cortex: activate ic50_workspace and submit graphical-ep phase 4

dashboard.html

@@ -236,7 +236,8 @@
 <p class="muted">One row per active project — the next thing, what it is holding, and when it last ruled.</p>
 <table class="map">
 <tr><th>Project</th><th>Next task</th><th>Awaiting</th><th>Running</th><th>Last ruling</th></tr>
-<tr><td><b>euclid_dr1_prelim</b></td><td>Is the ordered run's 8-nat gain ordering or library drift</td><td>0</td><td>3</td><td>-</td></tr>
+<tr><td><b>euclid_dr1_prelim</b></td><td>Is the ordered run's 8-nat gain ordering or library drift</td><td>0</td><td>4</td><td>-</td></tr>
+<tr><td><b>ic50_workspace</b></td><td>Does EP match the graphical joint fit end to end</td><td>0</td><td>1</td><td>-</td></tr>
 <tr><td><b>slope_hierarchy_scale</b></td><td>Does hierarchical slope recovery hold at N=25 to 50</td><td>0</td><td>1</td><td>-</td></tr>
 <tr><td><b>subhalo_validation</b></td><td>Does rectangular_adapt false-detect a subhalo on the clumpy source</td><td>0</td><td>5</td><td>2026-09-07</td></tr>
 </table>
@@ -246,7 +247,6 @@ <h3>Nothing open</h3>
 <tr><td>concr</td><td>dormant</td><td>none</td><td><button class="copy text" data-cmd="/cortex — retire the science project concr: confirm it is holding nothing live, then in PyAutoCortex run `python3 scripts/cortex.py retire concr --why &quot;&lt;one line on why&gt;&quot;`, `python3 scripts/cortex.py check`, `pyauto-brain cortex dashboard --apply`, and push the ledger. The row, its tasks and its rulings all stay — only the status and the note change." aria-label="Copy the retire prompt">📋 retire</button></td></tr>
 <tr><td>cowls_diana</td><td>dormant</td><td>none</td><td><button class="copy text" data-cmd="/cortex — retire the science project cowls_diana: confirm it is holding nothing live, then in PyAutoCortex run `python3 scripts/cortex.py retire cowls_diana --why &quot;&lt;one line on why&gt;&quot;`, `python3 scripts/cortex.py check`, `pyauto-brain cortex dashboard --apply`, and push the ledger. The row, its tasks and its rulings all stay — only the status and the note change." aria-label="Copy the retire prompt">📋 retire</button></td></tr>
 <tr><td>euclid</td><td>dormant</td><td>planned 3</td><td><button class="copy text" data-cmd="/cortex — retire the science project euclid: confirm it is holding nothing live, then in PyAutoCortex run `python3 scripts/cortex.py retire euclid --why &quot;&lt;one line on why&gt;&quot;`, `python3 scripts/cortex.py check`, `pyauto-brain cortex dashboard --apply`, and push the ledger. The row, its tasks and its rulings all stay — only the status and the note change." aria-label="Copy the retire prompt">📋 retire</button></td></tr>
-<tr><td>ic50_workspace</td><td>dormant</td><td>planned 1</td><td><button class="copy text" data-cmd="/cortex — retire the science project ic50_workspace: confirm it is holding nothing live, then in PyAutoCortex run `python3 scripts/cortex.py retire ic50_workspace --why &quot;&lt;one line on why&gt;&quot;`, `python3 scripts/cortex.py check`, `pyauto-brain cortex dashboard --apply`, and push the ledger. The row, its tasks and its rulings all stay — only the status and the note change." aria-label="Copy the retire prompt">📋 retire</button></td></tr>
 <tr><td>pj011646</td><td>dormant</td><td>none</td><td><button class="copy text" data-cmd="/cortex — retire the science project pj011646: confirm it is holding nothing live, then in PyAutoCortex run `python3 scripts/cortex.py retire pj011646 --why &quot;&lt;one line on why&gt;&quot;`, `python3 scripts/cortex.py check`, `pyauto-brain cortex dashboard --apply`, and push the ledger. The row, its tasks and its rulings all stay — only the status and the note change." aria-label="Copy the retire prompt">📋 retire</button></td></tr>
 <tr><td>profiling</td><td>dormant</td><td>none</td><td><button class="copy text" data-cmd="/cortex — retire the science project profiling: confirm it is holding nothing live, then in PyAutoCortex run `python3 scripts/cortex.py retire profiling --why &quot;&lt;one line on why&gt;&quot;`, `python3 scripts/cortex.py check`, `pyauto-brain cortex dashboard --apply`, and push the ledger. The row, its tasks and its rulings all stay — only the status and the note change." aria-label="Copy the retire prompt">📋 retire</button></td></tr>
 <tr><td>subhalo_simulations</td><td>dormant</td><td>none</td><td><button class="copy text" data-cmd="/cortex — retire the science project subhalo_simulations: confirm it is holding nothing live, then in PyAutoCortex run `python3 scripts/cortex.py retire subhalo_simulations --why &quot;&lt;one line on why&gt;&quot;`, `python3 scripts/cortex.py check`, `pyauto-brain cortex dashboard --apply`, and push the ledger. The row, its tasks and its rulings all stay — only the status and the note change." aria-label="Copy the retire prompt">📋 retire</button></td></tr>
