/************** Approximations to a Wiener Process **************
Author: Mico Loretan loretanm@frb.gov
Date: 19 Nov 96
This code is provided without guarantee for public
non-commercial use.

Approximations to a Wiener process on the [0,1] interval may be
accomplished in a number of different ways.  By far the easiest
method is based on the idea that the Wiener process is the limit
of a scaled random walk.  Recalling that a random walk is nothing
but the cumulative sum of iid increments, we can consider the
following code:
****************************************************************/
proc (2)=wiener(n);
        /*
        :: Generate an approximation to a Wiener process on the
        :: [0.1] interval by taking cumulative sum of standard
        :: normal random numbers.  The variable "n" defines the
        :: fineness of the grid.  The returned variables are x,
        :: the grid from 1/n to 1 at which the function values
        :: are computed, and y, the values of the (approximate)
        :: Wiener process.  The scaling required is simply
        :: sqrt(n).
        */
        local x,y,z;
        x = seqa(1/n,1/n,n);
        z = rndn(n,1);          /* n iid N(0,1) variates */
        y = cumsumc(z)/sqrt(n);
        retp( x,y );
endp;

/****************************************************************
The question that remains is how to choose n.  This will depend
in part on the nature of your application.  In most cases,
though, setting n=2500 or n=5000 will be more than sufficient.
****************************************************************/