[gecode-users] users Digest, Vol 124, Issue 6
Christian Schulte
cschulte at kth.se
Mon Feb 1 09:21:44 CET 2016
Hi,
I have a suspicion here. The easiest might be to send me your implementation of the constrain() member function so that I can have a look.
Best
Christian
--
Christian Schulte, www.gecode.org/~schulte
Professor of Computer Science, KTH, <mailto:cschulte at kth.se> cschulte at kth.se
Expert Researcher, SICS, <mailto:cschulte at sics.se> cschulte at sics.se
From: Juan Carlos Hernandez [mailto:jchernandez.madrid at gmail.com]
Sent: Sunday, January 31, 2016 09:10 AM
To: Alexander Schiendorfer; Christian Schulte
Subject: Re: [gecode-users] users Digest, Vol 124, Issue 6
Hi again,
I just wanted to say that the lexicographic search worked perfectly. Thanks!
Now, I am stuck at a segmentation fault that appears when I try to run it in parallel mode (say, using 8 threads). The error happens irregularly, but more often when I increase the thread count. I will see if I can make a small code that reproduces the error. Here is the Visual Studio trace, in case it is a known bug.
Unhandled exception at 0x0F5A81BA (GecodeInt-4-4-0-d-x86.dll) in cpsolver.exe: 0xC0000005: Access violation reading location 0x0194379C.
> GecodeInt-4-4-0-d-x86.dll!Gecode::Int::IntVarImp::RangeList::min() Line 103 C++
GecodeInt-4-4-0-d-x86.dll!Gecode::Int::IntVarImp::min() Line 230 C++
GecodeInt-4-4-0-d-x86.dll!Gecode::Int::IntView::min() Line 60 C++
GecodeInt-4-4-0-d-x86.dll!Gecode::Int::Rel::LexLqLe<Gecode::Int::IntView>::propagate(Gecode::Space & home, const int & __formal) Line 93 C++
GecodeKernel-4-4-0-d-x86.dll!Gecode::Space::status(Gecode::StatusStatistics & stat) Line 270 C++
GecodeSearch-4-4-0-d-x86.dll!Gecode::Search::Parallel::BAB::Worker::run() Line 120 C++
GecodeSupport-4-4-0-d-x86.dll!Gecode::Support::Thread::Run::exec() Line 64 C++
GecodeSupport-4-4-0-d-x86.dll!Gecode::Support::bootstrap(void * p) Line 48 C++
Thanks,
JC
2016-01-29 15:10 GMT+01:00 Alexander Schiendorfer <alexander.schiendorfer at informatik.uni-augsburg.de>:
Yes it's godly :)!
Let me jump on the bandwagon of showing off - http://isse-augsburg.github.io/constraint-relationships/ does really benefit from that feature (well, accessed by MiniSearch, but still!)
Cheers, have a good weekend
On 29.01.2016 14:05, Christian Schulte wrote:
Actually, this is the very reason that Gecode has a constrain() function as its primitive: it is way more powerful than a simple cost variable.
Just showing off a little ;-)
Cheers
Christian
--
Christian Schulte, <http://www.gecode.org/%7Eschulte> www.gecode.org/~schulte
Professor of Computer Science, KTH, <mailto:cschulte at kth.se> cschulte at kth.se
Expert Researcher, SICS, <mailto:cschulte at sics.se> cschulte at sics.se
From: Juan Carlos Hernandez [mailto:jchernandez.madrid at gmail.com]
Sent: Friday, January 29, 2016 01:39 PM
To: Christian Schulte <mailto:cschulte at kth.se> <cschulte at kth.se>; alexander.schiendorfer at informatik.uni-augsburg.de
Subject: Re: [gecode-users] users Digest, Vol 124, Issue 6
Dear Alex and Christian,
thanks a lot! A very elegant solution - I will try it out at once.
Cheers,
JC
2016-01-29 12:15 GMT+01:00 Christian Schulte <cschulte at kth.se>:
Alex is spot on here: but let me expand a little.
You might have used the cost() member function for B&B. However the cost()
member function is just implemented in terms of the constrain() member
function. So what I would do is the following:
- Have a integer variable array cost containing the two cost variables you
have in mind.
- implement a constrain(const Space& _b) member function: _b refers to the
so-far best solution. Cast this to your script type, say b. Then this just
posts
rel(*this, cost, IRT_LQ, b.cost);
This enforces that this->cost is lexicographically smaller then the so
far best cost b.cost.
That's it. More on the constrain() function you can find in Chapter 2 of
MPG.
