testing for uniquness in a large list

[bearophile]

Josiah:
>There are only 125,970 unique sequences of 12 items
>from 20 where order does not matter (20!)/(12!8!),

Right! I was thinking the same thing *after* writing my last post ^_-

With random sampling the generation of the last 10 combinations is a
bit slow, it require a couple of minutes.
There are direct ways to compute them:
http://www.netlib.org/toms/382

Bye,
bearophile