Development Documentation
Here is the documentation for key parts of the development side of the package.
FFTA.CallGraphNode — Type
struct CallGraphNodeNode of a call graph
Arguments
left: Offset to the left child noderight: Offset to the right child nodetype: Object representing the type of FFTsz: Size of this FFTs_in: The stride of the inputs_out: The stride of the output
FFTA.CallGraph — Type
struct CallGraph{T<:Complex}Object representing a graph of FFT Calls
Arguments
nodes: Nodes keeping track of the graphworkspace: Preallocated WorkspaceBLUESTEIN_CUTOFF: Minimum prime that will be FFTed with the Bluestein algorithm, below which the O(N^2) DFT is used.
FFTA.CallGraphNode! — Function
CallGraphNode!(
nodes::Vector{FFTA.CallGraphNode},
N::Int64,
workspace::Array{Vector{T}, 1},
BLUESTEIN_CUTOFF::Int64,
s_in::Int64,
s_out::Int64
) where T -> Int64Recursively instantiate a set of CallGraphNodes
Arguments
nodes: A vector (which gets expanded) ofCallGraphNodesN: The size of the FFTworkspace: A vector (which gets expanded) of preallocated workspacesBLUESTEIN_CUTOFF: Minimum prime that will be FFTed with the Bluestein algorithm, below which the O(N^2) DFT is used.s_in: The stride of the inputs_out: The stride of the output
FFTA.fft_dft! — Function
fft_dft!(
out::AbstractVector{T},
in::AbstractVector{T},
N::Int64,
start_out::Int64,
stride_out::Int64,
start_in::Int64,
stride_in::Int64,
d::FFTA.Direction
) where T<:ComplexDiscrete Fourier Transform, O(N^2) algorithm, in place.
Arguments
out: Output vectorin: Input vectorN: Size of the transformstart_out: Index of the first element of the output vectorstride_out: Stride of the output vectorstart_in: Index of the first element of the input vectorstride_in: Stride of the input vectord: Direction of the transform
FFTA.fft_pow3! — Function
fft_pow3!(
out::AbstractVector{T},
in::AbstractVector{U},
N::Int64,
start_out::Int64,
stride_out::Int64,
start_in::Int64,
stride_in::Int64,
minus120::T,
d::FFTA.Direction
) where {T, U}Power of 3 FFT, in place
Arguments
out: Output vectorin: Input vectorN: Size of the transformstart_out: Index of the first element of the output vectorstride_out: Stride of the output vectorstart_in: Index of the first element of the input vectorstride_in: Stride of the input vectorminus120: Depending on direction, perform either ∓120° rotationd: Direction of the transform