Rename function sparse_to_dense as to_dense and dense_to_sparse as to_sparse

CHANGELOG.md

@@ -7,6 +7,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased](https://github.com/qutip/QuantumToolbox.jl/tree/main)
 
+- Rename `sparse_to_dense` as `to_dense` and `dense_to_sparse` as `to_sparse`. ([#392])
+
 ## [v0.26.0]
 Release date: 2025-02-09
 
@@ -115,3 +117,4 @@ Release date: 2024-11-13
 [#386]: https://github.com/qutip/QuantumToolbox.jl/issues/386
 [#388]: https://github.com/qutip/QuantumToolbox.jl/issues/388
 [#389]: https://github.com/qutip/QuantumToolbox.jl/issues/389
+[#392]: https://github.com/qutip/QuantumToolbox.jl/issues/392

docs/src/resources/api.md

@@ -103,8 +103,8 @@ ket2dm
 expect
 variance
 LinearAlgebra.kron
-sparse_to_dense
-dense_to_sparse
+to_dense
+to_sparse
 vec2mat
 mat2vec
 ```

docs/src/users_guide/QuantumObject/QuantumObject.md

@@ -214,14 +214,14 @@ SparseMatrixCSC{Int64}(x_s)
 Matrix{Float64}(x_s)
 ```
 
-To convert between dense and sparse arrays, one can also use [`dense_to_sparse`](@ref) and [`sparse_to_dense`](@ref):
+To convert between dense and sparse arrays, one can also use [`to_sparse`](@ref) and [`to_dense`](@ref):
 
 ```@example Qobj
-x_d = sparse_to_dense(x_s)
+x_d = to_dense(x_s)
 ```
 
 ```@example Qobj
-dense_to_sparse(x_d)
+to_sparse(x_d)
 ```
 
 !!! note "Convert to GPU arrays"

ext/QuantumToolboxCUDAExt.jl

@@ -101,9 +101,9 @@ _change_eltype(::Type{T}, ::Val{32}) where {T<:AbstractFloat} = Float32
 _change_eltype(::Type{Complex{T}}, ::Val{64}) where {T<:Union{Int,AbstractFloat}} = ComplexF64
 _change_eltype(::Type{Complex{T}}, ::Val{32}) where {T<:Union{Int,AbstractFloat}} = ComplexF32
 
-QuantumToolbox.sparse_to_dense(A::MT) where {MT<:AbstractCuSparseArray} = CuArray(A)
+QuantumToolbox.to_dense(A::MT) where {MT<:AbstractCuSparseArray} = CuArray(A)
 
-QuantumToolbox.sparse_to_dense(::Type{T1}, A::CuArray{T2}) where {T1<:Number,T2<:Number} = CuArray{T1}(A)
-QuantumToolbox.sparse_to_dense(::Type{T}, A::AbstractCuSparseArray) where {T<:Number} = CuArray{T}(A)
+QuantumToolbox.to_dense(::Type{T1}, A::CuArray{T2}) where {T1<:Number,T2<:Number} = CuArray{T1}(A)
+QuantumToolbox.to_dense(::Type{T}, A::AbstractCuSparseArray) where {T<:Number} = CuArray{T}(A)
 
 end

src/deprecated.jl

@@ -15,11 +15,19 @@ end
 =#
 
 export FFTCorrelation
+export sparse_to_dense, dense_to_sparse
 
 FFTCorrelation() = error(
     "`FFTCorrelation` has been deprecated and will be removed in next major release, please use `spectrum_correlation_fft` to calculate the spectrum with FFT method instead.",
 )
 
+sparse_to_dense(args...) = error(
+    "`sparse_to_dense` has been deprecated and will be removed in next major release, please use `to_dense` instead.",
+)
+dense_to_sparse(args...) = error(
+    "`dense_to_sparse` has been deprecated and will be removed in next major release, please use `to_sparse` instead.",
+)
+
 correlation_3op_2t(
     H::QuantumObject{HOpType},
     ψ0::QuantumObject{StateOpType},

src/qobj/arithmetic_and_attributes.jl

@@ -298,7 +298,7 @@ LinearAlgebra.tr(
 
 Return the singular values of a [`QuantumObject`](@ref) in descending order
 """
-LinearAlgebra.svdvals(A::QuantumObject) = svdvals(sparse_to_dense(A.data))
+LinearAlgebra.svdvals(A::QuantumObject) = svdvals(to_dense(A.data))
 
