# Cumulative recorded validation

PF Simulation r34 retains **8,657 distinct passing automated parent-test identities**, represented by **64,845 passing parent-test executions** in **312 completed JUnit records** across **63 validation scopes**. Successful subtest executions are reported separately: **6,541**.

A separate numerical ledger retains **6,823 benchmark and verification check identities across 46 declared studies**. Parent tests, subtests, JavaScript checks, provider aggregates and numerical checks are different units. They are never combined into a grand total.

This cumulative evidence spans retained source revisions, from identifiable r9 records through the r34 workflow-reliability and Assistant Recovery, Loading and Opening consolidation of October 2026. It is not a new full rerun of the latest release. Missing earlier release evidence is not assigned invented counts.

## Automated validation breakdown

Each row gives identities first encountered in that release group after deduplication. All 5,595 identities in the Fix22 website ledger remain included; the subsequent completed release records add **3,062** identities and **17,883** passing parent executions. Within that increase, this 17 October Fix_24 refresh adds **138** identities and **441** passing parent executions to the 14 October Fix_23 ledger (8,519 identities / 64,404 executions). The 9 October Fix_23 ledger (8,206 identities / 63,432 executions) is preserved within that history. The former 3,102-identity adoption/private-workflow ledger is also entirely retained.

| Release group | Additional distinct passing identities |
|---|---:|
| Earlier retained releases (r9–r33) | 3,266 |
| r34 core, BYOK/BYOS, qualification and access controls | 1,157 |
| Private workflows, startup and lifecycle | 254 |
| Reporting and literature management | 73 |
| ISAES analysis, evidence and assistant tests | 516 |
| ISAES documentation and navigation | 31 |
| Autonomous ISAES investigations | 158 |
| Attachment and upload maintenance | 90 |
| Save controls and feedback maintenance | 50 |
| HF gateway and deployment budget | 226 |
| Native HF runtime and recovery | 346 |
| HF isolation and immutable payload compatibility | 195 |
| HF deployment documentation | 22 |
| Execution lifecycle and artifact publication | 343 |
| Runtime lease and heartbeat consolidation | 53 |
| Configurable runtime scaling and reliability | 233 |
| HF and shared deployment consolidation | 560 |
| Public service, SMTP readiness and Micius/Refresh fixes | 613 |
| Refresh and authentication controls | 20 |
| Execution logs and Orbit Viewer recovery | 123 |
| Tracking preparation and controller recovery | 114 |
| Configuration consistency and live-comparison controls | 50 |
| HF Astropy reference-cache initialization | 26 |
| Private-workflow reliability | 53 |
| Assistant Recovery, Loading and Opening consolidation | 85 |
| **Cumulative distinct identities** | **8,657** |

Coverage includes numerical and quantum models, standalone and hosted interfaces, permissions, account admission, storage/BYOK lifecycles, private workflows, reporting, ISAES, recovery, artifact delivery and release integrity. Re-executions contribute to the execution count but not the distinct-identity count.

## Numerical benchmark and verification breakdown

| Study family | Audited checks | Scope |
|---|---:|---|
| Independent QuTiP quantum comparisons | 3,829 | 748 cases across eight model families; repeated comparison receipts counted once. |
| QKD count distributions | 864 | 108 declared configurations. |
| Independent SatQuMA comparisons | 720 | 60 configurations × 12 independently compared quantities. |
| Nine quantum benchmark defaults | 459 | Bell, teleportation, swapping, purification, memory, repeater, HOM, tomography and QKD sweep. |
| Scientific integration preset studies | 535 | 27 completed numerical/reference-data preset studies; MATLAB and registration-only records excluded. |
| Portable numerical relativity | 143 | Minkowski, gauge-wave and link studies plus independent portable-solver verification. |
| Orbit and clock comparisons | 63 | 37 orbit checks and 26 published clock checks. |
| Qiskit phase-correlation study | 210 | The existing website study, retained once. |
| **Cumulative audited numerical ledger** | **6,823** | **46 declared studies** |

The 748 independent QuTiP cases comprise teleportation (300), swapping (100), purification (100), Bell states (60), mission memory (140), standalone memory (5), tomography (28) and Hong–Ou–Mandel interference (15). Their 3,829 case/metric checks count once despite repeated r22, r24 and r34 receipts. Computational independence under matched assumptions does not imply external human replication or physical-device certification.

The earlier website’s 357 highlighted checks—45 gauge-wave-link, two Hong–Ou–Mandel, 100 tomography and 210 phase-correlation checks—are already included. All 50 numerical evidence-file hashes remain unchanged in the latest package and retained website study.

