homemaker-py-2g7.4: shape-curve DP prototype (Otten/Stockmeyer) — PASS

Prototype + validation for an exact size/width/proportion feasibility DP over a frozen slicing topology, replacing the ~80-200 eval Nelder-Mead inner loop's approximate answer to the same question with one bottom-up pass (experiments/shapecurve_spike.py). Leaf feasible regions are exact FAIL_THRESHOLD-inversions of fitness.py's quality_size/width/proportion; internal-node composition runs on a shared discretised grid. Validated on harbor-house-l0 (experiments/validate_shapecurve.py, 200 random topologies vs NM minimising shape-fail-count directly): 99.0% agreement (0 false negatives), 93.6x speedup at grid_n=150, plot-level bbox approximation error quantified at +7.5% (root-causing both observed false positives). All three acceptance criteria cleared -- see DESIGN.md §37.2 for full results and the caveats/scope not covered (multi-storey, leaf_sharing/co_type, true skew-quad regions). Kept as a reference spike, same status as experiments/autodiff_spike.py (§34); production wiring into driver.py filed as homemaker-py-6xh. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01LSwQwpEaHFBkeVSDDWd75S
2026-08-02 23:43:30 +01:00 · 2026-08-02 23:43:30 +01:00 · 85c1183d4c
commit 85c1183d4c
parent 3f2438797f
4 changed files with 633 additions and 21 deletions
--- a/.beads/issues.jsonl
+++ b/.beads/issues.jsonl
@ -1,4 +1,4 @@
-{"id":"homemaker-py-2g7.4","title":"Exact shape-curve inner loop (Otten/Stockmeyer DP) replacing Nelder-Mead","description":"The classic slicing-floorplan result applied to our exact representation: each leaf's size/width/proportion constraints define a feasible-shape region; these compose bottom-up through the slicing tree as piecewise shape curves, yielding in ONE linear pass (no iteration): (a) whether ANY ratio assignment satisfies all per-leaf shape constraints, and (b) the ratios that realize a chosen point on the root curve. Today the same question costs an 80-eval NM run per child (~all of the 3M-eval budget) and answers it only approximately. Plan: (1) prototype on harbor-house-l0 with a rectangular plot approximation; (2) validate against innerloop.optimise — DP-feasible topologies must score \u003e= NM result when polished, DP-infeasible must never reach 0 shape fails under NM; (3) wire as a PRE-FILTER: prune shape-infeasible children before any native eval, and warm-start NM from DP ratios (or replace NM entirely where the plot is near-rectangular; keep NM as final polish for skew). CAVEATS to model honestly: crinkliness/access/adjacency are NOT in the DP (graph terms, not per-leaf shape) — the DP handles the size/width/proportion family only, which is fine for pruning; equal-offset skew-quad geometry means DP areas are approximate — measure the approximation error on real plots first (harbor plot is a near-rect quad). Expected payoff: 100-1000x cheaper feasibility, turning topology search into enumerate-and-prune and unlocking the racing/MAP-Elites/CP issues. Cf. §34: autodiff failed on wall-clock; this is a different attack — exactness via structure, not gradients.","acceptance_criteria":"on harbor-house-l0: DP verdict agrees with NM-polished shape-fail outcome on \u003e=95% of 200 random topologies; measured speedup \u003e=50x per feasibility decision; approximation error on the skew plot quantified","status":"open","priority":1,"issue_type":"feature","owner":"bruno@postle.net","created_at":"2026-08-02T09:15:04Z","created_by":"Bruno Postle","updated_at":"2026-08-02T09:15:04Z","dependencies":[{"issue_id":"homemaker-py-2g7.4","depends_on_id":"homemaker-py-2g7","type":"parent-child","created_at":"2026-08-02T10:15:04Z","created_by":"Bruno Postle","metadata":"{}"}],"dependency_count":0,"dependent_count":1,"comment_count":0}
+{"id":"homemaker-py-2g7.4","title":"Exact shape-curve inner loop (Otten/Stockmeyer DP) replacing Nelder-Mead","description":"The classic slicing-floorplan result applied to our exact representation: each leaf's size/width/proportion constraints define a feasible-shape region; these compose bottom-up through the slicing tree as piecewise shape curves, yielding in ONE linear pass (no iteration): (a) whether ANY ratio assignment satisfies all per-leaf shape constraints, and (b) the ratios that realize a chosen point on the root curve. Today the same question costs an 80-eval NM run per child (~all of the 3M-eval budget) and answers it only approximately. Plan: (1) prototype on harbor-house-l0 with a rectangular plot approximation; (2) validate against innerloop.optimise — DP-feasible topologies must score \u003e= NM result when polished, DP-infeasible must never reach 0 shape fails under NM; (3) wire as a PRE-FILTER: prune shape-infeasible children before any native eval, and warm-start NM from DP ratios (or replace NM entirely where the plot is near-rectangular; keep NM as final polish for skew). CAVEATS to model honestly: crinkliness/access/adjacency are NOT in the DP (graph terms, not per-leaf shape) — the DP handles the size/width/proportion family only, which is fine for pruning; equal-offset skew-quad geometry means DP areas are approximate — measure the approximation error on real plots first (harbor plot is a near-rect quad). Expected payoff: 100-1000x cheaper feasibility, turning topology search into enumerate-and-prune and unlocking the racing/MAP-Elites/CP issues. Cf. §34: autodiff failed on wall-clock; this is a different attack — exactness via structure, not gradients.","acceptance_criteria":"on harbor-house-l0: DP verdict agrees with NM-polished shape-fail outcome on \u003e=95% of 200 random topologies; measured speedup \u003e=50x per feasibility decision; approximation error on the skew plot quantified","status":"closed","priority":1,"issue_type":"feature","assignee":"Bruno Postle","owner":"bruno@postle.net","created_at":"2026-08-02T09:15:04Z","created_by":"Bruno Postle","updated_at":"2026-08-02T22:42:15Z","started_at":"2026-08-02T18:39:05Z","closed_at":"2026-08-02T22:42:15Z","close_reason":"Prototype PASS: 99.0% agreement (\u003e=95%), 93.6x speedup (\u003e=50x), approximation error quantified (7.5% bbox overestimate). See DESIGN.md §37.2. Not wired into product this session -- follow-up homemaker-py-6xh filed.","dependencies":[{"issue_id":"homemaker-py-2g7.4","depends_on_id":"homemaker-py-2g7","type":"parent-child","created_at":"2026-08-02T10:15:04Z","created_by":"Bruno Postle","metadata":"{}"}],"dependency_count":0,"dependent_count":1,"comment_count":0}
