# Changing domain for Fourier
**URL:** https://discourse.julialang.org/t/changing-domain-for-fourier/85331
**Category:** Numerics
**Tags:** question, approxfun, feature-request
**Created:** [August 5, 2022, 10:11am UTC](https://discourse.julialang.org/t/changing-domain-for-fourier/85331 "2022-08-05T10:11:44Z")
**Posts on this page:** 6
**Page:** 1
### Author:  [@andreasvarga](https://discourse.julialang.org/u/andreasvarga)
#### Post date: [August 5, 2022, 10:11am UTC](https://discourse.julialang.org/t/changing-domain-for-fourier/85331/1 "2022-08-05T10:11:44Z")
I am using [ApproxFun.jl](https://github.com/JuliaApproximation/ApproxFun.jl) as basis to define and manipulate periodic matrices within the [PeridicSystems.jl](https://github.com/andreasvarga/PeriodicSystems.jl), a collection of tools dedicated for handling periodic systems. I consider [ApproxFun.jl](https://github.com/JuliaApproximation/ApproxFun.jl) a valuable tool for my purposes, being able to address some of computational problems, which I was not able to solve within the Matlab version of this package. So I would like very much to continue using this package further for my developments and I would be very pleased to have some new functionality which would ease using this package for my purposes.
I encountered the following basic issue in handling periodic matrices with different, but still commensurate periods.
If I have two periodic function matrices `F1` of period `8π` and `F2` of period `2π` and I would like to compute their sum, this operation is mathematically possible, the sum `F1+F2` is simply a periodic function matrix of period `8π`. However, this operation is not supported within [ApproxFun.jl](https://github.com/JuliaApproximation/ApproxFun.jl), leading to either crashing Julia or stack overflow error:
```julia
julia> F1
Fun(2×2 ArraySpace:
ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}[Fourier(【0.0,25.132741228718345❫) Fourier(【0.0,25.132741228718345❫); Fourier(【0.0,25.132741228718345❫) Fourier(【0.0,25.132741228718345❫)],[-5.000560225745643e-18, 2.5363265666181656e-17, 3.497202527569243e-15, -2.720857218197605e-16, 3.497202527569243e-15, -2.720857218197605e-16, 3.552713678800501e-15, 2.3209220587652208e-15, -6.12323399573677e-17, 6.123233995736767e-17 … -4.73232564246473e-15, -2.3209220587652208e-15, -3.552713678800501e-14, 5.942090786205257e-16, -1.4782794558091705e-16, 1.790234627208065e-15, -1.519318030172918e-16, 1.790234627208065e-15, -1.519318030172918e-16, 1.0000000000000142])
julia> domain(F1)
【0.0,25.132741228718345❫
julia> F2
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❫)],[0.0, -1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0289288258043393e-16, 0.0, -0.0, 0.0, -0.0, 0.0, 1.0])
julia> domain(F2)
【0.0,6.283185307179586❫
julia> F1+F2
ERROR: StackOverflowError:
Stacktrace:
[1] map(f::ApproxFunBase.var"#288#289"{PeriodicSegment{Float64}}, v::Core.SimpleVector)
@ Base .\essentials.jl:622
[2] ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}(dom::PeriodicSegment{Float64})
@ ApproxFunBase C:\Users\Andreas\.julia\packages\ApproxFunBase\Hasa4\src\Spaces\SumSpace.jl:45
[3] SumSpace
@ C:\Users\Andreas\.julia\packages\ApproxFunFourier\MPNwF\src\ApproxFunFourier.jl:405 [inlined]
[4] canonicalspace
@ C:\Users\Andreas\.julia\packages\ApproxFunFourier\MPNwF\src\ApproxFunFourier.jl:97 [inlined]
[5] map
@ .\tuple.jl:222 [inlined]
[6] union_rule(S1sp::ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}, S2sp::ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64})
@ ApproxFunBase C:\Users\Andreas\.julia\packages\ApproxFunBase\Hasa4\src\Spaces\ProductSpaceOperators.jl:186
[7] union_by_union_rule(a::ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}, b::ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64})
@ ApproxFunBase C:\Users\Andreas\.julia\packages\ApproxFunBase\Hasa4\src\Space.jl:266
[8] union(a::CosSpace{PeriodicSegment{Float64}, Float64}, b::CosSpace{PeriodicSegment{Float64}, Float64})
@ ApproxFunBase C:\Users\Andreas\.julia\packages\ApproxFunBase\Hasa4\src\Space.jl:279
```
Note that I already raised basically the same issue in [#768](https://github.com/JuliaApproximation/ApproxFun.jl/issues/768) (about four months ago).
I would appreciate very much any idea how to overcome this error (e.g., by converting first ` F2` to period `8π` ?). I was not able to find a simple way to perform this (in my trials I crashed Julia several times). Sorry for my ignorance!
Furthermore, I would be very happy if an automatic conversion feature would be supported in a future release of [ApproxFun.jl](https://github.com/JuliaApproximation/ApproxFun.jl) to perform operations with periodic functions with commensurate periods.
