search

JuliaApproximation / ApproxFun.jl

Julia package for function approximation
http://juliaapproximation.github.io/ApproxFun.jl/
Other
528 stars 70 forks source link

Array concatenations produce unexpected results #924

Open andreasvarga opened 8 months ago

andreasvarga commented 8 months ago

The following illustrates the issue:

julia> f = Fun(Fourier(), [1,2,3,4])
Fun(Fourier(【0.0,6.283185307179586❫), [1, 2, 3, 4])

julia> g = Fun(Fourier(), [1,0,0,4])
Fun(Fourier(【0.0,6.283185307179586❫), [1, 0, 0, 4])

julia> Af = [f g]
Fun(1×2 ArraySpace:
ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}[Fourier(【0.0,6.283185307179586❫) Fourier(【0.0,6.283185307179586❫)], [1, 2, 1, 0, 3, 4, 0, 4])

julia> [Af Af]
1×2 Matrix{Fun{ApproxFunBase.ArraySpace{ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}, 2, PeriodicSegment{Float64}, Float64, Matrix{ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}}}, Int64, Vector{Int64}}}:
 Fun(1×2 ArraySpace:
SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}[Fourier(【0.0,6.283185307179586❫) Fourier(【0.0,6.283185307179586❫)], [1, 2, 1, 0, 3, 4, 0, 4])  …  Fun(1×2 ArraySpace:
SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}[Fourier(【0.0,6.283185307179586❫) Fourier(【0.0,6.283185307179586❫)], [1, 2, 1, 0, 3, 4, 0, 4])

julia> [Af; Af]
Fun(2-element ArraySpace:
ApproxFunBase.ArraySpace{ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}, 2, PeriodicSegment{Float64}, Float64, Matrix{ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}}}[1×2 ArraySpace:
ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}[Fourier(【0.0,6.283185307179586❫) Fourier(【0.0,6.283185307179586❫)], 1×2 ArraySpace:
ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}[Fourier(【0.0,6.283185307179586❫) Fourier(【0.0,6.283185307179586❫)]], [1, 2, 1, 0, 1, 2, 1, 0, 3, 4, 0, 4, 3, 4, 0, 4])

[Af Af] is hcat(Af,Af) and I would expect to produce a Fun(1×4 ArraySpace:..., while [Af; Af] is vcat(Af,Af) and I expect to produce a Fun(2×2 ArraySpace:..., as would result using concatenations of usual matrices. In both cases, arrays of matrices are generated, but I need to build the horizonatally or vertically concatenated matrices.

I appologize if I am wrong, and I would appreciate any hint how to obtain the (for me) expected results.

andreasvarga commented 8 months ago

The following is a possible fix to the above problems:

julia> Fun(t -> [Af(t) Af(t)],Fourier())
Fun(1×4 ArraySpace:
ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}[Fourier(【0.0,6.283185307179586❫) Fourier(【0.0,6.283185307179586❫) Fourier(【0.0,6.283185307179586❫) Fourier(【0.0,6.283185307179586❫)], [1.0, 2.0, 1.0, -1.49908e-16, 1.0, 2.0, 1.0, -1.49908e-16, 3.0, 4.0, -4.7375e-16, 4.0, 3.0, 4.0, -4.7375e-16, 4.0])

julia> Fun(t -> [Af(t); Af(t)],Fourier())
Fun(2×2 ArraySpace:
ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}[Fourier(【0.0,6.283185307179586❫) Fourier(【0.0,6.283185307179586❫); Fourier(【0.0,6.283185307179586❫) Fourier(【0.0,6.283185307179586❫)], [1.0, 2.0, 1.0, 2.0, 1.0, -1.49908e-16, 1.0, -1.49908e-16, 3.0, 4.0, 3.0, 4.0, -4.7375e-16, 4.0, -4.7375e-16, 4.0])

However, I realize that a more elaborate implementation ofhcat and vcat is probably necessary.