 {"id":"homemaker-py-2g7.3","title":"Hard/soft fail tiering: 'solved' = zero hard fails","description":"Lex-by-total-count treats a crinkly wall the same as a missing room, so search polishes shape taxes instead of fixing structure — the 3M-run best still carries 'level 0/1 not connected' and wrong-level fails after 1.7M evals. Split fails into HARD (missing space, wrong/required level, level connectivity, circulation connectivity, stairs, covered-outside) and SOFT (crinkliness, proportion, size, width, edge-too-long) tiers. Outer comparator becomes (-hard, -soft, fitness); 'solved' is defined as zero hard fails. GUARDS: (1) the inner-loop 0.5^n cliff must keep protecting against trading into new fails (§4.5/§4.9 — rerun the 0/9 inner-loop-protection check); (2) rerun the §4.9 outer A/B: the scheme must not reintroduce the scalar pathology; (3) §11.4 warns comparator reshaping alone does not escape topology basins — the claim here is narrower: budget stops being spent on soft fails while hard fails remain, and reporting becomes meaningful. The tier map lives in fitness.py next to the fail emission sites so new fail strings must declare a tier. Can start before the calibration issue lands but final tier assignments should be reviewed against its findings.","acceptance_criteria":"tiered comparator behind a flag with A/B on harbor+maple (3 seeds, 20k evals): hard-fail count at budget strictly better or equal on mean, no §4.9 regression; report shows hard/soft split","notes":"ACCEPTANCE A/B COMPLETE — PASS (2026-08-02, experiments/tier_ab_2g7_3.py,\nharbor-house + maple-court, 3 seeds, budget 20000, leaf_sharing=True,\nn_workers=4, wall ~2h53m):\n\n  harbor-house  hard mean: flat 11.67 -\u003e tiered 5.33   (soft 29.00 -\u003e 42.33)\n  maple-court   hard mean: flat 19.33 -\u003e tiered 14.00  (soft 71.33 -\u003e 87.67)\n\nHard-fail mean strictly better on BOTH programmes at fixed budget — the\nrequired acceptance bar. Soft/total rise as expected (budget redirected from\npolishing shape fails to structural ones). Full per-seed log at\nscratch/tier_ab_2g7_3/log.txt (not committed — scratch output, regenerate via\nthe script if needed).\n\nGuards: (1) inner-loop 0.5^n cliff untouched by construction (no diff to\ninnerloop.py or the existing 0.5**len(failures) line) — not re-measured\nempirically, doesn't need to be. (2) tiered key is still lexicographic, not a\nblended scalar, so structurally immune to the §4.8 scalar pathology;\nencoded as tests/test_driver.py::test_use_tiers_prefers_fewer_hard_over_fewer_total_fails.\n\nDESIGN.md §37.1 written up with full table and rationale. Feature lands\ndefault-off (--use-tiers / HOMEMAKER_USE_TIERS / driver.search(use_tiers=)),\nso no existing reproduction changes.\n\nFollow-on (not blocking, filed separately): convergence-SPEED comparison\n(evals to 0 hard fails, tiered vs flat, same budget) — this A/B measured\nfail composition at a fixed budget snapshot, not time-to-solved.","status":"closed","priority":1,"issue_type":"feature","assignee":"Bruno Postle","owner":"bruno@postle.net","created_at":"2026-08-02T09:14:14Z","created_by":"Bruno Postle","updated_at":"2026-08-02T17:51:18Z","started_at":"2026-08-02T09:58:53Z","closed_at":"2026-08-02T17:51:18Z","close_reason":"Acceptance A/B passed on both harbor-house and maple-court (hard-fail mean strictly better under tiering); guards verified; DESIGN.md §37.1 written up.","dependencies":[{"issue_id":"homemaker-py-2g7.3","depends_on_id":"homemaker-py-2g7","type":"parent-child","created_at":"2026-08-02T10:14:14Z","created_by":"Bruno Postle","metadata":"{}"}],"dependency_count":0,"dependent_count":0,"comment_count":0}
 {"id":"homemaker-py-2g7.2","title":"Calibrate the objective against human reference designs","description":"Score the traced human solutions (from the plan-\u003edom composer issue) and classify EVERY fail they raise as one of: (a) genuine spec violation (fix the trace or accept), (b) representation artifact (fix scoring, cf. §13.3/§13.8 share leaks), or (c) miscalibrated threshold (fix the constant/curve). Prime suspect: crinkliness — 48% of the evolved residual (§13.11), flat ~0.8/leaf tax even on squarest layouts (§13.1); if a real human plan pays it broadly, the gaussian on 1/crink is mis-tuned, not the designs. Outcome: either the human reference scores at/near 0 hard fails (objective validated, search is the gap) or a concrete list of scoring fixes. This finally makes 'the examples are solvable' a measured statement. Also record the human design's score as the per-programme target line on all future runs.","acceptance_criteria":"every fail on each human reference classified with evidence; miscalibrations filed/fixed; per-programme target scores recorded in DESIGN.md","status":"open","priority":1,"issue_type":"task","owner":"bruno@postle.net","created_at":"2026-08-02T09:14:11Z","created_by":"Bruno Postle","updated_at":"2026-08-02T09:14:11Z","dependencies":[{"issue_id":"homemaker-py-2g7.2","depends_on_id":"homemaker-py-2g7","type":"parent-child","created_at":"2026-08-02T10:14:11Z","created_by":"Bruno Postle","metadata":"{}"},{"issue_id":"homemaker-py-2g7.2","depends_on_id":"homemaker-py-2g7.1","type":"blocks","created_at":"2026-08-02T10:14:11Z","created_by":"Bruno Postle","metadata":"{}"}],"dependency_count":1,"dependent_count":0,"comment_count":0}