@@ -260,15 +260,22 @@ <h3>Nothing open</h3>
 <h3>euclid_dr1_prelim</h3>
 <p class="paths"><b>Local</b> <span class="pathchip">/mnt/c/Users/Jammy/Science/euclid_dr1_prelim</span></p>
 <p class="paths"><b>RAL</b> <span class="pathchip">/mnt/ral/jnightin/euclid_dr1_prelim</span></p>
-<p class="muted">active · both partition · tasks: gated 1 · running 2 · submitted 1</p>
+<p class="muted">active · both partition · tasks: running 2 · submitted 2</p>
 <div class="task"><button class="copy" data-cmd="Report where the runs of the PyAutoCortex task tasks/euclid_dr1_prelim/ordered_mge_control_unordered_102005065.md (342377_0) stand." aria-label="Copy the Claude command">📋</button><p><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/tasks/euclid_dr1_prelim/ordered_mge_control_unordered_102005065.md">Is the ordered run's 8-nat gain ordering or library drift</a><span class="tags"><span class="pill y">running</span></span><span class="facets"> — runs 342377_0 · wall 0:00 of 8:00 (0%)</span></p></div>
 <div class="task"><button class="copy" data-cmd="Report where the runs of the PyAutoCortex task tasks/euclid_dr1_prelim/ordered_mge_witness_102005065.md (342375_[0-1]) stand." aria-label="Copy the Claude command">📋</button><p><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/tasks/euclid_dr1_prelim/ordered_mge_witness_102005065.md">Does ordering the MGE bases fix labelling on one tile</a><span class="tags"><span class="pill y">running</span></span><span class="facets"> — runs 342375_[0-1] · wall 0:00 of 8:00 (0%)</span></p></div>
 <div class="task"><button class="copy" data-cmd="Report where the runs of the PyAutoCortex task tasks/euclid_dr1_prelim/dr1_prelim_10_lens_science_run.md (342301_[0-9], 342314_3) stand." aria-label="Copy the Claude command">📋</button><p><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/tasks/euclid_dr1_prelim/dr1_prelim_10_lens_science_run.md">Ten real DR1 lenses fitted from the pipeline repo alone</a><span class="tags"><span class="pill y">submitted</span></span><span class="facets"> — runs 342301_[0-9], 342314_3 · wall 0:00 of 48:00 (0%)</span></p></div>
-<div class="task"><button class="copy" data-cmd="python3 scripts/cortex.py gates   # then, once they have closed: move tasks/euclid_dr1_prelim/ordered_rerun_catalogue_parity.md ready
-# gates: euclid_strong_lens_modeling_pipeline#60" aria-label="Copy the Claude command">📋</button><p><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/tasks/euclid_dr1_prelim/ordered_rerun_catalogue_parity.md">Do ordered Nautilus reruns reproduce the euclid catalogue values?</a><span class="tags"><span class="pill n">gated</span></span><span class="facets"> — euclid_strong_lens_modeling_pipeline#60</span></p></div>
+<div class="task"><button class="copy" data-cmd="Report where the runs of the PyAutoCortex task tasks/euclid_dr1_prelim/ordered_rerun_catalogue_parity.md (342398) stand." aria-label="Copy the Claude command">📋</button><p><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/tasks/euclid_dr1_prelim/ordered_rerun_catalogue_parity.md">Do ordered Nautilus reruns reproduce the euclid catalogue values?</a><span class="tags"><span class="pill y">submitted</span></span><span class="facets"> — runs 342398 · wall 0:00 of 36:00 (0%)</span></p></div>
 <p class="muted"><a href="https://github.com/PyAutoLabs/PyAutoCortex/tree/main/tasks/euclid_dr1_prelim/">tasks/euclid_dr1_prelim/</a></p>
 </section>
 <section class="project">
+<h3>ic50_workspace</h3>
+<p class="paths"><b>Local</b> <span class="pathchip">/mnt/c/Users/Jammy/Science/ic50_workspace</span></p>
+<p class="paths"><b>RAL</b> <span class="pathchip">/mnt/ral/jnightin/ic50_workspace</span></p>
+<p class="muted">active · ral partition · tasks: submitted 1</p>
+<div class="task"><button class="copy" data-cmd="cd /mnt/c/Users/Jammy/Science/ic50_workspace &amp;&amp; hpc/sync jobs" aria-label="Copy the Claude command">📋</button><p><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/tasks/ic50_workspace/ep_scale_up.md">Does EP match the graphical joint fit end to end</a><span class="tags"><span class="pill y">submitted</span></span><span class="facets"> — runs 342408, 342409 · wall 0:00 of 48:00 (0%)</span></p></div>
+<p class="muted"><a href="https://github.com/PyAutoLabs/PyAutoCortex/tree/main/tasks/ic50_workspace/">tasks/ic50_workspace/</a> · <a href="https://github.com/Jammy2211/ic50_workspace/issues">issues</a></p>
+</section>
+<section class="project">
 <h3>slope_hierarchy_scale</h3>
 <p class="paths"><b>Local</b> <span class="pathchip">/mnt/c/Users/Jammy/Science/slope_hierarchy_scale</span></p>
 <p class="paths"><b>RAL</b> <span class="pathchip">/mnt/ral/jnightin/slope_hierarchy_scale</span></p>

