Skip to content

Make AN5 and JVODE caches resizable after resize! - #4759

Merged
ChrisRackauckas merged 6 commits into
SciML:masterfrom
ChrisRackauckas-Claude:nordsieck-resize-cache
Nov 2, 2026
Merged

ChrisRackauckas merged 6 commits into
SciML:masterfrom
ChrisRackauckas-Claude:nordsieck-resize-cache

Conversation

@ChrisRackauckas-Claude

@ChrisRackauckas-Claude ChrisRackauckas-Claude commented Oct 3, 2026 •

Copy link
Copy Markdown
Member

Summary

Enable in-place AN5, JVODE_Adams and JVODE_BDF integrators to resize their user buffers, Nordsieck history and Tsit5 stage buffers, then restart at order 1 with the caller-filled state. Require OrdinaryDiffEqCore 4.7, whose registered @cache extends SciMLBase.full_cache.

The resize regressions cover growth and shrinkage against a piecewise matrix-exponential reference, both u_modified!(integrator, true) and u_modified!(integrator, false), and NaN-poisoned resizing before the first step. All three algorithms are covered. Master is merged through b94e13a to include the JVODE controller recovery and BDF formula fixes.

Review fixes and local verification

Revision: eb80c56

All Julia runs use timeout 3600, one top-level run at a time, TMPDIR=<job tmpdir>, and JULIA_DEPOT_PATH=<job depot>:$HOME/.julia. <site tool dir> is removed from PATH. Commands below run from the checkout; probe scripts, scratch environments and logs are retained beside it in the job directory.

Registered Core compatibility

Raised the Core floor from 4.4 to 4.7. General's OrdinaryDiffEqCore/Versions.toml records 4.6.0 as tree c4035f06d7749c49347d8a200d3ba0a17085492e and 4.7.0 as c6acc3ee837cae1abc382569d9fb3254a7974e11; the resolved registered packages match those trees. Core 4.6.0's macro generates a caller-local full_cache, while 4.7.0 uses GlobalRef(SciMLBase, :full_cache).

The same Nordsieck source and resize/step probe were run against exact registered Core versions, with [sources] removed from scratch copies. The 4.6.0 copy retains the old Core compat so the resolver can reproduce the failure. Each probe tests AN5, JVODE_Adams and JVODE_BDF:

for version in 4.6.0 4.7.0; do
  timeout 3600 julia +1.12 --project="../scratch/codex-floor-${version}" -e 'using Pkg; Pkg.develop(path=joinpath(dirname(Base.active_project()), "Nordsieck")); include("../scratch/codex-floor-probe.jl")'
done
Julia 1.12.7; Core 4.6.0
has SciMLBase.full_cache: false (all three algorithms)
MethodError: no method matching full_cache(::OrdinaryDiffEqNordsieck.AN5Cache{...})
MethodError: no method matching full_cache(::OrdinaryDiffEqNordsieck.JVODECache{...})
registered Core resize | Error 3  Total 3

Julia 1.12.7; Core 4.7.0
has SciMLBase.full_cache: true (all three algorithms)
registered Core resize | Pass 3  Total 3

Logs: floor-4.6.0.log, floor-4.7.0.log.

Restart-branch regression discrimination

The callback test is parameterized over both modification flags. The exact committed test file was copied into each scratch package. Each scratch package changes only one restart branch in nordsieck_perform_step.jl; all other solver source matches the checkout.

GROUP=Core timeout 3600 julia +1.12 --project=../scratch/codex-mut-an5 -e 'using Pkg; Pkg.test()'
GROUP=Core timeout 3600 julia +1.12 --project=../scratch/codex-mut-jvode -e 'using Pkg; Pkg.test()'

Failing before (AN5 restart branch removed):

Nordsieck state resizing (AN5(), u_modified=false): Test Failed
Expression: sol.u[end] ≈ exact_resize(TF_resize) rtol=1e-7 atol=1e-10
Evaluated: [0.4081351587680127, -0.09753199041303871] ≈ [0.40813522033100175, -0.09753202519196175]
Nordsieck Tests | Pass 76  Fail 1  Total 77

Failing before (JVODE restart branch removed): both Adams and BDF false-flag cases fail successful_retcode, stop before shrink, and their grow interpolation checks throw DimensionMismatch: a has size (2,), b has size (3,).

Nordsieck Tests | Pass 69  Fail 6  Error 2  Total 77

Passing after (both branches intact): the complete Core group passes, including all six algorithm/flag combinations; final output is below. Logs: mutation-an5.log, mutation-jvode.log.

