Abstract: A non-transitory, computer-readable recording medium stores therein a reinforcement learning program that uses a value function and causes a computer to execute a process comprising: estimating first coefficients of the value function represented in a quadratic form of inputs at times in the past than a present time and outputs at the present time and the times in the past, the first coefficients being estimated based on inputs at the times in the past, the outputs at the present time and the times in the past, and costs or rewards that corresponds to the inputs at the times in the past; and determining second coefficients that defines a control law, based on the value function that uses the estimated first coefficients and determining input values at times after estimation of the first coefficients.

Abstract: A non-transitory, computer-readable recording medium stores a program of reinforcement learning by a state-value function. The program causes a computer to execute a process including calculating a temporal difference (TD) error based on an estimated state-value function, the TD error being calculated by giving a perturbation to each component of a feedback coefficient matrix that provides a policy; calculating based on the TD error and the perturbation, an estimated gradient function matrix acquired by estimating a gradient function matrix of the state-value function with respect to the feedback coefficient matrix for a state of a controlled object, when state variation of the controlled object in the reinforcement learning is described by a linear difference equation and an immediate cost or an immediate reward of the controlled object is described in a quadratic form of the state and an input; and updating the feedback coefficient matrix using the estimated gradient function matrix.

Abstract: A simulation method is performed by a computer by using an agent disposed in a virtual space. The method includes: controlling the agent so that the agent behaves in the virtual space under influence of each of a plurality of signs disposed in the virtual space; and determining the influence of each of the plurality of signs on the agent in accordance with an attribute relating to a display mode of the sign and an attribute of the agent.

Abstract: A non-transitory computer-readable recording medium has stored therein a program that causes a computer to execute a process including, classifying a plurality of data items included in outputted data output by carrying out a simulation using a plurality of agents into a plurality of groups based on difference between model elements of data output sources of the respective data items in the agents, converting the data items included in a group having a plurality of data items out of the groups into a smaller number of data items than a number of data items included in the group based on a predetermined rule, and identifying a combination of the data items having an appearance tendency equal to or higher than a predetermined value in the outputted data resulting from the converting.

Abstract: A simulation method is performed by a computer by using an agent disposed in a virtual space. The method includes: controlling the agent so that the agent behaves in the virtual space under influence of each of a plurality of signs disposed in the virtual space; and determining the influence of each of the plurality of signs on the agent in accordance with an attribute relating to a display mode of the sign and an attribute of the agent.

Abstract: A non-transitory, computer-readable recording medium stores a program of reinforcement learning by a state-value function. The program causes a computer to execute a process including calculating a TD error based on an estimated state-value function, the TD error being calculated by giving a perturbation to each component of a feedback coefficient matrix that provides a policy; calculating based on the TD error and the perturbation, an estimated gradient function matrix acquired by estimating a gradient function matrix of the state-value function with respect to the feedback coefficient matrix for a state of a controlled object, when state variation of the controlled object in the reinforcement learning is described by a linear difference equation and an immediate cost or an immediate reward of the controlled object is described in a quadratic form of the state and an input; and updating the feedback coefficient matrix using the estimated gradient function matrix.

Abstract: A non-transitory, computer-readable recording medium stores therein a reinforcement learning program that uses a value function and causes a computer to execute a process comprising: estimating first coefficients of the value function represented in a quadratic form of inputs at times in the past than a present time and outputs at the present time and the times in the past, the first coefficients being estimated based on inputs at the times in the past, the outputs at the present time and the times in the past, and costs or rewards that corresponds to the inputs at the times in the past; and determining second coefficients that defines a control law, based on the value function that uses the estimated first coefficients and determining input values at times after estimation of the first coefficients.

Abstract: A non-transitory computer readable recording medium has stored therein a simulation program that causes a computer to execute a process including arranging an agent in a virtual space that includes one or a plurality of places where guide information is set, the agent having perception information and behaving according to the perception information in the virtual space; updating the perception information of the agent according to guide information that is provided according to the position of the agent in the virtual space; and deteriorating the perception information, degree of the deteriorating being determined on the basis of at least any one of a behavior of the agent and an attribute of the agent.

Abstract: A planning method for planning to order a product, the planning method being executed by a computer, the planning method includes: generating a predicted value of a demand of the product at each plurality of prices based on a selling price and a number of sales in association with each other for each past sale date; calculating an order quantity that yields a highest profit, for each of the plurality of prices, using the predicted value of the demand of the product at the plurality of prices; storing, for the plurality of prices, the calculated order quantity and a profit in association with each other into a memory; and identifying a combination of a price and an order quantity that yields the highest profit, with reference to the memory.

