sci.op-research

On Mar 5, 8:16 am, Hans Mittelmann wrote:
> On Mar 5, 2:20 am, caokeg...@126.com wrote:
>
>
>
>
>
> > On 3=D4=C25=C8=D5, =CF=C2=CE=E75=CA=B115=B7=D6, caokeg...@126.com wrote:=

>
> > > today, I used cplex to solve a large scale MIP problem, and I met with=

> > > a difficult problem, "Error 1001: out of memory"! I searched in google=

> > > and cplex user mannual, I got when dealing with large scale problems,
> > > it may lack enough physical memory. according to suggested methods, I
> > > set several parameters as follows:
>
> > > cplex.setParam(IloCplex::TiLim,86400000);
> > > cplex.setParam(IloCplex::WorkDir,".");
> > > cplex.setParam(IloCplex::WorkMem,512000);
> > > cplex.setParam(IloCplex::MIPDisplay,0
> > > cplex.setParam(IloCplex::NodeFileInd,3);
>
> > > however, the problem still exists, and I have tried all the ways I can=

> > > think out.
>
> > > the physical memory is 1GB, the OS is windows XP.
>
> > > I'd appreciate anybody give me a useful suggestion. thank you!
>
> > I call cplex in vc2005, and parameters listed above have been changed
> > with different values, but it still doesn't work...
>
> 1 GB is apparently not sufficient. One can solve with close to 4GB on
> Linux 32-bit systems, maybe 2 or 3 GB with XP. For more you will need
> a 64-bit system.
> HOWEVER, since you are solving a MIP the tree could get too big. If
> that really is the case you would write it do disk.- Hide quoted text -
>
> - Show quoted text -

Try mipbooster from www.mipbooster.com. It uses very small memory
space.