Simon Byrne

Results 351 comments of Simon Byrne

@kanav99 @ChrisRackauckas: @thomasgibson and I decided that the best way to support this would be to add a 6-argument call interface to the function `f(du, u, param, t, α, β)`...

p.s. if you have opinions on the order of `α, β` in the argument list, now would be the time to tell us!

Okay, we're in the process of rewriting our solvers to be at least somewhat compatible, and will probably fork it off into a separate package.

Error appears to now be ``` ERROR: LoadError: MethodError: no method matching active_module() You may have intended to import Base.active_module The applicable method may be too new: running in world...

This might be due to https://github.com/JuliaLang/julia/pull/45838? @aviatesk Any suggestions on what we might be able to do here?

Ah, I just hit this: basically the problem is FFTW.jl. It has MKL_jll (for which the artifacts are lazy) in its Project.toml, but doesn't load it by default (lazy artifact...

The solution appears to be to use `include_transitive_dependencies=false`, which will avoid pulling in MKL_jll if it isn't actually loaded.

Perhaps it is worth changing this to be the default?

Actually it fails with just generic `ldiv!`: ```julia julia> ldiv!(BlockArray(rand(3,3)), BlockArray(rand(3))) ERROR: Overload materialize!(::Lmul{ArrayLayouts.AdjQRCompactWYQLayout{ArrayLayouts.DenseColumnMajor, ArrayLayouts.DenseColumnMajor}}) Stacktrace: [1] error(s::String) @ Base ./error.jl:33 [2] materialize!(M::ArrayLayouts.Lmul{ArrayLayouts.AdjQRCompactWYQLayout{ArrayLayouts.DenseColumnMajor, ArrayLayouts.DenseColumnMajor}, BlockArrays.BlockLayout{ArrayLayouts.DenseColumnMajor, ArrayLayouts.DenseColumnMajor}, Adjoint{Float64, LinearAlgebra.QRCompactWYQ{Float64, Matrix{Float64}}}, BlockVector{Float64,...

> Can you use a `PseudoBlockVector`? This should work Not really: for my particular use case I have my own `AbstractBlockVector` type: https://github.com/CliMA/ClimaCore.jl/blob/main/src/Fields/fieldvector.jl