lifecycle: register the two orphan complete/ records (fixes Lifecycle Drift CI)

benchmark_jax_gradient_based_optimizers_on_the.md and
next_wave_population_gradient_samplers_on_the.md were git-mv'd into complete/ by
09256cc as raw prompt files: underscore names (off the safe_name hyphen
convention every other record uses) and no complete.md entry, tripping the
`lifecycle.py check` invariant "every complete/ record has a complete.md entry".

Rename both to hyphenated slugs (jax-gradient-optimizer-benchmark,
next-wave-population-optimizers), add their complete.md entries
(autolens_workspace_developer #95/PR96 and #97/PR98, both merged), fix the
internal cross-ref, and regenerate complete/index.md. check + index --check now
pass.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Author: 2 people

complete.md

@@ -6138,3 +6138,17 @@
 - workspace-pr: https://github.com/PyAutoLabs/autofit_workspace/pull/95 (merged f23254bd, squash)
 - repos: autofit_workspace
 - summary: Phase 2 (redefined config→examples) of the multi-start gradient search promotion (Fit#1369). Added a MultiStartAdam section to scripts/searches/mle.py (JAX use_jax=True analysis, standard fit plot; ADABelief/Lion noted) + regenerated notebooks/searches/mle.ipynb. Original "config/packaged defaults" phase dropped (searches have no per-search config; new searches add zero config keys). Validated end-to-end (recovers 50/25/10). Finding: the searches need NO pytree registration in user scripts (instance built inside traced objective). --auto safe; Heart YELLOW human-acked. See project_multi_start_gradient_search_promotion.
+
+## jax-gradient-optimizer-benchmark
+- issue: https://github.com/PyAutoLabs/autolens_workspace_developer/issues/95 (closed)
+- completed: 2026-07-13
+- workspace-pr: https://github.com/PyAutoLabs/autolens_workspace_developer/pull/96 (merged)
+- repos: autolens_workspace_developer
+- summary: Workspace-only benchmark (searches_minimal/) of JAX gradient-based optimizers on the HST MGE lens likelihood (AnalysisImaging(use_jax=True), jax.value_and_grad; no PyAutoFit source change), grounded in Herculens / Enzi et al. 2026 (arXiv:2606.30620). Cold-start candidates optax Adam, ADABelief, jaxopt L-BFGS, numpyro SVI, jaxopt Levenberg-Marquardt against _metrics.MLTracker (evals/time-to-ML, robustness = distance to simulator-truth logL). Key finding: EVERY single cold-start method locks onto the WRONG basin (r_E ~3.5–5.0, logL ~-158000) while 12-start Adam (GIGA-Lens recipe) recovers truth (r_E 1.600, logL +31788, ~2/12 starts). Failure is basin selection, not optimizer/speed. Established the RAL A100 pipeline; motivated the next-wave follow-up. See project_jax_gradient_optimizer_benchmark.
+
+## next-wave-population-optimizers
+- issue: https://github.com/PyAutoLabs/autolens_workspace_developer/issues/97 (closed)
+- completed: 2026-07-13
+- workspace-pr: https://github.com/PyAutoLabs/autolens_workspace_developer/pull/98 (merged)
+- repos: autolens_workspace_developer
+- summary: Follow-up to jax-gradient-optimizer-benchmark (#95/PR96) — fast many-points OPTIMIZERS (MAP, not full Bayesian samplers) on the MGE lens likelihood, testing whether population/diversity buys robustness. Candidates (searches_minimal, same MGE MAP harness): multi-start of other local rules (L-BFGS/ADABelief/Lion/Levenberg-Marquardt), CMA-ES (evosax), SVGD-as-mode-finder (blackjax.svgd), plus a jaxns nested-sampling reference cameo. Baseline to beat = multi-start Adam (r_E 1.600, 2/12 starts, ~1254 s). Finding: robust MAP needs diversity + gradients; SVGD (A100) best (r_E 1.595). Full-sampling wave (tempered SMC/HMC/flowMC/converged Nautilus) explicitly deferred. RAL A100 pipeline used. See project_jax_gradient_optimizer_benchmark.

