For more details about restarting Nelder-Mead (e.g. fminsearch) to improve its convergence, see the description and file on http://www.mathworks.com/matlabcentral/fileexchange/33328

For more details about restarting Nelder-Mead (e.g. fminsearch) to improve its convergence, see the description and file on http://www.mathworks.com/matlabcentral/fileexchange/33328

Note that the same results are achieved by using fminsearch instead of fminunc in itaeoptimtf. fminsearch is the Matlab implementation of the Nelder-Mead algorithm (1965), a direct search method not using gradient information. This indicates that DS methods are adequate for time response shaping problems, and i used this for settling time minimization (for which gradients are not available and fminunc will fail at using finite differencing derivative estimations), see files #33022.

For more details about restarting Nelder-Mead (e.g. fminsearch) to improve its convergence, see the description and file on http://www.mathworks.com/matlabcentral/fileexchange/33328

For more details about restarting Nelder-Mead (e.g. fminsearch) to improve its convergence, see the description and file on http://www.mathworks.com/matlabcentral/fileexchange/33328

For more details about restarting Nelder-Mead (e.g. fminsearch) to improve its convergence, see the description and file on http://www.mathworks.com/matlabcentral/fileexchange/33328

For more details about restarting Nelder-Mead (e.g. fminsearch) to improve its convergence, see the description and file on http://www.mathworks.com/matlabcentral/fileexchange/33328

Second version available, heavily commented.
See the short preprint on http://arxiv.org/abs/1110.2615 for important points.
See the different preprint on http://arxiv.org/abs/1104.5183 for some other experiment results.

Comment only