Hi. I came across this: http://www.cs.nthu.edu.tw/~jang/book/addenda/matinv/matinv/
[I've implemented it, hoping that it would help me with a matrix that involved blocks having very large magnitude differences. But it turned out it didn't.] Would someone see a use-case/unit test where this algorithm would produce a more accurate result than a straight inversion (i.e. where the matrix is taken as a single block)? If so, I'd include the code in "MatrixUtils". Regards, Gilles --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