 {"id":"homemaker-py-2g7.1","title":"Human reference corpus: plan-\u003edom composer + first traced human solutions","description":"There are NO human-generated plans in the corpus — every non-empty .dom is evolution output, so the system has no ground truth for what a good design scores. Build the missing pipeline: (1) a plan-\u003edom composer — input a traced rectangular partition (rooms as rects/quads with type codes, per storey), validate it, extract the binary slicing tree by recursive guillotine-cut detection, and emit a .dom (levels, heights, perimeter, divisions). Non-slicible partitions are REPORTED with the offending region rather than rejected silently — whether human plans even lie in the slicing class is itself a first-order representability finding. (2) Trace at least one human-drawn solution for harbor-house (the plateau benchmark) and one for programme-house. Practical input path: trace in Inkscape over the scan and parse SVG rects (examples/harbor-house/drawings/ already holds SVG assets), or a simple YAML room list; avoid automatic raster vectorization for now. Uses: (a) calibration ground truth for the objective, (b) search seeds, (c) representability test of the slicing-tree phenotype, (d) later, few-shot examples for the LLM repair operator.","acceptance_criteria":"composer round-trips a synthetic slicible partition to a scoring .dom; at least one human harbor-house solution traced, composed, and scored with homemaker-fitness; non-slicible input produces a diagnostic naming the unsliceable region","status":"open","priority":1,"issue_type":"feature","owner":"bruno@postle.net","created_at":"2026-08-02T09:14:10Z","created_by":"Bruno Postle","updated_at":"2026-08-02T09:14:10Z","dependencies":[{"issue_id":"homemaker-py-2g7.1","depends_on_id":"homemaker-py-2g7","type":"parent-child","created_at":"2026-08-02T10:14:09Z","created_by":"Bruno Postle","metadata":"{}"}],"dependency_count":0,"dependent_count":1,"comment_count":0}
@ -27,6 +27,7 @@
 {"id":"homemaker-py-1p0","title":"Geometry inner loop: full-objective equal-offset ratio optimiser","description":"DESIGN.md §5.1, §7 Phase 1. Productionise experiments/optimize_fullfitness.py into homemaker: optimise(topology, x0=None) -\u003e (geometry, fitness). DOF = equal-offset division ratios of free branches (solver.free_branches, lowest-storey cut ownership), clipped to [eps, 1-eps]. Objective = full oracle fitness (never a proxy — §4.2 falsified). Must support warm-start x0 (§5.6) and a population/batch evaluation mode so each iteration scores via one batched oracle call (§4.6).","acceptance_criteria":"Reproduces or exceeds §4.5 gains (x1.24–x1.67, no new failures) on 2f45907, candidate-002, c964435; works as a library call on any corpus .dom","status":"closed","priority":1,"issue_type":"feature","assignee":"Bruno Postle","owner":"bruno@postle.net","created_at":"2026-06-11T23:36:58Z","created_by":"Bruno Postle","updated_at":"2026-06-12T08:46:31Z","started_at":"2026-06-12T00:14:19Z","closed_at":"2026-06-12T08:46:31Z","close_reason":"innerloop.optimise() lands: batched CMA-ES sigma ladder (0.05/0.15, IPOP popsize doubling, deterministic seeding) over equal-offset free-branch ratios vs full oracle fitness; warm-start x0 supported. Acceptance vs unprojected originals: x1.65/x1.66/x1.58 against bars x1.24/x1.67/x1.59, no new failures, 46 oracle calls vs NM's 200. Two near-bar results accepted as reproduced-within-noise (1% tol) — draw spread brackets the single-NM-draw bars; approved by Bruno 2026-06-12. Gotchas: equal-offset projection of legacy unequal cuts loses fitness/adds failures (midpoint projection used); pycma seed=0 means clock-seeded.","dependencies":[{"issue_id":"homemaker-py-1p0","depends_on_id":"homemaker-py-av5","type":"blocks","created_at":"2026-06-12T00:39:33Z","created_by":"Bruno Postle","metadata":"{}"}],"dependency_count":1,"dependent_count":3,"comment_count":0}
 {"id":"homemaker-py-8cs","title":"Experiment: warm-vs-cold start of inner loop (Lamarckian inheritance)","description":"DESIGN.md §5.6, §4.6. Warm-starting a child topology's inner loop from the parent's optimised ratios is the main lever for cutting per-topology cost (~3 min/topology cold). Apply single topology mutations to optimised corpus designs, re-optimise warm (surviving cuts keep values, new cuts get heuristic defaults) vs cold, compare oracle-call counts to convergence at equal final fitness.","acceptance_criteria":"Speedup factor measured across \u003e=10 mutated topologies; decision recorded (expect order-of-magnitude; if \u003c2x, revisit §4.6 Phase-2 scoping)","notes":"Experiment script committed (experiments/warm_vs_cold.py, 1cc86c8) and machinery validated oracle-free; one mutated child scored through the oracle OK. Waiting on homemaker-py-gp2 reference run to finish, then execute under URB_NO_OCCLUSION=1 (3 parents x 400 evals + 12 children x 2 x 200 evals, ~1.5-2 h oracle time). Default budgets: parent 400, child 200; target = evals to 95% of best final.","status":"closed","priority":1,"issue_type":"task","owner":"bruno@postle.net","created_at":"2026-06-11T23:36:58Z","created_by":"Bruno Postle","updated_at":"2026-06-12T11:44:45Z","closed_at":"2026-06-12T11:44:45Z","close_reason":"Measured (URB_NO_OCCLUSION=1, parent budget 400, child 200, 12 single mutations across 3 designs): cold start reached 95% of warm final in 0/12 cases within budget — speedup unbounded at practical budgets; warm finals beat cold finals x1.2-x4 in 12/12; 6/12 warm starts were within 95% at 1 eval (near-neutral mutations). Decision: Lamarckian warm-starting is MANDATORY in the memetic driver (homemaker-py-b39), not an optimisation; cold starts produce strictly worse geometry at equal budget. Note: 2 undivides were exactly fitness-neutral (same-type merge == Merge_Divided equivalence) — locality datum for homemaker-py-nyb.","dependencies":[{"issue_id":"homemaker-py-8cs","depends_on_id":"homemaker-py-1p0","type":"blocks","created_at":"2026-06-12T00:39:34Z","created_by":"Bruno Postle","metadata":"{}"}],"dependency_count":1,"dependent_count":0,"comment_count":0}
