I have an algorithm that outputs a vector 'v' for every iteration. With ever iteration the change in vector v will get smaller and smaller.
therefore my psuedo code should be
if vold-v < 0.001 end iteration
what is the right way to compare two vectors for this purpose? I believe i have to use SVD but can someone please give a tiny background on SVD and how i should implement it?
The mathematical thought is to use $$ \lVert v_{now} - v_{old} \rVert \lt \epsilon $$
Now you must choose a norm. If your language is Fortran, the command
norm(v)uses the Euclidean norm.Next, choose a threshold. In Fortran, the command
epsilonwill provide the $\epsilon$ factor for the indicated real precision.