logabsdet has a weird name

[ParadaCarleton]

It's not a bug per se, but logabsdet's name makes it sound like it's supposed to return the log-absolute value of the determinant. Would it make sense to create a log_det_sign or log_det_parity function and then document that it should be preferred to logabsdet?

[stevengj]

The name seems fine to me. It does return the log of the absolute value of the determinant; it just returns the sign too if you want it.

SAS calls it logabsdet. Numpy calls it slogdet (but documents one of the return values as logabsdet).

It's been around for quite some time in Julia (#3070) and hasn't seemed particularly problematic or controversial. It seems past the time for bikeshedding the name.

[ParadaCarleton]

Numpy calls it slogdet (but documents one of the return values as logabsdet).

slogdet would be a good name too. It follows the rule that function names should say what they do (sign and logdet).

SAS calls it logabsdet.

If it's common in other languages that seems fine to me.

It's been around for quite some time in Julia (JuliaLang/julia#3070) and hasn't seemed particularly problematic or controversial. It seems past the time for bikeshedding the name.

Allowing multiple names is not a breaking change.

[ararslan]

Allowing multiple names is not a breaking change.

...nor is it a good design.

[oscardssmith]

We don't seem likely to do this, so closing as not planned.