JVODE coverage and tolerance

Added JVODE_BDF to both the growth/shrink and initial-resize cases. Both JVODE wrappers use reltol=1e-8, abstol=1e-10, and a separate analytic-reference threshold of rtol=1e-6. The threshold bounds global solve and interpolation error separately from the local step tolerance. With the merged controller and BDF fixes, the measured relative errors are:

Algorithm After growth After shrink
AN5 1.16e-10 3.51e-10
JVODE_Adams 1.88e-8 1.39e-8
JVODE_BDF 5.02e-8 6.22e-8

Both modification flags produce the same errors. timeout 3600 julia +1.12 --project=lib/OrdinaryDiffEqNordsieck ../scratch/codex-accuracy.jl passes all 27 assertions; output is in accuracy-1.12.log.

Final test groups and static checks

GROUP=Core timeout 3600 julia +1.12 --project=lib/OrdinaryDiffEqNordsieck -e 'using Pkg; Pkg.test()'
GROUP=Core timeout 3600 julia +1.10 --project=../scratch/jl110 -e 'using Pkg; Pkg.test("OrdinaryDiffEqNordsieck")'
GROUP=QA timeout 3600 julia +1.12 --project=lib/OrdinaryDiffEqNordsieck -e 'using Pkg; Pkg.test()'

Julia 1.10.12 uses a separate local environment developing this checkout's package and sibling libraries; Julia 1.12.7 uses the sublibrary environment. Both test the same source.

Julia 1.12.7, GROUP=Core:
Nordsieck Tests   | Pass 77  Total 77
AN5 Startup Tests | Pass  6  Total  6
Testing OrdinaryDiffEqNordsieck tests passed

Julia 1.10.12, GROUP=Core:
Nordsieck Tests   | Pass 77  Total 77
AN5 Startup Tests | Pass  6  Total  6
Testing OrdinaryDiffEqNordsieck tests passed

Julia 1.12.7, GROUP=QA:
Allocation Tests | Pass 2  Broken 1  Total 3
JET Tests        | Pass 1  Total 1
Aqua             | Pass 21 Total 21
Testing OrdinaryDiffEqNordsieck tests passed

The allocation test's broken assertion is present in the starting head; this revision adds no skips or broken assertions. Logs: final-core-1.12.log, final-core-1.10.log, final-qa-1.12.log.

timeout 3600 julia +1.12 --project="$HOME/sandbox/goals/performance/runic-env" -e 'using Runic; exit(Runic.main(ARGS))' -- --check lib/OrdinaryDiffEqNordsieck/src/OrdinaryDiffEqNordsieck.jl lib/OrdinaryDiffEqNordsieck/src/nordsieck_caches.jl lib/OrdinaryDiffEqNordsieck/src/nordsieck_perform_step.jl lib/OrdinaryDiffEqNordsieck/test/nordsieck_tests.jl
typos lib/OrdinaryDiffEqNordsieck/Project.toml lib/OrdinaryDiffEqNordsieck/src/OrdinaryDiffEqNordsieck.jl lib/OrdinaryDiffEqNordsieck/src/nordsieck_caches.jl lib/OrdinaryDiffEqNordsieck/src/nordsieck_perform_step.jl lib/OrdinaryDiffEqNordsieck/test/nordsieck_tests.jl
git diff --check

All three exit 0 without findings. Logs: final-runic.log, final-typos.log.

Not verified

Out-of-place resize, GPU/non-Vector state, downstream packages, and long post-resize AN5 trajectories beyond the regression horizon t=0.8 were not tested. A docs build was not run; this revision changes a compatibility bound and tests, with no new public names or documentation changes. CI and independent review of this revision are pending. Historical reviews below apply to their stated heads and are preserved verbatim.

Please ignore until reviewed by @ChrisRackauckas.

Original implementation: Cursor Agent 2026.10.01-14929f9 (model: unknown, Cursor Auto), orchestrated by Claude Code (claude-opus-5-5[1m]), https://claude.ai/code/session_01LPHREnnonfLg1VcE1EJovv.