 @doc raw"""
     norm(A::QuantumObject, p::Real)
@@ -415,7 +415,7 @@ Matrix square root of [`QuantumObject`](@ref)
 !!! note
     `√(A)` (where `√` can be typed by tab-completing `\sqrt` in the REPL) is a synonym of `sqrt(A)`.
 """
-LinearAlgebra.sqrt(A::QuantumObject) = QuantumObject(sqrt(sparse_to_dense(A.data)), A.type, A.dimensions)
+LinearAlgebra.sqrt(A::QuantumObject) = QuantumObject(sqrt(to_dense(A.data)), A.type, A.dimensions)
 
 @doc raw"""
     log(A::QuantumObject)
@@ -425,7 +425,7 @@ Matrix logarithm of [`QuantumObject`](@ref)
 Note that this function only supports for [`Operator`](@ref) and [`SuperOperator`](@ref)
 """
 LinearAlgebra.log(A::QuantumObject{ObjType}) where {ObjType<:Union{OperatorQuantumObject,SuperOperatorQuantumObject}} =
-    QuantumObject(log(sparse_to_dense(A.data)), A.type, A.dimensions)
+    QuantumObject(log(to_dense(A.data)), A.type, A.dimensions)
 
 @doc raw"""
     exp(A::QuantumObject)
@@ -437,7 +437,7 @@ Note that this function only supports for [`Operator`](@ref) and [`SuperOperator
 LinearAlgebra.exp(
     A::QuantumObject{ObjType,DimsType,<:AbstractMatrix},
 ) where {ObjType<:Union{OperatorQuantumObject,SuperOperatorQuantumObject},DimsType} =
-    QuantumObject(dense_to_sparse(exp(A.data)), A.type, A.dimensions)
+    QuantumObject(to_sparse(exp(A.data)), A.type, A.dimensions)
 LinearAlgebra.exp(
     A::QuantumObject{ObjType,DimsType,<:AbstractSparseMatrix},
 ) where {ObjType<:Union{OperatorQuantumObject,SuperOperatorQuantumObject},DimsType} =

src/qobj/eigsolve.jl

@@ -462,10 +462,10 @@ function LinearAlgebra.eigen(
     kwargs...,
 ) where {OpType<:Union{OperatorQuantumObject,SuperOperatorQuantumObject}}
     MT = typeof(A.data)
-    F = eigen(sparse_to_dense(A.data); kwargs...)
+    F = eigen(to_dense(A.data); kwargs...)
     # This fixes a type inference issue. But doesn't work for GPU arrays
-    E::mat2vec(sparse_to_dense(MT)) = F.values
-    U::sparse_to_dense(MT) = F.vectors
+    E::mat2vec(to_dense(MT)) = F.values
+    U::to_dense(MT) = F.vectors
 
     return EigsolveResult(E, U, A.type, A.dimensions, 0, 0, true)
 end
@@ -478,7 +478,7 @@ Same as [`eigen(A::QuantumObject; kwargs...)`](@ref) but for only the eigenvalue
 LinearAlgebra.eigvals(
     A::QuantumObject{OpType};
     kwargs...,
-) where {OpType<:Union{OperatorQuantumObject,SuperOperatorQuantumObject}} = eigvals(sparse_to_dense(A.data); kwargs...)
+) where {OpType<:Union{OperatorQuantumObject,SuperOperatorQuantumObject}} = eigvals(to_dense(A.data); kwargs...)
 
