0.2.7 Progress Tracker

Current release reconciliation

Stable v0.2.7 is Owner-complete at 100/100. WP22 is closed as PASS_WITH_LIMITATIONS through R0 and Step 1. The earlier 46/50, 96/100, WP22 PLANNED and 0.2.7a0 entries retained in this chronological tracker are historical checkpoints, not the active release state.

This tracker records actual progress, not intent. WP01 through WP06 and T1-R record completed foundation controls; WP07 records the Terra-authorized elemental WEDGE6 kernel and WP08 records its bounded static vertical slice. Its public maturity remains EXPERIMENTAL; the slice does not imply modal, robustness or external-correlation qualification. WP09 records a bounded robustness campaign; its external outcomes are bounded and do not promote WEDGE6. WP10 records an independent consistent- mass modal evidence lot with bounded Code_Aster frequency and mode-shape correlation. WP10-FINAL qualifies the first three modes within its declared bounded scope; modes four to six remain diagnostic for refinement. WP12 records bounded large-scale readiness evidence for the existing structured TET4 route; its Owner decision remains pending. WP11 records a bounded maturity extension for existing small-strain J2 across TET4, TET10, HEX8 and HEX20; the existing qualification is kept and no universal increment-independence claim is added. WP20 closes the corresponding Owner review with the same bounded scope and no promotion. WP14 records the frozen large-scale execution contract for the Level-Up namespace. It is a governance/contract PASS only; it does not claim a 1M or 3M solve. WP17 closes the pinned PETSc/MPI route with limitations after consuming the controlled WP16 retry and WP18 Silver evidence. WP18 records a PASS_WITH_LIMITATIONS 3M ladder: Bronze preflight and Silver full solve passed on the declared PETSc route; Gold remains unattempted.

F1-F6 and R0 are closed with their recorded PASS or PASS_WITH_LIMITATIONS statuses. The full suite retains three visible failures in experimental or stale nonlinear paths; none was hidden or reclassified as PASS. Step 2B is the current reconciliation/CI checkpoint; the stable package and active qualification state are already recorded as 0.2.7 and 100/100.