Later scalable-runtime evidence repeats 110 existing checks: Bell defaults (62), FMU Dahlquist (3) and the gauge-wave link (45). The artifact-publication workload repeats 459 named checks across the nine existing quantum profiles. Its configuration was reconstructed from the posted workload request, with rounded Bell-angle serialization and repeater bookkeeping defaults; those differences do not establish new scientific studies. These are useful repeated executions, with **zero additional numerical identities or declared studies**. Operational kernel/isolation and artifact-publication probes are excluded from the numerical tally.

A numerical identity is a case/metric pair or a named check position in an explicitly declared study. The 46-study breakdown is: one independent QuTiP comparison registry, one QKD distribution registry, one SatQuMA registry, one orbit-convergence study, nine quantum defaults, 27 integration presets, three portable numerical-relativity studies, one independent portable-solver verification, one published-clock comparison and one website Qiskit study. The count represents recorded verification instances, not 6,823 distinct physical laws or independent experiments. Unindexed exploratory calculations are not claimed.

## Latest focused corrections

The latest full release includes the 15 October workflow-reliability correction and the 16 October Assistant consolidation. Their frozen receipts add **441 passing Python parent-test executions** and **138 distinct parent-test identities** to the prior website ledger.

| Validation item | Python parent tests | New distinct identities | Separately reported results |
|---|---:|---:|---|
| Workflow reliability | 136 | 53 | 152 UI cases and 39 assertion checks; 11 successful commands; no failures, errors or skips |
| Assistant consolidation | 305 | 85 | 70 successful subtests; 232 UI cases and 49 assertion checks; 14 successful commands; 7 skips and 1 deselection |
| **Refresh additions** | **441** | **138** | Units remain separate; no numerical-ledger additions |

The Assistant Python breakdown is 293 backend tests and 12 HTTP/BYOS tests. Seven PostgreSQL-dependent cases were skipped; one scientific-worker case was deliberately deselected. Its UI record contains 12 DOM/API suites. The workflow record contains three Python groups and eight JavaScript groups. These fixtures check behavior and boundaries; they do not establish deployed browser latency, real PostgreSQL concurrency, provider scientific accuracy or production throughput.

Earlier patch ZIP evidence is retained in the full release, but overlapping patch reruns are excluded from these current totals. Separate compact-handoff checks and website validation are also outside the full-release application ledger. No new numerical benchmark identity or declared study is claimed. Read [workflow reliability](workflow-reliability.md) and [Assistant consolidation](assistant-consolidation.md) for the implementation scope.

The Astropy correction retains its 106-test receipt and 26 added identities. Configuration maintenance retains 243 parent tests and 1,673 successful subtests. Its 212 declared topology/preset products (154 accepted, 58 rejected) and 102 natural starting configurations are matrix entries, not extra tests or simulated scientific results. The [configuration correction guide](configuration-fixes.md) explains that scope.

On **16 October 2026**, the operator reported that the excessive Assistant loading delay was resolved with the accepted patches. Subsequent supplied logs showed the consolidated full release installed and serving at controller revision `a9a912355f05ffc8085ea3474e57609fa2af59da`, with verified source/running identity and preserved configuration. This later deployment observation does not change frozen receipt flags or establish browser performance, production capacity or all-workflow scientific validity. Read the [current release and deployment scope](r34-validation.md).

## Release groups added since Fix22

These original completed receipts cover overlapping tests; their parent execution counts are not distinct coverage counts. Earlier completed release receipts are retained once. Superseded same-revision attempts, duplicate exports and aggregate shards are excluded.

