Skip to content

Bug fix/low rank compression #36

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

Merged
BaptisteLamic merged 3 commits into from
Jun 12, 2026
Merged

Bug fix/low rank compression #36

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
d8906e1
fix bug
May 20, 2026
391ff9d
Fix test
Jun 12, 2026
315f5f0
Remove dead code
Jun 12, 2026
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
48 changes: 32 additions & 16 deletions src/compression.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -132,21 +132,7 @@ function (Compression::HssCompression)(axis, f; stationary=false)
r = randcompress_adaptive(lm, cc, cc, atol=Compression.atol, rtol=Compression.rtol, kest=Compression.kest, leafsize=Compression.leafsize)
return r
end
function (Compression::HssCompression)(axis, f, g)
f00 = f(axis[1])
T, bs = eltype(f00), size(f00, 1)
u = Array{T}(undef, (bs, bs, length(axis)))
v = Array{T}(undef, (bs, bs, length(axis)))
for k in 1:length(axis)
u[:, :, k] .= f(axis[k])
v[:, :, k] .= g(axis[k])
end
n = length(axis) * bs
u = reshape(u, (n, bs))
v = reshape(u, (n, bs))
cc = bisection_cluster(length(axis) * bs, leafsize=Compression.leafsize)
return lowrank2hss(u, v, cc, cc)
end

function (Compression::HssCompression)(tab::HssMatrix)
return recompress!(tab, atol=Compression.atol, rtol=Compression.rtol, leafsize=Compression.leafsize)
end
Expand Down Expand Up @@ -214,7 +200,7 @@ end

function computeMatrixNorm(matrix::HssMatrix)
norm2 = real(tr(matrix' * matrix))
if norm2 <= 0
if norm2 <= 0
return zero(norm2)
else
return sqrt(norm2)
Expand Down Expand Up @@ -265,4 +251,34 @@ end
@test computeMatrixNorm(A) - computeMatrixNorm(hssA) < tol
@test abs(computeMatrixNorm(A) - computeMatrixNorm(hssA)) / computeMatrixNorm(A) < tol
end
end

@testitem "Low-rank kernel HSS compression correctness" begin
using LinearAlgebra
N, Dt, bs = 8, 1.0, 2
ax = LinRange(0, Dt, N)
for T in [Float64, ComplexF64, ComplexF32]
for causality in (Retarded, Advanced, Acausal)
tol = 100 * max(1E-12, eps(real(T)))
f(x) = T.([x 0; 0 x])
g(x) = T.([x^2 0; 0 x^2])
K = discretize_lowrank_kernel(
TrapzDiscretisation,
causality,
ax,
f,
g;
compression=HssCompression(atol=1e-12, rtol=1e-12)
)
_getMask(::Type{Acausal}) = (i, j) -> T(true)
_getMask(::Type{Retarded}) = (i, j) -> T(i >= j)
_getMask(::Type{Advanced}) = (i, j) -> T(i <= j)
mask = _getMask(causality)
K_ref = zeros(eltype(matrix(K)), N * bs, N * bs)
for i in 1:N, j in 1:N
K_ref[blockrange(i, bs), blockrange(j, bs)] .= f(ax[i]) * g(ax[j]) * mask(i, j)
end
@test norm(matrix(K) - K_ref) < tol
end
end
end
Loading