On 10/9/2013 10:21 AM, Vsh wrote: > Hello all, > > I'm working on a minimization problem on three different "types" of > variables of different physical nature. To be more specific, I have > stiffness tensor subjected to positive, definiteness constraints > derived from the elastic strain energy. This contains both linear and > non-linear parts. Secondly, I have density which can be constrained > (or bounded??) by a value between [0.8,1.6] g/cc. And thirdly i have > another isotropic elastic moduli that I can specify a reasonable range > for. I'm not sure how to set up the constrained function or in fact > the whole minimization problem given the multiple variable spaced > involved in this. Any help or references would greatly be appreciated. > Thanks.
Generally speaking, fmincon takes a single vector x as the unknowns. If you have a stiffness tensor T, a density p, and elastic modulii m, then take x = [T(:);p(:);m(:)]; What I mean is, turn your various matrices or tensors into column vectors with the (:) notation, and concatenate them to be a single vector of unknowns x. After you get your solution, you can disentangle the pieces by taking various ranges of x.