Skip to content

Handle Symmetric/Hermitian in PDSparseMat #230

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
timholy wants to merge 2 commits into
base: master
Choose a base branch
from
Open

Handle Symmetric/Hermitian in PDSparseMat #230

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
d951a7b
Handle Symmetric/Hermitian in PDSparseMat
timholy Dec 23, 2025
a9f3916
more testing
timholy Dec 23, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion Project.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
name = "PDMats"
uuid = "90014a1f-27ba-587c-ab20-58faa44d9150"
version = "0.11.37"
version = "0.11.38"

[deps]
LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
Expand Down
40 changes: 28 additions & 12 deletions src/pdsparsemat.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,23 +1,39 @@
const AbstractSparseLike{T<:Real, S<:AbstractSparseMatrix{T}} = Union{S, Symmetric{T,S}, Hermitian{T,S}}
# We can't define `convert(AbstractSparseLike{T}, mat)` here because of piracy
convert_abstractsparselike(::Type{T}, mat::AbstractSparseMatrix) where T<:Real = convert(AbstractSparseMatrix{T}, mat)
convert_abstractsparselike(::Type{T}, mat::SparseMatrixCSC) where T<:Real = convert(SparseMatrixCSC{T}, mat) # the AbstractSparseMatrix method doesn't work
convert_abstractsparselike(::Type{T}, mat::Symmetric{S,<:AbstractSparseMatrix{S}} where S) where T<:Real = Symmetric(convert_abstractsparselike(T, mat.data), Symbol(mat.uplo))
convert_abstractsparselike(::Type{T}, mat::Hermitian{S,<:AbstractSparseMatrix{S}} where S) where T<:Real = Hermitian(convert_abstractsparselike(T, mat.data), Symbol(mat.uplo))

"""
Sparse positive definite matrix together with a Cholesky factorization object.
"""
struct PDSparseMat{T <: Real, S <: AbstractSparseMatrix} <: AbstractPDMat{T}
struct PDSparseMat{T <: Real, S <: AbstractSparseLike{T}, C <: CholTypeSparse} <: AbstractPDMat{T}
mat::S
chol::CholTypeSparse

PDSparseMat{T, S}(m::AbstractSparseMatrix{T}, c::CholTypeSparse) where {T, S} =
new{T, S}(m, c) #add {T} to CholTypeSparse argument once #14076 is implemented
chol::C
# Note: cholesky(::SparseMatrixCSC{Float32}) returns CholTypeSparse{Float64}, so we can't enforce C <: CholTypeSparse{T}

function PDSparseMat{T, S, C}(m::AbstractSparseLike{T}, c::CholTypeSparse) where {T, S, C}
d = LinearAlgebra.checksquare(m)
size(c, 1) == d ||
throw(DimensionMismatch("Dimensions of mat and chol are inconsistent."))
return new{T, S, C}(m, c)
end
end
@deprecate PDSparseMat{T, S}(d::Int, m::AbstractSparseMatrix{T}, c::CholTypeSparse) where {T, S} PDSparseMat{T, S}(m, c)
@deprecate PDSparseMat{T, S}(d::Int, m::AbstractSparseMatrix{T}, c::CholTypeSparse) where {T, S} PDSparseMat{T, S, typeof(c)}(m, c)

function PDSparseMat(mat::AbstractSparseMatrix, chol::CholTypeSparse)
d = LinearAlgebra.checksquare(mat)
size(chol, 1) == d ||
throw(DimensionMismatch("Dimensions of mat and chol are inconsistent."))
return PDSparseMat{eltype(mat), typeof(mat)}(mat, chol)
PDSparseMat{T, S}(mat::AbstractSparseLike, chol::CholTypeSparse) where {T <: Real, S <: AbstractSparseLike{T}} = PDSparseMat{T, S, typeof(chol)}(mat, chol)
PDSparseMat{T}(mat::AbstractSparseLike{T}, chol::CholTypeSparse) where {T <: Real} = PDSparseMat{T, typeof(mat)}(mat, chol)
function PDSparseMat{T}(mat::AbstractSparseLike, chol::CholTypeSparse) where {T <: Real}
mat = convert_abstractsparselike(T, mat)::AbstractSparseLike{T} # prevent StackOverflowError if the eltype conversion lies
return PDSparseMat{T}(mat, chol)
end
PDSparseMat(mat::AbstractSparseLike{T}, chol::CholTypeSparse) where {T <: Real} = PDSparseMat{T}(mat, chol)

PDSparseMat{T}(mat::AbstractSparseLike{T,S}) where {T<:Real,S<:SparseMatrixCSC{T}} = PDSparseMat{T}(mat, cholesky(mat))
PDSparseMat{T}(mat::AbstractSparseLike) where T<:Real = PDSparseMat{T}(convert_abstractsparselike(T, mat))

