[Coin-bcpdiscuss] Core cuts and vars

Matteo Salani matteo.salani at epfl.ch
Wed Nov 8 11:51:17 EST 2006


Laszlo Ladanyi wrote:
> Hi Matteo,
>
> You seem to be doing things correctly (at least the idea is correct :-). So
> there is a bug either in BCP or in your code.
the latter with prob. 99,9%
>  The assign method picks up the
> number of cols and rows (i.e., the lengths of the other arrays) from the
> CoinPackedMatrix mat. So check first that everything is correct there. If you
>
>   
Ok so here's the problem!
since the core matrix of Bcp doesn't have any artificial variable for 
the m constraints (i.e. no "ones" from n to n+m) the CoinPackedMatrix 
will contain only indexes up to n.

if I got the point of CoinPackedMatrix is that  I will store only 
nonzero coefficients and indexes.  So why  the size of the LP should be 
taken from  the size of the CoinPackedMatrix which probably is not the 
right one?

Btw I'll try to add m artificial core variables and it should work 
unless I don't see why...

Many thanks for your quick answer!
Matteo
 



More information about the Coin-bcpdiscuss mailing list