I'm trying to use function "solve" to find the inverse of a matrix, which is positive definite in theory. But for some situations, I get error message because of computational singularity. I'm wondering if I can set the tolerance for detecting linear independence to an infinitely small value to avoid this problem. Is there any serious consequence by doing this? Are there other solutions to this problem? Thanks. -- View this message in context: http://www.nabble.com/Computational-singularity-of-a-positive-definite-matrix-tp17702809p17702809.html Sent from the R devel mailing list archive at Nabble.com.