Sheehan Olver

> While mathematically an infinity isn't a real number As you mentioned: Julia types are not the same as mathematical sets. I think they are best thought of as encoding...

If that’s the case then the help for ` norm ` should be updated o mention this plus a reference. Note Wikipedia has a discussion on how it is in...

It's been pointed out to me that $\lim_{p \rightarrow 0} ||x||_p^p$ does converge to the number of nonzero entries: ```julia julia> x = big.(randn(5)); julia> norm(x, 0.0001) 4.357267652943236549048218780568178713688531928129028744484491459811519165098176e+6989 julia> norm(x,...

Is the restriction to `TypedEndpointsInterval` needed anymore?

I've just tagged a new version, hopefully that fixes it

This discrepency is not present when I call `using AppleAccelerate` so it's presumably a change in OpenBLAS.... Probably it's not so critical so I can modify the tests in LazyArrays.jl...

I think the fact that x86 preserves order is suggesting it is indeed a bug. Though I agree it’s not so important

Here's another surprising inconsistency where LU solves seem to use a different implementation of forward/backward substitution than triangular solves: ```julia julia> A = [0.1 1 1; 2 4 8; 1...

You should include the outputs so we can see what the difference in timings are. Why do you expect it to "outperform"?

I believe the relevant code is here: https://github.com/JuliaArrays/LazyArrays.jl/blob/3ecdcd03d2ee23622c711e5362ae503a166f3a5e/src/lazybroadcasting.jl#L108 It might have to do with the use of `@simd`, that is, without conditionals it is SIMD-able. My only thought is that...