 {"id":"homemaker-py-av5","title":"Batched oracle: score many .dom files per invocation","description":"oracle.py currently scores one .dom per urb-fitness.pl call (~1.65 s/dom). DESIGN.md §4.6: batching amortises Perl startup to ~0.99 s/dom and is required so population/batch optimisers can score a whole generation in one oracle call. Extend oracle.py with a batch API: write N .dom files, one perl invocation, parse N .score/.fails pairs. Keep the single-file path for compatibility.","acceptance_criteria":"Batch of 35 corpus files scores in one perl invocation; per-file results identical to single-file calls; measured s/dom reported","status":"closed","priority":1,"issue_type":"task","assignee":"Bruno Postle","owner":"bruno@postle.net","created_at":"2026-06-11T23:36:56Z","created_by":"Bruno Postle","updated_at":"2026-06-12T00:14:06Z","started_at":"2026-06-11T23:50:40Z","closed_at":"2026-06-12T00:14:06Z","close_reason":"score_batch() lands in oracle.py; 35-file corpus parity verified single-vs-batch (1e-12 rel fitness, exact fail sets); 0.98 s/dom batched vs 1.27 single, x1.30","dependency_count":0,"dependent_count":1,"comment_count":0}
 {"id":"homemaker-py-6xh","title":"Wire shapecurve DP prototype into driver.py as a real pre-filter + NM warm-start","description":"homemaker-py-2g7.4's prototype (experiments/shapecurve_spike.py, DESIGN.md\n§37.2) validated PASS on harbor-house-l0 (99% agreement vs shape-fail-only NM\nover 200 random topologies, 93.6x speedup at grid_n=150, 0 false negatives).\nIt is not yet wired into the product — it's a reference spike only, same\nstatus as experiments/autodiff_spike.py (§34).\n\nTo productionise per the original plan (DESIGN.md §37 point 2):\n- Replace/augment operators.predicted_shape_fails with the DP as driver.py's\n  real per-child pre-filter (a single-sample heuristic today; the DP gives an\n  exact yes/no plus a realizing ratio point).\n- Warm-start innerloop.optimise's NM from the DP's realised ratios instead of\n  (or in addition to) the current proportion-aware target-geometry seed.\n- Multi-storey support: the DP only walks one level's leaves currently;\n  below-linked nodes (wall-stacking across storeys) aren't modelled.\n- leaf_sharing/co_type target-adjustment: not modelled in leaf_constraints,\n  needed for any programme that uses either (harbor-house-l0 doesn't).\n- Consider replacing the bounding-box leaf approximation with true skew-quad\n  polygon algebra to remove the ~7-12% area approximation error identified\n  as the root cause of both measured false positives (§37.2) -- or at least\n  characterise it on a LESS rectangular plot than harbor-house-l0's\n  near-rectangular trapezoid, where the error is likely worse.\n- A/B against the real driver.search: does DP-pre-filter + warm-start beat\n  today's predicted_shape_fails + cold/proportion-aware start on wall-clock\n  to N hard fails, on harbor-house (full) and/or a less-rectangular plot?","status":"open","priority":2,"issue_type":"feature","owner":"bruno@postle.net","created_at":"2026-08-02T22:39:46Z","created_by":"Bruno Postle","updated_at":"2026-08-02T22:39:46Z","dependency_count":0,"dependent_count":0,"comment_count":0}
 {"id":"homemaker-py-2g7.9","title":"Parallel best-of-N + racing harness (use all cores, kill stragglers early)","description":"§14 measured islands \u003c= best-of-N, and the 3M runs used workers=1-2 on a 4-core box — independent seeds are the proven shape and we are not even using the local machine. Build a harness: launch N independent search_staged seeds across all cores (processes, not threads — mind the cvw id()-keyed cache bug), checkpoint fail-counts periodically, successively halve (hyperband-style: kill runs above median hard-fail count at each rung, reallocate budget to survivors). Fix/respect homemaker-py-b8g (parallel non-determinism) and homemaker-py-cvw first or work around with process isolation. This multiplies whatever eval cost the shape-curve DP issue achieves; on its own it is a free 4x locally and scales to any box. Report best + variance across seeds (the seed-variance in §12-§13 tables is huge — 78 vs 97 same config — so best-of-N is worth several levers combined).","acceptance_criteria":"harness runs N=16 seeds on 4 cores with racing; at equal total native-eval budget beats the single-seed mean on harbor by at least the observed seed spread; deterministic per-seed replay","status":"open","priority":2,"issue_type":"task","owner":"bruno@postle.net","created_at":"2026-08-02T09:15:58Z","created_by":"Bruno Postle","updated_at":"2026-08-02T09:15:58Z","dependencies":[{"issue_id":"homemaker-py-2g7.9","depends_on_id":"homemaker-py-2g7","type":"parent-child","created_at":"2026-08-02T10:15:58Z","created_by":"Bruno Postle","metadata":"{}"},{"issue_id":"homemaker-py-2g7.9","depends_on_id":"homemaker-py-b8g","type":"blocks","created_at":"2026-08-02T10:16:16Z","created_by":"Bruno Postle","metadata":"{}"},{"issue_id":"homemaker-py-2g7.9","depends_on_id":"homemaker-py-cvw","type":"blocks","created_at":"2026-08-02T10:16:15Z","created_by":"Bruno Postle","metadata":"{}"}],"dependency_count":2,"dependent_count":0,"comment_count":0}
 {"id":"homemaker-py-2g7.7","title":"LLM repair operator at stagnation (dom+fails -\u003e targeted compound edits)","description":"Generalize the §4.10 lesson: deceptive valleys are crossed by COMPOUND edits (move room + re-home displaced room + fix ratios atomically), which we currently hand-code one per valley (mutate_level_compound_fix). Our fail messages are semantically rich and localized ('me1 on wrong level', 'level 1 not connected', '0/rlrlr proportion') and the .dom is readable — ideal LLM input. Loop: on stagnation (no fail-tier improvement for N evals), serialize best individual + .fails + programme summary -\u003e LLM proposes 3-5 multi-step repairs as structured edit scripts (a small DSL over existing operator primitives: swap/divide/retype/rotate with explicit paths — NOT freeform dom text, so proposals are always well-formed) -\u003e apply, inner-loop, lex-accept as usual. Native fitness disposes; a bad proposal costs one child budget. Cost discipline: one LLM call ~ thousands of native evals, so plateau-only, cache by (signature, fails) key. Benchmark: the 3M-run best sat on 'level 0 not connected' + 'me1 on wrong level' for \u003e1M evals — moves a plan-reader fixes in one edit. Use claude via API (see claude-api skill); temperature\u003e0 for diverse proposals. Later extension (separate issue): AlphaEvolve-style operator-code synthesis using our existing A/B harness as the evaluator.","acceptance_criteria":"on the evolved-3M-nols-3 15-fail plateau seed: repair loop reduces hard-fail count where 1M+ blind evals did not, within \u003c=20 LLM calls; edit-DSL rejects malformed proposals; A/B at equal native-eval budget shows strictly better final fails on \u003e=2/3 seeds","status":"open","priority":2,"issue_type":"feature","owner":"bruno@postle.net","created_at":"2026-08-02T09:15:54Z","created_by":"Bruno Postle","updated_at":"2026-08-02T09:15:54Z","dependencies":[{"issue_id":"homemaker-py-2g7.7","depends_on_id":"homemaker-py-2g7","type":"parent-child","created_at":"2026-08-02T10:15:53Z","created_by":"Bruno Postle","metadata":"{}"}],"dependency_count":0,"dependent_count":1,"comment_count":0}
