Re: [Bug-glpk] GLPSOL outputs MIP solution that is not LP optimal for fixed integers

[xypron]

Andrew Makhorin wrote:
> 
>> I would not have expected a MIP solution to yield an noninteger
>> objective for this model.
> 
> 'incommon' is declared as continuous:
> 
> var incommon {i in teams, j in teams: i < j}, >=0, <=1;
> 
> so I do not see a problem. Why do you think that all components of
> 'incommon' must be integral?
> 

Hello Andrew,

the feasibility pump provides assumedly feasible solution where all
binaries are either 0 or 1.

If I fix the values of the binaries and resolve the original problem
all components are integral.

Please, run the files 
http://www.nabble.com/file/p25430771/f1.mod f1.mod 
http://www.nabble.com/file/p25430771/f2.mod f2.mod 

(Please, check the ODBC connection data or use CSV.)

glpsol.exe --fpump --tmlim 30 -m f1.mod
glpsol.exe -m f2.mod

Compare the objective
f1: 1.041490000e+002
f2: 1.08

f1 saves the binary vlaues from the feasibility pump to the database.
f2 uses this values as fixed values.

Best regards

Xypron
-- 
View this message in context: 
http://www.nabble.com/GLPSOL-outputs-MIP-solution-that-is-not-LP-optimal-for-fixed-integers-tp25337983p25430771.html
Sent from the Gnu - GLPK - Bugs mailing list archive at Nabble.com.