dashboard.md

@@ -29,7 +29,8 @@ One row per active project — the next thing, what it is holding, and when it l
 
 | Project | Next task | Awaiting | Running | Last ruling |
 |---|---|---:|---:|---|
-| euclid_dr1_prelim | Is the ordered run's 8-nat gain ordering or library drift | 0 | 3 | - |
+| euclid_dr1_prelim | Is the ordered run's 8-nat gain ordering or library drift | 0 | 4 | - |
+| ic50_workspace | Does EP match the graphical joint fit end to end | 0 | 1 | - |
 | slope_hierarchy_scale | Does hierarchical slope recovery hold at N=25 to 50 | 0 | 1 | - |
 | subhalo_validation | Does rectangular_adapt false-detect a subhalo on the clumpy source | 0 | 5 | 2026-09-07 |
 
@@ -40,7 +41,6 @@ One row per active project — the next thing, what it is holding, and when it l
 | concr | dormant | none | retire ↓ |
 | cowls_diana | dormant | none | retire ↓ |
 | euclid | dormant | planned 3 | retire ↓ |
-| ic50_workspace | dormant | planned 1 | retire ↓ |
 | pj011646 | dormant | none | retire ↓ |
 | profiling | dormant | none | retire ↓ |
 | subhalo_simulations | dormant | none | retire ↓ |
@@ -69,14 +69,6 @@ One row per active project — the next thing, what it is holding, and when it l
 
 </details>
 
-<details><summary>📋 retire ic50_workspace</summary>
-
-```
-/cortex — retire the science project ic50_workspace: confirm it is holding nothing live, then in PyAutoCortex run `python3 scripts/cortex.py retire ic50_workspace --why "<one line on why>"`, `python3 scripts/cortex.py check`, `pyauto-brain cortex dashboard --apply`, and push the ledger. The row, its tasks and its rulings all stay — only the status and the note change.
-```
-
-</details>
-
 <details><summary>📋 retire pj011646</summary>
 
 ```
@@ -118,7 +110,7 @@ One row per active project — the next thing, what it is holding, and when it l
 
 **RAL** `/mnt/ral/jnightin/euclid_dr1_prelim`
 
-active · both partition · tasks: gated 1 · running 2 · submitted 1
+active · both partition · tasks: running 2 · submitted 2
 
 <details><summary>📋 <a href="tasks/euclid_dr1_prelim/ordered_mge_control_unordered_102005065.md">Is the ordered run's 8-nat gain ordering or library drift</a> — <b>running</b> — runs 342377_0 · wall 0:00 of 8:00 (0%)</summary>
 
@@ -144,17 +136,34 @@ Report where the runs of the PyAutoCortex task tasks/euclid_dr1_prelim/dr1_preli
 
 </details>
 
-<details><summary>📋 <a href="tasks/euclid_dr1_prelim/ordered_rerun_catalogue_parity.md">Do ordered Nautilus reruns reproduce the euclid catalogue values?</a> — <b>gated</b> — euclid_strong_lens_modeling_pipeline#60</summary>
+<details><summary>📋 <a href="tasks/euclid_dr1_prelim/ordered_rerun_catalogue_parity.md">Do ordered Nautilus reruns reproduce the euclid catalogue values?</a> — <b>submitted</b> — runs 342398 · wall 0:00 of 36:00 (0%)</summary>
 
