[Help-gsl] ran2 and random walks

[Joseph Wakeling]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I'm writing some code right now in which a parameter is meant to
fluctuate as a random walk.  I'm using ran2 as the rng.

Suppose I write a simple piece of code of the form,

{
    int i; double walk, increment;

    /* initialise variables */

    for(i=0;i<N;++i) {
        walk += (ran2_get_double() < 0.5)?-increment:increment;
        printf("%g\n",walk);
    }
}

This code produces a nice random walk.

Suppose now instead I add to this,

{
    int i; double walk, increment;

    /* initialise variables */

    for(i=0;i<N;++i) {
        walk += (ran2_get_double() < 0.5)?-increment:increment;
        printf("%g\n",walk);

        /* A load of other stuff now happens here,
           that also involves the generation of random
           numbers with ran2. */
    }
}

I've found that the random walk now has huge biases in it.

Can anyone explain why this should be so?  Am I doing something
outrageously stupid by letting the random walk and the other material be
generated by the same stream of pseudo-random numbers?

Many thanks,

    -- Joe

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFErjZtcjylL0sfzuERAuC4AJ9aWnsQxsevN8q/aijlks0z9HqkggCeI+9H
2rVJW7w0wOmpTbNUs6+Vr3o=
=JI6f
-----END PGP SIGNATURE-----