[Ipopt] Ipopt optimal solution is not feasible

[Carlo]

Hi,
I am trying to solve an optimal control problem with Ipopt in C++, but I am facing a particular issue, that seems to be undiscussed on this mail list.
I want to generate a trajectory with starting and ending condition. This works in some cases with Ipopt, and this makes me thinking that I have no bugs in the code.The problem is that for particular trajectory requirements, even if the program seems to find the optimal solution (i.e. inf_pr = 1e-10), this is not the case, since the optimized trajectory doesn't satisfy one or more constraints (usually not that much error, but definitely more than 1e-10 in my measurement unit). Could this be related to some of the Ipopt options? I am using CppAD for the automatic gradient.
Thank you in advance for your answers.
Best,Carlo 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/ipopt/attachments/20160112/bc5e890d/attachment.html>