[Ipopt-tickets] [Ipopt] #203: User-provided starting point is modified

[Ipopt]

#203: User-provided starting point is modified
----------------------+------------------------
Reporter:  baiyang11  |      Owner:  ipopt-team
    Type:  defect     |     Status:  new
Priority:  highest    |  Component:  Ipopt
 Version:  3.9        |   Severity:  major
Keywords:             |
----------------------+------------------------
 I'm trying to solve a NLP by Ipopt-3.9.2 using C. The idea is that I can
 get a near-optimal solution (feasible, of course) elsewhere for the
 problem, so I want to use it as a starting point, which may accelerate the
 computation to get the "optimal" solution, as I suppose.
 Therefore, before the function "IpoptSolve(nlp, x, NULL, &obj, NULL,
 mult_x_L, mult_x_U, NULL)" runs, I assign the corresponding value of the
 near-optimal solution to the vector x. However, when the IpoptSolve runs,
 output information shows that the objective of the iterate 0 is worse than
 that of near-optimal solution. I debug the code and find that before
 iterate 0 is running, the values of x have already been modified,
 especially that many variables I assign to 0 are modifed to the same value
 0.00999, which means the starting point is not the one I provide.
 I have no idea why this happens yet and how to realize my idea(using near-
 optimal solution as starting point to accelerate computation). I'am
 looking forward to your help! Thanks very much!

 PS.
 (1)Option "mu_strategy" is set to "adaptive";Option "linear_solver" is set
 to "MA27" .
 (2)I believe I provide the Jacobi and Hessian matrix right.

-- 
Ticket URL: <https://projects.coin-or.org/ticket/203>
Ipopt <http://projects.coin-or.org/Ipopt>
Interior-point optimizer for nonlinear programs.