…_jax_gradient_based_optimizers_on_the.md …6/07/jax-gradient-optimizer-benchmark.mdcomplete/2026/07/benchmark_jax_gradient_based_optimizers_on_the.md renamed to complete/2026/07/jax-gradient-optimizer-benchmark.md complete/2026/07/benchmark_jax_gradient_based_optimizers_on_the.md renamed to complete/2026/07/jax-gradient-optimizer-benchmark.md

@@ -16,7 +16,7 @@ Status: formalised
 
 ## The insight this builds on
 
-Follow-up to `experiment/workspaces/benchmark_jax_gradient_based_optimizers_on_the.md`
+Follow-up to `complete/2026/07/jax-gradient-optimizer-benchmark.md`
 (autolens_workspace_developer#95). That run established, on the HST MGE lens
 likelihood, that **every single cold-start method fails the same way** —
 optax Adam (r_E 4.89), ADABelief (5.01), jaxopt L-BFGS (4.42) and numpyro SVI

…e_population_gradient_samplers_on_the.md …26/07/next-wave-population-optimizers.mdcomplete/2026/07/next_wave_population_gradient_samplers_on_the.md renamed to complete/2026/07/next-wave-population-optimizers.md complete/2026/07/next_wave_population_gradient_samplers_on_the.md renamed to complete/2026/07/next-wave-population-optimizers.md

@@ -6,7 +6,7 @@ Token-light navigation over the finished-work records (schema:
 only then grep a dated bucket. Curators: edit the band between the CURATED
 markers; everything below GENERATED is rebuilt.
 
-558 records across 5 buckets.
+575 records across 5 buckets.
 
 <!-- CURATED:START -->
 ## Highlights
@@ -20,6 +20,7 @@ _(curate hard-won records here — survives regeneration.)_
 
