Quality of life fixes

README.md

@@ -17,11 +17,21 @@
 
 # ReservoirComputing.jl
 
-ReservoirComputing.jl provides an efficient, modular and easy to use implementation of Reservoir Computing models such as Echo State Networks (ESNs). For information on using this package please refer to the [stable documentation](https://docs.sciml.ai/ReservoirComputing/stable/). Use the [in-development documentation](https://docs.sciml.ai/ReservoirComputing/dev/) to take a look at at not yet released features.
+ReservoirComputing.jl provides an efficient, modular and easy to use
+implementation of Reservoir Computing models such as Echo State Networks (ESNs).
+For information on using this package please refer to the
+[stable documentation](https://docs.sciml.ai/ReservoirComputing/stable/).
+Use the
+[in-development documentation](https://docs.sciml.ai/ReservoirComputing/dev/)
+to take a look at not yet released features.
 
 ## Quick Example
 
-To illustrate the workflow of this library we will showcase how it is possible to train an ESN to learn the dynamics of the Lorenz system. As a first step we will need to gather the data. For the `Generative` prediction we need the target data to be one step ahead of the training data:
+To illustrate the workflow of this library we will showcase
+how it is possible to train an ESN to learn the dynamics of the
+Lorenz system. As a first step we gather the data.
+For the `Generative` prediction we need the target data
+to be one step ahead of the training data:
 
 ```julia
 using ReservoirComputing, OrdinaryDiffEq
@@ -52,7 +62,9 @@ target_data = data[:, (shift + 1):(shift + train_len)]
 test = data[:, (shift + train_len):(shift + train_len + predict_len - 1)]
 ```
 
-Now that we have the data we can initialize the ESN with the chosen parameters. Given that this is a quick example we are going to change the least amount of possible parameters. For more detailed examples and explanations of the functions please refer to the documentation.
+Now that we have the data we can initialize the ESN with the chosen parameters.
+Given that this is a quick example we are going to change the least amount of
+possible parameters:
 
 ```julia
 input_size = 3
@@ -63,14 +75,17 @@ esn = ESN(input_data, input_size, res_size;
     nla_type=NLAT2())
 ```
 
-The echo state network can now be trained and tested. If not specified, the training will always be ordinary least squares regression. The full range of training methods is detailed in the documentation.
+The echo state network can now be trained and tested.
+If not specified, the training will always be ordinary least squares regression:
 
 ```julia
 output_layer = train(esn, target_data)
 output = esn(Generative(predict_len), output_layer)
 ```
 
-The data is returned as a matrix, `output` in the code above, that contains the predicted trajectories. The results can now be easily plotted (for the actual script used to obtain this plot please refer to the documentation):
+The data is returned as a matrix, `output` in the code above,
+that contains the predicted trajectories.
+The results can now be easily plotted:
 
 ```julia
 using Plots
@@ -80,7 +95,8 @@ plot!(transpose(test); layout=(3, 1), label="actual")
 
 ![lorenz_basic](https://user-images.githubusercontent.com/10376688/166227371-8bffa318-5c49-401f-9c64-9c71980cb3f7.png)
 
-One can also visualize the phase space of the attractor and the comparison with the actual one:
+One can also visualize the phase space of the attractor and the
+comparison with the actual one:
 
 ```julia
 plot(transpose(output)[:, 1],
@@ -111,4 +127,8 @@ If you use this library in your work, please cite:
 
 ## Acknowledgements
 
-This project was possible thanks to initial funding through the [Google summer of code](https://summerofcode.withgoogle.com/) 2020 program. Francesco M. further acknowledges [ScaDS.AI](https://scads.ai/) and [RSC4Earth](https://rsc4earth.de/) for supporting the current progress on the library.
+This project was possible thanks to initial funding through
+the [Google summer of code](https://summerofcode.withgoogle.com/)
+2020 program. Francesco M. further acknowledges [ScaDS.AI](https://scads.ai/)
+and [RSC4Earth](https://rsc4earth.de/) for supporting the current progress
+on the library.