 @doc raw"""
     eigenenergies(A::QuantumObject; sparse::Bool=false, kwargs...)

src/qobj/functions.jl

@@ -4,7 +4,7 @@ Functions which manipulates QuantumObject
 
 export ket2dm
 export expect, variance
-export sparse_to_dense, dense_to_sparse
+export to_dense, to_sparse
 export vec2mat, mat2vec
 
 @doc raw"""
@@ -113,42 +113,42 @@ variance(O::QuantumObject{OperatorQuantumObject}, ψ::QuantumObject) = expect(O^
 variance(O::QuantumObject{OperatorQuantumObject}, ψ::Vector{<:QuantumObject}) = expect(O^2, ψ) .- expect(O, ψ) .^ 2
 
 @doc raw"""
-    sparse_to_dense(A::QuantumObject)
+    to_dense(A::QuantumObject)
 
 Converts a sparse QuantumObject to a dense QuantumObject.
 """
-sparse_to_dense(A::QuantumObject) = QuantumObject(sparse_to_dense(A.data), A.type, A.dimensions)
-sparse_to_dense(A::MT) where {MT<:AbstractSparseArray} = Array(A)
-sparse_to_dense(A::MT) where {MT<:AbstractArray} = A
+to_dense(A::QuantumObject) = QuantumObject(to_dense(A.data), A.type, A.dimensions)
+to_dense(A::MT) where {MT<:AbstractSparseArray} = Array(A)
+to_dense(A::MT) where {MT<:AbstractArray} = A
 
-sparse_to_dense(::Type{T}, A::AbstractSparseArray) where {T<:Number} = Array{T}(A)
-sparse_to_dense(::Type{T1}, A::AbstractArray{T2}) where {T1<:Number,T2<:Number} = Array{T1}(A)
-sparse_to_dense(::Type{T}, A::AbstractArray{T}) where {T<:Number} = A
+to_dense(::Type{T}, A::AbstractSparseArray) where {T<:Number} = Array{T}(A)
+to_dense(::Type{T1}, A::AbstractArray{T2}) where {T1<:Number,T2<:Number} = Array{T1}(A)
+to_dense(::Type{T}, A::AbstractArray{T}) where {T<:Number} = A
 
-function sparse_to_dense(::Type{M}) where {M<:SparseMatrixCSC}
+function to_dense(::Type{M}) where {M<:SparseMatrixCSC}
     T = M
     par = T.parameters
     npar = length(par)
     (2 == npar) || error("Type $M is not supported.")
     return Matrix{par[1]}
 end
 
-sparse_to_dense(::Type{M}) where {M<:AbstractMatrix} = M
+to_dense(::Type{M}) where {M<:AbstractMatrix} = M
 
 @doc raw"""
-    dense_to_sparse(A::QuantumObject)
+    to_sparse(A::QuantumObject)
 
 Converts a dense QuantumObject to a sparse QuantumObject.
 """
-dense_to_sparse(A::QuantumObject, tol::Real = 1e-10) = QuantumObject(dense_to_sparse(A.data, tol), A.type, A.dimensions)
-function dense_to_sparse(A::MT, tol::Real = 1e-10) where {MT<:AbstractMatrix}
+to_sparse(A::QuantumObject, tol::Real = 1e-10) = QuantumObject(to_sparse(A.data, tol), A.type, A.dimensions)
+function to_sparse(A::MT, tol::Real = 1e-10) where {MT<:AbstractMatrix}
     idxs = findall(@. abs(A) > tol)
     row_indices = getindex.(idxs, 1)
     col_indices = getindex.(idxs, 2)
     vals = getindex(A, idxs)
     return sparse(row_indices, col_indices, vals, size(A)...)
 end
-function dense_to_sparse(A::VT, tol::Real = 1e-10) where {VT<:AbstractVector}
+function to_sparse(A::VT, tol::Real = 1e-10) where {VT<:AbstractVector}
     idxs = findall(@. abs(A) > tol)
     vals = getindex(A, idxs)
     return sparsevec(idxs, vals, length(A))

