code cleanup and docstring fixes

Author: Maximilian-Stefan-Ernst

docs/src/tutorials/regularization/regularization.md

@@ -30,13 +30,12 @@ It can be used as
 ```julia
 SemOptimizerProximal(
     algorithm = ProximalAlgorithms.PANOC(),
-    options = Dict{Symbol, Any}(),
     operator_g,
     operator_h = nothing
     )
 ```
 
-The proximal operator (aka the regularization function) can be passed as `operator_g`, available options are listed [here](https://juliafirstorder.github.io/ProximalOperators.jl/stable/functions/).
+The proximal operator (aka the regularization function) can be passed as `operator_g`.
 The available Algorithms are listed [here](https://juliafirstorder.github.io/ProximalAlgorithms.jl/stable/guide/implemented_algorithms/).
 
 ## First example - lasso

src/additional_functions/helper.jl

@@ -14,84 +14,29 @@ function neumann_series(mat::SparseMatrixCSC; maxiter::Integer = size(mat, 1))
     return inverse
 end
 
-#=
-function make_onelement_array(A)
-    isa(A, Array) ? nothing : (A = [A])
-    return A
-end
- =#
-
-function semvec(observed, implied, loss, optimizer)
-    observed = make_onelement_array(observed)
-    implied = make_onelement_array(implied)
-    loss = make_onelement_array(loss)
-    optimizer = make_onelement_array(optimizer)
-
-    #sem_vec = Array{AbstractSem}(undef, maximum(length.([observed, implied, loss, optimizer])))
-    sem_vec = Sem.(observed, implied, loss, optimizer)
-
-    return sem_vec
-end
-
-skipmissing_mean(mat::AbstractMatrix) =
-    [mean(skipmissing(coldata)) for coldata in eachcol(mat)]
-
-function F_one_person(imp_mean, meandiff, inverse, data, logdet)
-    F = logdet
-    @. meandiff = data - imp_mean
-    F += dot(meandiff, inverse, meandiff)
-    return F
-end
-
-function remove_all_missing(data::AbstractMatrix)
-    keep = Vector{Int64}()
-    for (i, coldata) in zip(axes(data, 1), eachrow(data))
-        if any(!ismissing, coldata)
-            push!(keep, i)
-        end
-    end
-    return data[keep, :], keep
-end
-
 function batch_inv!(fun, model)
     for i in 1:size(fun.inverses, 1)
         fun.inverses[i] .= LinearAlgebra.inv!(fun.choleskys[i])
     end
 end
 
-#=
-function batch_sym_inv_update!(fun::Union{LossFunction, DiffFunction}, model)
-    M_inv = inv(fun.choleskys[1])
-    for i = 1:size(fun.inverses, 1)
-        if size(model.observed.patterns_not[i]) == 0
-            fun.inverses[i] .= M_inv
-        else
-            ind_not = model.observed.patterns_not[i]
-            ind = model.observed.patterns[i]
-
-            A = M_inv[ind_not, ind]
-            H = cholesky(M_inv[ind_not, ind_not])
-            D = H \ A
-            out = M_inv[ind, ind] - LinearAlgebra.BLAS.gemm('T', 'N', 1.0, A, D)
-            fun.inverses[i] .= out
-        end
-    end
-end =#
-
-function sparse_outer_mul!(C, A, B, ind) #computes A*S*B -> C, where ind gives the entries of S that are 1
+# computes A*S*B -> C, where ind gives the entries of S that are 1
+function sparse_outer_mul!(C, A, B, ind) 
     fill!(C, 0.0)
     for i in 1:length(ind)
         BLAS.ger!(1.0, A[:, ind[i][1]], B[ind[i][2], :], C)
     end
 end
 
-function sparse_outer_mul!(C, A, ind) #computes A*∇m, where ∇m ind gives the entries of ∇m that are 1
+# computes A*∇m, where ∇m ind gives the entries of ∇m that are 1
+function sparse_outer_mul!(C, A, ind) 
     fill!(C, 0.0)
     @views C .= sum(A[:, ind], dims = 2)
     return C
 end
 
-function sparse_outer_mul!(C, A, B::Vector, ind) #computes A*S*B -> C, where ind gives the entries of S that are 1
+# computes A*S*B -> C, where ind gives the entries of S that are 1
+function sparse_outer_mul!(C, A, B::Vector, ind) 
     fill!(C, 0.0)
     @views @inbounds for i in 1:length(ind)
         C .+= B[ind[i][2]] .* A[:, ind[i][1]]

src/additional_functions/simulation.jl

@@ -7,7 +7,7 @@ Return a new model with swaped observed part.
 
