Skip to content

Event re-solve: Newton with the problem's Jacobian first (0.7.1) - #89

Merged
JKRT merged 1 commit into
masterfrom
event-newton
Oct 18, 2026
Merged

JKRT merged 1 commit into
masterfrom
event-newton

Conversation

@JKRT

@JKRT JKRT commented Oct 18, 2026

Copy link
Copy Markdown
Member

At an event, the algebraic rows of a mass-matrix problem are re-solved by Newton with the problem's own Jacobian (the algebraic block, factorized sparse) before BrownFullBasicInit, which stays the fallback. At a flow reversal BrownFullBasicInit's finite-difference Jacobian straddles a relation without an event: MSL Fluid MeasuringTemperature spent 218 of 243 s in event re-solves; it now simulates in 24 s and validates.

Tests: the reduced flow reversal (BrownFullBasicInit 1015 right-hand side calls, Newton 4); a 3000-row algebraic block re-solved sparse.

Checked locally: tests 552/552; OM.jl tests pass (6 broken); OMLibraryTesting tests and smoke pass. MSL validate 3.2.3 351/425, 4.1.0 442/534 (before: 349, 441); simulate time 0.98x and 0.90x.

Version 0.7.1.

🤖 Generated with Claude Code

At an event, the algebraic rows of a mass-matrix problem are solved by
Newton with the problem's own Jacobian (the algebraic block, factorized
sparse) before BrownFullBasicInit. At a flow reversal BrownFullBasicInit's
finite-difference Jacobian straddles a relation without an event; MSL
Fluid MeasuringTemperature simulated 241 s, now 24 s.

Assisted-by: Claude Opus 5.5
@JKRT
JKRT merged commit c1d0a64 into master Oct 18, 2026
9 checks passed
@JKRT
JKRT deleted the event-newton branch October 23, 2026 18:44
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.

1 participant