 - [afa-wiki-currency-wiring-afa-skill-tranche](2026/07/afa-wiki-currency-wiring-afa-skill-tranche.md) — both autofit_assistant follow-ups shipped in one --auto session (one in-session Heart-YELLOW ack covering both…
 - [aggregator-output-contracts](2026/07/aggregator-output-contracts.md)
+- [aggregator-quick-fit-consolidation](2026/07/aggregator-quick-fit-consolidation.md) — guides/results tutorials ran own capped fits pruning to 2 samples vs deep [-10]/[9] IndexError → consolidate o…
 - [alma-interferometer](2026/07/alma-interferometer.md) — ALMA interferometer reduction live (phase 5) — first visibility-domain product family: calibrated MS → al.Inte…
 - [api-baseline-refresh](2026/07/api-baseline-refresh.md)
 - [assistant-benchmarks](2026/07/assistant-benchmarks.md)
@@ -44,7 +45,9 @@ _(curate hard-won records here — survives regeneration.)_
 - [colab-link-rot-libdocs](2026/07/colab-link-rot-libdocs.md)
 - [colab-maturity](2026/07/colab-maturity.md)
 - [colab-sim-verify-install](2026/07/colab-sim-verify-install.md)
+- [corrective-pr-heart-red](2026/07/corrective-pr-heart-red.md) — human-authorized corrective-PR exception for Heart RED — MERGED
 - [csv-api-lenstool](2026/07/csv-api-lenstool.md)
+- [database-latent-wheel-load](2026/07/database-latent-wheel-load.md) — database scripts failed on release wheels loading latent variables → AssertionError. Release-tail item D. Corr…
 - [delaunay-qhull-callback](2026/07/delaunay-qhull-callback.md)
 - [docs-theming-hub](2026/07/docs-theming-hub.md) — phases B+C of docs middle path — hub LIVE at pyautolabs.github.io (Pages; RETROFIT.md = paid-domain flip); sha…
 - [dpie-lenstool-param](2026/07/dpie-lenstool-param.md)
@@ -74,7 +77,12 @@ _(curate hard-won records here — survives regeneration.)_
 - [intake-census-dashboard](2026/07/intake-census-dashboard.md)
 - [intake-formalise](2026/07/intake-formalise.md)
 - [intake-reconcile](2026/07/intake-reconcile.md)
+- [interferometer-analysis-fitexception](2026/07/interferometer-analysis-fitexception.md) — interferometer (+point_source) log_likelihood_function lacked imaging's NumPy-path try/except→FitException gua…
+- [inversion-testmode-singular-guard](2026/07/inversion-testmode-singular-guard.md) — release-tail singular/non-PD inversion FAILs (slam/cpu_fast_modeling) were a flaky TEST_MODE artifact, not a c…
 - [jax-autodiff-gradients-audit](2026/07/jax-autodiff-gradients-audit.md) — 3-phase JAX autodiff gradients audit complete in one supervised session — every smooth likelihood FD-certified…
+- [jax-grad-param9-autodiff-fd-mismatch](2026/07/jax-grad-param9-autodiff-fd-mismatch.md) — jax_grad imaging pixelization parameter-9 autodiff-vs-FD mismatch → parameter-specific documented FD exclusion…
+- [jax-gradient-optimizer-benchmark](2026/07/jax-gradient-optimizer-benchmark.md)
+- [jax-pytree-partition-vars-no-dict](2026/07/jax-pytree-partition-vars-no-dict.md)
 - [jwst-frame-feasibility](2026/07/jwst-frame-feasibility.md) — JWST per-frame feasibility answered — GO, phased. Should: strongest case = NIRCam SW undersampling (subpixel d…
 - [jwst-frame-products](2026/07/jwst-frame-products.md) — JWST frame products live — 7 feasibility deltas implemented (_crf capture+consume, native MJy/sr + derived man…
 - [jwst-nircam-cosmos-web](2026/07/jwst-nircam-cosmos-web.md)
@@ -97,6 +105,8 @@ _(curate hard-won records here — survives regeneration.)_
 - [markdown-example-renderings](2026/07/markdown-example-renderings.md) — executed-markdown example pages, GitHub-browsable with real output images. generate_markdown.py (curated confi…
 - [markdown-renderings-howto](2026/07/markdown-renderings-howto.md) — batch 2b executed-markdown for HowTo chapter_1 (6/6/9). FAST (ch1 = few/no searches). Surfaced+fixed 2 tutoria…
 - [markdown-renderings-workspaces](2026/07/markdown-renderings-workspaces.md) — batch 2a executed-markdown for the 3 workspaces (autofit 3 overview, autogalaxy 18, autolens +21 → index 30). …
+- [mass-cse-jax-decompose](2026/07/mass-cse-jax-decompose.md) — incomplete Sérsic-stellar-mass CSE→JAX port (chaining.py FAIL) — thread xp through Sersic CSE deflection path,…
+- [matern-kernel-tfp-jax-incompat](2026/07/matern-kernel-tfp-jax-incompat.md)
 - [matrix-blackjax-and-smoke-guards](2026/07/matrix-blackjax-and-smoke-guards.md) — Final layers of the Python Version Matrix fix (after nufftax #382/#144 and jax #383/#1358/#497/#604). (1) blac…
 - [matrix-jax-skip](2026/07/matrix-jax-skip.md) — Second layer of the Python Version Matrix fix (after nufftax layer #382/#144). 11 unit tests across 4 repos ex…