Risk assessment

  • Risk: low
  • Blast radius: Only AN5 and JVODE_Adams/JVODE_BDF users in OrdinaryDiffEqNordsieck. The new code runs only after resize!(integrator, n), which used to throw MethodError for these solvers. The rest of the diff is the resize_restart flag (false unless a resize happened, so non-resize steps take the same path as before), @cache on AN5Cache/JVODECache, a renamed type parameter (dType → uNoUnitsType) and one new struct field. Those are internal cache types, not public API, so SemVer-wise this is a new capability (minor/patch), not a breaking change. No other sublibrary is touched.
  • Evidence: Head 139d471a: 127 checks pass, 4 fail, 1 skipped (matrix placeholder). All 4 failures are pre-existing. Each also fails on master HEAD 4a5c0e71 (4a5c0e7). For the first three I compared the failure signatures; for the fourth I compared the error on head and the job status on master:
  • Independent review: Devin CLI / fusion-claude-opus-5-5-high-sidekick-swe-2-medium rated it low, medium-high confidence: the change only adds a resize path that used to error, CI shows no new failures, and the imported internals (@cache, ODEIntegrator, reinit_controller!) are public in OrdinaryDiffEqCore on master. SciMLBase resize_non_user_cache! is in SciMLBase's public list at v3.39.0. No tests were weakened; the PR only adds tests.
  • Merge: needs human review. Findings:
    • The PR is a draft and says "Please ignore until reviewed by @ChrisRackauckas". There are no GitHub reviews or comments.
    • Accuracy concern, disclosed in the PR: after a resize, long AN5 runs can drift with EEst≈0.97 ("~t>0.9 after a 2→3 grow"). The test window is cut to tspan=(0,0.8) to avoid this, so the tests do not cover the known weakness and the fix may be only partly correct.
    • Not mentioned in the PR:
      • The compat floor stays at OrdinaryDiffEqCore = "4.4". At OrdinaryDiffEqCore-v4.4.0, @cache, ODEIntegrator and reinit_controller! exist but are not in Core's public list. Against the floor they are non-public imports, and the ExplicitImports QA check only passes because it runs against the monorepo's own Core.
      • The OrdinaryDiffEqNordsieck version is not bumped (2.3.1).
      • Δ, atmp, fsalfirst and ratetmp are resized twice: once by the @cache full_cache and again manually. This is harmless but redundant.
      • A test comment uses a bare (#4722) reference.
    • Also disclosed: JVODE is tested only at reltol=1e-4 because of a recursion bug in the separate PR 4758; there is a merge overlap with 4758 in nordsieck_perform_step.jl and nordsieck_tests.jl; and out-of-place and GPU resize are not tested.

🤖 Risk assessment posted by an AI agent (fleet master) — harness: Devin CLI / fusion-claude-opus-5-5-high-sidekick-swe-2-medium; dispatched by Devin CLI 3000.11.3 fleet-master caretaker, model Fusion (claude-opus-5-5 medium + swe-2 medium)
Conversation: local transcript (no session URL)

🤖 Automated comment from an AI agent running as @ChrisRackauckas — not written or reviewed by Chris.

Risk assessment (independent review, Codex gpt-6.1-sol, head 1be71a8)

  • Verdict: MERGE
  • Risk: medium-low (ML2: changes numerical restart and initialization for the peripheral AN5/JVODE algorithms, with independent analytic-reference tests failing before and passing after; L2 also applies to the added tests).
  • Blast radius: In-place AN5 and JVODE caches, including both JVODE wrappers, when integrator state dimensions change. The cache macro also enables generic state-buffer operations such as deletion and insertion. No shared Core implementation, default solver, exported name, dependency, tolerance, or existing assertion changes.
  • Evidence: Read pr.txt, the complete risk rules, and all of git diff origin/HEAD...HEAD; git diff --check origin/HEAD...HEAD passed. All Julia commands used Julia 1.13.2, JULIA_NUM_THREADS=4, JULIA_NUM_PRECOMPILE_TASKS=4, the requested run-local depot followed by $HOME/.julia, offline package resolution, and timeout 3600. Commands below use paths relative to the review run directory, with --compiled-modules=existing:
    • GROUP=Core julia --project=scratch/head/review-env --compiled-modules=existing scratch/head/lib/OrdinaryDiffEqNordsieck/test/runtests.jl: Nordsieck Tests 34/34 and AN5 Startup Tests 6/6 passed. This includes every new regression assertion.
    • julia --project=scratch/base/review-env --compiled-modules=existing scratch/regression.jl: the unchanged new regression failed on origin/HEAD (0139bdc23709266d93ebadf377d63484886d1478) with MethodError: no method matching full_cache(::AN5Cache{...}).
    • julia --project=scratch/unfixed/review-env --compiled-modules=existing scratch/diagnostic.jl scratch/regression-all.jl: using Nordsieck source from the merge base (59ba40e4b5bbae713d8df7b792f6ea62f033ff78), all three new testsets errored with missing full_cache methods, covering AN5, JVODE_Adams, and initial resizing. The scratch wrapper groups the original testsets so failures do not stop after AN5.
    • julia --project=scratch/head/review-env --compiled-modules=existing scratch/diagnostic.jl scratch/probes-refined.jl: 24/24 additional assertions passed across AN5, JVODE_Adams, and JVODE_BDF. Checks cover nonautonomous analytic accuracy after manual growth, deletion of a middle component, continuous-callback growth, and poisoned initial resizing without calling u_modified!.
    • Initial exploratory probes at fixed dt=0.001 failed eight JVODE accuracy assertions. No-resize controls reproduced approximately 1.05e-4 relative error on head and default; reducing dt decreased that error proportionally. All nine no-resize diagnostic rows matched head versus merge-base source. The refined probes reduce dt to 1e-5 (continuous callbacks: 3e-5) and retain the original assertion tolerances; no repository tests were changed.
    • Runtime Base.ispublic checks passed for all newly imported dependency names: @cache, ODEIntegrator, reinit_controller!, and resize_non_user_cache!. No new public names require added docstrings/docs entries, and no dependency or breaking API change was found.
    • G1 pass: coupled matrix-exponential references and nonautonomous analytic solutions independently check numerical accuracy. G2 not applicable: no GPU kernel, AD rule, Reactant tracing, or MPI implementation changes. G3 pass: grow/shrink reference tests and NaN-poisoned initial-resize tests exercise stale/uninitialized cache risks.
  • Merge: merge; no blocking findings on the reviewed head. The medium-low rating still requires the fleet's second independent review before automatic merging.