src/qobj/operators.jl

@@ -48,7 +48,7 @@ function rand_unitary(dimensions::Union{Dimensions,AbstractVector{Int},Tuple}, :
     # Because inv(Λ) ⋅ R has real and strictly positive elements, Q · Λ is therefore Haar distributed.
     Λ = diag(R) # take the diagonal elements of R
     Λ ./= abs.(Λ) # rescaling the elements
-    return QuantumObject(sparse_to_dense(Q * Diagonal(Λ)); type = Operator, dims = dimensions)
+    return QuantumObject(to_dense(Q * Diagonal(Λ)); type = Operator, dims = dimensions)
 end
 function rand_unitary(dimensions::Union{Dimensions,AbstractVector{Int},Tuple}, ::Val{:exp})
     N = prod(dimensions)
@@ -59,7 +59,7 @@ function rand_unitary(dimensions::Union{Dimensions,AbstractVector{Int},Tuple}, :
     # generate Hermitian matrix
     H = QuantumObject((Z + Z') / 2; type = Operator, dims = dimensions)
 
-    return sparse_to_dense(exp(-1.0im * H))
+    return to_dense(exp(-1.0im * H))
 end
 rand_unitary(dimensions::Union{Dimensions,AbstractVector{Int},Tuple}, ::Val{T}) where {T} =
     throw(ArgumentError("Invalid distribution: $(T)"))

src/qobj/quantum_object.jl

@@ -205,7 +205,7 @@ SciMLOperators.cache_operator(
     L::AbstractQuantumObject{OpType},
     u::AbstractVector,
 ) where {OpType<:Union{OperatorQuantumObject,SuperOperatorQuantumObject}} =
-    get_typename_wrapper(L)(cache_operator(L.data, sparse_to_dense(similar(u))), L.type, L.dimensions)
+    get_typename_wrapper(L)(cache_operator(L.data, to_dense(similar(u))), L.type, L.dimensions)
 
 function SciMLOperators.cache_operator(
     L::AbstractQuantumObject{OpType},

src/qobj/quantum_object_evo.jl

@@ -281,7 +281,7 @@ function QuantumObjectEvolution(
     end
 
     # Preallocate the SciMLOperator cache using a dense vector as a reference
-    v0 = sparse_to_dense(similar(op.data, size(op, 1)))
+    v0 = to_dense(similar(op.data, size(op, 1)))
     data = cache_operator(data, v0)
 
     return QuantumObjectEvolution(data, type, dims)

src/time_evolution/mesolve.jl

@@ -76,7 +76,7 @@ function mesolveProblem(
     check_dimensions(L_evo, ψ0)
 
     T = Base.promote_eltype(L_evo, ψ0)
-    ρ0 = sparse_to_dense(_CType(T), mat2vec(ket2dm(ψ0).data)) # Convert it to dense vector with complex element type
+    ρ0 = to_dense(_CType(T), mat2vec(ket2dm(ψ0).data)) # Convert it to dense vector with complex element type
     L = L_evo.data
 
     is_empty_e_ops = (e_ops isa Nothing) ? true : isempty(e_ops)

src/time_evolution/sesolve.jl

@@ -66,7 +66,7 @@ function sesolveProblem(
     check_dimensions(H_evo, ψ0)
 
     T = Base.promote_eltype(H_evo, ψ0)
-    ψ0 = sparse_to_dense(_CType(T), get_data(ψ0)) # Convert it to dense vector with complex element type
+    ψ0 = to_dense(_CType(T), get_data(ψ0)) # Convert it to dense vector with complex element type
     U = H_evo.data
 
     is_empty_e_ops = (e_ops isa Nothing) ? true : isempty(e_ops)

src/time_evolution/smesolve.jl

@@ -91,7 +91,7 @@ function smesolveProblem(
     dims = L_evo.dimensions
 
     T = Base.promote_eltype(L_evo, ψ0)
-    ρ0 = sparse_to_dense(_CType(T), mat2vec(ket2dm(ψ0).data)) # Convert it to dense vector with complex element type
+    ρ0 = to_dense(_CType(T), mat2vec(ket2dm(ψ0).data)) # Convert it to dense vector with complex element type
 
     progr = ProgressBar(length(tlist), enable = getVal(progress_bar))
 

src/time_evolution/ssesolve.jl

@@ -167,7 +167,7 @@ function ssesolveProblem(
     check_dimensions(H_eff_evo, ψ0)
     dims = H_eff_evo.dimensions
 