| Recorded release scope | Passing parent executions | Additional distinct identities | Separate subtests | Separate JavaScript commands |
|---|---:|---:|---:|---:|
| Hybrid Hugging Face gateway support and deployment-wide cost controls | 1,052 | 226 | 27 | 5 |
| Native Hugging Face SaaS execution and durable recovery integration | 1,444 | 346 | 67 | 5 |
| Hugging Face confinement and immutable private payload compatibility correction | 1,637 | 195 | 67 | 5 |
| Hugging Face full-release deployment guides and bundled Windows helper | 80 | 22 | 11 | 0 |
| Execution performance, sign-in and organization lifecycle maintenance | 821 | 140 | 43 | 7 |
| Execution performance, sign-in, organization lifecycle and HF Job response compatibility maintenance | 885 | 63 | 43 | 7 |
| Execution lifecycle, HF Job compatibility, PostgreSQL readiness and controller build maintenance | 901 | 16 | 61 | 7 |
| Execution lifecycle, HF Job compatibility, PostgreSQL readiness, WAL catch-up and controller build maintenance | 920 | 19 | 87 | 7 |
| Execution lifecycle, HF Job compatibility, PostgreSQL readiness, WAL catch-up, controller build and deployment documentation maintenance | 920 | 0 | 87 | 7 |
| Artifact publication reliability, batched storage, progress and execution lifecycle maintenance | 1,045 | 105 | 87 | 8 |
| Consolidated r34 lease reads and heartbeat cadence | 459 | 53 | 33 | 3 |
| r34 scalable runtime and reliable publication | 1,688 | 233 | 110 | 10 |
| r34 native HF fixes and shared deployment consolidation | 1,558 | 560 | 0 | 14 |
| r34 public service and permanent lifecycle integration | 2,798 | 577 | 0 | 14 |
| r34 public service SMTP readiness correction 2026-10-09 | 225 | 35 | 0 | 1 |
| r34 Micius preset and environment refresh correction 2026-10-09 | 17 | 1 | 0 | 1 |
| Refresh feedback and authentication controls maintenance | 20 | 20 | 0 | 4 |
| r34 execution, logs and Orbit Viewer recovery | 509 | 123 | 0 | 1 |
| r34 tracking preparation and controller recovery | 114 | 114 | 12 | 1 |
| r34 configuration consistency and live comparison controls | 243 | 50 | 1,673 | 2 |
| r34 HF Astropy reference-cache initialization | 106 | 26 | 0 | 0 |
| r34 private-workflow reliability corrections | 136 | 53 | 0 | 8 |
| r34 confirmed Assistant consolidation | 305 | 85 | 70 | 12 |

The preceding autonomous ISAES scope retains 1,495 parent tests, 130 subtests and four JavaScript commands. Attachment/upload maintenance retains 293 parent tests and five JavaScript commands; save-control maintenance retains 237 parent tests and ten JavaScript commands. Their identities remain part of the baseline union.

### Isolated PostgreSQL checks — separate aggregate

The retained isolated PostgreSQL record reports **144 passes across seven groups**: schema migration (2), managed lifecycle (12), assistant/store behavior (41), mail/account/query behavior (45), AI-account/migration semantics (42), actual worker regression (1) and physical recovery (1). The provider job was recorded as completed. These are isolated integration results, not a production load or external backup-restore certification.

Because the retained provider evidence does not include individual JUnit identities for canonical cross-run deduplication, these 144 aggregate passes are **not added** to the 8,657 distinct identities or 64,845 parent executions. The machine-readable release ledger preserves their separate count and scope.

## Reconciliation method

1. Inventory all **720** retained application JUnit XML files. Include **312** completed records selected by the original release receipts or the unchanged baseline selection.
2. Keep the Fix22 baseline and the 9 and 14 October Fix_23 identities and executions unchanged. Add final release receipts and explicitly retained preceding released-source records. Exclude failed/interrupted attempts, superseded same-revision runs, copied records and shard children already covered by their aggregate.
3. Resolve module/class/function/parameter identifiers to repository-relative parent-test identities. Keep named operational JUnit parent probes identifiable as such. Renamed tests remain different recorded identities; semantic equivalence is not guessed. The legacy machine field `cumulative_distinct_passed_pytest_identities` is retained for compatibility and represents this canonical automated parent-identity union.
4. Deduplicate raw receipt bytes and matching timestamp/host/identity exports. Exclude skipped cases from passing counts. Verify original JUnit hashes against their receipts and the latest package checksum inventory.
5. Keep **6,541** successful subtest executions separate; cross-run distinct subtest parameters are not consistently available. JavaScript commands/assertions, syntax checks, provider aggregates, numerical checks and website verification retain their own units.
6. Preserve the original source-era receipts. Hash consistency proves byte integrity, not independent certification.

Native Windows, representative researcher usability, production mixed-workload capacity/soak, actual fleet failover, external backup restore, live AI/BYOK/BYOS provider qualification and external scientific replication remain separate acceptance questions. Software tests and numerical agreement do not establish empirical validity of a new physical claim.

## Reproducible evidence

Application archive SHA-256:

`cfa7a2c3586735b97f4d957352e3e88eb224be2336a947049f8e1743fab33958`

Read the [cumulative identity ledger](/evidence/r34-validation-totals.json) and [release validation groups](/evidence/r34-release-validation-groups.json). The source archive contains the unchanged Fix22 accounting in `validation/isaes-cumulative-totals/`, the 9 October Fix_23 records in `validation/r34-live-update/`, the 14 October refresh in `validation/r34-20261014-update/`, and the current refresh records and accounting in `validation/website-release-refresh-20261017/`. Older baseline ledgers remain intact.

