Andrew, As always I am at your command sometimes, see http://rosettacode.org/wiki/Ramsey.
The impressive thing is that glpk can find a solution in 3 seconds. I had written some code in C++ to get an idea of the problem. Initially I tried cutting the graph into subgraphs of 6 solving those and then trying to combine them. I then started with 6 and tried adding a seventh, eigth etc to get some idea of the shape of this distribution. It rises in a manner that makes exponential seem gentle, peaks and then falls of a cliff with solutions very rare. The latter I simply tranferred into mathprog with scalar variables. -- Nigel Galloway [email protected] On Tue, Jan 17, 2012, at 08:28 PM, Andrew Makhorin wrote: > Hi Nigel, > > > Chrestomathy is my word for 2012. Don't worry I don't know what it > > means, looks like it may have escaped from Greek and is seeking refuge > > in English. http://rosettacode.org/wiki/Rosetta_Code is an example. > > Until 2011 it knew 424 langages. I've made that 425 by adding Mathprog: > > see http://rosettacode.org/wiki/Category:Mathprog > > > > I added a new Task for Mathprog to solve see > > http://rosettacode.org/wiki/Ramsey > > > > I added the problem in Mathprog see > > http://rosettacode.org/wiki/Execute_Ramsey_Mathprog > > > > and the solution see > > http://rosettacode.org/wiki/Solution_Ramsey_Mathprog > > > > Thank you for your information. > > I noticed that the MathProg model uses scalar variables and constraints > that makes it difficult to read. Why not to use indexed objects? > > > Andrew Makhorin > > -- http://www.fastmail.fm - Faster than the air-speed velocity of an unladen european swallow _______________________________________________ Help-glpk mailing list [email protected] https://lists.gnu.org/mailman/listinfo/help-glpk