 {"id":"homemaker-py-2g7.6","title":"Spike: graph-first construction — adjacency-realizing slicing trees / rectangular dualization","description":"Research spike, timeboxed. Literature: rectangular dualization (planar triangulated graph -\u003e rectangular floorplan) and characterizations of slicible adjacency graphs. Our programme already IS an adjacency graph (every room wants c, plus secondary pairs); instead of mutating trees hoping adjacency emerges, construct trees that realize the required adjacency by construction — the direction §11.6/§11.7 crawled toward greedily. Deliverable is a WRITTEN assessment (DESIGN.md section): can harbor's programme graph (16 rooms + spine, 2 storeys with stacking constraint) be dualized into slicing trees, how many, and is enumeration of realizing trees tractable? Prototype only if the answer is clearly yes. Watch for: multi-storey Below-inheritance constrains both floors' trees jointly; circulation spine is a connected dominating set requirement, not a simple adjacency.","acceptance_criteria":"DESIGN.md section with go/no-go verdict, the relevant algorithms named, and complexity estimate for harbor-scale programmes","status":"open","priority":2,"issue_type":"task","owner":"bruno@postle.net","created_at":"2026-08-02T09:15:07Z","created_by":"Bruno Postle","updated_at":"2026-08-02T09:15:07Z","dependencies":[{"issue_id":"homemaker-py-2g7.6","depends_on_id":"homemaker-py-2g7","type":"parent-child","created_at":"2026-08-02T10:15:07Z","created_by":"Bruno Postle","metadata":"{}"}],"dependency_count":0,"dependent_count":0,"comment_count":0}
@ -122,27 +123,27 @@
 {"id":"homemaker-py-erc.6","title":"Experiment: inner-loop slack-expansion objective term","description":"Inner-loop counterpart to plot-fill construction. If Diagnostic B shows the inner loop has room to expand leaves into slack but no objective gradient to do so (the scalar rewards hitting target area but not exceeding it where slack exists), add a term/incentive so the ratio optimiser pushes leaf boundaries out to consume neighbouring slack and satisfy size, rather than parking at target.\n\nCONDITIONAL on Diagnostic B: build this only if B localizes the gap to the inner loop (room to expand, no gradient); if B shows construction targets too-small dims, prefer the plot-fill construction sibling. Must preserve the §5.4 inner-loop cliff / §4.9 lexicographic protection — the term sits where it cannot displace the fail-count ordering. A/B vs §12.2 baseline, seeds 0/1/2, 20000 evals, staged, default-OFF. Record DESIGN.md §13.6.","notes":"DEPRIORITISED by Diagnostic B (§13.2). B shows the inner loop CANNOT repair undersize: the slack is depth-driven maldistribution baked into the frozen topology, and the equal-offset ratio DOF cannot shrink a 14x leaf to feed a starved one without trading into shape fails (0.5^n cliff). Wrong DOF and wrong direction — the blocker is slicing POSITION, not a missing expansion reward. Fix belongs upstream in construction/topology (erc.4 re-scoped, erc.3). Keep as a low-priority follow-up only if a depth-balanced construction still leaves a residual size gradient the inner loop could pick up.","status":"closed","priority":4,"issue_type":"feature","owner":"bruno@postle.net","created_at":"2026-06-22T23:16:24Z","created_by":"Bruno Postle","updated_at":"2026-06-28T13:22:22Z","closed_at":"2026-06-28T13:22:22Z","close_reason":"wont-fix (DESIGN §13.7): Diag B (§13.2) showed the inner loop cannot repair undersize (wrong DOF — slicing position, frozen-topology ratios). Superseded by depth-balanced construction (erc.4). Condition unmet.","dependencies":[{"issue_id":"homemaker-py-erc.6","depends_on_id":"homemaker-py-erc","type":"parent-child","created_at":"2026-06-23T00:16:23Z","created_by":"Bruno Postle","metadata":"{}"},{"issue_id":"homemaker-py-erc.6","depends_on_id":"homemaker-py-erc.2","type":"blocks","created_at":"2026-06-23T00:16:47Z","created_by":"Bruno Postle","metadata":"{}"}],"dependency_count":1,"dependent_count":0,"comment_count":0}
 {"id":"homemaker-py-erc.5","title":"Experiment: compactness-aware cuts (minimize leaf perimeter/area)","description":"Attacks the #1 factor, crinkliness (346) — a per-leaf perimeter/area property DISTINCT from proportion (aspect ratio). Proportion-aware seeding (leu.2) sizes splits but does not bias toward balanced, square-ish subdivision. Add a KD-tree-style 'keep both children compact' cut rule (prefer the cut orientation/position that minimises summed child perimeter/area) in construction.\n\nCONDITIONAL on Diagnostic A: if A shows per-leaf shape-fail is FLAT across densities (floor intrinsic to slicing density), better cuts at the same leaf count will not pay → this should be closed wont-fix in favour of leaf-sharing. Only build if A shows shape-fail RISES with density. A/B vs §12.2 baseline, seeds 0/1/2, 20000 evals, staged, default-OFF. Record DESIGN.md §13.5.","notes":"DEPRIORITISED by erc.1 verdict (§13.1): per-leaf shape-fail flat vs slicing density and cuts already squarest (_size_divisions_from_targets picks squarest rotation) yet still ~1.8 fails/leaf =\u003e little compactness headroom at fixed leaf count. Floor is intrinsic to leaf COUNT, not cut quality. Revisit only if leaf-sharing (erc.3) underdelivers.","status":"closed","priority":4,"issue_type":"feature","owner":"bruno@postle.net","created_at":"2026-06-22T23:16:21Z","created_by":"Bruno Postle","updated_at":"2026-06-28T13:22:17Z","closed_at":"2026-06-28T13:22:17Z","close_reason":"wont-fix (DESIGN §13.7): Diag A (§13.1) showed the floor is intrinsic to leaf COUNT not cut quality; revisit condition was 'only if leaf-sharing underdelivers' but leaf-sharing OVER-delivered (−32…−39%, §13.3). Condition unmet.","dependencies":[{"issue_id":"homemaker-py-erc.5","depends_on_id":"homemaker-py-erc","type":"parent-child","created_at":"2026-06-23T00:16:21Z","created_by":"Bruno Postle","metadata":"{}"},{"issue_id":"homemaker-py-erc.5","depends_on_id":"homemaker-py-erc.1","type":"blocks","created_at":"2026-06-23T00:16:43Z","created_by":"Bruno Postle","metadata":"{}"}],"dependency_count":1,"dependent_count":0,"comment_count":0}
