The script quasi_newton_dfp.m optimizes a general multi variable real valued function using DFP quasi Newton method. During the iterations if optimum step length is not possible then it takes a fixed step length as 1. For the theory any good book on optimization techniques can be consulted. This script is also useful in checking whether a given function is convex or concave and hence globally optimizable.
If the function is of dimension 2 and is convex then it works as classical Newton method giving one step convergence.
Hi, this script was very help- and useful for my use case. But one thing i couldn't solve by myself. Do somebody know who i could create the 3D-plot and the contour lines smoother? I couldn't found a property in the surfc-command which could solve my issue.
Please can you give us an example for using this script, I'm trying to understand something because i have to write a script for fminlbfgs, broyden method and another two and i'm trying to compare the results!
Not a function, so it pollutes the Command Window workspace. Prompts the user for all input including the function to be optimized! Does not require symbolic math toolbox as stated. Has at least one triple nested For loop. Does not appear to use any vectorization.