conv

cv = conv(u::AbstractArray{<:Number, N}, v::AbstractArray{<:Number, N}; shape=:full)

Convolution of two arrays.

If u,v are vectors returns the convolution of vectors u and v. If u,v are matrices returns the two-dimensional convolution of matrices A and B.

The keyword argument shape controls if we return the full convolution result (the default) or only the central part of the convolution, the same size as u. In the first case (for the 1D case), the length of the output is length(u) + length(v) - 1.

When we are using convolution to do filtering, we are normaly interested in recieving a result that is of the same size of u. In this case, use the option shape=:same (in fact, anything different from :full will work). But if we were multiplying polynomes, then we would want all convolution terms and shape=:full (again, the default) would give us that.

Example

julia> conv([-1, 2, 3, -2, 0, 1, 2], [2, 4, -1, 1], shape=:same)
7-element Vector{Int64}:
 15
  5
 -9
  7
  6
  7
 -1

Methods

# 1 method for generic function "conv" from GMT:
 [1] conv(u::AbstractArray{<:Number, N}, v::AbstractArray{<:Number, N}; shape) where N

See Also