Alfonso Landeros

Threaded Jacobians and Hessians

1

Hello! I am interested in using Polyster.jl to evaluate Jacobians and Hessians in parallel, similar to `threaded_gradient!` over [here](https://github.com/JuliaSIMD/Polyester.jl/blob/master/src/forwarddiff.jl#L45) using the ForwardDiff API. A parallel Jacobian is a higher priority...

`Pkg.add` changed and went unnoticed; we should use `PackageSpec`.

Branch `gh-pages` shows that everything was built correctly and "Latest" points to the correct version.

See #18. We should not require users to do `time = convert(Float64, ...)`. - [ ] Allow the keyword arguments to accept different types and handle as needed. - [...

Sometimes `svdsolve` throws a `DomainError` on v0.5.4 and now v0.6.0: ```julia using LinearAlgebra, KrylovKit, StableRNGs m, n = 4100, 4200 X = randn(StableRNG(1234), m, n) # should work for r

LoopVectorization fails on some operations?

3

Reproducing the issue: ```julia using SnpArrays mouse = SnpArray(SnpArrays.datadir("mouse.bed")) A = SnpLinAlg{Float64}(mouse) A*A' ``` This results in the following message: ``` ┌ Warning: #= /home/alanderos/.julia/packages/SnpArrays/WBzpL/src/linalg_direct.jl:760 =#: │ `LoopVectorization.check_args` on your...