-    ψ0 = sparse_to_dense(_CType(ψ0), get_data(ψ0))
+    ψ0 = to_dense(_CType(ψ0), get_data(ψ0))
 
     progr = ProgressBar(length(tlist), enable = getVal(progress_bar))
 

src/time_evolution/time_evolution.jl

@@ -397,12 +397,12 @@ function liouvillian_generalized(
     H_d = QuantumObject(Diagonal(complex(E)), type = Operator, dims = dims)
 
     Ω = E' .- E
-    Ωp = triu(dense_to_sparse(Ω, tol), 1)
+    Ωp = triu(to_sparse(Ω, tol), 1)
 
     # Filter in the Hilbert space
     σ = isnothing(σ_filter) ? 500 * maximum([norm(field) / length(field) for field in fields]) : σ_filter
     F1 = QuantumObject(gaussian.(Ω, 0, σ), type = Operator, dims = dims)
-    F1 = dense_to_sparse(F1, tol)
+    F1 = to_sparse(F1, tol)
 
     # Filter in the Liouville space
     # M1 = ones(final_size, final_size)
@@ -412,13 +412,13 @@ function liouvillian_generalized(
     Ω2 = kron(M1, Ω)
     Ωdiff = Ω1 .- Ω2
     F2 = QuantumObject(gaussian.(Ωdiff, 0, σ), SuperOperator, dims)
-    F2 = dense_to_sparse(F2, tol)
+    F2 = to_sparse(F2, tol)
 
     L = liouvillian(H_d)
 
     for i in eachindex(fields)
         # The operator that couples the system to the bath in the eigenbasis
-        X_op = dense_to_sparse((U'*fields[i]*U).data[1:final_size, 1:final_size], tol)
+        X_op = to_sparse((U'*fields[i]*U).data[1:final_size, 1:final_size], tol)
         if ishermitian(fields[i])
             X_op = (X_op + X_op') / 2 # Make sure it's hermitian
         end
@@ -452,8 +452,8 @@ function _liouvillian_floquet(
     L_0 = L₀.data
     L_p = Lₚ.data
     L_m = Lₘ.data
-    L_p_dense = sparse_to_dense(Lₚ.data)
-    L_m_dense = sparse_to_dense(Lₘ.data)
+    L_p_dense = to_dense(Lₚ.data)
+    L_m_dense = to_dense(Lₘ.data)
 
     S = -(L_0 - 1im * n_max * ω * I) \ L_p_dense
     T = -(L_0 + 1im * n_max * ω * I) \ L_m_dense
@@ -464,5 +464,5 @@ function _liouvillian_floquet(
     end
 
     tol == 0 && return QuantumObject(L_0 + L_m * S + L_p * T, SuperOperator, L₀.dimensions)
-    return QuantumObject(dense_to_sparse(L_0 + L_m * S + L_p * T, tol), SuperOperator, L₀.dimensions)
+    return QuantumObject(to_sparse(L_0 + L_m * S + L_p * T, tol), SuperOperator, L₀.dimensions)
 end

