Unconstrained optimization using Newton 1.0

Ok, yes, my expectation was that when I saw this, before I downloaded it I expected it to be a student's homework assigment. In fact, this is a reasonably carefully crafted tool.One serious issue is that this code requires you to provide explicit functional forms for the gradient and Hessian. There is a finite difference capability in the code, but it apparently does not run unless you provide at least a functional gradient or a Hessian matrix. (Unverified, because I could not get it to run when I tried it with an actual 2-d gradient function.) It called my function of two variables with a scalar input.Some minor comments:- While this claims to be a unconstrained solver, it actually allows bound constraints on the parameters.- I see that everywhere in the comments and help, there are references to this as an optimization code. But in no place does it actually say if it will minimize or maximize the objective. One would think this to be an important piece of information.Given the problems that I had getting this to run, it would seem that some working examples would be tremendously useful for most users.In general, if you have the optimization toolbox, I'd suggest you use it instead. If you don't have that TB, and you can get this function to run, it should do its job well. Given the time that the author has spent on writing this code, I do wish they had spent a little more time on making it usable by others.

Requirements:	No special requirements
Platforms:	Matlab
Keyword:	Bound,  Constraints,  Important,  Informationgiven,  Maximize,  Minimize,  Objective,  Optimization,  Parameters,  Piece,  Place,  Problems,  References,  Solver
Users rating:	0/10