[Coin-lpsolver] Pb of infeasibility? in BCP

[Francois Margot]

Mathieu:

On Thu, 4 May 2006, mathieu lacroix wrote:

> Hi,
> Thank you for your answer. I will test it. But I have some problems about 
> precision and tolerance with CLP.
> Is the precision of the LP solver (CLP in my case) function of the parameters 
> of BCP  (I have just change  these two parameters in Bcp :  BCP_Granularity 
> .99 and BCP_IntegerTolerance 1e-6) (That is parameters are automatically 
> defined relatively to BCP's parameters)? Or do I need to set the precision to 
> the lp solver or to OSI module?

I believe that there is (unfortunately) no transfer for precision settings
between Bcp and the Lp solver. In any case, neither the integer tolerance
nor the granularity would affect the Lp solver return code.

The default settings should be good enough. If you get abandoned Lp, 
something is likely wrong with the Lp formulation.

Francois