[gecode-users] Passing information to the master() function

[Christian Schulte]

Hi Kish,

Well, you get access to the master space: remember the master() function is
called on the _master_ space. That should be sufficient, or?

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: Thursday, October 03, 2013 7:35 PM
To: cschulte at kth.se
Cc: users at gecode.org
Subject: Re: [gecode-users] Passing information to the master() function

Hi Christian,

On 03/10/2013 09:51, Christian Schulte wrote:
> Hi Kish,
>
> Sorry for missing this.
>
> If you want to post the constraint then you would have to store the 
> y-variables at someplace such that the master() function (you would 
> have to re-define it) can access y.
>

Thanks for your reply. Am I correct in thinking that the restart search
facilities do not provide any way of passing extra current information from
the search back to the master() (for DFS; I guess I can (ab)use the
constrain() function in my case, as what I want to pass back is a
constraint, by changing the DFS search to BAB)?

Naively (because I don't know the details of Gecode), it seems to me that
providing the user with access to the master space during the search will
allow a simple way of passing information to the master space. I can see
that being useful if the user want to do things that depends on the state of
the search when restarting a search/finding a new solution.

> The second option is more meaningful: I think restricting restart to a 
> single solution does not hurt.

I am a little reluctant to do this, as this does not fit that well with how
I intend to provide the functionality in ECLiPSe, as I want to add restart
and no-goods as extra options to gfd's search routine, i.e. not as different
search methods.

Cheers,

Kish


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