WP	Status	Current test level	Start SHA	Evidence head	Owner decision	Blocker
WP01	PASS	T1 targeted	e99289aca40011ca0424944099e2d2093cf21a65	bb822839248b5ffb9faef5d79a6c83f288faefb3	release-truth foundation	-
WP02	PASS	T1 targeted	3058dbcf53967dc50f70814a71b4094d61023dda	pending commit	registry v2 contract	-
WP03	PASS	T1 targeted	ba6111a257ae567e496adcbcdc74de392dd66b6e	pending commit	descriptor and fail-closed preflight	-
WP04	PASS	T1 targeted	684c39c72191d43c53e1f21043dc746d213a561d	pending commit	declarative V&V harness v2	-
WP05	PASS	T1 targeted	fb102e649235a276096b3a37e19eb61e19a5b43f	PENDING_WP05_COMMIT	external oracle preflight bounded PASS; no WEDGE6 correlation	external tools local-only; QF WEDGE6 not implemented
WP06	PASS	T1 targeted	884637a60bc752c1d02644fe4d14ae056a2876b8	c3989df875bcb385bb8e3b144380526db8151d55	common diagnostic contract; no universal threshold	-
T1-R	PASS	T1 targeted	32e4e40bf18f0fdcd0a4ae9959d4f0df2b76892e	32e4e40bf18f0fdcd0a4ae9959d4f0df2b76892e	pre-WP07 formulation, mapping, face, quality and V&V contracts	Terra/Owner re-review required; kernel not authorized
T1-R3	PASS	T1 targeted	d4abc2f15b0e5167cd2faa4734e6a836bdb12514	pending commit	exact Jacobian certificate and strict external contract	Terra re-review required; kernel not authorized
WP07	PASS	T2 targeted	69b7d01beb81263fc2b87cfacb83985db10e3a82	e2e0de5a8df465d5f2254a954d1b2f5c97181cf0	technical kernel and elemental V&V PASS; public maturity EXPERIMENTAL	WP08 workflow, imports, face loads, reactions and external correlation deferred
WP08	PASS	T2 targeted	d4d2942a5fc31ffb97ef373a4466c46be34de991	8040909d6d65f740e1daf858ce572d250a87b39a	static workflow PASS; public maturity EXPERIMENTAL	WP09 robustness/external and WP10 modal evidence deferred
WP09	PASS_WITH_LIMITATIONS	T1 targeted + T2 external	2a27291bcc72e5819014fa172e3d056e80a87d43	4b2fcdc9ed51821b05b52851912be3ebbe764b14	WP09-FINAL: 12-case Code_Aster PENTA6 bounded external PASS; public maturity EXPERIMENTAL	CalculiX formulation mismatch; tolerance approval remains Owner review; no public WEDGE6 promotion
WP10	PASS_WITH_LIMITATIONS	T1/T2 targeted + Code_Aster	4e005423ba4fd87c6ab6ea2fe5c7a345c21d8e43	9d79dc8b306e6cc65f2f4ae2e77e00f676182b84	OWNER_APPROVED_BOUNDED_FIRST_THREE_MODES; modal maturity QUALIFIED_BOUNDED within declared scope	modes four to six remain diagnostic for refinement; no lumped mass or transfer to other dynamics
WP11	PASS_WITH_LIMITATIONS	T2 targeted	4d0ee14f4aa61b9337874a991263a93b4f9a8c73	94461602dfd1782be57c20e1801a0d5d8e262ef1	keep qualified bounded J2 scope; Owner-approved by WP20	no universal increment threshold; tangent symmetry diagnostic only; finite-kinematic J2 remains experimental
WP12	PASS_WITH_LIMITATIONS	T1 targeted + bounded scaling	4971ac4f6c1e5cff2ca48e40ca6db5e8147d0d0a	4971ac4f6c1e5cff2ca48e40ca6db5e8147d0d0a	proposed Owner review	300k assembly-only; 1M time-limited; SciPy/PETSc backend limits
WP13	PASS	T1 targeted	fcdde28a146a3a502972fdad30821f8e8a857da7	qualification/0_2_7/golden/evidence.json	golden numerical baseline and release truth	8 PASS + 1 EXPECTED_FAILURE_PASS; no maturity promotion
WP14	NOT_STARTED	T0 not run	-	-	-	-

Level-Up 1 historical extension

The active theme is Reproducible Large-Model Solving and Numerical Trust. 027-LEVEL-UP is CLOSED / ACCEPT_WITH_CONSOLIDATION as a scope decision; WP15 is PASS_WITH_LIMITATIONS on controlled subscale evidence, WP16 is PASS after the official PETSc retry, and WP17 is PASS_WITH_LIMITATIONS after the PETSc/MPI closure, and WP18 is PASS_WITH_LIMITATIONS after the Bronze/Silver ladder. WP19 is now PASS_WITH_LIMITATIONS on bounded adversarial and HEX8 diagnostic evidence; WP20 is now PASS_WITH_LIMITATIONS for the existing bounded J2 scope. WP21 is PASS_WITH_LIMITATIONS on surgical cleanup and does not promote a capability; WP22 is closed as PASS_WITH_LIMITATIONS through R0 and Step 1; neither package record alone constitutes an automatic publication approval. WP13 has its own controlled proof record; the remaining criteria and weights are authoritative in qualification/0_2_7/level_up_plan.json.

WP	Weight	Priority	Status	Rule
WP13	4%	MUST	PASS	release truth and golden baseline
WP14	5%	MUST	PASS	frozen large-scale execution contract; no solve claim
WP15	10%	MUST	PASS_WITH_LIMITATIONS	matrix-free TET4, SPD and preconditioning; subscale evidence complete, WP16 remains the 1M gate
WP16	10%	MUST	PASS	two reproducible PETSc CG/GAMG replays at 1,029,000 DOF satisfy the frozen WP14 contract
WP17	5%	SHOULD	PASS_WITH_LIMITATIONS	pinned PETSc/MPI route closed from two-replay 1M and 3M Silver evidence; host availability and AIJ memory remain limitations
WP18	7%	MUST	PASS_WITH_LIMITATIONS	Bronze preflight and two-replay Silver 3M solve; Gold unattempted
WP19	5%	MUST	PASS_WITH_LIMITATIONS	24-case fail-closed robustness corpus and bounded HEX8/C3D8 diagnostic
WP20	3%	SHOULD	PASS_WITH_LIMITATIONS	bounded existing four-family J2 closure; reused constitutive external evidence; no universal structural increment claim
WP21	3%	SHOULD	PASS_WITH_LIMITATIONS	surgical architecture/API/registry cleanup; broader redesign deferred
WP22	3%	MUST	PASS_WITH_LIMITATIONS	final release qualification

The 45% historical block, 82% current acquired/progress view and 100% total plan weights are separate governance measures and must not be added together.