src/wigner.jl

@@ -60,7 +60,7 @@ julia> wig = wigner(ψ, xvec, xvec);
 or taking advantage of the parallel computation of the `WignerLaguerre` method
 
 ```jldoctest wigner
-julia> ρ = ket2dm(ψ) |> dense_to_sparse;
+julia> ρ = ket2dm(ψ) |> to_sparse;
 
 julia> wig = wigner(ρ, xvec, xvec, method=WignerLaguerre(parallel=true));
 

test/core-test/eigenvalues_and_operators.jl

@@ -58,7 +58,7 @@
 
     # eigen solve for general matrices
     vals, _, vecs = eigsolve(L.data, sigma = 0.01, k = 10, krylovdim = 50)
-    vals2, vecs2 = eigen(sparse_to_dense(L.data))
+    vals2, vecs2 = eigen(to_dense(L.data))
     vals3, state3, vecs3 = eigsolve_al(L, 1 \ (40 * κ), k = 10, krylovdim = 50)
     idxs = sortperm(vals2, by = abs)
     vals2 = vals2[idxs][1:10]

test/core-test/generalized_master_equation.jl

@@ -15,12 +15,12 @@
     Tlist = [0, 0.0]
 
     E, U, L1 = liouvillian_generalized(H, fields, Tlist, N_trunc = N_trunc, tol = tol)
-    Ω = dense_to_sparse((E'.-E)[1:N_trunc, 1:N_trunc], tol)
+    Ω = to_sparse((E'.-E)[1:N_trunc, 1:N_trunc], tol)
 
-    H_d = Qobj(dense_to_sparse((U'*H*U)[1:N_trunc, 1:N_trunc], tol))
-    Xp = Qobj(Ω .* dense_to_sparse(triu((U'*(a+a')*U).data[1:N_trunc, 1:N_trunc], 1), tol))
-    a2 = Qobj(dense_to_sparse((U'*a*U).data[1:N_trunc, 1:N_trunc], tol))
-    sm2 = Qobj(dense_to_sparse((U'*sm*U).data[1:N_trunc, 1:N_trunc], tol))
+    H_d = Qobj(to_sparse((U'*H*U)[1:N_trunc, 1:N_trunc], tol))
+    Xp = Qobj(Ω .* to_sparse(triu((U'*(a+a')*U).data[1:N_trunc, 1:N_trunc], 1), tol))
+    a2 = Qobj(to_sparse((U'*a*U).data[1:N_trunc, 1:N_trunc], tol))
+    sm2 = Qobj(to_sparse((U'*sm*U).data[1:N_trunc, 1:N_trunc], tol))
 
     # Standard liouvillian case
     c_ops = [sqrt(0.01) * a2, sqrt(0.01) * sm2]
@@ -33,12 +33,12 @@
     Tlist = [0.2, 0.0]
 
     E, U, L1 = liouvillian_generalized(H, fields, Tlist, N_trunc = N_trunc, tol = tol)
-    Ω = dense_to_sparse((E'.-E)[1:N_trunc, 1:N_trunc], tol)
+    Ω = to_sparse((E'.-E)[1:N_trunc, 1:N_trunc], tol)
 
-    H_d = Qobj(dense_to_sparse((U'*H*U)[1:N_trunc, 1:N_trunc], tol))
-    Xp = Qobj(Ω .* dense_to_sparse(triu((U'*(a+a')*U).data[1:N_trunc, 1:N_trunc], 1), tol))
-    a2 = Qobj(dense_to_sparse((U'*a*U).data[1:N_trunc, 1:N_trunc], tol))
-    sm2 = Qobj(dense_to_sparse((U'*sm*U).data[1:N_trunc, 1:N_trunc], tol))
+    H_d = Qobj(to_sparse((U'*H*U)[1:N_trunc, 1:N_trunc], tol))
+    Xp = Qobj(Ω .* to_sparse(triu((U'*(a+a')*U).data[1:N_trunc, 1:N_trunc], 1), tol))
+    a2 = Qobj(to_sparse((U'*a*U).data[1:N_trunc, 1:N_trunc], tol))
+    sm2 = Qobj(to_sparse((U'*sm*U).data[1:N_trunc, 1:N_trunc], tol))
 
     @test abs(expect(Xp' * Xp, steadystate(L1)) - n_thermal(1, Tlist[1])) / n_thermal(1, Tlist[1]) < 1e-4
 

test/core-test/negativity_and_partial_transpose.jl

@@ -24,7 +24,7 @@
     @testset "partial_transpose" begin
         # A (24 * 24)-matrix which contains number 1 ~ 576
         A_dense = Qobj(reshape(1:(24^2), (24, 24)), dims = (2, 3, 4))
-        A_sparse = dense_to_sparse(A_dense)
+        A_sparse = to_sparse(A_dense)
         PT = (true, false)
         for s1 in PT
             for s2 in PT