src/esn/esn_inits.jl

@@ -13,6 +13,9 @@ a range defined by `scaling`.
   - `T`: Type of the elements in the reservoir matrix.
     Default is `Float32`.
   - `dims`: Dimensions of the matrix. Should follow `res_size x in_size`.
+
+# Keyword arguments
+
   - `scaling`: A scaling factor to define the range of the uniform distribution.
     The matrix elements will be randomly chosen from the
     range `[-scaling, scaling]`. Defaults to `0.1`.
@@ -55,6 +58,9 @@ elements distributed uniformly within the range [-`scaling`, `scaling`] [^Lu2017
   - `T`: Type of the elements in the reservoir matrix.
     Default is `Float32`.
   - `dims`: Dimensions of the matrix. Should follow `res_size x in_size`.
+
+# Keyword arguments
+
   - `scaling`: The scaling factor for the weight distribution.
     Defaults to `0.1`.
   - `return_sparse`: flag for returning a `sparse` matrix.
@@ -106,6 +112,9 @@ Create an input layer for informed echo state networks [^Pathak2018].
   - `T`: Type of the elements in the reservoir matrix.
     Default is `Float32`.
   - `dims`: Dimensions of the matrix. Should follow `res_size x in_size`.
+
+# Keyword arguments
+
   - `scaling`: The scaling factor for the input matrix.
     Default is 0.1.
   - `model_in_size`: The size of the input model.
@@ -167,6 +176,9 @@ is randomly determined by the `sampling` chosen.
   - `T`: Type of the elements in the reservoir matrix.
     Default is `Float32`.
   - `dims`: Dimensions of the matrix. Should follow `res_size x in_size`.
+
+# Keyword arguments
+
   - `weight`: The weight used to fill the layer matrix. Default is 0.1.
   - `sampling_type`: The sampling parameters used to generate the input matrix.
     Default is `:bernoulli`.
@@ -239,7 +251,10 @@ function minimal_init(rng::AbstractRNG, ::Type{T}, dims::Integer...;
             rng,
             T)
     else
-        error("Sampling type not allowed. Please use one of :bernoulli or :irrational")
+        error("""\n
+            Sampling type not allowed. 
+            Please use one of :bernoulli or :irrational\n
+        """)
     end
     return layer_matrix
 end
@@ -282,7 +297,7 @@ function _create_irrational(irrational::Irrational, start::Int, res_size::Int,
     return T.(input_matrix)
 end
 
-"""
+@doc raw"""
     chebyshev_mapping([rng], [T], dims...;
         amplitude=one(T), sine_divisor=one(T),
         chebyshev_parameter=one(T), return_sparse=true)
@@ -292,14 +307,15 @@ using a sine function and subsequent rows are iteratively generated
 via the Chebyshev mapping. The first row is defined as:
 
 ```math
-    w(1, j) = amplitude * sin(j * π / (sine_divisor * n_cols))
+    W[1, j] = \text{amplitude} \cdot \sin(j \cdot \pi / (\text{sine_divisor} 
+        \cdot \text{n_cols}))
 ```
 
 for j = 1, 2, …, n_cols (with n_cols typically equal to K+1, where K is the number of input layer neurons).
 Subsequent rows are generated by applying the mapping:
 
 ```math
-    w(i+1, j) = cos(chebyshev_parameter * acos(w(i, j)))
+    W[i+1, j] = \cos( \text{chebyshev_parameter} \cdot \acos(W[pi, j]))
 ```
 
 # Arguments
@@ -364,22 +380,23 @@ function chebyshev_mapping(rng::AbstractRNG, ::Type{T}, dims::Integer...;
 end
 
 @doc raw"""
-    logistic_mapping(rng::AbstractRNG, ::Type{T}, dims::Integer...;
-        amplitude=0.3, sine_divisor=5.9, logistic_parameter = 3.7,
+    logistic_mapping([rng], [T], dims...;
+        amplitude=0.3, sine_divisor=5.9, logistic_parameter=3.7,
         return_sparse=true)
 
