Abstract: When numerically integrating an integrand function A over an unbounded domain, a vector map f converts an m (m≧1)-dimensional vector into an m-dimensional vector wherein a multidimensional density function &rgr; of the limiting distribution resulting from repeatedly applying the map f to a predetermined m-dimensional vector u is analytically solvable. A first storage unit stores an m-dimensional vector x, a second storage unit stores a scalar value w, a first computing unit computes a vector x′=f(x), a second computing unit computes a scalar value w′=A(x)/&rgr;(x), an update unit updates values in the first and second storage units and by storing the vector x′ on the first storage unit and adding the scalar value w′ to a value to be stored in the second storage unit, and an output unit computes a scalar value s=w/(c+1) when the number of update times by the update unit becomes c (c≧1) and outputs the result.