 # Arguments
 - `model::AbstractSemSingle`: model to swap the observed part of.
-- `kwargs`: additional keyword arguments; typically includes `data = ...`
+- `kwargs`: additional keyword arguments; typically includes `data` and `specification`
 - `observed`: Either an object of subtype of `SemObserved` or a subtype of `SemObserved`
 
 # Examples

src/additional_functions/start_val/start_fabin3.jl

@@ -18,7 +18,7 @@ end
 # SemObservedMissing
 function start_fabin3(observed::SemObservedMissing, implied, args...; kwargs...)
     if !observed.em_model.fitted
-        em_mvn(observed; kwargs...)
+        em_mvn!(observed; kwargs...)
     end
 
     return start_fabin3(implied.ram_matrices, observed.em_model.Σ, observed.em_model.μ)
@@ -45,24 +45,6 @@ function start_fabin3(
     )
     @assert length(F_var2obs) == size(F, 1)
 
-    # check in which matrix each parameter appears
-
-    #=     in_S = length.(S_ind) .!= 0
-        in_A = length.(A_ind) .!= 0
-        A_ind_c = [linear2cartesian(ind, (n_var, n_var)) for ind in A_ind]
-        in_Λ = [any(ind[2] .∈ F_ind) for ind in A_ind_c]
-
-        if !isnothing(M)
-            in_M = length.(M_ind) .!= 0
-            in_any = in_A .| in_S .| in_M
-        else
-            in_any = in_A .| in_S
-        end
-
-        if !all(in_any)
-            @warn "Could not determine fabin3 starting values for some parameters, default to 0."
-        end =#
-
     # set undirected parameters in S
     S_indices = CartesianIndices(S)
     for j in 1:nparams(S)
@@ -79,7 +61,6 @@ function start_fabin3(
 
     # set loadings
     A_indices = CartesianIndices(A)
-    # ind_Λ = findall([is_in_Λ(ind_vec, F_ind) for ind_vec in A_ind_c])
 
     # collect latent variable indicators in A
     # maps latent parameter to the vector of dependent vars

src/additional_functions/start_val/start_simple.jl

@@ -63,7 +63,6 @@ function start_simple(
     nparams(ram_matrices)
 
     start_val = zeros(n_par)
-    n_obs = nobserved_vars(ram_matrices)
     n_var = nvars(ram_matrices)
 
     C_indices = CartesianIndices((n_var, n_var))

src/frontend/common.jl

@@ -1,7 +1,7 @@
 # API methods supported by multiple SEM.jl types
 
 """
-    params(semobj) -> Vector{Symbol}
+    params(partable::ParameterTable) -> Vector{Symbol}
 
 Return the vector of SEM model parameter identifiers.
 """
@@ -42,7 +42,7 @@ nlatent_vars(semobj) = length(latent_vars(semobj))
 """
     param_indices(semobj)
 
-Returns a dict of parameter names and their indices in `semobj`.
+Returns a dict of parameter labels and their indices in `semobj`.
 
 # Examples
 ```julia

src/frontend/fit/fitmeasures/fit_measures.jl

@@ -14,6 +14,6 @@ end
 """
     fit_measures(sem_fit, args...)
 
-Return a default set of fit measures or the fit measures passed as `arg...`.
+Return a default set of fit measures or the fit measures passed as `args...`.
 """
 function fit_measures end

src/frontend/fit/fitmeasures/minus2ll.jl

@@ -41,7 +41,7 @@ end
 function minus2ll(observed::SemObservedMissing)
     # fit EM-based mean and cov if not yet fitted
     # FIXME EM could be very computationally expensive
-    observed.em_model.fitted || em_mvn(observed)
+    observed.em_model.fitted || em_mvn!(observed)
 
     Σ = observed.em_model.Σ
     μ = observed.em_model.μ

src/frontend/fit/summary.jl

@@ -15,15 +15,15 @@ function details(sem_fit::SemFit; show_fitmeasures = false, color = :light_cyan,
     println("Number of data samples:      $(nsamples(sem_fit))")
     print("\n")
     printstyled(
-        "----------------------------------- Model ----------------------------------- \n";
+        "----------------------------------- Model ------------------------------------ \n";
         color = color,
     )
     print("\n")
     print(sem_fit.model)
     print("\n")
     if show_fitmeasures
         printstyled(
-            "--------------------------------- Fitmeasures --------------------------------- \n";
+            "-------------------------------- Fitmeasures --------------------------------- \n";
             color = color,
         )
         print("\n")
@@ -51,7 +51,7 @@ function details(
     if show_variables
         print("\n")
         printstyled(
-            "--------------------------------- Variables --------------------------------- \n";
+            "---------------------------------- Variables --------------------------------- \n";
             color = color,
         )
         print("\n")
@@ -242,9 +242,6 @@ function details(
         )
         print("\n")
     end
-
-    #printstyled("""No need to copy and paste results, you can use CSV.write(DataFrame(my_partable), "myfile.csv")"""; hidden = true)
-
 end
 
 function details(
@@ -297,9 +294,6 @@ function details(
             show_columns = show_columns,
         )
     end
-
-    # printstyled("""No need to copy and paste results, you can use CSV.write(DataFrame(my_partable), "myfile.csv")"""; hidden = true)
-
 end
 
 function check_round(vec; digits)

src/frontend/pretty_printing.jl

@@ -1,3 +1,7 @@
+##############################################################
+# Some helpers to implement show methods for SEM.jl objects
+##############################################################
+
 function print_field_types(io::IO, struct_instance)
     fields = fieldnames(typeof(struct_instance))
     types = [typeof(getproperty(struct_instance, field)) for field in fields]
@@ -25,7 +29,7 @@ function print_type(io::IO, struct_instance)
 end
 
 ##############################################################
-# Loss Functions, Implied,
+# Loss Function, Implied, Observed, Optimizer
 ##############################################################
 
 function Base.show(io::IO, struct_inst::SemLossFunction)