[Scip] epsilon for functions isPositive, isZero, etc

[Timo Berthold]

Hi Alessia,

You can use SCIPchgRealParam() to change the parameter numerics/epsilon.
Note that at many places in the code, SCIP uses SCIPisFeas...() checks
which compare against the numerics/feastol parameter (and not the
epsilon).

Ciao
Timo

> Hello Matthias,
>
> thanks! Is there a function to change the epsilon once for all those
> functions? it would be quicker to test different values!
>
> Alessia
>
>
> On 01/27/2014 11:59 AM, Matthias Miltenberger wrote:
>> Hi Alessia,
>>
>> SCIP uses 1e-9 as default epsilon for these functions. It can be changed
>> to a value in the range of 1e-20 to 1e-3.
>>
>> best
>> Matthias
>>
>> On 27.01.2014 11:17, Alessia Violin wrote:
>>> Hello!
>>>
>>> I was wondering which is the "epsilon" used in the functions
>>> SCIPisPosite, isZero, isEQ, etc? I tried to find this information in
>>> the
>>> developer's manual but I didn't succeed.
>>>
>>> I have a branch-and-cut-and-price code with quite a lot of other things
>>> (stabilisation, heuristics, ...), and if the epsilon is too small the
>>> convergence of column generation is very slow, and if it's too big
>>> there
>>> are some particular configurations of parameters for which optimal
>>> columns are not generated. Moreover this also depends on the magnitude
>>> of the data for different set of instances and I am trying to figure
>>> out
>>> the best epsilon to use.
>>>
>>> Thanks in advance!
>>>
>>> Alessia
>>>
>>>
>
> --
> Alessia Violin
> Service Graphes et Optimisation Mathématique (G.O.M.)
> Université Libre de Bruxelles
> C.P. 210/01
> Boulevard du Triomphe
> B-1050 BRUXELLES
> Tel: 02 650 58 80 - Fax: 02 650 59 70
> Email: aviolin at ulb.ac.be
> Webpage: http://homepages.ulb.ac.be/~aviolin/
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip
>