🤖 Posted by an AI agent — harness: Codex CLI 0.162.1 · model: gpt-6.1-sol
Conversation: local headless review run (no public session URL); transcript kept by the fleet master.

🤖 Automated comment from an AI agent running as @ChrisRackauckas — not written or reviewed by Chris.

Risk assessment (independent review, Devin CLI 3000.11.92 / fusion-claude-opus-5-5-high-sidekick-swe-2-medium, head 1be71a8)

  • Verdict: CHANGES
  • Risk: medium-low (ML2: adds restart/initialization numerics for the non-default AN5 and JVODE (Adams/BDF) solvers after a state-size change. It has analytic-reference evidence, and default and already-working results are bit-identical. L6 also matches: the old behavior was MethodError.)
  • Blast radius: Only in-place AN5, JVODE_Adams and JVODE_BDF integrators in OrdinaryDiffEqNordsieck, and only after resize!/deleteat!/addat!. Those calls used to throw MethodError. These are peripheral solvers, not an OrdinaryDiffEq main lane. Non-resize code takes the same instructions: resize_restart stays false. I confirmed 24/24 bitwise-identical solutions against master, including same-size callbacks and out-of-place solves. No dependency, export, public name or tolerance changes. AN5Cache/JVODECache gain a field and a renamed type parameter; both are internal types.
  • Evidence: Julia 1.13.2, JULIA_NUM_THREADS=4, run-local depot, one process at a time, all in scratch copies (checkout untouched).
    • Fail-before: master 98fa0512e source plus the PR's test file, GROUP=Core Pkg.test(). Result: 50 pass, 3 errored, all MethodError: no method matching full_cache(::AN5Cache…) / (::JVODECache…).
    • Pass-after: head merged onto current master 98fa0512e. git merge-tree is clean; I built the tree with git merge-file, because master has since landed the JVODE_BDF rewrite (Use BDF formulas in JVODE_BDF #4786). GROUP=Core Pkg.test(): Nordsieck Tests 60/60, AN5 Startup Tests 6/6.
    • Bitwise check (scratch/bitwise.jl) on master and merged: AN5, JVODE_Adams and JVODE_BDF × {in-place linear, out-of-place linear, Lorenz, same-size u .*= 1.1 callback} × reltol {1e-4, 1e-8}. All 24 hashes of (t, u) and every nf are identical.
    • Probes against analytic exp(A t) (scratch/probes.jl, merged tree):
      • JVODE_Adams grow/shrink: 6.3e-9 / 2.4e-9 at rt=1e-8.
      • JVODE_BDF grow/shrink: 4.6e-8 / 5.0e-8 at rt=1e-8.
      • Resize between step! calls, deleteat!, and a NaN-poisoned resize before the first step without u_modified! are all accurate or finite on all three solvers.
    • Compat floor (scratch/floor_probe.jl): the PR's Nordsieck source with [sources] stripped resolves against registered OrdinaryDiffEqCore 4.6.0, OrdinaryDiffEqTsit5 2.1.0, SciMLBase 3.39.1 and DiffEqBase 7.13.1. Result: OrdinaryDiffEqNordsieck.full_cache === SciMLBase.full_cache is false (a local function with 2 methods), and resize! throws MethodError for both AN5 and JVODE.
    • Mutation (scratch/mut): with head's nordsieck_perform_step.jl replaced by master's, all 10 new assertions still pass. Only resize plus u_modified!(integrator, false) exposes the change. Without the hunks, JVODE aborts with dt_epsilon/DimensionMismatch and AN5's error goes from 6.7e-9 to 1.5e-5.
    • Gates:
      • G1: pass. Tests and my probes compare against matrix-exponential references the code did not produce. The AN5 post-shrink drift is a pre-existing AN5 bug: a fresh solve with no resize reproduces it (finding 4).
      • G2: not applicable. No GPU, AD, Reactant or MPI code changes.
      • G3: fail (partial). The grow/shrink tests catch the main stale-cache risk, but nothing tests the new resize_restart cache flag (finding 2).
  • Independent review: Devin CLI 3000.11.92 / fusion-claude-opus-5-5-high-sidekick-swe-2-medium rated it medium-low (ML2) at head 1be71a8, high confidence. It is a peripheral-solver fix with analytic evidence and bit-identical existing paths, but the compat floor does not support the @cache it relies on, and the restart-flag code is untested.
  • Merge: needs changes (raise the OrdinaryDiffEqCore compat floor to 4.7; add a test that exercises resize_restart (resize + u_modified!(integrator, false)); also still draft, and the PR body says "Please ignore until reviewed by @ChrisRackauckas")

