Optimization device

Claim 2

An optimization device comprising: means for inputting an objective function which includes a parameter to be optimized and is an object for searching an optimal solution, a required precision indicating a precision required in searching the optimal solution and a search region for searching the optimal solution for the objective function to make said objective function into a convex function; and means for inputting said objective function made into the convex function to detect a search start point for starting the search of the optimal solution from said search region of the optimal solution.

Claim 3

An optimization device according to claim 1, wherein said means for making said objective function into the convex function comprising; means for having a very large value only at a point where a solution of said objective function is equal to said required precision and for making a prescribed function having a sufficiently small value at other points, and means for giving a diffusing operation to said prescribed function to make the convex objective function.

Claim 4

An optimization device according to claim 1, wherein said means for detecting the search start point comprising; means for determining a maximum value in said search region from the solutions of said convex objective function, means for detecting the position of a point having the maximum value of said convex objective function, and means, when said determined convex objective function value is equal to or above a prescribed threshold value and the point having said maximum value is within the search region, for outputting said point as a search start point.

Claim 5

An optimization device comprising: means for inputting an objective function which includes a parameter to be optimized and is an object for searching an optimal solution, a required precision indicating a precision required in searching the optimal solution and a search region for searching the optimal solution for the objective function to make said objective function into a convex function; means for inputting said convex objective function to detect a search start point for starting the search of the optimal solution from said search region of the optimal solution; and means for detecting the optimal solution based on the detected search start point.

Claim 6

An optimization device comprising: means for inputting an objective function which includes a parameter to be optimized and is an object for searching an optimal solution, a required precision indicating a precision required in searching the optimal solution and a search region for searching the optimal solution for the objective function to make said objective function into a convex function; means for inputting said convex objective function to detect a search start point for starting the search of the optimal solution from said search region of the optimal solution; and means for detecting the optimal solution based on the detected search start point.

Claim 7

An optimization device according to claim 4, wherein said means for making said objective function into the convex function comprising; means for having a very large value only at a point where a solution of said objective function is equal to said required precision and for making a prescribed function having a sufficiently small value at other points, and means for giving a diffusing operation to said prescribed function to make the convex objective function.

Claim 8

An optimization device according to claim 4, wherein said means for detecting the search start point comprising; means for determining a maximum value in said search region from the solutions of said convex objective function, means for detecting the position of a point having the maximum value of said convex objective function, and means, when said determined convex objective function value is equal to or above a prescribed threshold value and the point having said maximum value is within the search region, for outputting a point on said objective function corresponding to said point as a search start point.

Claim 9

An optimization device according to claim 4, wherein said means for detecting the optimal solution comprising; means for inputting said objective function, for starting to calculate a value of said objective function from said search start point, and for determining a maximum value or a minimum value in said search region from the solutions of said objective function, means for detecting the position of a point having the maximum value or the minimum value of said objective function, and means, when the point having said maximum value or a minimum value is within the search region, for outputting said point as an optimal solution.

Claim 10

An optimization device according to claim 4, wherein said means for making said objective function into the convex function comprising; means for having a very large value only at a point where a solution of said objective function is equal to said required precision and for making a prescribed function having a sufficiently small value at other points, and means for giving a diffusing operation to said prescribed function to make the convex objective function, said means for detecting the search start point comprising; means for determining a maximum value in said search region from the solutions of said convex objective function, means for detecting the position of a point having the maximum value of said convex objective function, and means, when said determined convex objective function value is equal to or above a prescribed threshold value and the point having said maximum value is within the search region, for outputting a point on said objective function corresponding to said point as a search start point, and said means for detecting the optimal solution comprising; means for inputting said objective function, for starting to calculate a value of said objective function from said search start point, and for determining a maximum value or a minimum value in said search region from the solutions of said objective function, means for detecting the position of a point having the maximum value or the minimum value of said objective function, and means, when the point having said maximum value or a minimum value is within the search region, for outputting said point as an optimal solution.

Claim 11

An optimization device comprising: means for inputting an objective function which includes a parameter to be optimized and is an object for searching an optimal solution, a required precision indicating a precision required in searching the optimal solution and a search region for searching the optimal solution for the objective function to make said objective function into a convex function; means for inputting said convex objective function to detect a search start point for starting the search of the optimal solution from said search region of the optimal solution; and means for detecting a global optimum based on the detected search start point.

Claim 12

An optimization device comprising: means for inputting an objective function which includes a parameter to be optimized and is an object for searching an optimal solution, a required precision indicating a precision required in searching the optimal solution and a search region for searching the optimal solution for the objective function to make said objective function into a convex function; means for inputting said convex objective function to detect a search start point for starting the search of the optimal solution from said search region of the optimal solution; and means for detecting a global optimum based on the detected search start point.

Claim 13

An optimization device according to claim 9, wherein said means for making said objective function into the convex function comprising; means for having a very large value only at a point where a solution of said objective function is equal to said required precision and for making a prescribed function having a sufficiently small value at other points, and means for giving a diffusing operation to said prescribed function to make the convex objective function.

Claim 14

An optimization device according to claim 9, wherein said means for detecting the search start point comprising; means for determining a maximum value in said search region from the solutions of said convex objective function, means for detecting the position of a point having the maximum value of said convex objective function, and means, when said determined convex objective function value is equal to or above a prescribed threshold value and the point having said maximum value is within the search region, for outputting said point as a search start point.

