These are chat archives for thunder-project/thunder

Apr 2015
Johannes Friedrich
Apr 29 2015 00:17
Hi, has anybody an idea how to left-multiply a RowMatrix with a diagonal matrix, or equivalently multiply each row/record by a different scalar? For right-multiplication I simply do a RowMatrix.dotTimes(v) where v is the vector (np.array) of diagonal elements.
Coded up factor analysis for the case that each row is a sample and the columns features. I’d like to be able to also regard each row as feature, but want to avoid to transpose the big RowMatrix. Maybe I ran into a similar issue as @rhofour.