The source archive retains the accounting scripts, original completed receipts, selected JUnit records and preservation checks. Follow `validation/website-release-refresh-20261017/README.md` to reproduce the current refresh:

```sh
python validation/website-release-refresh-20261017/reconcile.py
```

Earlier scripts in `validation/r34-live-update/` reproduce the dated 9 October ledger, not the current figures.

These accounting steps read completed evidence; they do not rerun the application or scientific models.

## Complete retained scope ledger

| Retained scope | Passing parent executions | Additional distinct identities |
|---|---:|---:|
| `validation_qkd` | 973 | 806 |
| `validation_benchmarks` | 1,030 | 100 |
| `validation_gui` | 1,047 | 123 |
| `validation_gui_r12` | 221 | 23 |
| `validation_gui_r13` | 189 | 23 |
| `validation_precision` | 1,212 | 71 |
| `validation_assurance` | 1,475 | 171 |
| `validation_config` | 1,866 | 179 |
| `validation_results` | 283 | 11 |
| `validation_physics_fixes` | 1,714 | 101 |
| `validation_competitive` | 2,116 | 223 |
| `validation_mission_r20` | 1,973 | 143 |
| `validation_saas` | 2,144 | 98 |
| `validation_r22` | 443 | 57 |
| `validation_research_r22` | 4,596 | 54 |
| `validation_spacetime_r221` | 469 | 29 |
| `validation_cosmetic_r222` | 735 | 0 |
| `validation_run_comparison_r23` | 547 | 85 |
| `validation_r24` | 2,362 | 122 |
| `validation_r25` | 595 | 68 |
| `validation_r26` | 441 | 153 |
| `validation_r27` | 249 | 93 |
| `validation_r28` | 455 | 68 |
| `validation_r29` | 731 | 83 |
| `validation_r30` | 572 | 110 |
| `validation_r31` | 739 | 74 |
| `validation_r32` | 806 | 67 |
| `validation_r33` | 937 | 131 |
| `validation_r34` | 1,101 | 164 |
| `validation_r34_byok` | 1,413 | 311 |
| `validation_r34_byos` | 1,806 | 393 |
| `validation_r34_refresh` | 1,132 | 78 |
| `validation_r34_adoption` | 3,206 | 175 |
| `validation_r34_gui_bugfix` | 203 | 36 |
| `validation_r34_private_workflows` | 379 | 130 |
| `validation_r34_startup` | 582 | 41 |
| `validation_r34_lifecycle` | 897 | 83 |
| `validation_r34_reporting` | 183 | 22 |
| `validation_r34_literature` | 200 | 17 |
| `validation_r34_isaes` | 527 | 219 |
| `validation_r34_literature_management` | 204 | 34 |
| `validation_r34_isaes_completion` | 784 | 223 |
| `validation_r34_isaes_ai` | 1,306 | 74 |
| `validation_r34_isaes_documentation` | 94 | 31 |
| `validation_r34_autonomous` | 1,495 | 158 |
| `validation_r34_upload_fix` | 293 | 90 |
| `validation_r34_save_controls` | 237 | 50 |
| `validation_r34_huggingface` | 1,052 | 226 |
| `validation_r34_huggingface_native` | 1,444 | 346 |
| `validation_r34_huggingface_compatibility` | 1,637 | 195 |
| `validation_r34_huggingface_documentation` | 80 | 22 |
| `validation_r34_execution_lifecycle` | 5,492 | 343 |
| `validation_r34_runtime_consolidation` | 459 | 53 |
| `validation_r34_scalable_runtime` | 1,688 | 233 |
| `validation_r34_hf_shared_consolidation` | 1,558 | 560 |
| `validation_r34_public_service` | 3,040 | 613 |
| `validation_r34_ui_controls` | 20 | 20 |
| `validation_r34_execution_recovery` | 509 | 123 |
| `validation_r34_tracking_recovery` | 114 | 114 |
| `validation_r34_configuration_fixes` | 243 | 50 |
| `validation_r34_astropy_cache_fix` | 106 | 26 |
| `validation_r34_workflow_reliability` | 136 | 53 |
| `validation_r34_assistant_consolidation` | 305 | 85 |
| **Total** | **64,845** | **8,657** |