Update rules

• Record the exact source SHA before a lot starts and the evidence SHA after a lot is committed.
• Keep SUPPORTED, TESTED, VERIFIED and QUALIFIED_BOUNDED separate.
• Record SKIPPED, RESOURCE_LIMITED, NOT_COMPARABLE and EXPECTED_FAILURE explicitly; never convert them to PASS.
• Update the machine-readable progress record in the same commit as a status change.
• A work package can move to READY_FOR_OWNER_REVIEW only with the evidence listed in the gate matrix and a clean, reproducible replay.

Current baseline note

The only baseline evidence inherited at foundation start is the controlled 0.2.6 release at e839373b6aef291a93292186d7553ba5cd12af55. It is a reference point, not a 0.2.7 result. WP01 is the first completed foundation control; WP02 is complete for the registry control; WP03 is complete for the descriptor and preflight control; WP04 is complete for the additive V&V harness control; WP05 has completed the controlled external-oracle preflight. WP06 has completed the common mesh-quality diagnostic contract. T1-R has prepared the remaining pre-WP07 contracts and asymmetric fixtures. Terra/Owner review authorized WP07, whose elemental kernel and targeted V&V are complete. WP08 has completed the bounded static vertical slice: Gmsh Prism 6 import, declared TRI3/QUAD4 face loads, common static assembly, equilibrium and post-processing are evidenced through the V&V v2 path. WEDGE6 remains EXPERIMENTAL; WP09 has completed its internal robustness evidence with external limitations recorded, and WP10 has completed its separate modal evidence gate.

WP09 preserves its 22-case internal robustness corpus: 18 passes and four expected fail-closed outcomes covering inverted geometry, wrong node order, malformed Gmsh and singular boundary conditions. Its WP09-FINAL external campaign adds 12 Code_Aster PENTA6 cases covering affine tension/compression, shear, bending, TRI3/QUAD4 pressure, prescribed displacement, multi-element, declared distortion and three refinement levels. All 12 primary displacement, total-reaction and strain-energy comparisons pass with tolerances fixed before execution; final external replay is deterministic. CalculiX C3D6 remains explicitly NOT_FORMULATION_COMPATIBLE with the QF six-point production quadrature, and WEDGE6 remains EXPERIMENTAL with public qualification deferred.

WP12 has completed a bounded readiness campaign at declared 100k, 300k, 500k, 750k and 1M target levels. Matrix-free TET4 solves completed through 750141 DOF, a separate 311469-DOF assembly-only probe recorded sparse storage and resource use, and the 1M attempt was classified RESOURCE_LIMITED_TIME. SciPy CG and direct routes retain their explicit configured/resource limits; PETSc/SLEPc were unavailable. The evidence is ready for Owner review and does not claim universal 1M or multi-million-DOF support.

WP10 preserves the original 16 controlled modal cases as historical evidence. WP10-FINAL adds the declared refinement/MAC catalog and records a four-level 4/8/16/32 prism sequence. The first three frequencies meet the predeclared final-change rule; all requested modes are finite and positive, deterministic, and have normalized residual at most 2.40e-11. Code_Aster 18.1.0/PENTA6 matches 24/24 modes over four same-mesh cases, with maximum frequency error 1.927e-13 and minimum MAC 0.9999999999999991. The active modal maturity is QUALIFIED_BOUNDED for the declared consistent-mass WEDGE6 scope; modes four to six remain diagnostic for refinement and other dynamic routes stay outside the claim.

