[Ipopt] How to reduce number of gradient/jacobian calls

[Praveen C]

Hello
I am using ipopt for shape optimization using cfd and adjoint method to
compute gradients. Since gradient/jacobians are expensive to compute, is
there any way to reduce to number of times ipopt requests them, by choosing
appropriate options in ipopt ?

My problem involves minimizing a nonlinear function with one nonlinear
equality constraint. There are 20 design variables and I use the
quasi-newton approximation. The jacobian is full.

For mu_strategy, I have tried between "monotone" and "adaptive" and I find
monotone is faster for me, needs less number of function/gradient
evaluations (seems opposite to what I have read).

Thanks
praveen
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.coin-or.org/pipermail/ipopt/attachments/20090915/a45bfd5f/attachment.html