Open jucor opened 11 years ago
Come to think of it, I wonder if it is really a good idea, since the inverse is not numerically stable: it might be better to just use the pivot method every time as we currently do. Have to double check with @shakirm for his thoughts.
Ping @dpfau : we agreed on Friday that this was probably indeed a good idea :) I just realize I made this note to myself above. Any thoughts on it? Regardless, since many packages offer this, we should implement it.
Agreed it would be nice - more control over when you need to do a matrix inversion is never a bad thing.
Issue by jucor from Friday Nov 01, 2013 at 18:30 GMT Originally opened as https://github.com/jucor/torch-randomkit/issues/6
Suggested by Shakir.