Cheers
Christian
--
Christian Schulte, www.gecode.org/~schulte <http://www.gecode.org/%7Eschulte>
Professor of Computer Science, KTH, cschulte at kth.se
Expert Researcher, SICS, cschulte at sics.se
-----Original Message-----
From: users-bounces at gecode.org [mailto:users-bounces at gecode.org] On Behalf
Of Alexander Schiendorfer
Sent: Friday, January 29, 2016 12:06 PM
To: users at gecode.org
Subject: Re: [gecode-users] users Digest, Vol 124, Issue 6
Hi Juan,
it seems to me that you're searching for lexicographic branch-and-bound.
There is a global constraint lex_lesseq that you might want to consider for
posting in the constrain method of a Space.
Cheers,
Alex
On 29.01.2016 12:00, users-request at gecode.org wrote:
> Send users mailing list submissions to
> users at gecode.org
>
> To subscribe or unsubscribe via the World Wide Web, visit
> http://www.gecode.org/cgi-bin/mailman/listinfo/users
> or, via email, send a message with subject or body 'help' to
> users-request at gecode.org
>
> You can reach the person managing the list at
> users-owner at gecode.org
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of users digest..."
>
>
> Today's Topics:
>
> 1. Best way to implement multi-criterion optimization in Gecode?
> (Juan Carlos Hernandez)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Thu, 28 Jan 2016 12:30:54 +0100
> From: Juan Carlos Hernandez <jchernandez.madrid at gmail.com>
> To: users at gecode.org
> Subject: [gecode-users] Best way to implement multi-criterion
> optimization in Gecode?
> Message-ID:
> <CAFd=f3EAU+XPqJo02KPT9SWibPjxkifw-F2so-4mE+x4Kv7Q5A at mail.gmail.com <mailto:f3EAU%2BXPqJo02KPT9SWibPjxkifw-F2so-4mE%2Bx4Kv7Q5A at mail.gmail.com> >
> Content-Type: text/plain; charset="utf-8"
>
> Hi everyone,
>
>
> I am trying to use Gecode to solve a multi-criterion problem
>
>
> minimize (F1(x,z), F2(x,z) )
>
> subject to constraints on x,z
>
>
> The objectives are ordered in order of importance, so I would like to
> first solve the problem P1
>
>
> minimize F1(x,z)
>
> subject to original constraints on x,z
>
>
> and then consider the modified problem P2
>
>
> minimize F2(x,z)
>
> subject to original constraints on x,z
>
> F1(x,z) = F1opt
>
>
> where F1opt is the optimal value of P1.
>
>
> What is the most efficient way to implement this in Gecode?
>
>
>
> I am now doing two completely separate searches, in the sense that I
> create a new Space and run BAB on it for each of the problems. I am
> sure that this cannot be the most efficient way of dealing with the
> problem, but am unable to figure out how to do it "the right way".
> Essentially, when solving P2 I should be able to make use of the work done
while solving P1...
>
>
> Any feedback is very much appreciated!
>
>
> Thanks,
>
> JC
> -------------- next part -------------- An HTML attachment was
> scrubbed...
> URL:
> <http://www.gecode.org/pipermail/users/attachments/20160128/06e6ff75/a
> ttachment-0001.html>
>
> ------------------------------
>
> _______________________________________________
> Gecode users mailing list
> users at gecode.org
> https://www.gecode.org/mailman/listinfo/gecode-users
>
>
> End of users Digest, Vol 124, Issue 6
> *************************************
--
Alexander Schiendorfer, M.Sc.
Institute for Software & Systems Engineering Raum 3043, Gebäude N
Universität Augsburg
Web:
http://www.informatik.uni-augsburg.de/lehrstuehle/swt/se/staff/aschiendorfer
/
Tel.: +49 (821) 598 - 2196 <tel:%2B49%20%28821%29%20598%20-%202196>
E-Mail: alexander.schiendorfer at informatik.uni-augsburg.de
_______________________________________________
Gecode users mailing list
users at gecode.org
https://www.gecode.org/mailman/listinfo/gecode-users
_______________________________________________
Gecode users mailing list
users at gecode.org
https://www.gecode.org/mailman/listinfo/gecode-users
--
Alexander Schiendorfer, M.Sc.
Institute for Software & Systems Engineering
Raum 3043, Gebäude N
Universität Augsburg
Web: http://www.informatik.uni-augsburg.de/lehrstuehle/swt/se/staff/aschiendorfer/
Tel.: +49 (821) 598 - 2196 <tel:%2B49%20%28821%29%20598%20-%202196>
E-Mail: alexander.schiendorfer at informatik.uni-augsburg.de
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gecode.org/pipermail/users/attachments/20160201/22ab11a5/attachment-0001.html>
More information about the users
mailing list