Expand the documentation

Project.toml

@@ -9,6 +9,7 @@ LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 
 [compat]
 FFTW = "1"
+LinearAlgebra = "1"
 julia = "1.6"
 
 [extras]

README.md

@@ -4,7 +4,7 @@ This is a fork of https://github.com/HolyLab/RFFT.jl with a new UUID such that i
 
 ----
 
-In-place real FFTs for Julia
+In-place real FFTs for Julia. Supports "plans" to optimize the algorithm for transformations that you'll perform many times.
 
 For example
 ```julia
@@ -14,7 +14,14 @@ a = rand(Float64, 100, 150)
 
 # initialize a buffer 'RCpair' that contains a real and complex space
 buf = RFFT.RCpair{Float64}(undef, size(a))
+```
+
+`real(buf)` views the underlying memory buffer as an array reals, while `complex(buf)` views the same
+memory buffer as an array of complexes. The user is responsible for keeping track of which view is currently relevant.
+
+If you'll be performing lots of FFTs on this buffer, it's best to create an optimized plan.
 
+```julia
 # create the plan
 plan = RFFT.plan_rfft!(buf; flags=FFTW.MEASURE)
 
@@ -24,3 +31,5 @@ copy!(buf, new)
 new_fft = plan(buf)
 
 ```
+
+`RCpair` can be used to implement fast convolutions and many other fourier-based operations on real-valued data.

src/RFFT.jl

@@ -1,3 +1,14 @@
+"""
+# RFFT
+
+Highlights of the RFFT package:
+- [`RCpair`](@ref): a buffer for in-place real-to-complex FFTs
+- [`plan_rfft!`](@ref): create a plan for in-place real-to-complex FFTs
+- [`plan_irfft!`](@ref): create a plan for in-place complex-to-real FFTs
+
+Use `real(RC)` and `complex(RC)` to access the real and complex views of the buffer, respectively.
+`copy!(RC, A)` can be used to copy data (either real- or complex-valued) into the buffer.
+"""
 module RFFT
 
 using FFTW, LinearAlgebra
@@ -6,12 +17,34 @@ export RCpair, plan_rfft!, plan_irfft!, rfft!, irfft!, normalization
 
 import Base: real, complex, copy, copy!
 
-mutable struct RCpair{T<:AbstractFloat,N,RType<:AbstractArray{T,N},CType<:AbstractArray{Complex{T},N}}
+struct RCpair{T<:AbstractFloat,N,RType<:AbstractArray{T,N},CType<:AbstractArray{Complex{T},N}}
     R::RType
     C::CType
     dims::Vector{Int}
 end
 
+"""
+    RCPair{T<:AbstractFloat}(undef, realsize::Dims, dims=1:length(realsize))
+
+Create a buffer for performing in-place fourier transforms (and inverse) on real-valued data.
+A single underlying buffer can be viewed either as real data or as complex data:
+
+```julia
+RC = RCpair{Float64}(undef, (10, 10))
+real(RC) # 10×10 real array
+complex(RC) # 6×10 complex array
+```
+
+`dims` can be used to control which dimensions are transformed.
+
+The user is responsible to keep track of the current state of the buffer:
+
+```julia
+copy!(RC, rand(10, 10))   # copies real-valued data into the buffer; `real(RC)` is the relevant view
+rfft!(RC)                 # computes the FFT of the real-valued data; now `complex(RC)` is the relevant view
+irfft!(RC)                # computes the inverse FFT of the complex-valued data; now `real(RC)` is the relevant view
+```
+"""
 function RCpair{T}(::UndefInitializer, realsize::Dims{N}, dims=1:length(realsize)) where {T<:AbstractFloat,N}
     sz = [realsize...]
     firstdim = dims[1]
@@ -41,6 +74,16 @@ rplan_fwd(R, C, dims, flags, tlim) =
     FFTW.rFFTWPlan{eltype(R),FFTW.FORWARD,true,ndims(R)}(R, C, dims, flags, tlim)
 rplan_inv(R, C, dims, flags, tlim) =
    FFTW.rFFTWPlan{eltype(R),FFTW.BACKWARD,true,ndims(R)}(R, C, dims, flags, tlim)
+
+"""
+    plan = plan_rfft!(RC::RCpair; flags=FFTW.ESTIMATE, timelimit=FFTW.NO_TIMELIMIT)
+
+Create a plan for performing the real-to-complex FFT on the data in `RC`.
+Perform the FFT with `plan(RC)`.
+
+Planning allows you to optimize the performance of (I)FFTs for particular sizes of arrays.
+See the FFTW documentation for more information about planning.
+"""
 function plan_rfft!(RC::RCpair{T}; flags::Integer = FFTW.ESTIMATE, timelimit::Real = FFTW.NO_TIMELIMIT) where T
     p = rplan_fwd(RC.R, RC.C, RC.dims, flags, timelimit)
     return Z::RCpair -> begin
@@ -49,6 +92,16 @@ function plan_rfft!(RC::RCpair{T}; flags::Integer = FFTW.ESTIMATE, timelimit::Re
         return Z
     end
 end
+
+"""
+    plan = plan_rfft!(RC::RCpair; flags=FFTW.ESTIMATE, timelimit=FFTW.NO_TIMELIMIT)
+
+Create a plan for performing the real-to-complex FFT on the data in `RC`.
+Perform the FFT with `plan(RC)`.
+
+Planning allows you to optimize the performance of (I)FFTs for particular sizes of arrays.
+See the FFTW documentation for more information about planning.
+"""
 function plan_irfft!(RC::RCpair{T}; flags::Integer = FFTW.ESTIMATE, timelimit::Real = FFTW.NO_TIMELIMIT) where T
     p = rplan_inv(RC.C, RC.R, RC.dims, flags, timelimit)
     return Z::RCpair -> begin