Path: utzoo!utgpu!water!watmath!clyde!bellcore!rutgers!cmcl2!nrl-cmf!ames!joyce!sri-unix!quintus!ok
From: ok@quintus.uucp (Richard A. O'Keefe)
Newsgroups: comp.lang.prolog
Subject: Re: Random Clause Selection (was: re:probabilstic or)
Keywords: meta interpreter, random, probabilistic, completeness, loops
Message-ID: <308@quintus.UUCP>
Date: 24 Aug 88 08:53:47 GMT
References: <1171@tuhold>
Sender: news@quintus.UUCP
Reply-To: ok@quintus.UUCP (Richard A. O'Keefe)
Organization: Quintus Computer Systems, Inc.
Lines: 22

In article <1171@tuhold> thom@tuhold (Thom Fruehwirth) writes:
>In <256@quintus.uucp> Richard O'Keefe sketched how to try clauses
>in random order:
>>
>> p(~Args~):-
>>	random_permutation([1,2,3,4,5,6,7], Perm),
>>	member(Index, Perm),
>>	p(Index, ~Args~).
>
>As Jamie Andrews in <628@etive.ed.ac.uk> pointed out, O'Keefe probably
>intended to have a cut between random_permutation/2 and member/2.

NO WAY!  What would it do?
random_permutation/2 generates *ONE* random permutation between its two
arguments.  This is the only clause of p(~Args~), and random_permutation/2
doesn't leave any choice points, so there is nothing for a cut to do.
What this code fragment does is to try the (original) clauses of p in a
random order; for that we want *ONE* random permutation.

random(-X) generates *ONE* pseudo-random float uniformly distributed
between 0.0 and 1.0.  You wouldn't expect it to generate another if you
backtrack into it.  Why should random_permutation/2 be any different?