🤖 Posted by an AI agent — harness: Devin CLI 3000.11.92 · model: fusion-claude-opus-5-5-high-sidekick-swe-2-medium
Conversation: local headless Devin review run (no public session URL); transcript kept by the fleet master.

🤖 Automated comment from an AI agent running as @ChrisRackauckas — not written or reviewed by Chris.

Risk assessment (independent review, Claude Opus 5.5 (claude-opus-5-5), head 1be71a8)

  • Verdict: MERGE
  • Risk: medium-low (ML2: changes the restart/initialization logic of the non-default AN5/JVODE algorithms, with independent evidence that fails before and passes after. Default-algorithm and no-resize results are unchanged. L6 and ML8 also match, and neither is higher.)
  • Blast radius: Only in-place AN5, JVODE_Adams and JVODE_BDF in OrdinaryDiffEqNordsieck. These are peripheral, non-main-lane solvers. The new branches run only after resize!/deleteat!/addat!, which used to throw MethodError. resize_restart is false otherwise, and I showed the resulting trajectories are bit-identical to base. The internal cache structs gain a field and rename a type parameter; neither struct is public. No exports, defaults, dependencies or tolerances change, and no other sublibrary is touched.
  • Evidence: All runs used Julia with JULIA_DEPOT_PATH=<run>/depot:<local path> and timeout ≥ 2400 s, on trees extracted with git archive into scratch/.
    • Head Core group: GROUP=Core Pkg.test() in lib/OrdinaryDiffEqNordsieck passed: Nordsieck Tests 34/34, AN5 Startup 6/6.
    • Base (merge-base 59ba40e) with the PR's test file: the 3 new testsets error with MethodError: no method matching full_cache(::AN5Cache{...}), 24 pass. Fails before, passes after.
    • QA on head: GROUP=QA passed. AllocCheck: 2 pass plus 1 pre-existing broken. JET 1/1, Aqua 21/21.
    • Bit-identity, base vs head: AN5, JVODE_Adams and JVODE_BDF on Lorenz and a forced linear system, each with no callback and with a same-size u-modifying DiscreteCallback, at reltol 1e-7 and 1e-4. The hash((sol.t, sol.u)), step counts and nf all match exactly in every case.
    • Independent accuracy, head only: a nonlinear, non-autonomous system grows 2→4 at t=0.3, then deleteat! at t=0.6, out to t=3. It is checked against a Vern9 solve at reltol 1e-14 with the same callback.
      • AN5 at reltol 1e-8: relative error ≤ 9.5e-7 at all checkpoints. Manual resize!/deleteat! between step! calls: 4.1e-8 at t=3.
      • JVODE_Adams at reltol 1e-3: relative error 2e-3 to 0.35, about 4× a fresh 3-d solve right after the deleteat, within tolerance scaling elsewhere.
      • Initial resize with uprev[3]=NaN and 1000 fixed steps against exp(A t): AN5 relerr 4.7e-15, JVODE_Adams/BDF 1.7e-5.
    • G1 pass: checked against matrix-exponential and Vern9 references, not the code's own output.
    • G2 not applicable: no GPU, AD, Reactant or MPI code changes.
    • G3 pass: the grow/shrink tests and the NaN-poisoned initial-resize test would catch stale or uninitialized cache state. Finding 1 notes that they don't discriminate the PR's own copyto!, because Core already syncs uprev.
  • Merge: merge. No blocking findings. Findings 1–3 are optional cleanups. Medium-low still needs the second independent review that the merge rule requires.