 - [matrix-nufftax-py312](2026/07/matrix-nufftax-py312.md) — Fixed the chronic PyAutoBuild "Python Version Matrix" weekly red. 38 failures/job = nufftax ModuleNotFoundErro…
@@ -105,7 +115,10 @@ _(curate hard-won records here — survives regeneration.)_
 - [multi-shared-state-core-api](2026/07/multi-shared-state-core-api.md)
 - [multi-shared-state-examples](2026/07/multi-shared-state-examples.md) — EPIC — phases 1-4 + follow-up
 - [multi-shared-state-examples-phase-3-workspace-examples](2026/07/multi-shared-state-examples-phase-3-workspace-examples.md)
+- [multi-start-gradient-examples](2026/07/multi-start-gradient-examples.md) — Phase 2 (redefined config→examples) of the multi-start gradient search promotion (Fit#1369). Added a MultiStar…
+- [multi-start-gradient-search](2026/07/multi-start-gradient-search.md) — Phase 1 of promoting the benchmark-winning multi-start gradient MAP optimizer to first-class PyAutoFit searche…
 - [nautilus-nn-bottleneck](2026/07/nautilus-nn-bottleneck.md) — Nautilus overhead research — anatomy on fast likelihoods (NN training 40-52% + bound geometry 25-38% of wall; …
+- [next-wave-population-optimizers](2026/07/next-wave-population-optimizers.md)
 - [nightly-loud-failure](2026/07/nightly-loud-failure.md) — nightly driver hardening — token probe, fetch errors page (never 💤), anchor validation w/ 24h fallback, no anc…
 - [nightly-release-activity-gate](2026/07/nightly-release-activity-gate.md)
 - [nightly-release-implementation](2026/07/nightly-release-implementation.md)
@@ -131,6 +144,7 @@ _(curate hard-won records here — survives regeneration.)_
 - [psf-oversample-workspace](2026/07/psf-oversample-workspace.md)
 - [psf-visible-input](2026/07/psf-visible-input.md)
 - [pyautogut-organ](2026/07/pyautogut-organ.md) — Gut organ + hygiene drive seam — complete
+- [pyautolens-jax-joss-paper](2026/07/pyautolens-jax-joss-paper.md) — PyAutoLens-JAX JOSS paper scaffold — MERGED
 - [pyautoreduce-hst-design](2026/07/pyautoreduce-hst-design.md)
 - [pyautoscientist-3b-clone](2026/07/pyautoscientist-3b-clone.md) — clone (Mitosis) agent v0 live per DESIGN.md phasing — analyze mode emits CloneDecision (domain analysis via as…
 - [pyautoscientist-3b-config](2026/07/pyautoscientist-3b-config.md) — config extraction 4a — Heart version_skew map → config/repos.yaml `version_skew:` block (incl. unpolled assist…
@@ -145,10 +159,12 @@ _(curate hard-won records here — survives regeneration.)_
 - [refactor-conductor](2026/07/refactor-conductor.md)
 - [refactor-post-phase3](2026/07/refactor-post-phase3.md) — PyAutoReduce post-phase-3 refactor — MERGED
 - [regen-workspace-notebooks](2026/07/regen-workspace-notebooks.md)
+- [release-advisory-tier-slow-scripts](2026/07/release-advisory-tier-slow-scripts.md) — mode=release integrate kept going RED not on bugs but on a shifting perf-flake tail of slow real-search script…
 - [release-datasets-group-b](2026/07/release-datasets-group-b.md) — bare-`dataset/` repos — purged & MERGED
 - [release-docs-polish-learn-paths](2026/07/release-docs-polish-learn-paths.md)
 - [release-ships-simulated-datasets](2026/07/release-ships-simulated-datasets.md) — Group A — all 4 legs complete & MERGED; Group B separate
 - [release-stamping-slim](2026/07/release-stamping-slim.md)
+- [release-validation-tail-burndown](2026/07/release-validation-tail-burndown.md) — EPIC/capstone. Burned down the 18f/5t mode=release tail. All real correctness bugs fixed+merged+validated on w…
 - [remove-nss-sampler](2026/07/remove-nss-sampler.md) — ALL phases MERGED + dev-clean
 - [remove-pulse-compat](2026/07/remove-pulse-compat.md)
 - [review-faculty](2026/07/review-faculty.md)
@@ -170,6 +186,7 @@ _(curate hard-won records here — survives regeneration.)_
 - [stpsf-tier2b](2026/07/stpsf-tier2b.md) — JWST per-frame PSF coverage lifted 3/6 -> 6/6 (F115W validated) — STPSF tier-2b at frame DETECTOR + target pos…
 - [stub-skill-recipes-fill-in](2026/07/stub-skill-recipes-fill-in.md) — retired by user decision
 - [user-defaults-discovery](2026/07/user-defaults-discovery.md) — batch finale (task 6, rescoped discovery-half) — start-new-project step 0 environment-discovery preflight: rea…
+- [verify-install-check-f-autosimulate-and-dep-pin](2026/07/verify-install-check-f-autosimulate-and-dep-pin.md)
 - [version-check-compat-floor](2026/07/version-check-compat-floor.md)
 - [version-pinning-design-review](2026/07/version-pinning-design-review.md)
 - [weak-likelihood-function](2026/07/weak-likelihood-function.md)