Re: [Help-glpk] [Bug-glpk] [bug] glpk cycling

[Andrew Makhorin]

Hi Chris,
> 
> I did some testing with this version of dual simplex: I tried several
> problems from miplib 2010, including some that had trouble with the
> previous version and it seems to be a big improvement. The only issue
> I run into was with primal simplex, similar to the case that was
> reported earlier [1].
> 

Thank you for testing.

I don't know why I missed this common case when all objective
coefficients are zero, and it is unclear how the dual solver could solve
such instances at all, because in this case every dual step is
degenerate until a dual feasible ray is found. A similar situation (i.e.
for primal conic constraints Ax = 0, x >= 0) may happen in the primal
simplex solver, though such instances are rarely appear in practice.

Best regards,

Andrew Makhorin