尋找函數的全局極小值，global minimization of contrast function with random restarts the data are assumed whitened (i.e. with identity covariance matrix). The output is such that Wopt*x are the independent sources.
Included are the files wav1.m, wav2.m, wavecoef.mat and readme.
wav2 function implements the tree structured wavelet transform of the input matrix, up to the given level of decomposition. Wav2 uses another function called wav1, which takes the well known wavelet transform of the given matrix. Daubechies wavelet coefficients are used for wavelet transform operation wahich is saved in wavcoeff.mat.
使用pso求最小化一函數
matlab程式碼，寫的非常簡潔(不到100行)，且還包括了2維的圖形展示，和大家分享參考!!!
一起學習matlab和各種optimize methods
最小化：(x-15)^2+(y-20)^2
The swarm matrix is
swarm(index, [location, velocity, best position, best value], [x, y components or the value component])
Author: Wesam ELSHAMY (wesamelshamy@yahoo.com) MSc Student, Electrical Enginering Dept., Faculty of Engineering Cairo University, Egypt
平均因子分解法，適用于正定矩陣First, let s recall the definition of the Cholesky decomposition: Given a symmetric positive definite square matrix X, the Cholesky decomposition of X is the factorization X=U U, where U is the square root matrix of X, and satisfies:
(1) U U = X
(2) U is upper triangular (that is, it has all zeros below the diagonal).
It seems that the assumption of positive definiteness is necessary. Actually, it is "positive definite" which guarantees the existence of such kind of decomposition.
The MatlabBGL library fills a hole in Matlab s suite of algorithms. Namely, it provides a rich set of algorithms to work with graphs, as in graph theory graphs. The MatlabBGL package uses Matlab s native sparse matrix type as a graph and provides algorithms that work
