[Ipopt] acceptable tolerance options

[Tony Kelman]

In order for Ipopt to exit with the “Solved to acceptable level” message, the tolerance from acceptable_tol as well as all the other acceptable_*_tol options (see http://www.coin-or.org/Ipopt/documentation/node41.html#SECTION000112070000000000000) have to be satisfied for acceptable_iter iterations in a row. You can try reducing acceptable_iter, and/or loosening some of the acceptable_*_tol options.

-Tony


From: Naumann, Steffi 
Sent: Wednesday, December 18, 2013 5:25 AM
To: ipopt at list.coin-or.org 
Subject: [Ipopt] acceptable tolerance options

Hi,

 

I’m new with Ipopt and I’m trying to find good termination criteria for my optimization problem. In some cases the problem got solved with success. In other cases solving the problem exceeds maximum CPU-time, but the results seem to be adequate. In these cases I tried to introduce the “acceptable tolerance” options. When I set the acceptable tolerances to values which get fulfilled in the optimization-run without acceptable tolerance options, the solver does not stop with the considered “Solved to acceptable level” statement. Instead of that it exceeds again the time-limit and ends with much worse results.

 

Is it possible that the acceptable tolerance options have an influence to the result of the optimization?

 

Thanks

Steffi

 

M.Sc. Steffi Naumann (geb. Pflugradt)

 

Institutsteil Angewandte Systemtechnik (AST) des Fraunhofer IOSB

Am Vogelherd 50

D-98693 Ilmenau

 

Tel.:            +49 3677 461-104 
Fax:            +49 3677 461-100

E-Mail:       steffi.naumann at iosb-ast.fraunhofer.de

Internet:   www.iosb-ast.fraunhofer.de

 



--------------------------------------------------------------------------------
_______________________________________________
Ipopt mailing list
Ipopt at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/ipopt
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/ipopt/attachments/20131218/78149989/attachment.html>