 ```
-python3 scripts/cortex.py gates   # then, once they have closed: move tasks/euclid_dr1_prelim/ordered_rerun_catalogue_parity.md ready
-# gates: euclid_strong_lens_modeling_pipeline#60
+Report where the runs of the PyAutoCortex task tasks/euclid_dr1_prelim/ordered_rerun_catalogue_parity.md (342398) stand.
 ```
 
 </details>
 
 [tasks/euclid_dr1_prelim/](https://github.com/PyAutoLabs/PyAutoCortex/tree/main/tasks/euclid_dr1_prelim/)
 
+### ic50_workspace
+
+**Local** `/mnt/c/Users/Jammy/Science/ic50_workspace`
+
+**RAL** `/mnt/ral/jnightin/ic50_workspace`
+
+active · ral partition · tasks: submitted 1
+
+<details><summary>📋 <a href="tasks/ic50_workspace/ep_scale_up.md">Does EP match the graphical joint fit end to end</a> — <b>submitted</b> — runs 342408, 342409 · wall 0:00 of 48:00 (0%)</summary>
+
+```
+cd /mnt/c/Users/Jammy/Science/ic50_workspace && hpc/sync jobs
+```
+
+</details>
+
+[tasks/ic50_workspace/](https://github.com/PyAutoLabs/PyAutoCortex/tree/main/tasks/ic50_workspace/) · [issues](https://github.com/Jammy2211/ic50_workspace/issues)
+
 ### slope_hierarchy_scale
 
 **Local** `/mnt/c/Users/Jammy/Science/slope_hierarchy_scale`

projects.yaml

@@ -211,10 +211,10 @@ ic50_workspace:
   ral_root: /mnt/ral/jnightin/ic50_workspace
   mirror: none
   sync_cli: hpc/sync
-  sync_verbs: [pull]
-  ledger: README.md
+  sync_verbs: [push, pull, logs, sync, push-data-init, pull-full, status, submit, push-submit, jobs, sacct, cancel, wait-and-pull, tail, du, check, clear-logs]
+  ledger: NOTES.md
   assistant: none
   witness_file: results/**/*.json
-  partition: both
-  status: dormant
-  note: "personal remote, recorded as a fact — not a PyAutoLabs repo; ral_root planned, sync_verbs unverified"
+  partition: ral
+  status: active
+  note: "personal remote, recorded as a fact — not a PyAutoLabs repo; activated 2026-09-09 for graphical-ep phase 4, ral_root created and first runs submitted; NOTES.md predates the graphical/HPC work and needs a refresh"

tasks/ic50_workspace/ep_scale_up.md

@@ -2,11 +2,11 @@
 
 Project: ic50_workspace
 Summary: Does EP match the graphical joint fit end to end
-State: planned
+State: submitted
 Gates:
 Witness: one committed EP-vs-graphical parity table (means ± errors) at small N, EP running at the largest N reached with per-rung timings committed, and any derived-variable or scaling blocker recorded as its own filed prompt
 Budget: 48:00
-Runs:
+Runs: 342408, 342409
 Ruling:
 Review-minutes: 25
 Epic: graphical-ep
@@ -15,7 +15,12 @@ Migrated-from: PyAutoMind/draft/research/graphical_ep/ic50_ep_scale_up.md
 
 ## Question
 
-Ready when: graphical-ep phase 2 (Mind) is issued — add its ref to `Gates:` and move to `gated`.
+Started 2026-09-09. Phases 1 and 2 of the campaign shipped on 2026-09-02 (the whole
+D1–D6 PyAutoFit fix wave: #1558, #1560, #1562, #1572, #1573, #1574, #1576, #1578,
+#1580), so the gate this task was waiting on is cleared and it went
+`planned → ready → submitted` directly. The RAL PyAutoFit mirror was verified to
+contain #1580 before submitting, so unlike phase 3's EP arm these runs carry the
+stale-mask fixed-point fix.
 
 The IC50 cancer use case is the scale target: the end goal is graphical + EP fits at 10 000+
 datasets, with a clear demonstration that EP matches the graphical joint fit at small N before
@@ -55,9 +60,14 @@ Acceptance, verbatim from the Mind prompt:
 - Mind `draft/feature/autofit/ep_lbfgs_jax.md` — the development half (the speed lever)
 - Mind `research/graphical_ep/ep_campaign.md` phase 4 — the campaign row this feeds
 - Mind `research/autofit/autofit_profiling_bootstrap.md` — where the wrapper-overhead evidence goes
+- `scripts/compare_graphical_ep.py` (in the project) — writes the parity table this
+  task's witness asks for, to `results/graphical_ep_comparison_<name>.{txt,json}`
 
 ## Runs
 
+- 342408: submitted — ral — submitted 2026-09-09 — wall 0:00 — EP sim, nlive 150, max_steps 12
+- 342409: submitted — ral — submitted 2026-09-09 — wall 0:00 — graphical sim joint Dynesty, nlive 150
+
 ## Ruling
 
 (none)