 {"id":"homemaker-py-2g5","title":"Rebuild occlusion/daylight/sun subsystem in Python (post-Phase-5, after optimisation fully native)","description":"DESIGN.md §6 port scope — a whole subsystem, not a term. quality_daylight (Leaf.pm:281-296) needs Urb::Misc::Sun + Urb::Field::Occlusion (+CIESky); quality_uncrinkliness also takes the occlusion object. Indoor spaces return 1 for daylight; cost is outdoor spaces + crinkliness. Port Sun_horizontal (262980-minute normalisation) and the occlusion wall set from Dom-\u003eWalls.","acceptance_criteria":"Daylight and crinkliness factors match Perl (float tolerance) across the corpus, including multi-storey cases","notes":"Re-scoped 2026-06-12: occlusion disabled in the Urb oracle instead of ported (see homemaker-py-gp2). Native fitness ships with simple crinkliness (illumination factor = 1, in homemaker-py-gnw). This issue is now the eventual Python occlusion rebuild, only after optimisation works entirely in Python. Restores outdoor-daylight and shaded-wall selection pressure.\nReframed 2026-06-17: orthogonal to epic homemaker-py-c4c. This is fitness FIDELITY (restoring daylight + shaded-wall selection pressure to match Perl), not search CAPABILITY — it changes what 'good' means, not the search's ability to find good. It will NOT improve final designs in the sense currently sought. Stays P4, deferred until the topology-search-quality epic lands and optimisation is fully native.","status":"open","priority":4,"issue_type":"feature","owner":"bruno@postle.net","created_at":"2026-06-11T23:38:25Z","created_by":"Bruno Postle","updated_at":"2026-06-17T19:14:48Z","dependency_count":0,"dependent_count":0,"comment_count":0}
-{"_type":"memory","key":"experiment-harness-gotcha-the-leaf-sharing-relaxed-objective","value":"Experiment harness gotcha: the leaf-sharing RELAXED objective (§13.3) is injected ONLY by monkeypatching fitness.load_config in the parent process (run_staged_search.py / probe scripts). This is parent-process-only and does NOT propagate into ProcessPoolExecutor workers (n_workers\u003e1), which re-import fitness fresh and score under the STRICT on-disk patterns.config -\u003e r.n_fails MISMATCH (worker strict vs parent relaxed re-score). ALL §13.x floor runs were therefore SERIAL. Any future PARALLEL leaf-sharing experiment will silently mis-score until leaf_sharing lives on disk/CLI (tracked: homemaker-py-x3b). The parallel driver itself is correct; both paths score via load_config(programme_dir)."}
+{"_type":"memory","key":"ld2-13-6-interior-o-seed-diagnostic-all","value":"ld2/§13.6 interior-O seed diagnostic: ALL crinkliness fails in the constructed bal+share seed are UNDER-exposed (crink\u003c0.62, landlocked rooms with no facade + no uncovered-O neighbour) — zero over-exposed sliver fails. So the erc crinkliness residual is genuine under-daylighting, validating the interior light-well premise. Default outside_divisor=6 was too sparse (null: harbor 147-\u003e142, crinkliness even rose). odiv=3 is the seed-optimal joint setting: harbor seed fails 147-\u003e129 (-18), maple 219-\u003e206 (-14), landlocked fails drop, at cost of more leaves (harbor +4, maple +8). Because it ADDS leaves it carries the §13.4 wash-out risk; A/B to convergence pending."}
 {"_type":"memory","key":"strategy-decision-2026-06-12-bruno-occlusion-daylight","value":"Strategy decision 2026-06-12 (Bruno): occlusion/daylight is ORTHOGONAL to building a scalable optimiser. Disable it in Urb (env flag, homemaker-py-gp2) rather than port it; native fitness uses simple crinkliness (illumination factor = 1); rebuild occlusion in Python only after optimisation is fully native (homemaker-py-2g5, now P4). Consequence: all scores change when the flag flips — re-baseline corpus/.score, DESIGN \\$4.5 gains, gate bars at one clean boundary AFTER homemaker-py-1p0 closes; Phase-2 urb-evolve benchmark must run with the same flag."}
 {"_type":"memory","key":"cli-tool-style-prefer-python-m-homemaker-module","value":"CLI tool style: prefer python -m homemaker.module --parameters pattern, installable via pip install -e . with pyproject.toml entry_points. Not standalone bin/ scripts."}
 {"_type":"memory","key":"collapse-global-94g-and-any-label-usage-optimisation","value":"collapse_global (94g) and any label/usage optimisation CANNOT fix geometry-intrinsic fails. The harbor-house 15-fail best layout contains long-thin cells that are useless whatever room usage is assigned — their width/proportion/crinkliness fails are shape-bound, not label slack. Two consequences: (1) do not over-claim collapse gains — only ~2-3 of that layout's fails are reclaimable relabel slack, the rest are geometry- or building-level bound; (2) the threshold objective must not be tuned to 'pass' a degenerate cell via a permissive room type — a metric-pass on a physically useless space is gaming, not a fix. Real remedies for these are geometry/topology search (cell shape) and circulation placement, filed separately, not the collapse."}
