[Coin-osi-devel] loadProblem / default row parameters => different constraints

[Matthew Saltzman]

On Thu, 2007-07-26 at 22:46 -0400, Laszlo Ladanyi wrote:
> No conscious decision, just what is aesthetically pleasing as default for each
> kind. For lb/ub you want symmetry, that's why it's -inf/inf. For sense/rhs
> range, a 0 rhs apeals the most. So now we have a discrepancy :-(. 

I suspect the sense/RHS/range version was designed to act like CPLEX or
XPRESS.

> 
> I think that they should behave the same way. I'd say we should go for >=0,
> since that's what happens if rhs is not specified in an MPS file.

That's probably the right way to go.  Note that this affects the user
interface though, so it becomes a versioning issue.

		Matt

> 
> --Laci 
> 
> On Thu, 26 Jul 2007, Lou Hafer wrote:
> 
> > Folks,
> > 
> > 	While polishing the Osi unitTest, this little anomaly slapped me in the
> > face.
> > 
> > 	There are two versions of loadProblem.  One describes row bounds with an
> > upper and lower bound; the other with a sense, rhs, range triplet.  The default
> > values specified in the documentation will produce different constraints:
> > 
> >   lb		-infinity	=>	-inf <= ax <= inf
> >   ub		 infinity
> > 
> >   sense		 'G' ( >= )
> >   rhs		 0		=>	ax >= 0
> >   range		 0
> > 
> > This isn't necessarily bad, but I'm wondering if it was a conscious design 
> > choice. Does anyone's memory go this far back?
> > 
> > 							Lou
> > 
> > _______________________________________________
> > Coin-osi-devel mailing list
> > Coin-osi-devel at list.coin-or.org
> > http://list.coin-or.org/mailman/listinfo/coin-osi-devel
> > 
> 
> _______________________________________________
> Coin-osi-devel mailing list
> Coin-osi-devel at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/coin-osi-devel
-- 
                Matthew Saltzman

Clemson University Math Sciences
mjs AT clemson DOT edu
http://www.math.clemson.edu/~mjs