Re[5]: [Help-glpk] glpk 4.5 release information

[Michael Hennebry]

On Wed, 21 Jul 2004, Andrew Makhorin wrote:

> >Why not a value indicating OK?
>
> MIP_E_OK indicates that the tree is completely explored, in which case
> lpx_integer sets the status LPX_I_OPT (integer optimal solution found)
> or LPX_I_NOFEAS (problem has no integer feasible solution). However,
> if the search is prematurely terminated, the status must be LPX_I_FEAS
> (integer feasible solution found but its optimality not proven) or
> LPX_I_INFEAS (no integer feasible solution found yet).

I think that ideally lpx_integer would return MIP_E_OK
and the status would be indicate a possibly suboptimal
solution, but one as good as requested.
Second best would be MIP_E_OK and LPX_I_FEAS.
Third best would be MIP_E_OK and LPX_I_OPT.
The solution is optimal to the desired precision.

Any of the above combinations would indicate
to the user that the result was useable.
A return code of MIP_E_TMLIM would indicate that the
result might not be useable and the user would have
no way to know better unless the status were LPX_I_OPT.

-- 
Mike   address@hidden
"Nothing says it like words if you know how to use them."
                    --  the Professional Organization of English Majors