sssets v2.0 (#164)

* sssets v2.0

* rename dataset to ssset

* use oldsize for backwards compat code

* use size in the test suite

* use size in general

CHANGELOG.md

@@ -1,5 +1,9 @@
 # RecurrenceAnalysis.jl changelog
 
+## v2.1.0
+
+Updated to StateSpaceSets.jl v2.
+
 ## v2.0.0
 
 - [**BREAKING**] The `@windowed` macro is removed. It was some incredibly complicated 250 lines of code that offer little benefit. Instead of using this macro, use the new `windowed` function. It does something simple: replaces
@@ -50,11 +54,11 @@
 - Now `AbstractRecurrenceMatrix` types have a parameter with the possible values:
     * `WithinRange`: when neighboring points are determined by a given distance in the phase space.
     * `NeighborNumber`: when each point has a determined number of neighbors.
-- Input arguments of `RecurrenceMatrix`, etc. are generalized to `AbstractDataset`.
+- Input arguments of `RecurrenceMatrix`, etc. are generalized to `AbstractStateSpaceSet`.
 
 ## v1.3.2
 - Bugfixes of `transitivity`
-- Deprecate `AbstractMatrix` inputs to `RecurrenceMatrix` etc., in favor of `Dataset`.
+- Deprecate `AbstractMatrix` inputs to `RecurrenceMatrix` etc., in favor of `StateSpaceSet`.
 
 ## v1.3.0
 - Increased performance of `RecurrenceMatrix`.
@@ -91,12 +95,12 @@
   create a dedicated struct for each kind of matrix.
 
 ## v0.6.0
-* Rework how `recurrencematrix` is computed: now a conversion to `Dataset` is done for small dimensions. In addition, the default metric is now `Euclidean`.
+* Rework how `recurrencematrix` is computed: now a conversion to `StateSpaceSet` is done for small dimensions. In addition, the default metric is now `Euclidean`.
 
 ## v0.5.0
 * With this update core functionality of delay embedding and estimating embedding dimension is moved (and used from) the Julia package `DelayEmbeddings`.
 * `RecurrenceAnalysis` is also joining JuliaDynamics and **DynamicalSystems.jl** from 0.5.0 onwards.
-* Extensions have been implemented so that most functionality works also with `Dataset`s.
+* Extensions have been implemented so that most functionality works also with `StateSpaceSet`s.
 
 ## 23-09-2018 - v0.4.0
 

Project.toml

@@ -1,7 +1,7 @@
 name = "RecurrenceAnalysis"
 uuid = "639c3291-70d9-5ea2-8c5b-839eba1ee399"
 repo = "https://github.com/JuliaDynamics/RecurrenceAnalysis.jl.git"
-version = "2.0.8"
+version = "2.1.0"
 
 [deps]
 DelimitedFiles = "8bb1440f-4735-579b-a4ab-409b98df4dab"
@@ -20,7 +20,7 @@ DelimitedFiles = "1"
 Distances = "0.8, 0.9, 0.10"
 Graphs = "1.6"
 Reexport = "1"
-StateSpaceSets = "1"
+StateSpaceSets = "2"
 Statistics = "1"
 UnicodePlots = "0.3, 1, 2, 3"
 julia = "1.5"

src/RecurrenceAnalysis.jl

@@ -10,6 +10,14 @@ end RecurrenceAnalysis
 using Reexport
 @reexport using StateSpaceSets
 
+# for backwards compatibility with SSSets.jl v1
+# TODO: Cleanup: go through the instances in the codebase
+# where `oldsize` is called: replace it with `size` if the
+# input is a `Matrix`, and with `dimension` if the input is a `StateSpaceSet`.
+oldsize(s::AbstractStateSpaceSet) = (length(s), dimension(s))
+oldsize(s) = size(s)
+oldsize(s, i::Int) = oldsize(s)[i]
+
 Array_or_SSSet = Union{AbstractArray{<:Real}, AbstractStateSpaceSet}
 Vector_or_SSSet = Union{AbstractVector{<:Real}, AbstractStateSpaceSet}
 