🤖 Posted by an AI agent — harness: Claude Code · model: claude-opus-5-5
Conversation: local headless review run (no public session URL); transcript kept by the fleet master.

Risk assessment (review-fix revision)

  • Risk: medium, conservatively pending independent review of this revision.
  • Blast radius: In-place AN5 and JVODE_Adams/JVODE_BDF integrators after a state-size change. The review fix changes only the Core compatibility bound and tests; the merged master supplies the JVODE controller and BDF fixes. No new public names or dependencies are introduced.
  • Evidence: Exact registered Core 4.6.0 fails all three resize probes; 4.7.0 passes 3/3. Removing the AN5/JVODE restart branches makes the new false-flag tests fail. Intact source passes Core on Julia 1.10 and 1.12, QA, Runic and typos; numerical error measurements and complete output counts are above.
  • Independent review: Historical reviews under their original headings apply to head 1be71a8; the Devin CLI / fusion-claude-opus-5-5-high-sidekick-swe-2-medium review requested the three changes addressed here. Independent review of the new head is pending. Review history: Make AN5 and JVODE caches resizable after resize! #4759.
  • Merge: needs human review; the PR remains a draft and requires review by @ChrisRackauckas. CI and independent review of this revision are pending.

🤖 Updated by an AI agent using Codex CLI 0.163.0 (model: gpt-6.1-sol). Session: local session, transcript <local path> on a CI worker host.

🤖 Automated review section from an AI agent running as Chris Rackauckas's account — not written or reviewed by Chris.

Independent current-head review of eb80c56

CHANGES for required original/continuation session attribution. Strong numerical independent evidence: Nordsieck77/77/startup6, base with newtests50pass7errors; removing AN5restart76pass1fail and JVODErestart69pass6fail2errors. QA unchanged with polling resolves EMFILE; Core4.7floor6/6, independent matrix exponential controls. Separate clean default AN5 adaptive/restart defect gives5pass1fail/error0.126 despite tight tolerances, plus no-callback control; queue separate bisection, not a new resize regression. Do not claim original unavailable author scripts or full monorepo coverage.

Risk assessment

  • Risk: medium-low (ML2, ML4, ML8).
  • Blast radius: source paths and metadata limitations stated above; higher recorded assessments retained.
  • Evidence: independent Git/object/runtime/log checks and actual local executions explicitly distinguished above.
  • Independent review: Codex CLI0.163.0, gpt-6.1-sol, local session 01a15730-32c7-7103-87f1-855ba8be6fae (no session URL), rated it medium-low (ML2, ML4, ML8), VERDICT: CHANGES at head eb80c56.
  • Merge: held for findings and live gates; medium and above require Chris.

Links:


🤖 Posted by an AI agent — harness: Codex CLI 0.163.0 · model: gpt-6.1-sol
Conversation: local session 01a154f0-8986-78e1-8af2-1119e8cdea4a (no session URL)

🤖 Automated review section from an AI agent running as Chris Rackauckas's account — not written or reviewed by Chris.

Assigned correction of eb80c56

This correction is queued to a fresh Codex CLI worker, model gpt-6-luna, under the SciML submaster’s pipeline ownership. Scope: Correct original/continuation session attribution and footer from verified runtime records only; preserve source and human history, stop if signed ancestry or missing original identity prevents a safe correction. Existing ML2 numerical assessment and separate AN5 baseline investigation remain.

This assignment permits metadata work only, with no Julia/test execution or code changes. Historical attribution corrections require verified original identities, complete source/author/date/message preservation proofs and a guarded expected-head update. Missing records must be reported honestly; no identity or review is fabricated.

The worker does not merge, approve or close PRs. Any pushed revision needs a fresh independent counted review; existing source reviews are not silently repinned. This assignment does not change the PR’s risk level or resolve any finding.

Links:


🤖 Posted by an AI agent — harness: Codex CLI 0.163.0 · model: gpt-6.1-sol
Conversation: local session 01a154f0-8986-78e1-8af2-1119e8cdea4a (no session URL)

ChrisRackauckas and others added 3 commits October 3, 2026 04:31
…order 1.

AN5Cache and JVODECache had no full_cache, so resize!(integrator, n) threw MethodError. Nested z/Tsit5 arrays also need resizing, and the history is invalid after a size change.

Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
Co-Authored-By: Cursor Agent <noreply@cursor.com>
Agent-Harness: Cursor Agent 2026.10.01-14929f9
Agent-Model: unknown (Cursor auto)
Agent-Session: local session, transcript /home/crackauc/sandbox/goals/performance/jobs/pr-nordsieck-resize/log.txt on amdci2
Setting reeval_fsal in the resize hook skipped Core's discontinuity
sync at iter==0, so startup consumed uninitialized history. Restrict
the extra FSAL eval to a resize restart so same-size callbacks are
unchanged.

Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
Co-Authored-By: Cursor Agent <noreply@cursor.com>
Agent-Harness: Cursor Agent 2026.10.01-14929f9
Agent-Model: unknown (Cursor auto)
Agent-Session: local session, transcript /home/crackauc/sandbox/goals/performance/jobs/pr-nordsieck-resize/log.txt on amdci2
Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
Co-Authored-By: Devin <noreply@cognition.ai>
Agent-Harness: Devin CLI 3000.11.3 (conflict resolution), Claude Code 2.1.285 (orchestration)
Agent-Model: fusion-claude-opus-5-5-high-sidekick-swe-2-medium
Agent-Session: local Claude session 3c311569-dda6-4687-bc9f-65febb212c33 (fleet master); Devin log /Users/chrisrackauckas/sandbox/master/fusion-fixes/SciML_OrdinaryDiffEq.jl-4759/log.txt
@ChrisRackauckas

Copy link
Copy Markdown
Member

🤖 Automated comment from an AI agent running as @ChrisRackauckas — not written or reviewed by Chris.

Merged the default branch into this PR (clean merge, no conflicts; head 9cac8fa) so CI runs on current master. Its earlier failures in SciMLSensitivity Core1 and the downgrade jobs were fixed on master after this PR's last CI run.


🤖 Posted by an AI agent — harness: Devin CLI 3000.11.3 (merge), Claude Code 2.1.285 (posting) · model: fusion-claude-opus-5-5-high-sidekick-swe-2-medium
Conversation: local Claude session 3c311569-dda6-4687-bc9f-65febb212c33 (fleet master)

ChrisRackauckas and others added 3 commits October 16, 2026 07:07
Include the JVODE controller recovery and BDF formula fixes needed by the resize regression tests.

Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
Co-Authored-By: Codex <noreply@openai.com>
Agent-Harness: Codex CLI 0.163.0
Agent-Model: gpt-6.1-sol
Agent-Session: local session, transcript /home/crackauc/sandbox/goals/performance/jobs/pr-nordsieck-resize/log-codex.txt on amdci2
Cover both modification flags and JVODE_BDF in the analytic resize regression, and tighten JVODE solve and reference tolerances.

Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
Co-Authored-By: Codex <noreply@openai.com>
Agent-Harness: Codex CLI 0.163.0
Agent-Model: gpt-6.1-sol
Agent-Session: local session, transcript /home/crackauc/sandbox/goals/performance/jobs/pr-nordsieck-resize/log-codex.txt on amdci2
@tently-review

tently-review Bot commented Oct 18, 2026

Copy link
Copy Markdown

🛡️ Tently review

✅ No design-intent regressions or decision drift found.

Posted by Tently

@ChrisRackauckas-Claude

Copy link
Copy Markdown
Member Author

🤖 Automated comment from an AI agent running as @ChrisRackauckas — not written or reviewed by Chris.

Independent review (Codex CLI 0.166.0 / gpt-6.1-sol), session 01a19834-a9a7-7d80-9de9-55255a433487, reviewed head eb80c56

  • Verdict: MERGE
  • Risk: medium-low (ML2: non-default solver restart numerics have independent analytic evidence and failing-before/passing-after results; ML8 and ML4 also apply).
  • Round / slot: 2 / 1
  • Blocking priorities: none

Review written by the reviewer named above in a session other than the PR author's, and copied here by the posting agent without change to its findings. The posting agent did not re-run the reviewer's commands.

Summary: Adds resizing and history restart support for in-place AN5 and both JVODE variants, with a Core compatibility floor and analytic regression tests. The added regressions fail on the merge-base and pass on the exact head; no blocking defect was found.

Blast radius: In-place AN5, JVODE_Adams and JVODE_BDF state resizing, including their history and nested Tsit5 buffers. No shared Core implementation or default solver changes.

Evidence: Head targeted tests pass 83/83; the identical regression file on merge-base b94e13a gives 50 passes and seven errors from missing full_cache methods. Nordsieck CI Core, QA and downgrade checks pass. The sole failing PR check matches an exact master Enzyme failure. The additional adversarial probe failed during reviewer harness setup, before executing assertions; its coverage is not claimed.

Blocking findings

None.

Non-blocking findings

None.

