Non-Negative Least Squares - What is $A^P$?

Question

I have been reading up on the NNLS algorithm and I couldn't understand the notation of A^P. Does it mean that only the values common between A and P will be used in A. How will the resultant matrix be? https://en.wikipedia.org/wiki/Non-negative_least_squares#Algorithms

Answer 1

In the algorithm you mention in the link, $A\in\mathbb{R}^{m\times n}$ and $P\subseteq\{1,\ldots,n\}$. $A^{P}$ denotes a $m\times|P|$ matrix constructed using columns of $A$ indexed by $P$.