PDSparseMat(mat::SparseMatrixCSC) = PDSparseMat(mat, cholesky(mat))
PDSparseMat(mat::AbstractSparseLike{T,S}) where {T<:Real,S<:SparseMatrixCSC{T}} = PDSparseMat{T}(mat)
PDSparseMat(fac::CholTypeSparse) = PDSparseMat(sparse(fac), fac)

function Base.getproperty(a::PDSparseMat, s::Symbol)
Expand Down
2 changes: 1 addition & 1 deletion src/utils.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@ macro check_argdims(cond)
end
end

function _addscal!(r::Matrix, a::Matrix, b::Union{Matrix, SparseMatrixCSC}, c::Real)
function _addscal!(r::Matrix, a::Matrix, b::Union{M, Symmetric{T, <:M}, Hermitian{T, <:M}}, c::Real) where {T<:Real, M <: Union{Matrix, SparseMatrixCSC}}
if c == one(c)
for i in eachindex(a)
@inbounds r[i] = a[i] + b[i]
Expand Down
18 changes: 17 additions & 1 deletion test/pdmtypes.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,8 @@ using Test

@testset "test the functionality" begin
M = convert(Array{T, 2}, [4.0 -2.0 -1.0; -2.0 5.0 -1.0; -1.0 -1.0 6.0])
MS = Symmetric(convert(Matrix{T}, [4.0 NaN NaN; -2.0 5.0 NaN; -1.0 -1.0 6.0]), :L)
MH = Hermitian(convert(Matrix{T}, [4.0 NaN NaN; -2.0 5.0 NaN; -1.0 -1.0 6.0]), :L)
V = convert(Array{T, 1}, [1.5, 2.5, 2.0])
X = convert(T, 2.0)
f64M = Float64.(M)
Expand All @@ -25,6 +27,8 @@ using Test
test_pdmat(PDMat(M), M, cmat_eq = true, verbose = 1)
test_pdmat(PDMat{Float64}(M), f64M, cmat_eq = true, verbose = 1)
test_pdmat(PDMat{Float64, Matrix{Float64}}(M), f64M, cmat_eq = true, verbose = 1)
test_pdmat(PDMat(MS), M, cmat_eq = true, verbose = 1)
test_pdmat(PDMat(MH), M, cmat_eq = true, verbose = 1)
@test_throws TypeError PDMat{Float32, Matrix{Float64}}(M)
end
@testset "PDMat from PDMat" begin
Expand Down Expand Up @@ -57,6 +61,16 @@ using Test
end
@testset "PDSparseMat" begin
test_pdmat(PDSparseMat(sparse(M)), M, cmat_eq = true, verbose = 1, t_eig = false)
test_pdmat(PDSparseMat{Float64}(sparse(M)), Float64.(M), cmat_eq = true, verbose = 1, t_eig = false)
test_pdmat(PDSparseMat{Float64}(sparse(M), cholesky(sparse(M))), Float64.(M), cmat_eq = true, verbose = 1, t_eig = false)
sparseMS = Symmetric(sparse(MS.data), :L)
test_pdmat(PDSparseMat(sparseMS), M, cmat_eq = true, verbose = 1, t_eig = false)
test_pdmat(PDSparseMat{Float64}(sparseMS), Float64.(M), cmat_eq = true, verbose = 1, t_eig = false)
test_pdmat(PDSparseMat{Float64}(sparseMS, cholesky(sparseMS)), Float64.(M), cmat_eq = true, verbose = 1, t_eig = false)
sparseMH = Hermitian(sparse(MH.data), :L)
test_pdmat(PDSparseMat(sparseMH), M, cmat_eq = true, verbose = 1, t_eig = false)
test_pdmat(PDSparseMat{Float64}(sparseMH), Float64.(M), cmat_eq = true, verbose = 1, t_eig = false)
test_pdmat(PDSparseMat{Float64}(sparseMH, cholesky(sparseMH)), Float64.(M), cmat_eq = true, verbose = 1, t_eig = false)
end
end

Expand Down Expand Up @@ -222,7 +236,9 @@ using Test
@test Mat32 isa Matrix{Float32}
@test Mat32 ≈ Float32.(A)

M = @inferred AbstractPDMat(cholesky(sparse(A)))
# sparse(::SparseArrays.CHOLMOD.Factor) is inferred to return a Union{Symmetric{T, <:SparseMatrixCSC{T}}, SparseMatrixCSC{T}}
# Now that we support Symmetric directly, the following is no longer inferrable
M = AbstractPDMat(cholesky(sparse(A)))
@test M isa PDSparseMat
@test Matrix(M) ≈ A
end
Expand Down
Loading