$title FcNetStat Management program for fcnet model $version 140 goto OK $abort 'need newer GAMS version' $label OK $set tblheaders InputFileName,NumberOfVariables,NumberOfEquations,NumberOfNonZeros,ModelStatus,SolverTime,ObjectiveValue,ObjectiveValueEstimate,xxx set th headers / %tblheaders%/ $escape = $onecho > rdata.gms table %=1(*,th) $ondelim %tblheaders% $include %=1.txt $offdelim ; $offecho $escape % * Read the tracefiles from the different runs: $offlisting $batinclude rdata xpress $batinclude rdata cplex $batinclude rdata cplexcuts $batinclude rdata cplexcutsheur $onlisting alias (u,*) Set m(u) all models ms(u) simple models md(u) difficult models; m(u) = cplex(u,'NumberOfVariables'); ms(u) = cplex(u,'NumberOfVariables') and not cplexcuts(u,'NumberOfVariables'); md(u) = cplexcuts(u,'NumberOfVariables'); set hth / '#Eqns', '#Vars', '#NZ', 'BestInt', 'BestBnd', 'rel.Gap' / parameter rep; rep(m,'#Eqns') = cplex(m,'NumberOfEquations'); rep(m,'#Vars') = cplex(m,'NumberOfVariables'); rep(m,'#NZ') = cplex(m,'NumberOfNonZeros'); * Simple Models from CPLEX and XPRESS run rep(ms,'BestInt') = round(min(cplex(ms,'ObjectiveValue'),xpress(ms,'ObjectiveValue')),4); rep(ms,'BestBnd') = round(max(cplex(ms,'ObjectiveValueEstimate'),xpress(ms,'ObjectiveValueEstimate')),4); * Difficult model from all runs rep(md,'BestInt') = round(min(cplex (md,'ObjectiveValue'), xpress (md,'ObjectiveValue'), cplexcuts (md,'ObjectiveValue'), cplexcutsheur(md,'ObjectiveValue')),4); rep(md,'BestBnd') = round(max(cplex (md,'ObjectiveValueEstimate'), xpress (md,'ObjectiveValueEstimate'), cplexcuts (md,'ObjectiveValueEstimate'), cplexcutsheur(md,'ObjectiveValueEstimate')),4); rep(m,'rel.Gap') = round(100*(rep(m,'BestInt') - rep(m,'BestBnd'))/rep(m,'BestBnd'),2); file fhtm / fcnetstat.htm /; put fhtm '' /; $onput FCNetLib Model Statistics

Fixed Cost Network Flow Library (FCNetLib)

FCNetLib is a collection of instances for the fixed cost network flow problem. The purpose of the collection is to provide a unified presentation of the data instances together with some GAMS models that implement a straight forward IP formulation and a more sophisticated version that utilizes the GAMS/BCH facility. A class of known cuts (dicuts) is implemented as well as the local branching heuristic within the BCH framework.

The original models were in XPRESS-Modeler format and collected by Laurence Wolsey and collagues at Universite Catholique de Louvain. They were translated by Alexey Koptsevich into GAMS.

Reference

Download

Download FCNetLib.zip

Additional Information

FCNet GAMS Models

Helper GAMS Programs

Trace files from pure CPLEX, pure XPRESS, and CPLEX with dicuts and heuristic

FCNetLib Model Statistics

$offput $set tdmacro ALIGN=RIGHT BGCOLOR="# loop(hth, put '' /) put '' set c / c1 fefefe, c2 e0e0e0 / mc(u,c); scalar ordm /0/; loop(m, ordm=ordm+1; mc(m,'c1')$(mod(ordm,2)=1) = yes; mc(m,'c2')$(mod(ordm,2)=0) = yes; ); set hth0(hth) /'#Eqns', '#Vars', '#NZ'/, hth2(hth) / 'BestBnd' /; $set hth0 put$hth0(hth) rep(m,hth):0:0; $set hth2 put$hth2(hth) rep(m,hth):0:2; $set hthgap put$(sameas(hth,'rel.Gap') and rep(m,hth)) rep(m,hth):0:2; $set hthBI put$sameas(hth,'BestInt') '' rep(m,hth):0:2 ''; loop(mc(m,c), put / ''/; put ''/; loop(hth, put '' /) put '' / ) put '
Name' hth.tl:0 ' 
' m.tl:0 ''; %hth0% %hth2% %hthBI% %hthgap% put '  
';