[Ipopt] question about convergence problem of Ipopt

Andreas Waechter andreasw at watson.ibm.com
Fri May 21 09:18:42 EDT 2010


Hi,

The answer is "NO".  The reason range from the problem not satisfying 
certain assumptions to numerical difficulties to possible bugs in 
imperfect code.  This is true for an nonlinear optimization code, and 
doesn't depend on whether the algorithm is looked for a local or a global 
solution.

Andreas

On Fri, 21 May 2010, Lqc wrote:

> Hi all:
         I used the Ipopt 3.6.1 for my application.When I used different 
data sets for my problem, sometimes Ipopt can't solve the problem. While 
used other optimization solver on the same dataset, the problem can be 
solved.
        My question is that if a problem can be solved, could the Ipopt be 
sure to solve the problem? The answer is "YES" because of the "global 
convergence" of Ipopt? Or the answer is "NO" because of the impact of 
other unconspicuous matter?
       Thank you very much in advance!

Regards

LiuFuxing



More information about the Ipopt mailing list