[Ipopt] Using Ipopt for problems without analytical derivatives

Ariyan Kabir akabir at usc.edu
Mon Jul 16 14:03:27 EDT 2018


I have formulated some of my problems as *constrained optimization* and
some as *unconstrained optimization*. In both cases, the upper and lower
bounds for the variables are defined. However, I *do not have analytical
gradient, jacobian, and hessian available* for my problems.

I have some initial results using interior point method from MATLAB's fmincon.
I am looking for a c++ library for faster computation time. I came across
Ipopt and it looks promising for my problem.

I am new to Ipopt. I have installed the library and played with the example
problems. I need some help before I can proceed with using Ipopt for my

   - Is there any way I could *use Ipopt without writing the analytical
   gradient, jacobian, and hessian*?
   - Does the library currently support *numerical approximation* for them?
   If yes, could you please point me to the right documentation/example?
   - Moreover, can I use Ipopt for *unconstrained optimization* problems?

I will highly appreciate if you could help me kick-start using Ipopt in my
project. I look forward to your reply.

Best regards,

Ariyan M Kabir
PhD Student and Graduate Research Assistant
Department of Aerospace and Mechanical Engineering
Viterbi School of Engineering
University of Southern California
Los Angeles, CA 90089
Email: akabir at usc.edu, ariyan.kabir at gmail.com
*+1 (703) 474-2784*
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/ipopt/attachments/20180716/d1c2d48f/attachment-0001.html>

More information about the Ipopt mailing list