@@ -27,7 +35,7 @@ export RecurrenceMatrix, CrossRecurrenceMatrix, JointRecurrenceMatrix
 export embed,
        reconstruct,
        recurrence_threshold,
-       Dataset,
+       StateSpaceSet,
        distancematrix,
        textrecurrenceplot,
        recurrenceplot,

src/deprecate.jl

@@ -23,24 +23,24 @@ end
 
 for T in (:WithinRange, :NeighborNumber)
     @eval function RecurrenceMatrix{$T}(x::AbstractMatrix, ε; kwargs...)
-        @warn string("`RecurrenceMatrix{", $T, "}(x::AbstractMatrix, ε; kwargs...)` is deprecated, use `RecurrenceMatrix{", $T, "}(Dataset(x), ε; kwargs...)`")
-        RecurrenceMatrix{$T}(Dataset(x), ε; kwargs...)
+        @warn string("`RecurrenceMatrix{", $T, "}(x::AbstractMatrix, ε; kwargs...)` is deprecated, use `RecurrenceMatrix{", $T, "}(StateSpaceSet(x), ε; kwargs...)`")
+        RecurrenceMatrix{$T}(StateSpaceSet(x), ε; kwargs...)
     end
 
     for call in (:CrossRecurrenceMatrix, :JointRecurrenceMatrix)
         @eval function $call{$T}(x::AbstractMatrix, y, ε; kwargs...)
-            @warn string("`", $call, "{", $T, "}(x::AbstractMatrix, y, ε; kwargs...)` is deprecated, use `", $call, "{", $T, "}(Dataset(x), y, ε; kwargs...)`")
-            $call{$T}(Dataset(x), y, ε; kwargs...)
+            @warn string("`", $call, "{", $T, "}(x::AbstractMatrix, y, ε; kwargs...)` is deprecated, use `", $call, "{", $T, "}(StateSpaceSet(x), y, ε; kwargs...)`")
+            $call{$T}(StateSpaceSet(x), y, ε; kwargs...)
         end
 
         @eval function $call{$T}(x, y::AbstractMatrix, ε; kwargs...)
-            @warn string("`", $call, "{", $T, "}(x, y::AbstractMatrix, ε; kwargs...)` is deprecated, use `", $call, "{", $T, "}(x, Dataset(y), ε; kwargs...)`")
-            $call{$T}(x, Dataset(y), ε; kwargs...)
+            @warn string("`", $call, "{", $T, "}(x, y::AbstractMatrix, ε; kwargs...)` is deprecated, use `", $call, "{", $T, "}(x, StateSpaceSet(y), ε; kwargs...)`")
+            $call{$T}(x, StateSpaceSet(y), ε; kwargs...)
         end
 
         @eval function $call{$T}(x::AbstractMatrix, y::AbstractMatrix, ε; kwargs...)
-            @warn string("`", $call, "{", $T, "}(x::AbstractMatrix, y::AbstractMatrix, ε; kwargs...)` is deprecated, use `", $call, "{", $T, "}(Dataset(x), Dataset(y), ε; kwargs...)`")
-            $call{$T}(Dataset(x), Dataset(y), ε; kwargs...)
+            @warn string("`", $call, "{", $T, "}(x::AbstractMatrix, y::AbstractMatrix, ε; kwargs...)` is deprecated, use `", $call, "{", $T, "}(StateSpaceSet(x), StateSpaceSet(y), ε; kwargs...)`")
+            $call{$T}(StateSpaceSet(x), StateSpaceSet(y), ε; kwargs...)
         end
     end
 end

src/matrices/distance_matrix.jl

@@ -7,7 +7,7 @@
 Create a matrix with the distances between each pair of points of the
 time series `x` and `y` using `metric`.
 
-The time series `x` and `y` can be `AbstractDataset`s or vectors or matrices with data points
+The time series `x` and `y` can be `AbstractStateSpaceSet`s or vectors or matrices with data points
 in rows.
 The data point dimensions (or number of columns) must be the same for `x` and `y`.
 The returned value is a `n×m` matrix, with `n` being the length (or number of rows)
@@ -17,18 +17,18 @@ The metric can be any of the `Metric`s defined in
 the [`Distances` package](https://github.com/JuliaStats/Distances.jl)
 and defaults to `Euclidean()`.
 """
-distancematrix(x, metric::Union{Metric,String}=DEFAULT_METRIC, parallel = size(x)[1] > 500) = _distancematrix(x, getmetric(metric), Val(parallel))
+distancematrix(x, metric::Union{Metric,String}=DEFAULT_METRIC, parallel = oldsize(x)[1] > 500) = _distancematrix(x, getmetric(metric), Val(parallel))
 
 # For 1-dimensional arrays (vectors), the distance does not depend on the metric
-distancematrix(x::AbstractVector{<:Real}, y::AbstractVector{<:Real}, metric=DEFAULT_METRIC, parallel = size(x)[1] > 500) = abs.(x .- y')
+distancematrix(x::AbstractVector{<:Real}, y::AbstractVector{<:Real}, metric=DEFAULT_METRIC, parallel = oldsize(x)[1] > 500) = abs.(x .- y')
 
