add logabstanh
#86
add logabstanh
#86
Conversation
src/basicfuns.jl
Outdated
| The implementation ensures `logabstanh(-x) = logabstanh(x)`. | ||
| """ | ||
| function logabstanh(x::Real) | ||
| return log1p(-2/((exp(2abs(x))+1))) |
There was a problem hiding this comment.
As indicated by the special cases close to 0 for Float32 and Float64, maybe a safer default would be
| return log1p(-2/((exp(2abs(x))+1))) | |
| return log(abs(tanh(x))) |
There was a problem hiding this comment.
I think the version I wrote stays more accurate overall but I'll do a test
There was a problem hiding this comment.
I don't think it's possible to do such an analysis? You can't realistically test any subtype of Real.
There was a problem hiding this comment.
for any AbstractFloat log(tanh(x)) will be inaccurate for large x since tanh(x) will be very close to 1.
Co-authored-by: David Widmann <[email protected]>
Co-authored-by: David Widmann <[email protected]>
Co-authored-by: David Widmann <[email protected]>
Co-authored-by: David Widmann <[email protected]>
|
Something missing here? Or can this be merged ? |
|
|
Should I re-open and rebase to add Float32 and Float64 fast-paths? |
|
It might be simpler to just revert 39127b6 instead of rebasing. For a start I mean. |
|
@oscardssmith: choose whatever is most convenient for you, but having these methods would be very welcome. The new testing framework should be sufficient for tests. |
|
can you reopen? I seem to not have permission to. |
|
Unfortunately I am unable, Github says "that these commits are already merged." |
|
Yeah, all of them are part of the merged PR. |
|
ah. Will re-create then. |
5 participants
as requested by @c05 on slack. This has 15 ULP for Float64, and 5 ULP for Float32.