...and here is the page Christian
-- Christian Schulte, www.ict.kth.se/~cschulte/ > -----Original Message----- > From: Christian Schulte [mailto:cschu...@kth.se] > Sent: Tuesday, July 23, 2013 2:29 PM > To: 'Kish Shen' > Cc: 'users@gecode.org' > Subject: RE: [gecode-users] Using IntActivity with space > > Just added initialization of Activity by a function to the trunk, see the single page > from MPG for details. > > Christian > > -- > Christian Schulte, www.ict.kth.se/~cschulte/ > > > > -----Original Message----- > > From: users-boun...@gecode.org [mailto:users-boun...@gecode.org] On > > Behalf Of Kish Shen > > Sent: Tuesday, July 16, 2013 6:11 PM > > Cc: users@gecode.org > > Subject: Re: [gecode-users] Using IntActivity with space > > > > On 16/07/2013 10:09, Christian Schulte wrote: > > > Yes, changes to recomputation can change the variable ordering but > > > that's the best we can do.I think quite some variable selection > > > strategies will also work differently on different hardware/software > > > platforms due to different rounding... > > > > > > > To me, changing the amount of recomputation is somewhat different from > > running on different hardware/software platforms. Am I correct that > > besides activity, changing the amount of recomputation will only > > affect memory/execution performance? > > > > Is it not possible to turn off the gathering of activity information > > during recomputation? I assume that (like in our interface code), you > > know when you are performing recomputation, and if so, can the > > gathering of activity information be turned off for the duration? > > > > > How about one can give an initialization function for Activity such > > > that the activity value will be initialized for each variable. That should be > easy. > > > > This sound like a good idea! > > > > Cheers, > > > > Kish > > > > > > > > _______________________________________________ > > Gecode users mailing list > > users@gecode.org > > https://www.gecode.org/mailman/listinfo/gecode-users
MPG 144.pdf
Description: Adobe PDF document
_______________________________________________ Gecode users mailing list users@gecode.org https://www.gecode.org/mailman/listinfo/gecode-users