 Generate an input weight matrix using a logistic mapping [^wang2022].The first
 row is initialized using a sine function:
 
 ```math
-    W(1, j) = amplitude * sin(j * π / (sine_divisor * in_size))
+    W[1, j] = \text{amplitude} \cdot \sin(j \cdot \pi / 
+        (\text{sine_divisor} \cdot in_size))
 ```
 
 for each input index `j`, with `in_size` being the number of columns provided in `dims`. Subsequent rows
 are generated recursively using the logistic map recurrence:
 
 ```math
-    W(i+1, j) = logistic_parameter * W(i, j) * (1 - W(i, j))
+    W[i+1, j] = \text{logistic_parameter} \cdot W(i, j) \cdot (1 - W[i, j])
 ```
 
 # Arguments
@@ -389,7 +406,8 @@ are generated recursively using the logistic map recurrence:
     Default is `Float32`.
   - `dims`: Dimensions of the matrix. Should follow `res_size x in_size`.
 
-# keyword arguments
+# Keyword arguments
+
   - `amplitude`: Scaling parameter used in the sine initialization of the
     first row. Default is 0.3.
   - `sine_divisor`: Parameter used to adjust the phase in the sine initialization.
@@ -452,14 +470,15 @@ as follows:
 - The first element of the chain is initialized using a sine function:
 
 ```math
-      W(1,j) = amplitude * sin( (j * π) / (factor * n * sine_divisor) )
+      W[1,j] = \text{amplitude} \cdot \sin( (j \cdot \pi) / 
+          (\text{factor} \cdot \text{n} \cdot \text{sine_divisor}) )
 ```
   where `j` is the index corresponding to the input and `n` is the number of inputs.
 
 - Subsequent elements are recursively computed using the logistic mapping:
 
 ```math
-      W(i+1,j) = logistic_parameter * W(i,j) * (1 - W(i,j))
+      W[i+1,j] = \text{logistic_parameter} \cdot W[i,j] \cdot (1 - W[i,j])
 ```
 
 The resulting matrix has dimensions `(factor * in_size) x in_size`, where
@@ -474,7 +493,8 @@ the number of rows is overridden.
     Default is `Float32`.
   - `dims`: Dimensions of the matrix. Should follow `res_size x in_size`.
 
-# keyword arguments
+# Keyword arguments
+
   - `factor`: The number of logistic map iterations (chain length) per input,
     determining the number of rows per input.
   - `amplitude`: Scaling parameter A for the sine-based initialization of
@@ -563,6 +583,9 @@ and scaled spectral radius according to `radius`.
   - `T`: Type of the elements in the reservoir matrix.
     Default is `Float32`.
   - `dims`: Dimensions of the reservoir matrix.
+
+# Keyword arguments
+
   - `radius`: The desired spectral radius of the reservoir.
     Defaults to 1.0.
   - `sparsity`: The sparsity level of the reservoir matrix,
@@ -590,7 +613,10 @@ function rand_sparse(rng::AbstractRNG, ::Type{T}, dims::Integer...;
     rho_w = maximum(abs.(eigvals(reservoir_matrix)))
     reservoir_matrix .*= radius / rho_w
     if Inf in unique(reservoir_matrix) || -Inf in unique(reservoir_matrix)
-        error("Sparsity too low for size of the matrix. Increase res_size or increase sparsity")
+        error("""\n
+            Sparsity too low for size of the matrix.
+            Increase res_size or increase sparsity.\n
+          """)
     end
 