WP11 has completed its all-family small-strain J2 characterization without changing the formulation: material paths, tangent FD, multi-element, cycles, energy, rollback, increment sensitivity, Newton and failure-mode records are in the controlled evidence artifact. The existing qualified J2 scope is retained with explicit limitations; finite-kinematic J2 and unrelated dynamics gaps remain outside this work package. WP16 has a completed 1,029,000-DOF PETSc qualification, and WP17-R has recorded the pinned PETSc/MPI route and its AIJ memory limitation. WP18 has completed its Bronze/Silver ladder: the 3M Silver solve passed twice under the frozen contract; Gold remains NOT_ATTEMPTED because no restart/checkpoint or distinct second physical case was run. WP19 has completed its bounded adversarial and HEX8 diagnostic campaign. WP20 closes the Owner review with the existing bounded scope; WP21 has completed its surgical compatibility and release-truth scope with PASS_WITH_LIMITATIONS; WP22 is closed as PASS_WITH_LIMITATIONS through the final Owner release checkpoint.

WP19 - Adversarial robustness and HEX8 diagnostic

WP19 is recorded as PASS_WITH_LIMITATIONS from execution source SHA dc5975b78727d9dca6d0a48b716e60f355b8799f, with the lot starting at 7f7ffbaf0b3fdda7d3ad31ba95f20a54e4719a53. The 24-case T1 corpus produced 10 positive PASS results and 14 EXPECTED_FAILURE_PASS results. There were no unexpected failures or invalid evidence records. Replay was deterministic, failure paths were fail-closed, and no NaN/Inf result was accepted.

The HEX8 study contains three axial-refinement, three slenderness and three transverse-resolution rows. Six same-mesh CalculiX 2.20/C3D8 displacement comparisons pass the existing one-percent diagnostic threshold, with maximum full-displacement relative error 1.997130986610937e-06. External reactions and strain energy are NOT_COMPARABLE because the inherited deck requests displacement only. QF and C3D8 agree while both deviate from the slender-beam Euler diagnostic; the result is classified as LOW_ORDER_LIMITATION with secondary MESH_DEPENDENCE. Locking is compatible with the observation but not proven, and no HEX8R/SRI/B-bar formulation is evaluated or promoted.

The controlled records are qualification/0_2_7/wp19_state.json, qualification/0_2_7/wp19_runtime/wp19_robustness_summary.json, qualification/0_2_7/wp19_runtime/wp19_hex8_diagnostic.json and qualification/0_2_7/wp19_runtime/wp19_golden_replay.json. WEDGE6 and all previous gate decisions remain unchanged. WP20 is closed with bounded limitations and WP21 is closed with PASS_WITH_LIMITATIONS; WP22 is closed through the final release checkpoint.

WP20 - Residual J2 and external V&V closure

WP20 is PASS_WITH_LIMITATIONS from review start SHA 26a734d1656c1c824c27f4708a8783abfddde17c against evidence source SHA 94461602dfd1782be57c20e1801a0d5d8e262ef1. The Owner decision is OWNER_APPROVED_BOUNDED_KEEP_EXISTING_SCOPE: TET4, TET10, HEX8 and HEX20 remain KEEP within the existing QUALIFIED_BOUNDED small-strain J2 scope.

Return mapping, yield detection, unload/reload, simple cycling, tangent FD, increment characterization, rollback, energy, cross-family checks, explicit failure modes, no NaN/Inf and deterministic replay are recorded as passing or bounded evidence. The tangent FD maximum is 2.120472111937634E-10 against the existing 1E-6 limit. No universal structural increment threshold is claimed; tangent symmetry and modified Newton behavior remain diagnostics.

External V&V is PARTIAL_REUSED_CONTROLLED_EVIDENCE: Code_Aster 18.1.0 G06 constitutive evidence is reused for the four families, with no new structural campaign and no post-result tolerance retuning. Finite-kinematic J2 remains experimental/not qualified. The authoritative records are qualification/0_2_7/wp20_state.json, qualification/0_2_7/wp11_j2_evidence.json and docs/verification/0_2_7/0_2_7_wp20_j2_closeout.md. WP21 is closed with PASS_WITH_LIMITATIONS; WP22 is closed through the final release checkpoint.

