Is there a library for a generic convolution function?

General Usage
, ,
1

I need to do a convolution of BigInts and DSP says that BigFloat isn’t supported when I try.

2

You just need to load a package implementing a BigFloat FFT, and then DSP.conv will work. For example, using GenericFFT.jl:

julia> using DSP, GenericFFT

julia> setprecision(6); # lower the precision to reduce the amount of output below

julia> u, v = [rand(BigFloat, 10) for _=1:2]
2-element Vector{Vector{BigFloat}}:
 [0.0156, 0.422, 0.812, 0.438, 0.125, 0.984, 0.109, 0.875, 0.812, 0.625]
 [0.109, 0.5, 0.719, 0.797, 0.422, 0.0312, 0.938, 0.156, 0.469, 0.688]

julia> DSP.conv(u,v)
19-element Vector{BigFloat}:
 -0.119
  0.0298
  0.484
  0.766
  1.75
  1.19
  1.53
  1.59
  2.31
  2.38
  2.5
  2.88
  1.25
  1.69
  1.59
  1.03
  1.28
  0.688
  0.594
4 Likes