Skip to content

add PosDefCholeskyFactor #38

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.

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
wants to merge 3 commits into
base: master
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
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
1 change: 1 addition & 0 deletions docs/src/index.md
Original file line number Diff line number Diff line change
Expand Up @@ -98,6 +98,7 @@ as𝕀
```@docs
UnitVector
CorrCholeskyFactor
PosDefCholeskyFactor
```

# Defining custom transformations
Expand Down
62 changes: 61 additions & 1 deletion src/special_arrays.jl
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
export UnitVector, CorrCholeskyFactor
export UnitVector, CorrCholeskyFactor, PosDefCholeskyFactor

"""
(y, r, ℓ) = $SIGNATURES
Expand Down Expand Up @@ -129,3 +129,63 @@ function inverse!(x::RealVector, t::CorrCholeskyFactor, U::UpperTriangular)
end
x
end



"""
PosDefCholeskyFactor(n)

Cholesky factor of a symmetric positive-definite matrix of size `n`.

Transforms ``n×(n+1)/2`` real numbers to an ``n×n`` upper-triangular matrix `Ω`, such that
`Ω'*Ω` is a positive definite matrix.
"""
@calltrans struct PosDefCholeskyFactor <: VectorTransform
n::Int
function PosDefCholeskyFactor(n)
@argcheck n ≥ 1 "Dimension should be positive."
new(n)
end
end

dimension(t::PosDefCholeskyFactor) = (t.n*(t.n+1)) ÷ 2

function transform_with(flag::LogJacFlag, t::PosDefCholeskyFactor, x::RealVector)
@unpack n = t
T = extended_eltype(x)
ℓ = logjac_zero(flag, T)
U = Matrix{T}(undef, n, n)
index = firstindex(x)
@inbounds for col in 1:n
for row in 1:(col-1)
U[row, col] = x[index]
index += 1
end
if flag isa NoLogJac
Copy link
Owner

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You could avoid code duplication here by using transform_with(flag, ...). When flag::NoLogJac, it is "summed" automatically as a no-op.

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

transform_with doesn't seem to work for scalar transforms with scalar args:

julia> TransformVariables.transform_with(TransformVariables.NoLogJac(), asℝ₊, 1.0)
ERROR: MethodError: no method matching transform_with(::TransformVariables.NoLogJac, ::TransformVariables.ShiftedExp{true,Float64}, ::Float64)
Closest candidates are:
  transform_with(::TransformVariables.NoLogJac, ::TransformVariables.ScalarTransform, ::AbstractArray{T,1} where T<:Real) at /Users/simon/.julia/dev/TransformVariables/src/scalar.jl:15
  transform_with(::TransformVariables.LogJac, ::TransformVariables.ScalarTransform, ::AbstractArray{T,1} where T<:Real) at /Users/simon/.julia/dev/TransformVariables/src/scalar.jl:18
  transform_with(::TransformVariables.LogJacFlag, ::UnitVector, ::AbstractArray{T,1} where T<:Real) at /Users/simon/.julia/dev/TransformVariables/src/special_arrays.jl:45
  ...
Stacktrace:
 [1] top-level scope at none:0

U[col, col] = transform(asℝ₊, x[index])
else
U[col, col], ℓi = transform_and_logjac(asℝ₊, x[index])
ℓ += ℓi
end
index += 1
end
UpperTriangular(U), ℓ
end

inverse_eltype(t::PosDefCholeskyFactor, U::UpperTriangular) = extended_eltype(U)

function inverse!(x::RealVector, t::PosDefCholeskyFactor, U::UpperTriangular)
@unpack n = t
@argcheck size(U, 1) == n
@argcheck length(x) == dimension(t)
index = firstindex(x)
@inbounds for col in 1:n
for row in 1:(col-1)
x[index] = U[row,col]
index += 1
end
x[index] = inverse(asℝ₊, U[col, col])
index += 1
end
x
end
11 changes: 11 additions & 0 deletions test/runtests.jl
Original file line number Diff line number Diff line change
Expand Up @@ -339,3 +339,14 @@ end
@test_nowarn @inferred transform(t, x)
@test_nowarn @inferred transform_and_logjac(t, x)
end

@testset "positive definite cholesky factor" begin
t = PosDefCholeskyFactor(4)
d = dimension(t)

v = randn(d)
U = transform(t, v)
@test U isa UpperTriangular
@test size(U) = (4,4)
Copy link
Owner

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You probably want a == here.

@test inverse(t,U) ≈ v
end
Copy link
Owner

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please

  • also test for U being positive (semi) definite (eg just check the diagonal),
  • check the log jacobian using ForwardDiff (you find examples of this in the tests),
  • test that the results are @inferred

Happy to help if necessary.