     return return_sparse ? sparse(reservoir_matrix) : reservoir_matrix
@@ -609,6 +635,9 @@ Create and return a delay line reservoir matrix [^Rodan2010].
   - `T`: Type of the elements in the reservoir matrix.
     Default is `Float32`.
   - `dims`: Dimensions of the reservoir matrix.
+
+# Keyword arguments
+
   - `weight`: Determines the value of all connections in the reservoir.
     Default is 0.1.
   - `return_sparse`: flag for returning a `sparse` matrix.
@@ -640,8 +669,10 @@ julia> res_matrix = delay_line(5, 5; weight=1)
 function delay_line(rng::AbstractRNG, ::Type{T}, dims::Integer...;
         weight=T(0.1), return_sparse::Bool=true) where {T <: Number}
     reservoir_matrix = DeviceAgnostic.zeros(rng, T, dims...)
-    @assert length(dims) == 2&&dims[1] == dims[2] "The dimensions
-    must define a square matrix (e.g., (100, 100))"
+    @assert length(dims) == 2&&dims[1] == dims[2] """\n
+        The dimensions must define a square matrix
+        (e.g., (100, 100))
+      """
 
     for i in 1:(dims[1] - 1)
         reservoir_matrix[i + 1, i] = weight
@@ -652,7 +683,7 @@ end
 
 """
     delay_line_backward([rng], [T], dims...;
-        weight = 0.1, fb_weight = 0.2, return_sparse=true)
+        weight=0.1, fb_weight=0.2, return_sparse=true)
 
 Create a delay line backward reservoir with the specified by `dims` and weights.
 Creates a matrix with backward connections as described in [^Rodan2010].
@@ -664,6 +695,9 @@ Creates a matrix with backward connections as described in [^Rodan2010].
   - `T`: Type of the elements in the reservoir matrix.
     Default is `Float32`.
   - `dims`: Dimensions of the reservoir matrix.
+
+# Keyword arguments
+
   - `weight`: The weight determines the absolute value of
     forward connections in the reservoir. Default is 0.1
   - `fb_weight`: Determines the absolute value of backward connections
@@ -709,7 +743,7 @@ end
 
 """
     cycle_jumps([rng], [T], dims...; 
-        cycle_weight = 0.1, jump_weight = 0.1, jump_size = 3, return_sparse=true)
+        cycle_weight=0.1, jump_weight=0.1, jump_size=3, return_sparse=true)
 
 Create a cycle jumps reservoir with the specified dimensions,
 cycle weight, jump weight, and jump size.
@@ -721,6 +755,9 @@ cycle weight, jump weight, and jump size.
   - `T`: Type of the elements in the reservoir matrix.
     Default is `Float32`.
   - `dims`: Dimensions of the reservoir matrix.
+
+# Keyword arguments
+
   - `cycle_weight`:  The weight of cycle connections.
     Default is 0.1.
   - `jump_weight`: The weight of jump connections.
@@ -779,7 +816,7 @@ end
 
 """
     simple_cycle([rng], [T], dims...; 
-        weight = 0.1, return_sparse=true)
+        weight=0.1, return_sparse=true)
 
 Create a simple cycle reservoir with the specified dimensions and weight.
 
@@ -789,6 +826,9 @@ Create a simple cycle reservoir with the specified dimensions and weight.
     from WeightInitializers.
   - `T`: Type of the elements in the reservoir matrix. Default is `Float32`.
   - `dims`: Dimensions of the reservoir matrix.
+
+# Keyword arguments
+
   - `weight`: Weight of the connections in the reservoir matrix.
     Default is 0.1.
   - `return_sparse`: flag for returning a `sparse` matrix.
@@ -831,7 +871,7 @@ end
 
 """
     pseudo_svd([rng], [T], dims...; 
-        max_value=1.0, sparsity=0.1, sorted = true, reverse_sort = false,
+        max_value=1.0, sparsity=0.1, sorted=true, reverse_sort=false,
         return_sparse=true)
 
 Returns an initializer to build a sparse reservoir matrix with the given
@@ -844,6 +884,9 @@ Returns an initializer to build a sparse reservoir matrix with the given
   - `T`: Type of the elements in the reservoir matrix.
     Default is `Float32`.
   - `dims`: Dimensions of the reservoir matrix.
+
+# Keyword arguments
+
   - `max_value`: The maximum absolute value of elements in the matrix.
     Default is 1.0
   - `sparsity`: The desired sparsity level of the reservoir matrix.