What I ran

Read the complete rules and diff before author context; recorded and hashed INITIAL.md. Inspected the generic resize/delete interfaces, cache macro, controllers, restart code and tests. Git commands exited 0: git -C src fetch origin '+refs/pull/4759/head:refs/remotes/pr/4759' master, git -C src rev-parse HEAD, git -C src merge-base origin/master HEAD, git -C src diff origin/master...HEAD, and git -C src diff --check origin/master...HEAD. The reviewed head equals the target. The diff merge-base and PR baseRefOid are b94e13a; fetched origin/master is d72c550.

Julia 1.13.2, one process at a time, with JULIA_NUM_THREADS=4, OPENBLAS_NUM_THREADS=1, JULIA_NUM_PRECOMPILE_TASKS=4, JULIA_PKG_PRECOMPILE_AUTO=0, an isolated tmp directory, and the assigned shared depot followed by default depots. Absolute environment paths are omitted. Scratch packages retain identical Julia source; only their Project.toml sources tables are removed. The head environment develops local Nordsieck, Core, Tsit5, DiffEqBase and DiffEqDevTools, and adds Test, LinearAlgebra and ODEProblemLibrary using offline resolution. The base environment reuses exactly that manifest, changing only the Nordsieck source path to the merge-base snapshot.

  • timeout 3600 julia +1.13.2 --startup-file=no --compiled-modules=existing run-head.jl — exit 0. Includes the entire head nordsieck_tests.jl and an5_startup_tests.jl under one outer testset: Head targeted files | 83 / 83. The changed file contributes 77 assertions and startup contributes six.
  • timeout 3600 julia +1.13.2 --startup-file=no --compiled-modules=existing --project=env-base run-base.jl — exit 1. Includes the exact head nordsieck_tests.jl against merge-base Nordsieck source: Base with head regression tests | Pass 50 Error 7 Total 57. All six algorithm/modification-flag resize cases and the initial-resize testset error with MethodError: no method matching full_cache(::AN5Cache...) or full_cache(::JVODECache...). The initial-resize loop stops at AN5 on base, so the two initial JVODE cases are not separately reached there. All added assertions pass on head.
  • timeout 3600 julia +1.13.2 --startup-file=no --compiled-modules=existing --project=env-head probe.jl — exit 1 before any probe assertions. The reviewer harness attempted to load base source into a separate module with include_string; its nested include("algorithms.jl") resolved against the working directory and raised SystemError: opening file algorithms.jl: No such file or directory. This is a harness failure, not a solver test failure. The three-run limit was reached and no retry was made.

Read live checks with gh pr checks 4759 --repo SciML/OrdinaryDiffEq.jl, failed logs with gh run view 39111088286 --repo SciML/OrdinaryDiffEq.jl --log-failed, and five recent completed master IntegrationTest runs with gh run list. Compared failed logs from master run 41319752001. The default cache directory was read-only; redirecting XDG_CACHE_HOME to review scratch allowed retrieval. Both PR and master fail at test/Core4/ensembles.jl:19 in SciMLSensitivity, with AssertionError: mi2 !== nothing in Enzyme src/rules/parallelrules.jl:233, and identical 15-pass/1-error/2-broken summaries. This is verified baseline failure, not an assumption based on job status.

Inspected CI URLs:

G1 is satisfied by the added piecewise matrix-exponential accuracy reference. G3 is satisfied by growth followed by shrinkage under both modification flags, plus NaN-poisoned initialization. G2 has no changed GPU kernel, AD rule, Reactant or MPI implementation to exercise.

What I did not verify

The planned extra analytic nonautonomous manual-resize, middle-deletion, continuous-callback and bitwise base/head controls did not execute because the third-run harness failed. Runtime public-name assertions in that probe likewise did not execute. No whole-package or downstream suites were run locally. GPU/non-Vector resizing, out-of-place resizing, AD through a resize, Reactant and MPI were not exercised locally; these implementations are not changed by this PR. Exact registered Core 4.7 resolution, Julia 1.10, QA and documentation builds were not independently rerun. Long post-resize AN5 trajectories beyond the regression interval were not tested. This is a code merge recommendation; the PR remains a draft.

The PR description returned embedded historical reviews and current-head review summaries after the blind assessment was hashed. The separate prior review file remained unread until the complete review was frozen; this limits the later phase's independence from review text embedded in author context.

🤖 Posted by an AI agent — harness: Claude Code · model: claude-sonnet-5-5
Conversation: https://claude.ai/code/session_01379heiiB35TtywhR7uSU74

@ChrisRackauckas-Claude
ChrisRackauckas-Claude marked this pull request as ready for review November 2, 2026 02:04
@ChrisRackauckas
ChrisRackauckas merged commit 9d1d788 into SciML:master Nov 2, 2026
131 of 132 checks passed
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