Jishnu Bhattacharya
Jishnu Bhattacharya
```julia julia> B = BandedMatrix(-3=> 2ones(2), -1=>ones(4), 0=>2ones(5)); Buplo = :L; SB = Symmetric(B, Buplo); julia> cholesky(Matrix(SB), check = false) Failed factorization of type Cholesky{Float64, Matrix{Float64}} julia> cholesky(SB, check =...
This reduces TTFX to some extent (although more seems possible) On master ```console $ julia --project -e '@time using BandedMatrices' 2.627211 seconds (5.68 M allocations: 537.898 MiB, 14.21% compilation time)...
I noticed that the [permissions of GITHUB_TOKEN](https://github.com/JuliaLinearAlgebra/BandedMatrices.jl/actions/runs/3578304165/jobs/6018304263) are different in this repo compared to others. In this repo: ```console GITHUB_TOKEN Permissions Contents: write Metadata: read ``` In, say, [StaticArrays](https://github.com/JuliaArrays/StaticArrays.jl/actions/runs/3569185857/jobs/5998877411), where...
```julia julia> B = BandedMatrix(1=>Float64.(1:7)) 8×8 BandedMatrix{Float64} with bandwidths (-1, 1): ⋅ 1.0 ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ 2.0 ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅...
### Proposal: Add functions `diagonal`, `tridiagonal` etc that construct appropriate banded matrices. Eg: ```julia julia> diagonal(d) = BandedMatrix(0=>d); julia> diagonal(ones(4)) 4×4 BandedMatrix{Float64} with bandwidths (0, 0): 1.0 ⋅ ⋅ ⋅...
The following works on v1.9, but not on v1.10 and above, as `AbstractQ` isn't an `AbstractArray` anymore: ```julia julia> using LinearAlgebra julia> Q = qr(rand(3,4)).Q; julia> similar(Q, 1, 1) 1×1...
My understanding was that a token would not be required when creating a PR from a fork to a public, open-source project. However, in the run https://github.com/JuliaMath/FFTW.jl/actions/runs/7854977161/job/21436255573?pr=295, I encounter ```...
E.g. https://github.com/FluxML/Zygote.jl/blob/17ca911b82134c4a765822cd2b7ee19e959cc8e4/src/lib/array.jl#L783 This looks like it should belong to an extension in `FillArrays.jl`
```julia julia> a = 1:2 1:2 julia> conv(a, a) 3-element Vector{Int64}: 1 4 4 julia> conv(OffsetArray(a), OffsetArray(a)) 3-element OffsetArray(::Vector{Int64}, 2:4) with eltype Int64 with indices 2:4: 1 4 4 julia>...
This might help in dispatching to efficient methods for `StridedArrays`: ```julia julia> B = PseudoBlockArray(collect(reshape(1:36,6,6)), [1,2,3], [1,2,3]) 3×3-blocked 6×6 PseudoBlockMatrix{Int64}: 1 │ 7 13 │ 19 25 31 ───┼──────────┼──────────── 2...