[Clp] Infinite loop in Clp

William H. Patton pattonwh at comcast.net
Wed May 7 18:48:50 EDT 2008


>>Is there a way to see what coin rescales my problem to?
I see in clp\src
presolveToFile  controlled by param parser
	    case PRESOLVE:
	      if (action==0)
		preSolve = 5;
	      else if (action==1)
		preSolve=0;
	      else if (action==2)
		preSolve=10;
	      else
		preSolveFile=true;
	      break;
and
	      solveOptions.setSolveType(method);
	      if(preSolveFile)
		presolveOptions |= 0x40000000;
	      solveOptions.setSpecialOption(4,presolveOptions);

	bool presolveToFile = (presolveOptions&0x40000000)!=0;

-----Original Message-----
From: Kim Hansen [mailto:kimhanse at gmail.com] 
Sent: Wednesday, May 07, 2008 3:53 AM
To: pattonwh at comcast.net
Cc: Clp
Subject: Re: [Clp] Infinite loop in Clp

On Wed, May 7, 2008 at 5:55 AM, William H. Patton <pattonwh at comcast.net>
wrote:
> It is nice to avoid inf loops. But I think the issue was difficult scaling
> in the original problem.  So it could well show up again if you change the
> e-008 to e-010 on C0000081 and C0000085.

I expected Coin to be able to rescale my problem for me, but that
might be a misunderstanding?

I believe that my problem should be possible to rescale even though it
has very large and very small numbers, they should be grouped together
such that rescaling should be possible.

Is there a way to see what coin rescales my problem to?

-- 
Kim Hansen
Vadgårdsvej 3, 2.tv
2860 Søborg
Fastnet: 3956 2437 -- Mobil: 3091 2437





More information about the Clp mailing list