[gecode-users] AFC decay with global AFC in search engines

[Christian Schulte]

After some head scratching I found out what the problem is: when you call
INT_VAR_AFC_* and do not give it an argument, it will reset the afc-decay to
1.0 (so that means no decay). When you call afc_decay(d) after
INT_VAR_AFC_*, the d is used!

Cheers
Christian


--
Christian Schulte, Professor of Computer Science, KTH,
www.ict.kth.se/~cschulte/


-----Original Message-----
From: users-bounces at gecode.org [mailto:users-bounces at gecode.org] On Behalf
Of Kish Shen
Sent: Tuesday, May 13, 2014 12:01 AM
To: users at gecode.org
Subject: Re: [gecode-users] AFC decay with global AFC in search engines

On 12/05/2014 22:39, Kish Shen wrote:
> For activity, I set select to INT_VAR_ACTIVITY_MAX(act) before the 
> call to branch() (as I was doing for afc), and this seems to work.

I guess this might not be very clear: what I do for activity is:

    IntActivity act(*solver, vars, actdecay);
    ....
    select = INT_VAR_ACTIVITY_MAX(act);
    ....
    branch(*solver vars, select, choice);

where actdecay is the activity decay, and by working I mean that changing
the value of actdecay does seem to affect the search.

The difference with AFC is that I can use AFC outside of the search engines
(i.e. if the search is done in ECLiPSe), because there is a "global" AFC
associated with the space, which is what I use. For activity, I can only use
it with a Gecode search engine, by declaring an IntActivity act for the
variables involved in the search (vars).

Cheers,

Kish




_______________________________________________
Gecode users mailing list
users at gecode.org
https://www.gecode.org/mailman/listinfo/gecode-users