 {"_type":"memory","key":"deceptive-valleys-in-topology-search-when-every-single","value":"Deceptive valleys in topology search: when every single-step mutation from a target state passes through a high-fail intermediary (e.g. level_fix displaces a room into 5+ new fails), a compound operator that atomically applies two coordinated changes can escape. Design compound operators to land on the low-fail state directly, bypassing the deceptive gradient. Programme-house example: level_compound_fix atomically moves the level-constrained room AND re-inserts the displaced room adjacent to C in one step (operators.py, 2026-06-14)."}
 {"_type":"memory","key":"run-to-run-reproducibility-in-homemaker-layout-serial","value":"Run-to-run reproducibility in homemaker-layout: serial search (workers=1) is byte-for-byte deterministic; parallel (workers\u003e1) is now deterministic too AFTER fixing driver._run_batch to admit futures in submission order (was as_completed/completion order, bug xcy). Reproducibility holds only for a FIXED worker count — serial vs parallel differ because children-per-iteration is 1 vs n_workers (different batch granularity), which is expected, not a bug. The constructive seeder was NEVER nondeterministic: _assign_adjacency_aware has unique idx tiebreaks; comparing topologies with Python builtin hash() of the signature STRING is invalid (PYTHONHASHSEED salts str hashing per process) — use a stable hash (sha1) or genome.signature equality."}
 {"_type":"memory","key":"9o5-multi-use-leaves-is-path-a-superposition","value":"9o5 multi-use leaves is path (a) — superposition as SEARCH RELAXATION that COLLAPSES to specific usage at the end, NOT path (b) loose-fit/no-collapse. Bruno's intent: codes with SIMILAR leaf requirements form an interchangeable equivalence class; during evolution the solver doesn't commit which leaf serves which specific usage (smoother landscape, no fighting over exact leaf usage); at the end the layout is CONDENSED to specific usages by brute-forcing the in-class assignment (3 interchangeable usages over 3 leaves = 3! = 6 combinations to check, pick best). 'Derive automatically' compatibility = requirement-similarity grouping. This reverses the issue's stated 'path b preferred' note."}
 {"_type":"memory","key":"correction-to-urb-fitness-bug-memory-bruno-2026","value":"CORRECTION to urb-fitness-bug memory (Bruno, 2026-06-12): 'C' is NOT a 'covered' type — Is_Covered is a geometric predicate (indoor space above). Urb's generic types are canonically UPPERCASE: C=circulation, O=outside, S=sahn (get_space_types qw/C O S/; corpus is 100% uppercase, never 'c'/'o' leaves). The mixed-case designs that fired the latent ratio_type first-match bug were created by homemaker's own operator type pool emitting lowercase 'c'/'o' — fixed: driver/operators now emit uppercase generics only, and class checks use t[0].lower() in 'cos'. The Urb class-sum patch stays as defensive hardening (zero impact on canonical designs). Native port (3y7/gnw): treat type classes case-insensitively, generics canonically uppercase."}
 {"_type":"memory","key":"homemaker-py-pythonpath-set-pythonpath-home-bruno-src","value":"homemaker-layout PYTHONPATH: package installed as 'homemaker-layout' via pip install -e . so 'import homemaker_layout' works from anywhere without PYTHONPATH. For running tests use 'python -m pytest' from project root /home/bruno/src/homemaker-layout (pyproject.toml adds src/ automatically). Never try pip show homemaker — that's the old homemaker-addon conflict."}
 {"_type":"memory","key":"urb-fitness-bug-found-fixed-2026-06-12","value":"Urb fitness bug found+fixed 2026-06-12 (patch in /home/bruno/src/urb, uncommitted): ProgrammeDriven.pm ratio_o/ratio_type grepped case-insensitively over the ratios hash and took the FIRST key — nondeterministic (x4.5 score swings) for designs with mixed-case type classes (both 'c' circulation and 'C' covered). Fixed to SUM the class (matches Is_Circulation//Is_Outside semantics); 35/35 corpus scores unchanged. CRITICAL for homemaker-py-3y7/gnw: the native port must implement class-SUM ratios. Building.pm has the same unpatched pattern (site-driven path, not used by our oracle). Also: the memetic search reward-hacked this bug before the fix — search results predating it are noise artifacts."}
 {"_type":"memory","key":"urb-oracle-nondeterminism-urb-fitness-pl-output-varies","value":"Urb oracle nondeterminism: urb-fitness.pl output varies run-to-run from Perl hash-order randomisation — .fails line ORDER shuffles (compare sorted, use oracle.Score.fail_lines) and the score float can flip by ~1 ULP (compare with math.isclose rel_tol=1e-12, never ==). Not a batching artifact; affects single runs too. Matters for the Phase 3 native-fitness parity gate (homemaker-py-uxz)."}
 {"_type":"memory","key":"homemaker-py-3l6-fix-leaf-sharing-evolve-runs","value":"homemaker-py-3l6 fix: leaf-sharing evolve runs now auto-finish before write via driver.polish_finish — unfold_shared_leaves() then a warm-started leaf_sharing=False polish search (--polish-budget, default budget//2). Makes the written .dom honest under canonical homemaker-fitness (internal==canonical when leaf_sharing off). Interrupt path forces polish_budget=0 (unfold+rescore only). This is yaa's unfold-then-polish, made automatic; Schedule B annealing is still kpu."}
 {"_type":"memory","key":"island-model-psk-14-is-a-null-priming","value":"Island model (psk, §14) is a NULL: priming a population from N converged independent elites + crossover-heavy migration does not beat best-of-N at equal total budget (maple island 124 vs control 116). The child_probe instrument shows WHY: area-matched crossover across independently-converged elites almost never synthesizes (1-3 of ~64 children beat the better parent, max drop 2-5) because the slicing encoding is non-canonical (9gp), so splices are disruptive not combinatorial. Search-machinery null #3 after graded-objective and niching/restarts; residual stays geometry/shape-bound."}
