James Lucas

I haven't tested myself but it looks like there is some mention of [OS X in the install docs](https://github.com/gperftools/gperftools/blob/master/INSTALL). I'm afraid that is all the help I can offer -...

Long comment on dead thread incoming, sorry! I agree with @t-ae that the current gradient penalty code does not enforce a 1-Lipschitz discriminator over patches. In this case, the discriminator...

I had a little time to look into this issue. It's really hard to find details of the implementation anywhere - but I checked out the [netlib/eispack](http://netlib.org/eispack/svd.f) code. Their algorithm...

Ticked off the problems tackled in #46 .

Thanks! Ticked that off too

I'd been thinking about the first alternative too - this is indeed the most common use case and may be worth exposing. However, I don't think it is a clean...

I meant the first solution only - in that case I think it would preferable to just support _column-major_/_stride wrapping_ across all operations instead of having a single exception for...

The real concern for me is how we handle the decomposition algorithms for both row-major and column-major. I think the only real solution is writing them twice - we can't...

I think this is a solid idea and a good way to stretch into supporting opt-in blas (and others) in the future. A few comments: > Functions should never allocate...

> but this still makes us have to deal with unsafe pointer-based APIs spread across the code base, which is also not entirely ideal. I agree that this wouldn't be...