[Ipopt] infeasible start-solution

Stefan Vigerske stefan at math.hu-berlin.de
Sun Feb 14 09:54:36 EST 2010


Ipopt might move your initial point inside the box before starting to
iterate, see the options bound_frac and bound_push in


Am 12.02.2010 16:16, schrieb Claudia Stangl:
> hallo,
> I've got a lack of understanding:
> The objective function of my problem is only 1, so there is nothing to 
> optimize and I'am only interested in feasibility.
> When solving the problem I get a feasible solution.
> If I use this feasible solution as starting point in the function 
> get_starting_point IPOPT needs further 8 iterations to solve the problem 
> and to get (another) feasible solution. I gave the starting point to 
> IPOPT with setprecision(20) so that I think that it is correct enough.
> Further more you can see in the output that the mistake before iteration 
> 0 is less than 1-9 and then it gets bigger. Why is this (feasible) point 
> not feasible bevore iteration0?
> Can somebody explain this phenomenon?
> I use Ipopt version 3.8.1, running with linear solver mumps.
> It would be very nice to get an answer!
> kind regards
> claudia stangl
> _______________________________________________
> Ipopt mailing list
> Ipopt at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/ipopt

More information about the Ipopt mailing list