Re: donlp2 problem
- From: spellucci@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx (Peter Spellucci)
- Date: Wed, 25 Jan 2006 12:14:52 +0000 (UTC)
In article <1138128737.285967.194400@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
MechYouDee@xxxxxxxxx writes:
>Hi,
>
>I am using donlp2 to solve a very simple problem. My objective function
>is just the sum of the variables. I modified the code so that it would
>print "fx" after the optimization. However, almost every time I got the
>result that fx = 0. I add the following line to function solchk()
>
> printf("\noptimal value of f = %lf\n",fx);
>
>To make sure that the objective function works, I add another line
>
>printf("Test here");
>
>in the function ef(), where I defined my objective function.
>
>And I can NOT see the test words. So it seems that the function of ef()
>did not work at all. Does anybody know why? Thanks in advance.
>
>Best,
>Ming
>
it may be running in the Phase1, where f is never called. switch on te0
you should see the phase1 running on stdout, maybe it fails to find a
sufficiently feasible solution. sometimes increasing tau0 helps
or better a better initial guess. there are many harmless looking examples
even well defined, for which a bad initial guess produces failure in any
NLP solver (e.g. if penalization does not improve feasiblility)
hth
peter
.
- Follow-Ups:
- Re: donlp2 problem
- From: MechYouDee
- Re: donlp2 problem
- References:
- donlp2 problem
- From: MechYouDee
- donlp2 problem
- Prev by Date: Re: Newton Solver library needed
- Next by Date: Re: Preliminary advice on finite volume methods
- Previous by thread: Re: donlp2 problem
- Next by thread: Re: donlp2 problem
- Index(es):
Relevant Pages
|
