-described configuration, the present disclosure first performs a solution process for an optimization problem for which parameters have been set within a preset solution time, determines whether or not the evaluation target information satisfies the conditions based on the transition of the solution within the solution time, and changes the parameters according to the evaluation result. This allows appropriate parameters to be set in a short time, and then performs further solution process, thereby shortening the overall solution processing time.

[0049] In addition, at least one or more of the functions of the above-mentioned solution-finding unit 121, judgment unit 122, and setting unit 123 may be executed by an information processing device installed and connected anywhere on the network, that is, they may be executed by so-called cloud computing.

[0050] The above-described program can be stored and supplied to a computer using various types of non-transitory computer-readable media. Non-transitory computer-readable media include various types of tangible storage media. Examples of non-transitory computer-readable media include magnetic recording media (e.g., flexible disks, magnetic tapes, hard disk drives), magneto-optical recording media (e.g., magneto-optical disks), CD-ROMs (Read Only Memory), CD-Rs, CD-RWs, and semiconductor memories (e.g., mask ROMs, PROMs (Programmable ROMs), EPROMs (Erasable PROMs), flash ROMs, and RAMs (Random Access Memory)). The program can also be supplied to a computer by various types of transitory computer-readable media. Examples of transitory computer-readable media include electrical signals, optical signals, and electromagnetic waves. The transitory computer-readable media can be supplied to a computer via wired communication paths such as electric wires and optical fibers, or via wireless communication paths.

[0051] Although the present disclosure has been described above with reference to the above-described embodiments, the present disclosure is not limited to the above-described embodiments. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure. Furthermore, each of the above-described embodiments can be combined with other embodiments as appropriate.

