[Bonmin] size of integer variables

[Pierre Bonami]

Dear Francois,
I am not sure what the answer is. I think that the answer is really 
problem dependent (in MILP it is very easy to construct small examples 
where any solver not doing exact arithmetic will fail and Bonmin is not 
doing exact arithmetic).
I would say you have to be careful (and I am trying to be very careful 
in my answer). At the very least your variable value should be small 
enough so that Bonmin can test if it is integer or not I would say with 
an absolute precision of 10^-6.
Maybe somebody has a better answer.
Best,
Pierre



Francois Dionne wrote:
> Dear mailing list,
>  
> I'm working on a series of small nonconvex MINLPs, and I have a 
> question: what is the maximum allowed value of an integer variable?
>  
> Right now and in the near future, I expect my integer variables to fit 
> inside the mantissa of a double floating point, but I am not sure 
> about the number of guard digits I should allow after the decimal point.
>  
> Thank you in advance.
>  
> ------------------------------------------------------------------------
>
> _______________________________________________
> Bonmin mailing list
> Bonmin at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/bonmin
>   

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.coin-or.org/pipermail/bonmin/attachments/20080930/654be9c2/attachment.html