[Ipopt] Fwd: help for restoration failure

[Tony Kelman]

You know more about your model than I do, and I’ve never used APMonitor for modeling. You may want to ask there. The fact that it says your model has negative 2 degrees of freedom sounds bad though. If random initial guesses are giving the “Problem may be infeasible” exit message, then again it’s indicative that finding a feasible point for your optimization problem is challenging. If there are simpler calculations you can do to obtain a close-to-feasible solution to start from (and I mean taking into account the structure and constraints of your problem, something much more sophisticated than all zeros or a random guess), you will likely need to do that to get reliable behavior on this particular problem.



From: Mostafa Safdarnejad 
Sent: Saturday, July 19, 2014 3:21 PM
To: Tony Kelman 
Cc: ipopt 
Subject: Re: [Ipopt] Fwd: help for restoration failure

Tony,  

Thanks for your response. I am sure the problem is feasible because I have other scenarios running perfectly fine. I changed some concepts of my model and I got this infeasibility. Although the model is infeasible now, the trends that variables are making are quite reasonable (however I am not sure how reliable trends for an infeasible case can be).  I am using zero for almost all of the initial values (as other feasible scenarios) but using other random initial values got me to " Converged to a point of local infeasibility. Problem may be infeasible." but no restoration failed error at the end of the output (however there is "r" next to the iteration number. The output of solver is attached. Do you have any further insight on what is going on? 

Thanks

Mostafa Safdarnejad



On Sat, Jul 19, 2014 at 3:58 PM, Tony Kelman <kelman at berkeley.edu> wrote:

  Essentially this means Ipopt had difficulty obtaining a feasible point for your problem. Are you sure your problem is feasible? Would it be possible for you to calculate a better initial starting point for the optimization?


  From: Mostafa Safdarnejad 
  Sent: Friday, July 18, 2014 11:55 AM
  To: ipopt at list.coin-or.org 
  Subject: [Ipopt] Fwd: help for restoration failure

  Hi, I am using IPOPT solver for a NLP optimization and I got an infeasibility error due to restoration failure (attached file). I am not familiar with programming in IPOPT. So, I would appreciate if anyone can help me with specific details on how to fix it? I am using APMonitor modeling language and Matlab as my interface to run.  

  Thanks



------------------------------------------------------------------------------
  _______________________________________________
  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/20140719/348fd6bd/attachment.html>