[Scip] Integrality tolerance

Stefan Heinz heinz at zib.de
Wed Sep 29 11:35:42 MEST 2010


Hi,

in scip.c there is the function SCIPisIntegral()

http://scip.zib.de/doc2/html/scip_8h.html#a3022a89ce7a688e8474ffd3a179f76e0

if you follow that function in the code you see that SCIP uses the value of 
the parameter "numerics/epsilon" as tolerance. Changing that parameter 
influences this comparison.

Best Stefan

On Wednesday 29 September 2010 11:15:33 Emilie Danna wrote:
> Hi,
> 
> I'm looking for an integrality tolerance parameter in SCIP, i.e. a value
> epint such that
> round(x) - epint <= x <= round(x) + epint implies that x is considered an
> integer value by SCIP. I couldn't find such a parameter in the list of
> parameters at http://scip.zib.de/doc/html/PARAMETERS.html and looking into
> the code, it seems that EPSISINT is called with a variety of epsilon
>  values. What is the closest value or parameter that could be used as a
> proxy? SCIP_DEFAULT_EPSILON? The LP feasibility tolerance?
> 


More information about the Scip mailing list