Andreas Noack
Andreas Noack
Reducing allocations is generally a good thing but my question is how much is reduced there. For a general toeplitz, you'd just swap the two vectors (e.g. in the real...
Sure. Wrapped types just add a lot of complexity and also potential compilation overhead so it's just good to avoid them when it free or cheap enough.
Superseded by https://github.com/JuliaStats/StatsBase.jl/pull/816
> So if a user provides two tstops which are closer than dtmin, then it will error. Notice that this is not what happens here. The user only provides a...
In only see `x_tol`, `f_tol`, and `g_tol` at https://julianlsolvers.github.io/Optim.jl/stable/user/config/#General-Options. None of the `rel` or `abs` versions. Also, is it true that `x_tol` is relative and `f_tol` is absolute?
Adding one more example here ```julia julia> A = @SMatrix([-1.0 0.0; 1.0 -0.1875]) 2×2 SMatrix{2, 2, Float64, 4} with indices SOneTo(2)×SOneTo(2): -1.0 0.0 1.0 -0.1875 julia> exp(2016*A) 2×2 SMatrix{2, 2,...
I'm sorry for the silence here. This contributions is highly appreciated. I'm a bit stretched for the time being but will try to get this reviewed over the summer.
I don't think the solvers are restricted to `Vector`. Have you tried with a `SharedVector`?
There are some conflicts here now. Are the changes still relevant?