[0052] <Supplementary Notes> Some or all of the above embodiments can also be described as in the following supplementary notes. Below, an outline of the configurations of an optimization processing device, an optimization processing method, and a program according to the present disclosure will be described. However, the present disclosure is not limited to the following configurations. (Supplementary Note 1) An optimization processing device comprising: a solution-finding unit that performs solution-finding processing within a preset solution-finding time for an optimization problem for which parameters used in solution transition processing according to solution evaluation during solution search are set; a determination unit that determines whether preset judgment target information identified based on solution transition during the solution-finding processing within the solution-finding time satisfies a preset condition; and a setting unit that changes the parameter and sets it for the optimization problem according to whether the judgment target information within the solution-finding time satisfies the condition. (Supplementary Note 2) The optimization processing device according to Supplementary Note 1, wherein the parameter is set so that a larger value worsens the evaluation of the solution; and the setting unit changes the parameter to a larger value when the judgment target information within the solution-finding time does not satisfy the condition. (Supplementary Note 3) The optimization processing device according to Supplementary Note 1, wherein the parameter is set so that a larger value worsens the evaluation of the solution, and the setting unit changes and sets the parameter to a smaller value if the information to be judged in the solution search process within the solution search time satisfies the condition. (Supplementary Note 4) The optimization processing device according to Supplementary Note 3, wherein the setting unit changes and sets the solution search time to a longer time if the information to be judged in the solution search process satisfies the condition within the solution search time, and the solution search unit performs the solution search process within the changed solution search time for the optimization problem for which the changed parameter is set. (Supplementary Note 5) The optimization processing device according to Supplementary Note 1, wherein a constraint condition for a solution is set for the optimization problem, and the parameter is set as a weight of an evaluation value corresponding to whether a solution satisfies the constraint condition when searching for a solution, and the determination unit determines whether the information to be judged, which is a solution to be transitioned to in the solution search process within the solution search time, satisfies the constraint condition.(Supplementary Note 6) The optimization processing device according to Supplementary Note 5, wherein the parameter is set so that the larger the value is, the worse the evaluation value becomes when a solution does not satisfy the constraint condition during a solution search, and the setting unit changes and sets the parameter to a larger value when the solution to be transitioned to does not satisfy the constraint condition. (Supplementary Note 7) The optimization processing device according to Supplementary Note 5, wherein the parameter is set so that the larger the value is, the worse the evaluation value becomes when a solution does not satisfy the constraint condition during a solution search, and the setting unit changes and sets the parameter to a smaller value when the solution to be transitioned to satisfies the constraint condition. (Supplementary Note 8) The optimization processing device according to Supplementary Note 7, wherein the setting unit changes and sets the solution finding time to a longer time when the solution to be transitioned to satisfies the constraint condition, and the solution finding unit performs the solution finding process for the optimization problem for which the changed parameter is set within the changed solution finding time. (Supplementary Note 9) The optimization processing device according to Supplementary Note 1, wherein the parameter is set to be used when determining whether or not to transition a solution in accordance with an evaluation of the solution during a solution search, and the determination unit determines whether or not the determination target information, which represents a transition status of the solution in the solution search process within the solution search time, satisfies the condition. (Supplementary Note 10) The optimization processing device according to Supplementary Note 9, wherein the parameter is set so that the larger the value of the parameter, the higher the probability of the solution transitioning in a direction in which the evaluation of the solution deteriorates during a solution search, the determination unit determines whether or not the determination target information, which represents a transition determination rate of the solution in the solution search process within the solution search time, is higher than a threshold, and the setting unit changes and sets the parameter to a larger value when the transition determination rate during the solution search time is smaller than the threshold.(Supplementary Note 11) The optimization processing device according to Supplementary Note 9, wherein the parameter is set so that the larger the value thereof, the higher the probability of a solution transition in a direction that worsens the evaluation of the solution during a solution search, the determination unit determines whether the determination target information representing a transition determination rate of a solution in the solution search process within the solution search time is higher than a threshold, and the setting unit changes and sets the parameter to a smaller value when the transition determination rate during the solution search time is higher. (Supplementary Note 12) The optimization processing device according to Supplementary Note 9, wherein the parameter is set so that the larger the value thereof, the higher the probability of a solution transition during a solution search, the determination unit determines whether the determination target information representing a transition determination rate of a solution in the solution search process within the solution search time is within a predetermined range, and the setting unit changes and sets the solution search time to a longer time when the transition determination rate during the solution search time is within the predetermined range, and the solution search unit performs the solution search process within the changed solution search time. (Supplementary Note 13) An optimization processing method comprising: performing a solution processing within a predetermined solution time for an optimization problem, for which parameters used in solution transition processing according to solution evaluation during a solution search are set; determining whether predetermined judgment target information identified based on solution transitions in the solution processing within the solution search time satisfies a predetermined condition; and changing the parameter for the optimization problem according to whether the judgment target information within the solution search time satisfies the condition. (Supplementary Note 14) An optimization processing method according to Supplementary Note 13, wherein the parameter is set so that a larger value of the parameter worsens the evaluation of the solution; and if the judgment target information within the solution search time does not satisfy the condition, changing the parameter to a larger value and setting it. (Supplementary Note 15) An optimization processing method according to Supplementary Note 13, wherein the parameter is set so that a larger value of the parameter worsens the evaluation of the solution; and if the judgment target information in the solution search processing within the solution search time satisfies the condition, changing the parameter to a smaller value and setting it.(Supplementary Note 16) The optimization processing method according to Supplementary Note 13, wherein a constraint condition for a solution is set for the optimization problem, the parameter is set as a weight of an evaluation value according to whether or not a solution satisfies the constraint condition during a solution search, and it is determined whether or not the determination target information, which is a destination solution obtained in the solution search process within the solution search time, satisfies the constraint condition. (Supplementary Note 17) The optimization processing method according to Supplementary Note 16, wherein the parameter is set so that the evaluation value becomes worse when a solution does not satisfy the constraint condition during a solution search as the value increases, and when the destination solution does not satisfy the constraint condition, the parameter is changed to a larger value and set. (Supplementary Note 17.1) The optimization processing method according to Supplementary Note 17, wherein the parameter is set so that the evaluation value becomes worse when a solution does not satisfy the constraint condition during a solution search as the value increases, and when the destination solution satisfies the constraint condition, the parameter is changed to a smaller value and set. (Supplementary Note 17.2) The optimization processing method according to Supplementary Note 17.1, wherein if the solution to be transitioned satisfies the constraint, the solution time is changed and set to a longer time, and the solution search process is performed within the changed solution time for the optimization problem for which the changed parameters have been set. (Supplementary Note 18) The optimization processing method according to Supplementary Note 13, wherein the parameters are set to be used when determining whether or not to transition a solution depending on an evaluation of the solution when searching for a solution, and it is determined whether the judgment target information representing a solution transition status in the solution search process within the solution time satisfies the condition.(Supplementary Note 19) An optimization processing method according to Supplementary Note 18, wherein the parameter is set so that the larger the value thereof, the higher the probability of a solution transition in a direction that worsens the evaluation of the solution during a solution search, determining whether the judgment target information representing a transition determination rate of the solution during the solution search process within the solution search time is higher than a threshold, and if the transition determination rate during the solution search time is lower than the threshold, changing and setting the parameter to a larger value. (Supplementary Note 19.1) An optimization processing method according to Supplementary Note 18, wherein the parameter is set so that the larger the value thereof, the higher the probability of a solution transition in a direction that worsens the evaluation of the solution during a solution search, determining whether the judgment target information representing a transition determination rate of the solution during the solution search process within the solution search time is higher than a threshold, and if the transition determination rate during the solution search time is higher than the threshold, changing and setting the parameter to a smaller value. (Supplementary Note 19.2) An optimization processing method according to Supplementary Note 18, wherein the parameter is set so that the larger the value, the higher the solution transition probability during a solution search, determining whether or not the judgment target information indicating a solution transition determination rate in the solution search process within the solution search time is within a predetermined range, changing and setting the solution search time to a longer time if the transition determination rate during the solution search time is within the predetermined range, and performing the solution search process within the changed solution search time. (Supplementary Note 20) A computer-readable storage medium storing a program that causes a computer to execute processes of: performing a solution search process within a predetermined solution search time for an optimization problem for which parameters used in solution transition processing according to solution evaluation during a solution search are set, determining whether or not predetermined judgment target information specified based on solution transitions during the solution search process within the solution search time satisfies a predetermined condition, and changing the parameter depending on whether the judgment target information during the solution search time satisfies the condition, and setting the parameter for the optimization problem.