Active global accounting

The historical Level-Up portfolio is closed as Level-Up 1 at 50/50. The historical pre-closure Level-Up 2 checkpoint was 46/50, giving CURRENT_GLOBAL_PROGRESS = 96/100. The active Owner-complete program is 50/50 and CURRENT_GLOBAL_PROGRESS = 100/100. The earlier 45/55, 82/85 and 32/50 views remain historical accounting records only; they are not added to the current program and do not replace any evidence. The active machine-readable records are qualification/0_2_7/level_up_2_plan.json, qualification/0_2_7/level_up_2_state.json and qualification/0_2_7/level_up_2_index.json.

Level-Up 2 active plan

Work package	Weight	Priority	Status	Rule
LU2-WP01	4%	MUST	PASS	evidence and performance observatory
LU2-WP02	9%	MUST	PASS_WITH_LIMITATIONS	CPU/MPI/GAMG freeze before large runs
LU2-WP03	9%	MUST	PASS_WITH_LIMITATIONS	distinct second 3M workload and two replays
LU2-WP04	5%	MUST	PASS	real 5M Bronze completed with global readiness, finalization and evidence
LU2-WP05	9%	MUST	PASS	complete two-replay 5M Silver solve under the frozen route
LU2-WP06	4%	MUST	PASS_WITH_LIMITATIONS	recovery, diagnostics and fail-closed execution; bounded to existing checkpoint routes
LU2-WP07	4%	SHOULD	PASS_WITH_LIMITATIONS	existing route maturity retained at bounded combination scope; no promotion or demotion
LU2-WP08	2%	SHOULD	PASS_WITH_LIMITATIONS	explicit decisions; deferred/research routes remain unqualified
LU2-WP09	4%	MUST	PASS_WITH_LIMITATIONS	final registry, CI, package and Owner qualification

LU2-WP02 evidence is defined by the predeclared contract qualification/0_2_7/wp02_execution_contract.json, the controlled run index qualification/0_2_7/wp02_runtime/wp02_evidence_index.json, the frozen configuration qualification/0_2_7/wp02_runtime/wp02_config_freeze.json and the closeout qualification/0_2_7/wp02_state.json. The 3M structured TET4 route passed at 2, 4 and 8 ranks with replay coverage at 2 and 8 ranks. GAMG and contiguous partitioning were selected from characterized subscale alternatives. Preflight, redistribution, communication and I/O are explicitly unmeasured because the legacy runner does not expose separate boundaries; no phase is inferred. The evidence is therefore bounded to the recorded host, Docker image, input and solver configuration.

LU2-WP03 is closed as PASS_WITH_LIMITATIONS with 3M_GOLD_COMPUTE = PASS. Workload A is the preserved WP18 Silver control; Workload B is a materially distinct 3,000,000-DOF structured TET4 case with two PASS replays under the unchanged freeze. The controlled evidence is indexed by qualification/0_2_7/wp03_runtime/wp03_evidence_index.json and the bounded metrics are recorded in docs/verification/0_2_7/0_2_7_wp03_3m_gold_compute.md. Preflight, redistribution, communication and I/O remain NOT_MEASURED; no performance speedup is inferred between the distinct workloads. LU2-WP04 was completed as PASS on a real 5,012,640-DOF TET4 model, with global readiness, all-rank finalization and preserved evidence. LU2-WP05 then completed the two-replay 5M Silver solve under the unchanged freeze. The earlier owner-interrupted attempt remains historical forensic evidence and is not the active gate status; C1 was not confirmed.

LU2-WP08 is closed as PASS_WITH_LIMITATIONS by the controlled decision matrix qualification/0_2_7/lu2_wp08_decision_matrix.json. Mixed TET/WEDGE/HEX remains a partial technical path without an end-to-end qualification and is deferred. WEDGE15 and PYRAMID5 are not supported or active capabilities and are deferred. Existing HEX8 remains bounded under WP19; HEX8R, SRI and B-bar are research-only, and hourglass control is deferred with reduced integration. No numerical source, capability maturity or WP04 forensic status changed. The supervised WP04 retry remains the next operational action.

