Re: [Help-glpk] GLPK and MIPLIB 2003 manna81.mps
Hello Jen-Min, the problem file can be found at http://miplib.zib.de/miplib2003/manna81.php The optimum solution is described in http://www.dei.unipd.it/~fisch/papers/lexGomory_cuts.pdf as LP opt = -13297 MILP opt = -13164. -13164 is the first integer solution found by GLPK (after 2 minutes on a rather slow Linux server) with a gap of 1 %. This matches a report concerning the Dash optimizer http://www.ima.umn.edu/2004-2005/W7.25-29.05/activities/Perregaard-Michael/cbranch.ppt manna81: solved on root with no branching. Could You, please, identify which is the termination criterion for COIN_OR, e.g does it simply stop if the gap is less then 1 %? Glpsol will only stop at a much smaller gap if You do not use the command line parameter --mipgap. Which was the optimum found by COIN_OR? Do You get the same result on Windows: MILP opt = -13164 for gap = 1 %? Best regards Xypron Original-Nachricht > Datum: Wed, 28 May 2008 06:45:57 -0700 > Von: "Cheng,Jen-Min" <[EMAIL PROTECTED]> > An: help-glpk@gnu.org > Betreff: [Help-glpk] GLPK and MIPLIB 2003 manna81.mps > > I generated glpsol.exe 4.2.8 using Microsoft Visual Studio 2005. It did > not stop after over 10 hours when manna81.mps of MIPLIB 2003 was being > solved. However COIN_OR could solve manna81.mps in less than 10 > seconds. > > Will any please tell me what happened. Thanks. > > > > > ___ > Help-glpk mailing list > Help-glpk@gnu.org > http://lists.gnu.org/mailman/listinfo/help-glpk -- GMX startet ShortView.de. Hier findest Du Leute mit Deinen Interessen! Jetzt dabei sein: http://www.shortview.de/[EMAIL PROTECTED] ___ Help-glpk mailing list Help-glpk@gnu.org http://lists.gnu.org/mailman/listinfo/help-glpk
[Help-glpk] GLPK and MIPLIB 2003 manna81.mps
I generated glpsol.exe 4.2.8 using Microsoft Visual Studio 2005. It did not stop after over 10 hours when manna81.mps of MIPLIB 2003 was being solved. However COIN_OR could solve manna81.mps in less than 10 seconds. Will any please tell me what happened. Thanks. ___ Help-glpk mailing list Help-glpk@gnu.org http://lists.gnu.org/mailman/listinfo/help-glpk
Re: [Help-glpk] Building GLPK under VS2005 for x64
Hi Peter, Looks like I have found a bug. Please replace line 242 in file glplib.h: #define align_boundary sizeof(double) by the following line: #define align_boundary 16 I made this replacement in the project on your machine, and the bug disappeared (at least the test program now runs correctly). Please keep me informed about further bugs/problems that may appear in 64-bit version of glpk. Best regards, Andrew Makhorin ___ Help-glpk mailing list Help-glpk@gnu.org http://lists.gnu.org/mailman/listinfo/help-glpk
Re: [Help-glpk] set of sets
> I am writing a timetabling program in GMPL and I #39;d like to > write some incompatibility constraints of courses generally, e.g. I > want to say that no more than 1 of {A1, A2, A3} can appear in the same > slot, and no more than 1 of {B1, B2} can appear in the same slot > something like.. > set incompatitible := (A1, A2, A3) (B1, B2); > min. > s.t. no_incomp{s in SLOTS}: sum{s in SLOTS, i in incomp} > allocation[slot, course in i] <= 1; /* allocation is binary decision > variable > Question: > 1. Can I even have a set of sets of different size? > 2. How can I express the above? > I have read the manual for quite a while but I can #39;t seem to > find something similar. I #39;d appreciate if you can point to the > right direction. You might declare a set (array) of sets as follows: set incomp[1..N]; where incomp[1] is a first set of incompatible courses, incomp[2] is a second set, etc., assuming the following initialization in the data section: set incomp[1] := A1 A2 A3; set incomp[2] := B1 B2; . . . ___ Help-glpk mailing list Help-glpk@gnu.org http://lists.gnu.org/mailman/listinfo/help-glpk