Skip to content

nlstep tests: unbreak NSA(TrustRegion) order assertions that now pass - #3897

Closed
singhharsh1708 wants to merge 1 commit into
SciML:masterfrom
singhharsh1708:nlstep-unbreak-stale-test-broken
Closed

nlstep tests: unbreak NSA(TrustRegion) order assertions that now pass#3897
singhharsh1708 wants to merge 1 commit into
SciML:masterfrom
singhharsh1708:nlstep-unbreak-stale-test-broken

Conversation

@singhharsh1708

Copy link
Copy Markdown
Contributor

Four @test_broken order assertions in modelingtoolkit/nlstep_tests.jl now pass, so the MTK lane reports them as Unexpected Pass and errors out:

Unexpected Pass
Expression: abs(sim.𝒪est[:l∞] - 2) < 0.2
Got correct result, please change to @test if no longer broken.

All four are NonlinearSolveAlg(TrustRegion()) convergence-order checks — TRBDF2 and QNDF2, in both the non-autonomous and autonomous Robertson blocks. They were the failures behind #3807 and #3818, and the recent NonlinearSolve releases (SciML/NonlinearSolve.jl#1010 recomputing the trust-region radius on reinit!, plus #1008) fixed them upstream.

This flips those four to @test. The two KenCarp4 order-4 assertions still fail and stay @test_broken.

Verified on a clean checkout of master (8ed2f7a): before, 19 passed, 0 failed, 4 errored, 2 broken; after, 23 passed, 2 broken, 0 errored. The same 4 errors show up on the MTK lane of every open PR that touches this sublibrary, which is what makes that lane red.

AI Disclosure

Claude assisted with this work.

@ChrisRackauckas

Copy link
Copy Markdown
Member

I already got these running

@singhharsh1708
singhharsh1708 deleted the nlstep-unbreak-stale-test-broken branch July 13, 2026 11:37
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