[Ipopt] Fmincon vs Ipopt

Andreas Waechter waechter at iems.northwestern.edu
Wed Jun 9 14:27:55 EDT 2021

Hi Hisham,

I would run Ipopt''s derivative checker to make sure there are no errors 
in the first derivatives.



On 6/6/21 9:20 AM, Hisham Abdulla wrote:
> Dear all,
> I'm trying to solve a nonlinear convex minimization problem with 
> linear constraints which I have
> solved successfully using MATLAB fmincon.
> In fmincon, I used the  option configuration  as below
> *options=optimset('Algorithm','sqp','ScaleProblem','obj-and-constr','DerivativeCheck','off','GradObj','on','TolX',1e-16,'TolFun',1e-16,'MaxFunEvals',8000,'MaxIter',6000); 
> *
> and I got local minima after 465 iterations.
> In Ipopt, I have modelled my objective function, constraints and its 
> gradients without errors, but couldn't solve the problem. I used 
> various linear solvers and scaling. The magnitude of variables in the 
> problem ranges from 10^-3 to 10^3 and I need to get the value of local 
> minima below 10^-9.
> The problem doesn't converge to local minima even after 200000 
> iterations.
> Even after I used the solution which I got from matlab as initial 
> values, Ipopt still iterates and doesn't converge.
> Since I am not from a mathematical background I don't have much 
> knowledge about Ipopt options.
> Could anyone help me please.
> Thank you
> Hisham Abdulla
> _______________________________________________
> Ipopt mailing list
> Ipopt at list.coin-or.org
> https://list.coin-or.org/mailman/listinfo/ipopt

Andreas Waechter

Department of Industrial Engineering and Management Sciences
McCormick School of Engineering
Northwestern University
Evanston, IL 60208

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/ipopt/attachments/20210609/3974c63b/attachment.html>

More information about the Ipopt mailing list