Re: [Help-glpk] glpk 4.65 release information

[Andrew Makhorin]

Hi Chris,


> In order to test the release, I tried the 2010 MIPLIB benchmark set,
> calling glpsol with "--pcost --cuts" and a two hour time limit. The
> final result is in the attached file, but the quick summary is that 20
> problems from this set were solved, while there were three aborts.

Thank you very much for your efforts. I think your results are very
useful.

I'd like to note that sometimes '--cuts' is not a best choice. For
example, air04 and air05 (hard set partitioning problems) from miplib
3.0 are solved quickly (about 10 mins on my machine) with '--pcost',
but with cutting planes the solution takes more 6 hours. Probably this
is because cutting planes increase fractionality of node subproblems.


Best regards,

Andrew Makhorin

> 
> 
> Of the three aborts, problem mspp16 just run out of memory, while
> problems msc98-ip and ns1688347 failed in the primal simplex.
> 
> 
> Specifically, msc98-ip failed with:
> 
> Warning: numerical instability (dual simplex, phase II)
> Warning: dual simplex failed due to excessive numerical instability
> Assertion failed: teta_lim >= 0.0
> Error detected in file simplex/spxprim.c at line 665
> 
> while ns1688347 failed with:
> Error: basis matrix is singular to working precision (cond = 3.07e+18)
> Assertion failed: teta_lim >= 0.0
> Error detected in file simplex/spxprim.c at line 665
> 
> 
> 
> 
> Best Regards,
> 
> 
> Chris Matrakidis
>