search

SciML / LinearSolve.jl

LinearSolve.jl: High-Performance Unified Interface for Linear Solvers in Julia. Easily switch between factorization and Krylov methods, add preconditioners, and all in one interface.
https://docs.sciml.ai/LinearSolve/stable/
Other
248 stars 53 forks source link

Expand OperatorAssumptions to allow for choosing conditioning #289

Closed ChrisRackauckas closed 1 year ago

ChrisRackauckas commented 1 year ago

This is the last thing of https://github.com/SciML/LinearSolve.jl/issues/283 to be fixed up. It allows for different levels of conditioning assumptions, in order to allow for making a trade-off between performance and numerical stability. Right now I've handled the very ill-conditioned cases but not the well-conditioned case (which is the case the issue is interested in).

codecov[bot] commented 1 year ago

Codecov Report

Merging #289 (d4f8a8c) into main (6cbe99d) will decrease coverage by 3.07%. The diff coverage is 33.33%.

@@            Coverage Diff             @@
##             main     #289      +/-   ##
==========================================
- Coverage   68.78%   65.71%   -3.07%     
==========================================
  Files          14       14              
  Lines         897      913      +16     
==========================================
- Hits          617      600      -17     
- Misses        280      313      +33
Impacted Files Coverage Δ
src/LinearSolve.jl 82.75% <ø> (-13.80%) :arrow_down:
src/default.jl 42.26% <11.11%> (-11.95%) :arrow_down:
src/common.jl 80.32% <100.00%> (-9.51%) :arrow_down:

... and 2 files with indirect coverage changes

:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more