main.m

% -------------------------------------------------------------------
% Non-Linear Optimization
% Author: Thiago Lima Silva (thiagolims@gmail.com)
% -------------------------------------------------------------------

% Make sure the workspace is clean before we start
clc
clear
clear global

addpath(genpath('library'));
addpath(genpath('algorithms'));
addpath(genpath('testFunctions'));

%% 
% intial point
x0=[-1; 2]; %%


% Ghraphic: plot the function to be minimized
xmin = -2;
xmax = 2;
dx = .05;
ymin = -1;  
ymax = 3;
dy = .05;

[z,y]= meshgrid(xmin:dx:xmax, ymin:dy:ymax);
banana=10*z.^4 - 20*(z.^2).*y + 10*y.^2 + z.^2 -2*z + 5;
contour(z,y,banana,30);
axis('square');
hold all;

%% Creating an automatic diff object
% [x,y] = initVariablesADI(x0(1),x0(2));
% xk = [x;y]; %% initial point with ADI

xk = x0;

%% Anonymous functions 
fbanana = @(xk) fban(xk);
gbanana = @(xk) gban(xk);
hbanana = @(xk) hban(xk);

fsquare = @(xk) squareX(xk);

%% Search direction method
optType = 4; % SteepestDescent = 1, Newton = 2, DFP = 3, BFGS = 4, SR1 = 5 

%% Line search type
lsType = 1; % Armijo = 1, ArmijoGoldstein=2, Polynomial = 3

%% Invoking optimization algorithm
tic 
% for i=1:100
    [xs, fs, gs] = gradientDescent(fbanana, gbanana, hbanana, xk, optType, lsType);
% end
toc