[0053] REFERENCE SIGNS LIST 10 Optimization processing device 11 Solution-finding unit 12 Determination unit 13 Setting unit 14 Problem storage unit 100 Optimization processing device 101 CPU 102 ROM 103 RAM 104 Program group 105 Storage device 106 Drive device 107 Communication interface 108 Input / output interface 109 Bus 110 Storage medium 111 Communication network 121 Solution-finding unit 122 Determination unit 123 Setting unit

Claims

1. For an optimization problem with parameters set for solution transition processing according to the evaluation of a solution during solution search, a solving unit that performs a solving process within a preset solving time, a determination unit that determines whether preset determination target information specified based on solution transition in the solving process within the solving time satisfies preset conditions, and a setting unit that changes the parameters according to whether the determination target information within the solving time satisfies the conditions and sets them for the optimization problem. An optimization processing device comprising:

2. The optimization processing device according to claim 1, wherein the parameter is set such that the larger the value, the worse the evaluation of the solution, and the setting unit changes and sets the parameter to a larger value when the determination target information within the solving time does not satisfy the conditions. An optimization processing device.

3. The optimization processing device according to claim 1, wherein the parameter is set such that the larger the value, the worse the evaluation of the solution, and the setting unit changes and sets the parameter to a smaller value when the determination target information in the solving process within the solving time satisfies the conditions. An optimization processing device.

4. The optimization processing device according to claim 3, wherein the setting unit changes and sets the solving time to a longer time when the determination target information within the solving time satisfies the conditions, and the solving unit performs the solving process within the changed solving time for the optimization problem with the changed parameter set. An optimization processing device.

5. The optimization processing device according to claim 1, wherein the optimization problem has solution constraint conditions set, the parameter is set as a weight of an evaluation value according to whether a solution satisfies the constraint conditions during solution search, and the determination unit determines whether the determination target information, which is the solution to be the transition destination obtained in the solving process within the solving time, satisfies the constraint conditions. An optimization processing device.

6. The optimization processing apparatus according to claim 5, wherein the parameter is set such that the larger the value, the worse the evaluation value when the solution does not satisfy the constraint condition during solution search, and the setting unit changes and sets the parameter to a larger value when the solution that is the transition destination does not satisfy the constraint condition. Optimization processing apparatus.

7. The optimization processing apparatus according to claim 5, wherein the parameter is set such that the larger the value, the worse the evaluation value when the solution does not satisfy the constraint condition during solution search, and the setting unit changes and sets the parameter to a smaller value when the solution that is the transition destination satisfies the constraint condition. Optimization processing apparatus.