LU2-WP01 evidence is defined by qualification/0_2_7/observatory_contract.json and the controlled sample qualification/0_2_7/wp01_observatory_sample.json. LU2 weights sum to 50 %. C1, C2 and C3 are installed zero-weight conditional gates. C1 remains DORMANT because the WP04 time-budget overrun was observed during an owner-interrupted run without persisted progress/resource telemetry; C2 and C3 remain dormant. LU2-WP03 is complete, LU2-WP04 remains incomplete, and a supervised WP04 retry is the next active action; its pre-LU2 baseline is 8f08bfb5a6d4dedcd24966f5474e8c12cbfa5bc3.

The following paragraph preserves the earlier WP06 closeout snapshot; it is historical accounting, not the current LU2 state.

LU2-WP06 is closed as PASS_WITH_LIMITATIONS at 4%. The additive lifecycle and diagnostic taxonomy are covered by qualification/0_2_7/wp06_execution_contract.json and tests/unit/test_execution_contract.py; existing nonlinear and Newmark checkpoint tests remain the recovery evidence. Recovery is deliberately bounded to those routes, with no universal distributed or fault-tolerance claim. This closeout raised LU2 accounting from 24/50 to 28/50 and global progress from 74/100 to 78/100. WP04 remains USER_INTERRUPTED_INCONCLUSIVE and its supervised retry remains the next operational action.

LU2-WP07 closes as PASS_WITH_LIMITATIONS through the machine-readable matrix qualification/0_2_7/lu2_wp07_maturity_matrix.json and state qualification/0_2_7/lu2_wp07_state.json. Existing route maturity was audited against the registry and reused WP06, WP19 and WP20 evidence; no new external campaign, numerical source change or maturity promotion was performed. The historical pre-closure accounting was 46/50 and 96/100 globally. The active Owner-complete accounting is 50/50 and 100/100. WP04 and WP05 are closed; LU2-WP09 is closed through the final release qualification checkpoint.

S1 installs the optional rank-zero JSONL assembly telemetry contract for that retry. It records flushed chunk progress, rates, ETA when available, phases, resource fields and one-million-element milestones; it does not change the WP04 status or global progress.

F5 packaging and compatibility audit

F5 starts from clean source SHA 124c61f6492eee351a34e3542d198a13c00c2874 and closes as PASS_WITH_LIMITATIONS. Wheel and sdist artifacts for the historical 0.2.7a0 candidate were built and checked in isolated output directories, then installed from outside the repository on Windows Python 3.12 and 3.13. Public imports, a lightweight TET4 API workflow, CLI help/version/check/solve and packaged runtime resources passed. A missing-resource P1 in the release-vv package path was fixed in MANIFEST.in and pyproject.toml; the command now emits a controlled report when its legacy bundled evidence is not a current 0.2.7 qualification.

The historical candidate was 0.2.7a0, with no tag, PyPI publication or GitHub release at that checkpoint. The stable package identity is reconciled in the current release records. Python 3.10/Linux are represented by CI declarations, Python 3.11 and macOS remain unverified in this audit, and optional PETSc/MPI/HDF5 routes retain their explicit environment requirements. Global qualification accounting is unchanged and F6 is the next audit; no maturity or numerical claim moved. The controlled record is qualification/0_2_7/f5_packaging_compatibility_audit.json.

F6 numerical and performance regression audit

F6 starts from f6cfde036f5866c15e688bce70be5ed21b493ff1 and closes as PASS_WITH_LIMITATIONS. The controlled regression audit retained zero P0 and zero P1 findings. Targeted element, solver, BC/load/material, post-processing, MPI/PETSc and governance checks passed. The full suite result was 2147 passed, 3 failed, 184 skipped, 2 warnings; the three failures are the same pre-existing F4 failures in experimental or stale nonlinear paths, with no new skip or xfail.

The active 1M, 3M and 5M replay records, bounded C3 10M record and structured AIJ preallocation evidence remain valid. No numerical source, tolerance, baseline or maturity changed. A new 5M/10M run was not required because F6 introduced no relevant source change and this gate is a controlled regression audit rather than a new qualification campaign. F6 evidence is recorded in qualification/0_2_7/f6_numerical_performance_regression_audit.json and docs/verification/0_2_7/0_2_7_f6_numerical_performance_regression_audit.md.

F6 is ready for the separate Owner R0 decision; R0 was not started.