Hessian in fmincon

Dear Members,

   I am using fmincon for the optimization of parameters. fmincon has nicely evaluated the optimized parameters. For the sensitivity analysis, I need to compute the Hessian. I have observed that the Hessian in fmincon is the Hessian of the Lagrangian, but I want to have the Hessian as the matrix of second derivatives of the objective function f (as in fminunc). How Can I evaluate this matrix using fmincon. I would be really thankful to you for giving me the answer. Any idea is welcome.

Thanks

Best Regards

Answer 1

[Karthik Vemireddy]

select the option under derivatives as 'Gradient and Hessian supplied' for Trust region reflective algorithm or 'Gradient supplied' for the other algorithms.

Write your objective fcn as described in the following doc (see section 'including derivatives') :

http://www.mathworks.de/help/toolbox/optim/ug/brhkghv-3.html#brhkghv-4

You can use Trust region algorithm r interior point algorithm, if they suit ur problem

 r try meddling with the hessian fcn by giving your own fcn handle r forcefully giving  a user supplied fcn (just a suggestion, not sure if it wud work)

http://www.mathworks.de/help/toolbox/optim/ug/fmincon.html#f186882