8. The optimization processing apparatus according to claim 7, wherein the setting unit changes and sets the solution time to a longer time when the solution that is the transition destination satisfies the constraint condition, and the solution unit performs the solution processing within the changed solution time for the optimization problem with the changed parameter set. Optimization processing apparatus.

9. The optimization processing apparatus according to claim 1, wherein the parameter is set to be used when determining whether to transition the solution according to the evaluation of the solution during solution search, and the determination unit determines whether the determination target information indicating the solution transition status in the solution processing within the solution time satisfies the condition. Optimization processing apparatus.

10. The optimization processing apparatus according to claim 9, wherein the parameter is set such that the larger the value, the higher the probability of solution transition in the direction of deterioration of solution evaluation during solution search, and the determination unit determines whether the determination target information indicating the solution transition determination rate in the solution processing within the solution time is higher than a threshold value, and the setting unit changes and sets the parameter to a larger value when the transition determination rate within the solution time is smaller than the threshold value. Optimization processing apparatus.

11. The optimization processing apparatus according to claim 9, wherein the parameter is set such that the higher the value, the higher the transition probability of the solution in the direction of deterioration of the evaluation of the solution during the search for the solution, the determination unit determines whether the determination target information representing the transition determination rate of the solution in the solution search process within the solution search time is higher than a threshold value, and the setting unit, when the transition determination rate within the solution search time is greater than the threshold value, changes and sets the parameter to a smaller value. Optimization processing apparatus.

12. The optimization processing apparatus according to claim 9, wherein the parameter is set such that the higher the value, the higher the transition probability of the solution during the search for the solution, the determination unit determines whether the determination target information representing the transition determination rate of the solution in the solution search process within the solution search time is a value within a predetermined range, and the setting unit, when the transition determination rate within the solution search time is a value within the predetermined range, changes and sets the solution search time to a longer time, and the solution search unit performs the solution search process within the changed solution search time. Optimization processing apparatus.

13. For an optimization problem with a parameter set for the transition process of the solution according to the evaluation of the solution during the search for the solution, perform a solution search process within a preset solution search time, determine whether preset determination target information specified based on the transition of the solution in the solution search process within the solution search time satisfies a preset condition, and change the parameter according to whether the determination target information within the solution search time satisfies the condition and set it for the optimization problem. Optimization processing method.

14. The optimization processing method according to claim 13, wherein the parameter is set such that the higher the value, the more the evaluation of the solution deteriorates, and when the determination target information within the solution search time does not satisfy the condition, the parameter is changed and set to a larger value. Optimization processing method.

15. The optimization processing method according to claim 13, wherein the parameter is set such that the larger the value, the worse the evaluation of the solution, and when the determination target information in the solution solving process within the solution solving time satisfies the condition, the parameter is changed to a smaller value and set.

16. The optimization processing method according to claim 13, wherein in the optimization problem, constraint conditions of the solution are set, the parameter is set as a weight of an evaluation value according to whether the solution satisfies the constraint conditions during the search for the solution, and it is determined whether the determination target information, which is the solution serving as the transition destination obtained in the solution solving process within the solution solving time, satisfies the constraint conditions.

17. The optimization processing method according to claim 16, wherein the parameter is set such that the larger the value, the worse the evaluation value when the solution does not satisfy the constraint conditions during the search for the solution, and when the solution serving as the transition destination does not satisfy the constraint conditions, the parameter is changed to a larger value and set.

18. The optimization processing method according to claim 13, wherein the parameter is set to be used when determining whether to transition the solution according to the evaluation of the solution during the search for the solution, and it is determined whether the determination target information representing the solution transition status in the solution solving process within the solution solving time satisfies the condition.

19. The optimization processing method according to claim 18, wherein the parameter is set such that the larger the value, the higher the probability of solution transition in the direction of deteriorating solution evaluation during the search for the solution, it is determined whether the determination target information representing the solution transition determination rate in the solution solving process within the solution solving time is higher than a threshold value, and when the transition determination rate within the solution solving time is smaller than the threshold value, the parameter is changed to a larger value and set.

20. For an optimization problem with parameters set for solution transition processing according to the evaluation of a solution during solution search, perform a solution search process within a preset solution search time, determine whether preset determination target information specified based on the solution transition in the solution search process within the solution search time satisfies preset conditions, and change the parameters according to whether the determination target information within the solution search time satisfies the conditions and set them for the optimization problem. A computer-readable storage medium storing a program for causing a computer to execute the process.

Images