 # If the metric is supplied as a string, get the corresponding Metric from Distances
-distancematrix(x, y, metric::String, parallel = size(x)[1] > 500) = distancematrix(x, y, getmetric(metric), parallel)
+distancematrix(x, y, metric::String, parallel = oldsize(x)[1] > 500) = distancematrix(x, y, getmetric(metric), parallel)
 
 const MAXDIM = 9
-function distancematrix(x::Tx, y::Ty, metric::Metric=DEFAULT_METRIC, parallel = size(x)[1] > 500) where
-         {Tx<:Union{AbstractMatrix, AbstractDataset}} where {Ty<:Union{AbstractMatrix, AbstractDataset}}
-    sx, sy = size(x), size(y)
+function distancematrix(x::Tx, y::Ty, metric::Metric=DEFAULT_METRIC, parallel = oldsize(x)[1] > 500) where
+         {Tx<:Union{AbstractMatrix, AbstractStateSpaceSet}} where {Ty<:Union{AbstractMatrix, AbstractStateSpaceSet}}
+    sx, sy = oldsize(x), oldsize(y)
     @assert sx[2] == sy[2] """
         The dimensions of the data points in `x` and `y` must be equal!
         Found dim(x)=$(sx[2]), dim(y)=$(sy[2]).
@@ -39,7 +39,7 @@ function distancematrix(x::Tx, y::Ty, metric::Metric=DEFAULT_METRIC, parallel =
     elseif Tx <: Matrix && Ty <: Matrix && metric == Chebyshev()
         return _distancematrix(x, y, metric, Val(parallel))
     else
-        return _distancematrix(Dataset(x), Dataset(y), metric, Val(parallel))
+        return _distancematrix(StateSpaceSet(x), StateSpaceSet(y), metric, Val(parallel))
     end
 end
 
@@ -48,7 +48,7 @@ _distancematrix(x::AbstractMatrix, y::AbstractMatrix, metric::Metric, ::Val{fals
 
 # First we define the serial versions of the functions.
 # Core function for AbstractDatasets
-function _distancematrix(x::AbstractDataset{S,Tx}, y::AbstractDataset{S,Ty},
+function _distancematrix(x::AbstractStateSpaceSet{S,Tx}, y::AbstractStateSpaceSet{S,Ty},
     metric::Metric, ::Val{false}) where {S, Tx, Ty}
 
     x = x.data
@@ -62,7 +62,7 @@ end
 
 # Now, we define the parallel versions.
 
-function _distancematrix(x::AbstractDataset{S,Tx}, y::AbstractDataset{S,Ty},
+function _distancematrix(x::AbstractStateSpaceSet{S,Tx}, y::AbstractStateSpaceSet{S,Ty},
     metric::Metric, ::Val{true}) where {S, Tx, Ty}
 
     x = x.data
@@ -150,7 +150,7 @@ function _distancematrix(x::Vector{T}, metric::Metric, ::Val{true}) where T
 
 end
 
-function _distancematrix(x::AbstractDataset{S, T}, metric::Metric, ::Val{true}) where T where S
+function _distancematrix(x::AbstractStateSpaceSet{S, T}, metric::Metric, ::Val{true}) where T where S
     d = zeros(T, length(x), length(x))
 
     Threads.@threads for k in partition_indices(length(x))

src/rna/graphs.jl

@@ -34,7 +34,7 @@ Theory and Best Practices*, Springer, pp. 101-165 (2015).
 """
 function Graphs.SimpleGraphs.SimpleGraph(R::AbstractRecurrenceMatrix)
     graph = SimpleGraph(R.data)
-    delta = SimpleGraphFromIterator(Edge(v,v) for v in 1:size(graph, 1))
+    delta = SimpleGraphFromIterator(Edge(v,v) for v in 1:oldsize(graph, 1))
     graph = difference(graph, delta)
     return graph
 end

