SPAMS note

m x N K the row lendth of dictionay m verbose , if comment verbose mean show progress L the column length of dictionary (dont have to) N

looks like simple trach and me use 1e0 / 1e15 larger one but not small one 1e-9

————— used in my project to train GPS
if exist('limit','var') == 0, limit = 8; end;

param.K = limit ; % change limit*2 to limit so the size square
param.iter = 500; %
param.modeParam = 0;
param.mode = 0; % done the testing with 1,2,3,4. do not produce good results.
param.posAlpha = 0;
param.posD = 0;
param.whiten = 0;
param.L = param.K * 2;


X = xData;

% for i = 1:size(xData,2)
% X(:,i) = X(:,i) - mean(X(:,i));
% end

num_bases = limit * 2;

param.lambda = 1e9;

Psi = mexTrainDL(X, param);

Psi = full(Psi);

———————————