Jishnu Bhattacharya
Jishnu Bhattacharya
This shows whether the tests pass on the latest Julia nightly, and may be used as an early warning indicator of potential regressions. Clicking on the badge redirects one to...
On master ```julia julia> r = big(10)^400 .+ (0:1); julia> chebyshevtransform(r) 2-element Vector{Float64}: Inf NaN ``` This PR fixes this.
This fixes the inverse Chebyshev transform for immutable input vectors. The following works now: ```julia julia> ichebyshevtransform(1.0:4.0) 4-element Vector{Float64}: 6.499813138042575 -4.05147160887461 1.808830921755324 -0.25717245092328955 ``` This throws an error on master,...
The `PermutedDimsArray` constructor is type-unstable if the permutation is not a compile-time constant, so it's best to avoid this if possible. In this PR, I have replaced the broadcast operations...
This was left over from a type-stability hack, but isn't needed any more. In any case, the function `_fftKtype` isn't doing anything, and may be safely removed.
This fixes ```julia julia> pochhammer(3,1:5) 5-element Vector{Int64}: 3 12 60 360 2520 ```
This is a robust alternative to the problems that #220 and #221 are trying to solve, as this removes the dependency on the internal type parameters of FFTW altogether (albeit...
Syntactically, all of these seem to reduce unnecessary allocations, although I have not verified the actual impact.
https://github.com/JuliaApproximation/FastTransforms.jl/blob/d7be08cbaec7879314701fd706bccf374e046a84/src/FastTransforms.jl#L6-L8 This adds additional names to the user's namespace, as well as makes this package brittle. Perhaps it's better to avoid re-exporting them, and ask users to import them as...
```julia julia> x = rand(1000); julia> y1 = @btime FastTransforms.cheb2ultra($x, 2); 24.224 ms (2 allocations: 7.98 KiB) julia> y2 = @btime ApproxFunBase.mul_coefficients(Conversion(Chebyshev(), Ultraspherical(2)), $x); 45.568 μs (13 allocations: 63.27 KiB)...