Matt J
Posts:
4,994
Registered:
11/28/09


Re: Nonlinear optimization
Posted:
Mar 7, 2013 3:48 PM


"Toan Cao" <toancv3010@gmail.com> wrote in message <khatls$bf4$1@newscl01ah.mathworks.com>... > > Hi Matt J, > > Thanks for your consideration! > Yes, i know what you mean in order to implement easily for my problem. However, in limited space of the text, i can not explain why i need rotation matrices to describe movements of points. It is reason i just explain main idea of the cost function. > If you would like to know more, i can show the paper i read and want to implement, its title: "Embedded deformation for shape manipulation". Its cost function is expressed in section 4. > So, when applying LSQNONLIN, i am confused which option (1 or 2) i use for f(x)=[f{1}(x),f{2}(x),...,f{n}(x)]' ? > I am willing to receive any your suggestion for my case. =======
Toan,
The square roots are bad, because they introduce nondifferentiabilities into the cost function.

