Re: [Help-glpk] Solving a large LP problem

[Samik Raychaudhuri]

Thanks Andrew, I will try submitting the problem to NEOS solver.
I had one question though. My output seems to be showing that the 
presolve stage is already over and its trying to solve the presolved 
problem. In that case, will --nopresol be of any use?
I will try to find out the system settings as mentioned by you.
Regards,
-Samik

On 7/15/2004 12:05 PM, Andrew Makhorin wrote:

> > This is the output I got when I solved with a considerably smaller problem.
> > -------------------------
> > $ glpsol --mps new.mps
> > load_mps: reading LP data from `new.mps'...
> > load_mps: name `BLANK'
> > load_mps: 243742 rows
> > load_mps: 364514 columns
> > load_mps: 1148311 non-zeros
> > load_mps: 1 right-hand side vector(s)
> > load_mps: 0 range vector(s)
> > load_mps: 0 bound vector(s)
> > load_mps: 932145 cards were read
> > lpx_simplex: original LP has 243742 rows, 364514 columns, 1148311 non-zeros
> > lpx_simplex: presolved LP has 90344 rows, 203072 columns, 487180 non-zeros
> > lpx_adv_basis: size of triangular part = 89732
> > umalloc: size = 3614184; malloc failed
> > ------------------------------------
> > Is this b/c I have less system memory available, or b/c of something to 
> > do with the code? How can I check if system memory is the problem.
> >    
>
> The last message means that some glpk routine attempted to allocate a
> memory block of 3,614,184 bytes long, however, the control program
> could not satisfy the request (i.e. malloc returned NULL). As a rule
> this happens if the machine has not enough ram memory. You can try to
> specify the option `--nopresol', because presolver data structures
> eat up much ram memory (although for your instance using the presolver
> has a significant effect). The error also may happen due to some system
> settings which limit the amount of ram memory allocated to application
> program.
>
> Btw, glpsol 4.4 is installed on the NEOS Server:
> http://www-neos.mcs.anl.gov/neos/solvers/MILP:GLPK/
> You can try to submit your mps file to the server at least to see
> whether glpsol is able to solve your instance.
>
> Andrew Makhorin
>
>
>