src/rqa/histograms.jl

@@ -112,7 +112,7 @@ function diagonalhistogram(x::Union{ARM,AbstractMatrix}; lmin::Integer=2, theile
     (theiler < 0) && throw(ErrorException(
         "Theiler window length must be greater than or equal to 0"))
     (lmin < 1) && throw(ErrorException("lmin must be 1 or greater"))
-    m,n = size(x)
+    m,n = oldsize(x)
     rv = rowvals(x)[:]
     dv = colvals(x) .- rowvals(x)
     loi_hist = Int[]
@@ -158,7 +158,7 @@ function verticalhistograms(x::Union{ARM,AbstractMatrix};
     (theiler < 0) && throw(ErrorException(
         "Theiler window length must be greater than or equal to 0"))
     (lmin < 1) && throw(ErrorException("lmin must be 1 or greater"))
-    m,n=size(x)
+    m,n=oldsize(x)
     rv = rowvals(x)
     cv = colvals(x)
     return _linehistograms(rv,cv,lmin,theiler,distances)

src/rqa/radius.jl

@@ -32,7 +32,7 @@ function sorteddistances(x; theiler::Integer=0, scale=1, kwargs...)
     scale = (typeof(scale) <: Function) ? scale(dm) : scale
     dm /= scale
     # Browse upper triangle after Theiler window
-    n = size(x,1)
+    n = oldsize(x,1)
     nd = (n+1)*n/2
     ntheiler = theiler*n - theiler*(theiler-1)/2
     distarray = zeros(round(Int, nd-ntheiler))

src/rqa/rqa.jl

@@ -77,15 +77,15 @@ _rrdenominator(R::ARM; theiler=0, kwargs...) = length(R)
 function _rrdenominator(R::AbstractMatrix; theiler=0, kwargs...)
 
     (theiler == 0) && (return length(R))
-    k = size(R,1) - theiler
+    k = oldsize(R,1) - theiler
     return k*(k+1)
 end
 
 function _rrdenominator(R::M; theiler=0, kwargs...) where
     M<:Union{RecurrenceMatrix,JointRecurrenceMatrix}
 
     (theiler == 0) && (return length(R))
-    k = size(R,1) - theiler
+    k = oldsize(R,1) - theiler
     return k*(k+1)
 end
 
@@ -128,12 +128,12 @@ macro histogram_params(keyword, description, hist_fun)
                 points inside the Theiler window (see [`rqa`](@ref) for the
                 default values and usage of the keyword argument `theiler`).
 
-                Notes: This metric was first proposed in the paper "Exploiting Nonlinear Recurrence 
-                and Fractal Scaling Properties for Voice Disorder Detection" as Recurrence Period 
-                Density Entropy or Recurrence Probability Density Entropy (RPDE). 
-                It is a normalized dimensionless metric in the range [0,1]. 
-                In the 2018 article "Recurrence threshold selection for obtaining robust recurrence 
-                characteristics in different embedding dimensions", the indicator RPDE is explicitly 
+                Notes: This metric was first proposed in the paper "Exploiting Nonlinear Recurrence
+                and Fractal Scaling Properties for Voice Disorder Detection" as Recurrence Period
+                Density Entropy or Recurrence Probability Density Entropy (RPDE).
+                It is a normalized dimensionless metric in the range [0,1].
+                In the 2018 article "Recurrence threshold selection for obtaining robust recurrence
+                characteristics in different embedding dimensions", the indicator RPDE is explicitly
                 called Recurrence Time Entropy (RTE). Here RPDE and RTE are clearly the same indicator.
                 """
         else
@@ -257,7 +257,7 @@ Compute the diagonalwise recurrence rate `τ-RR` from a recurrence matrix `R`.
 Note that this only works for squared recurrence matrices.
 """
 function tau_recurrence(R::Union{ARM,AbstractMatrix})
-    n = minimum(size(R))
+    n = minimum(oldsize(R))
     rv = rowvals(R)
     rr_τ = zeros(n)
     @inbounds for col in 1:n

src/rqa/windowed_function.jl

@@ -10,7 +10,7 @@ function windowed(rmat::Union{ARM,AbstractMatrix}, f::Function, width::Integer,
         "Window width must be must be greater than or equal to 2"))
   (step < 1) && throw(ErrorException(
         "Step size must be greater than or equal to 1"))
-  windows = 1:step:(size(rmat, 1)-width+1)
+  windows = 1:step:(oldsize(rmat, 1)-width+1)
   map(1:length(windows)) do i
     f(rmat[windows[i]:(windows[i]+width-1), windows[i]:(windows[i]+width-1)]; kwargs...)
   end

test/rmatrix_analytic.jl

@@ -10,7 +10,7 @@ using Distances: Chebyshev
 t = range(0, 2π; length = 20) # length is crucial and decides distance and thresholds
 c = cos.(t)
 s = sin.(t)
-X = Dataset(c, s)
+X = StateSpaceSet(c, s)
 Y = X[1:10]
 
 dmat = distancematrix(X)