Skip to content

Extend vdim to support vector-valued kernels#141

Open
maltezfaria wants to merge 37 commits into
mainfrom
vec-valued-vdim
Open

Extend vdim to support vector-valued kernels#141
maltezfaria wants to merge 37 commits into
mainfrom
vec-valued-vdim

Conversation

@maltezfaria

Copy link
Copy Markdown
Member

No description provided.

@codecov

codecov Bot commented Jan 19, 2026

Copy link
Copy Markdown

Codecov Report

❌ Patch coverage is 92.49292% with 53 lines in your changes missing coverage. Please review.
✅ Project coverage is 80.82%. Comparing base (1bb113f) to head (d926754).
⚠️ Report is 10 commits behind head on main.

Files with missing lines Patch % Lines
src/kernels.jl 89.51% 15 Missing ⚠️
ext/IntiFMM2DExt.jl 81.81% 14 Missing ⚠️
src/vdim.jl 97.65% 7 Missing ⚠️
src/nystrom.jl 60.00% 6 Missing ⚠️
src/api.jl 82.75% 5 Missing ⚠️
src/bdim.jl 96.42% 3 Missing ⚠️
src/entities.jl 50.00% 3 Missing ⚠️
Additional details and impacted files
@@            Coverage Diff             @@
##             main     #141      +/-   ##
==========================================
- Coverage   83.41%   80.82%   -2.59%     
==========================================
  Files          25       25              
  Lines        3985     4553     +568     
==========================================
+ Hits         3324     3680     +356     
- Misses        661      873     +212     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

maltezfaria and others added 8 commits May 22, 2026 17:22
What this commit does is execute X over scalar densities
regarding the monomials as an FMM "charge" and request the FMM-based
Hessian of the resulting potential, rather than before where we
applied X to a dipole vector and took a gradient. The latter
gradient-of-dipole computation is still the way that the `online' phase
of the algorithm works, as there we truly do have a vector `g` input.
`HessianKernel` now accepts two modes of dispatch, charge vs. dipole,
and the FMM, if supported for that mode, then executes respectively, a
scalar->SMatrix computation or an SVector->SVector computation (the
latter a dipole contraction of the input vector with the full Hessian).
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants