[Ipopt-tickets] [Ipopt] #213: Repeatability of optimization runs with IPOPT

[Ipopt]

#213: Repeatability of optimization runs with IPOPT
--------------------------+------------------------
Reporter:  florian.saupe  |      Owner:  ipopt-team
    Type:  defect         |     Status:  new
Priority:  normal         |  Component:  Ipopt
 Version:  3.11           |   Severity:  normal
Keywords:  repeatability  |
--------------------------+------------------------
 Dear IPOPT Team,
 I am currently using your code for an MPC control and I have issues with
 the repeatability of the results obtained with IPOPT.
 I interface IPOPT v3.11.1 using the OPTI Toolbox on an 64bit windows
 machine with an i7 processor.
 Everytime I solve my problem I get slightly different results. Over the
 horizon of the MPC problem, they add up such that in the end they are
 significantly different. I uses another solver interfaced via the OPTI
 Toolbox which does not show this issue (but it is much slower). Are there
 operations in IPOPT which include random numbers? Can they be deactivated?

 For one problem setup, if I run the 'derivative_test', it seems to yield
 different results each time. In that setup I am not confident that I have
 calculated the gradient correctly. After reading the documentation I still
 did not now how to choose the point at which the gradient is tested and I
 do not know if that test is supposed to be deterministic i.e if that test
 uses random test points.

 Thank you in advance for your help.
 Regards,
 Florian Saupe

-- 
Ticket URL: <https://projects.coin-or.org/Ipopt/ticket/213>
Ipopt <http://projects.coin-or.org/Ipopt>
Interior-point optimizer for nonlinear programs.