sparse-linear-algebra icon indicating copy to clipboard operation
sparse-linear-algebra copied to clipboard

Either Floats are valid elements or they aren't.

Open gilgamec opened this issue 7 years ago • 1 comments

The Elt class includes Floats, while they PrintDense class works only for things over Doubles. This means that some functions, like lu, which requires only Elts, work for matrices of Floats, while e.g. qr (needing PrintDense) works only for matrices of Doubles. Some of the constructors, e.g. vr and vc, also only work with Doubles.

Could you either allow Floats as elements everywhere, or disallow them completely? Or at the very least, more cleanly delineate exactly what operations (exact solvers / iterative solvers / other stuff) they will be allowed in?

gilgamec avatar Sep 05 '18 17:09 gilgamec

@gilgamec Hi, good catch; this is inconsistent. The API has been in need of attention for quite some time now. If you have an idea on how to fix this I'll happily merge a PR.

ocramz avatar Sep 06 '18 07:09 ocramz