Claim 15

An optimization device according to claim 9, wherein said means for detecting a global optimum comprising; means for entering said objective function and starting to calculate a value of said objective function from said search start point to determine a maximum value or a minimum value in said search region from the solutions of said objective function, means for detecting the position of a point having a maximum value or a minimum value of said objective function, means for outputting said point as an optimal solution when a point having a maximum value or a minimum value of said objective function is within the search region, means for repeatedly searching the optimal solution by decreasing a value of said required precision when said optimal solution conforming to said required precision given initially is present, and means for outputting as the global optimum said optimal solution conforming to said required precision immediately before said optimal solution is eliminated.

Claim 16

An optimization device according to claim 9, wherein said means for detecting a global optimum comprising; means for entering said objective function and starting to calculate a value of said objective function from said search start point to determine a maximum value or a minimum value in said search region from the solutions of said objective function, means for detecting the position of a point having a maximum value or a minimum value of said objective function, means for outputting said point as an optimal solution when a point having a maximum value or a minimum value of said objective function is within the search region, means for repeatedly searching the optimal solution by increasing a value of said required precision when said optimal solution conforming to said required precision given initially is not present, and means for outputting as the global optimum said optimal solution conforming to said required precision detecting said optimal solution for the first time.

Claim 17

An optimization device according to claim 9, wherein said means for making said objective function into the convex function comprising; means for having a very large value only at a point where a solution of said objective function is equal to said required precision and for making a prescribed function having a sufficiently small value at other points, and means for giving a diffusing operation to said prescribed function to make said objective function into a convex function, said means for detecting the search start point comprising; means for determining a maximum value in said search region from the solutions of said objective function made into the convex function, means for detecting the position of a point having the maximum value of said objective function made into the convex function, and means, when said determined objective function value made into the convex function is equal to or above a prescribed threshold value and the point having said maximum value is within the search region, for outputting said point as a search start point, and said means for detecting the global optimum comprising; means for entering said objective function and starting to calculate a value of said objective function from said search start point to determine a maximum value or a minimum value in said search region from the solutions of said objective function, means for detecting the position of a point having a maximum value or a minimum value of said objective function, means for outputting said point as an optimal solution when a point having a maximum value or a minimum value of said objective function is within the search region, means for repeatedly searching the optimal solution by decreasing a value of said required precision when said optimal solution conforming to said required precision given initially is present, and means for outputting as the global optimum said optimal solution conforming to said required precision immediately before said optimal solution is eliminated.

Claim 18

An optimization device according to claim 9, wherein said means for making said objective function into the convex function comprising; means for having a very large value only at a point where a solution of said objective function is equal to said required precision and for making a prescribed function having a sufficiently small value at other points, and means for giving a diffusing operation to said prescribed function to make said objective function into the convex function, said means for detecting the search start point comprising; means for determining a maximum value in said search region from the solutions of said objective function made into the convex function, means for detecting the position of a point having the maximum value of said objective function made into the convex function, and means, when said determined objective function value made into the convex function is equal to or above a prescribed threshold value and the point having said maximum value is within the search region, for outputting said point as a search start point, and said means for detecting the global optimum comprising; means for entering said objective function and starting to calculate a value of said objective function from said search start point to determine a maximum value or a minimum value in said search region from the solutions of said objective function, means for detecting the position of a point having a maximum value or a minimum value of said objective function, means for outputting said point as an optimal solution when a point having a maximum value or a minimum value of said objective function is within the search region, means for repeatedly searching the optimal solution by increasing a value of said required precision when said optimal solution conforming to said required precision given initially is not present, and means for outputting as the global optimum said optimal solution conforming to said required precision detecting said optimal solution for the first time.

Claim 19

An optimization method comprising: a step for inputting an objective function which includes a parameter to be optimized and is an object for searching an optimal solution, a required precision indicating a precision required in searching the optimal solution and a search region for searching the optimal solution for the objective function to make said objective function into a convex function; a step for inputting said convex objective function to detect a search start point for starting the search of the optimal solution from said search region of the optimal solution, and a step for detecting the optimal solution based on the detected search start point.

Claim 20

An optimization method comprising: a step for inputting an objective function which includes a parameter to be optimized and is an object for searching an optimal solution, a required precision indicating a precision required in searching the optimal solution and a search region for searching the optimal solution for the objective function to make said objective function into a convex function; a step for inputting said convex objective function to detect a search start point for starting the search of the optimal solution from said search region of the optimal solution, and a step for detecting the optimal solution based on the detected search start point.

Claim 21

An optimization method according to claim 16, wherein said step for making said objective function into the convex function including; a step for having a very large value only at a point where a solution of said objective function is equal to said required precision and for making a prescribed function having a sufficiently small value at other points, and a step for giving a diffusing operation to said prescribed function to make the convex objective function, said step for detecting the search start point including; a step for determining a maximum value in said search region from the solutions of said convex objective function, a step for detecting the position of a point having the maximum value of said convex objective function, and a step, when said determined convex objective function value is equal to or above a prescribed threshold value and the point having said maximum value is within the search region, for outputting said point as a search start point.

Claim 22

An optimization method according to claim 16, wherein said step for detecting the optimal solution including; a step for inputting said objective function, for starting to calculate a value of said objective function from said search start point, and for determining a maximum value or a minimum value in said search region from the solutions of said objective function, a step for detecting the position of a point having the maximum value or the minimum value of said objective function, and a step, when the point having said maximum value or a minimum value is within the search region, for outputting said point as an optimal solution.