---
### Author:  [@andreasvarga](https://discourse.julialang.org/u/andreasvarga)
#### Post date: [August 5, 2022, 11:29am UTC](https://discourse.julialang.org/t/changing-domain-for-fourier/85331/2 "2022-08-05T11:29:26Z")
I think I got it. The following works:
```julia
julia> F1+Fun(t->F2(t),domain(F1))
Fun(2×2 ArraySpace:
ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}[Fourier(【0.0,25.132741228718345❫) Fourier(【0.0,25.132741228718345❫); Fourier(【0.0,25.132741228718345❫) Fourier(【0.0,25.132741228718345❫)],[-3.561673020442948e-17, 5.072653133236335e-17, 3.497202527569243e-15, -2.720857218197605e-16, 3.497202527569243e-15, -2.720857218197605e-16, 3.18903060143427e-15, 2.8161797819056647e-15, -1.2246467991473535e-16, 1.224646799147353e-16 … -4.73232564246473e-15, -2.565902665598115e-15, -3.481523346342104e-14, 1.1257283409025332e-15, -2.9565589116183405e-16, 1.790234627208065e-15, -1.519318030172918e-16, 1.790234627208065e-15, -1.519318030172918e-16, 2.000000000000014])
```
And
```julia
julia> F2+Fun(t->F1(t),domain(F2))
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❫)],[-1.294720671475603e-16, -2.0000000000000004, 2.4384757919829034e-15, -4.181839259808612e-15, 2.4384757919829034e-15, -4.181839259808612e-15, 1.6933257206756878e-14, -3.283993251760897e-14, 6.008022436061937e-16, 5.466932223891189e-17, 2.3058884893441655e-15, 3.493507550589483e-16, 2.3058884893441655e-15, 3.493507550589483e-16, 2.000000000000005])
```
works too.
---
### Author:  [@DNF](https://discourse.julialang.org/u/DNF)
#### Post date: [August 5, 2022, 3:11pm UTC](https://discourse.julialang.org/t/changing-domain-for-fourier/85331/3 "2022-08-05T15:11:01Z")
> [@andreasvarga](#):
>
> `t->F2(t)`
This anonymous function shouldn’t be necessary. You can just write `F2`, as in
```julia
F1+Fun(F2, domain(F1))
```
This a general observation about functions, nothing to do with ApproxFun.jl
---
### Author:  [@andreasvarga](https://discourse.julialang.org/u/andreasvarga)
#### Post date: [August 5, 2022, 4:04pm UTC](https://discourse.julialang.org/t/changing-domain-for-fourier/85331/4 "2022-08-05T16:04:07Z")
Unfortunately, this doesn’t work:
```julia
julia> F1+Fun(F2, domain(F1))
ERROR: StackOverflowError:
Stacktrace:
[1] map(f::ApproxFunBase.var"#288#289"{PeriodicSegment{Float64}}, v::Core.SimpleVector)
@ Base .\essentials.jl:622
[2] ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64}(dom::PeriodicSegment{Float64})
@ ApproxFunBase C:\Users\Andreas\.julia\packages\ApproxFunBase\Hasa4\src\Spaces\SumSpace.jl:45
[3] SumSpace
@ C:\Users\Andreas\.julia\packages\ApproxFunFourier\MPNwF\src\ApproxFunFourier.jl:405 [inlined]
[4] canonicalspace
@ C:\Users\Andreas\.julia\packages\ApproxFunFourier\MPNwF\src\ApproxFunFourier.jl:97 [inlined]
[5] defaultcoefficients(f::Vector{Float64}, a::CosSpace{PeriodicSegment{Float64}, Float64}, b::ApproxFunBase.SumSpace{Tuple{CosSpace{PeriodicSegment{Float64}, Float64}, SinSpace{PeriodicSegment{Float64}, Float64}}, PeriodicSegment{Float64}, Float64})
@ ApproxFunBase C:\Users\Andreas\.julia\packages\ApproxFunBase\Hasa4\src\Space.jl:332
```
---
### Author:  [@DNF](https://discourse.julialang.org/u/DNF)
#### Post date: [August 5, 2022, 4:30pm UTC](https://discourse.julialang.org/t/changing-domain-for-fourier/85331/5 "2022-08-05T16:30:16Z")
Hm. Is it because `F2` is a “Fun”, not a function, and `t->F2(t)` turns it into one? Too bad, `x->foo(x)` is a bit of an eyesore.
---
### Author:  [@jishnub](https://discourse.julialang.org/u/jishnub)
#### Post date: [August 9, 2022, 5:52am UTC](https://discourse.julialang.org/t/changing-domain-for-fourier/85331/6 "2022-08-09T05:52:02Z")
This requires defining the union of domains properly. I am working on [a fix](https://github.com/JuliaApproximation/ApproxFunFourier.jl/pull/41) for this. It’s not fully ready yet, but will hopefully be included in ApproxFun soon.