Abstract: A non-transitory computer-readable recording medium stores an operation planning program that causes a computer to execute a process including: for a set including elements that are users, for which a plan of operation including ride sharing is to be generated, performing ordering of the elements of the set by using indices indicating highness of possibilities that subadditivity is fulfilled, determining whether a combination of the elements in descending order of the ordering fulfills the subadditivity, the number of the elements in the combination being equal to or less than a predetermined number, and partitioning the set into subsets, for which the ride sharing is to be operated, by adding the combination of elements fulfilling the subadditivity to the subsets; and generating the plan of operation by using the partitioned subsets.

Abstract: A non-transitory computer-readable recording medium having stored therein a simulation program, for simulating a movement of an intruder in a security zone by using an intruder agent in a model corresponding to the security zone, that causes a processor to execute a process includes calculating a movement route of the intruder agent in the model on the basis of identified information of the intruder agent, wherein the identified information of the intruder agent includes information of security guard deployment in the model and information of past security guard deployment in the model.

Abstract: A control server according to an embodiment sorts a plurality of notebook PCs into a plurality of groups so that the total value of the remaining energy is a value similar to the total value of the remaining energy of the rechargeable batteries of a plurality of notebook PCs included in a different group. The control server according to the embodiment performs local search individually on the sorted groups, and generates a control plan for the individual notebook PCs.

Abstract: A plan determination method of which process is executed by a computer, the process includes receiving lead time for each raw material for a product, the lead time being indicative of time interval between a time at which an order of the each raw material is ordered and a time of arrival of the each raw material; and calculating an order quantity of the each raw material and a production quantity of the product which cause a cost relating to manufacturing of the product to be minimized, by using the received lead time for the each raw material and a predicted demand quantity of the product.

Abstract: The disclosed method includes: generating, from model expressions, each representing a relationship between input parameters and output evaluation indicators and a range of residuals for each model expression, at least one of a first problem to calculate a region that can be feasible with the model expressions, value ranges of the input parameters and the range of the residuals for each model expression and a second problem to calculate a region that is always feasible with the model expressions, the value ranges of the input parameters and the influence of the aforementioned range; calculating a feasible region(s) for a generated problem(s) to obtain data of the feasible region(s); and generating visualization data of the feasible region(s) to output the visualization data.

Abstract: A planning method for planning to order a product, the planning method being executed by a computer, the planning method includes: generating a predicted value of a demand of the product at each plurality of prices based on a selling price and a number of sales in association with each other for each past sale date; calculating an order quantity that yields a highest profit, for each of the plurality of prices, using the predicted value of the demand of the product at the plurality of prices; storing, for the plurality of prices, the calculated order quantity and a profit in association with each other into a memory; and identifying a combination of a price and an order quantity that yields the highest profit, with reference to the memory.

Abstract: A system includes: circuitry configured to receive a condition regarding a constraint condition of a product, acquire past requirement values for the product, predict, for each of a plurality of periods, requirement value for the product by calculating the requirement value for each of the plurality of periods based on the acquired past requirement values, generate, based on the predicted requirement value for each of the plurality of periods, a probability distribution of the constraint condition for each of a plurality of requested arrangements each of which indicates requested quantities of the product for each of the plurality of periods, and output at least one of the plurality of requested arrangements, based on the generated probability distribution and the received condition regarding the constraint condition.

Abstract: In this disclosure, equations to be solved in the model predictive control are transformed by using an off-line algebraic simplification method into a matrix operational expression representing a product of a coefficient matrix and a vector regarding solution inputs within a control horizon is equal to a function vector regarding target values of output states and the output states. The size of the coefficient matrix is reduced compared with the conventional matrix. Then, the matrix operational expression is solved in an online plant control apparatus with present output states and present target values of the output stats of a plant to be controlled, by the direct method, to output the solution to the plant.

Abstract: A parameter determination method is disclosed. Information of specification of the output is received. A first circuit constant and a second circuit constant to set in elements forming an equivalent circuit of the predetermined circuit is received. A first range of a plurality of the parameters which are to be set in a compensator that compensates the output is specified based on the information of the specification and the first circuit constant. A second range of a plurality of parameters which are to be set in the compensator is specified based on the information of the specification and the second circuit constant. At least one of a parameter included in both the first range and the second range.

Abstract: An order quantity determination method includes: accepting lead time from product order to arrival; calculating a stock quantity of the product by a processor based on an arrival quantity of the product and a demand forecast value of the product, the arrival quantity of the product is calculated based on the accepted lead time and order time of the product; and calculating an order quantity of the product by a processor based on a cost for holding the calculated stock quantity of the product, a price of the product, and the demand forecast value of the product.

Abstract: In this disclosure, equations to be solved in the model predictive control are transformed by using an off-line algebraic simplification method into a matrix operational expression representing a product of a coefficient matrix and a vector regarding solution inputs within a control horizon is equal to a function vector regarding target values of output states and the output states. The size of the coefficient matrix is reduced compared with the conventional matrix. Then, the matrix operational expression is solved in an online plant control apparatus with present output states and present target values of the output stats of a plant to be controlled, by the direct method, to output the solution to the plant.