Message2268

Author malte
Recipients malte, mikko
Date 2012-06-18.10:07:33
Content
Thanks, Mikko! I'm surprised to see that the special case for the goals is not
mentioned in the paper -- it is intentional, although maybe it goes too far.

The main reason why we have the special rule is that it never makes sense (in
the CG heuristic) to have a non-goal variable as the highest-level variable,
since all variables that occur after the last goal variable have no influence on
the heuristic value. But of course that doesn't mean that *all* goal variable
ought to occur right at the end.

I'll look into it, but it requires a lot of testing to make sure that
performance is not adversely affected in some configurations.
History
Date User Action Args
2012-06-18 10:07:34maltesetmessageid: <1340006854.24.0.675416463268.issue341@gmail.com>
2012-06-18 10:07:34maltesetrecipients: + malte, mikko
2012-06-18 10:07:34maltelinkissue341 messages
2012-06-18 10:07:33maltecreate