 {"_type":"memory","key":"unfold-strategy-for-shared-leaves-homemaker-py-8iv","value":"Unfold strategy for shared leaves (homemaker-py-8iv, resolved 2026-07-16): use the BALANCED GRID (operators._grow_balanced/_size_subtree_equal), NOT circulation-aware slicing. Slicing a shared leaf perpendicular to its access edge so every child touches the corridor was implemented + A/B-tested and LOST decisively (150k-eval warm-start polish from evolved-3M: slice 41 fails/3.5e-14 vs grid 25 fails/2.4e-09, grid ahead at every milestone). Reason: k rooms all touching one wall are intrinsically thin slices; that geometric debt (proportion/long/width) is unfixable without topology change, while the grid's squarer children let local search re-route access cheaply via level_retype/place_missing/level_fix. Lesson: at the sharing-\u003eno-sharing transition, prioritise squarer children and leave access to local search; do not reintroduce slicing in Schedule B (kpu)."}
 {"_type":"memory","key":"user-preference-bruno-this-is-a-fedora-system","value":"User preference (Bruno): this is a Fedora system — NEVER install Python packages via pip without asking first; always ask whether to install the rpm via dnf (e.g. python3-cma) before considering pip. Applies to any dependency additions."}
 {"_type":"memory","key":"programme-house-optimisation-result-2026-06-14-15","value":"Programme-house optimisation result (2026-06-14/15): best achievable is 1 fail (l1 wrong level, score ~0.005). 0 fails is geometrically impossible: l1 (min 27m²) must occupy ll (~23m²) at level 0, which eliminates the t3-adj-C provider; dividing ll into lll(l1)+llr(C) gives llr proportion ~6:1 (fails). Python memetic optimizer achieves 1 fail in 50k evals vs Perl optimiser's 2-3 fails. Winning topology: TWO C nodes at level 0 — ll(C) for t3-adj-C via geometric contact, rl(C) for staircase via tree-sibling adjacency to rrr(O). Best .dom: scratch/from-warmstart-fixed.dom and scratch/from-compound3-fixed.dom."}
 {"_type":"memory","key":"proportion-aware-constructive-seeding-leu-2-12-2","value":"Proportion-aware constructive seeding (leu.2/§12.2): sizing seed cuts from target AREAS only regresses (thin slivers wreck aspect); you must ALSO pick each cut's rotation for child squareness. It is a convergence ACCELERATOR via a deeper local optimum around the constructed topology: wins where that topology is roughly right and budget is scarce (harbor -13%, maple -10% at 20k evals) but DELAYS small programmes where the seed must be restructured by undivide (programme-house regresses at fixed budget, yet reaches the floor given budget - speed, not asymptote). Default-on. Also: n_storeys must honour storey_minimum, not just level: keys (programme-house storey_minimum:2, all rooms level:0 - was seeded 1 storey short; cq1)."}
 {"_type":"memory","key":"never-use-corpus-filenames-candidate-001-dom-candidate","value":"Never use corpus filenames (candidate-001.dom, candidate-002.dom, generated.dom, init.dom, etc.) as --output targets when running experiments. These are test fixtures. Always write experimental outputs to scratch/ or a timestamped path. Lesson from 2026-06-14: warm-start runs overwrote candidate-001/002.dom and broke graph tests."}
 {"_type":"memory","key":"adjacency-in-binary-slicing-tree-is-structural-not","value":"Adjacency in binary slicing tree is structural, not geometric: the inner-loop NM cannot fix topological adjacency failures. Two paths exist: (1) tree-sibling adjacency — a node is adjacent to its sibling in the tree; (2) cross-zone geometric adjacency — leaves from different subtrees that happen to share a boundary. Staircase/adjacency fails require a topology mutation that changes which nodes are siblings or which zones touch. This was proved empirically on programme-house: staircase fail from rot=0 layout could not be fixed by NM but was fixed by level_retype creating a two-C topology (2026-06-14/15)."}
 {"_type":"memory","key":"experiment-seeding-pitfall-run-search-scaled-py-s","value":"Experiment seeding pitfall: run_search_scaled.py's default PH_SEED (c964…dom) is a FINISHED programme-house design — passing it warm-starts and floors at ~3 fails, NOT a blank-slate topology search. For blank-slate runs comparable to §11.5/§11.6 baselines, seed from examples/programme-house/init.dom (a bare undivided plot; driver bootstrap auto-triggers only on bare plots). Bit the 6zy sweep — first pass used c964 and falsely showed 3-fail floor across the whole grid."}
 {"_type":"memory","key":"multi-storey-staircase-consistency-when-dividing-or-retyping","value":"Multi-storey staircase consistency: when dividing or retyping a circulation (C) leaf at one level, the same structural change should be propagated to the matching leaf on ALL other storeys so the stair core path is maintained. The optimizer cannot fix staircase disruptions through trial-and-error geometry alone — it requires a synchronized multi-level operator that applies the same topology change to every storey simultaneously."}
 {"_type":"memory","key":"urb-fitness-bug-found-fixed-2026-06-12","value":"Urb fitness bug found+fixed 2026-06-12 (patch in /home/bruno/src/urb, uncommitted): ProgrammeDriven.pm ratio_o/ratio_type grepped case-insensitively over the ratios hash and took the FIRST key — nondeterministic (x4.5 score swings) for designs with mixed-case type classes (both 'c' circulation and 'C' covered). Fixed to SUM the class (matches Is_Circulation//Is_Outside semantics); 35/35 corpus scores unchanged. CRITICAL for homemaker-py-3y7/gnw: the native port must implement class-SUM ratios. Building.pm has the same unpatched pattern (site-driven path, not used by our oracle). Also: the memetic search reward-hacked this bug before the fix — search results predating it are noise artifacts."}
 {"_type":"memory","key":"proportion-aware-constructive-seeding-leu-2-12-2","value":"Proportion-aware constructive seeding (leu.2/§12.2): sizing seed cuts from target AREAS only regresses (thin slivers wreck aspect); you must ALSO pick each cut's rotation for child squareness. It is a convergence ACCELERATOR via a deeper local optimum around the constructed topology: wins where that topology is roughly right and budget is scarce (harbor -13%, maple -10% at 20k evals) but DELAYS small programmes where the seed must be restructured by undivide (programme-house regresses at fixed budget, yet reaches the floor given budget - speed, not asymptote). Default-on. Also: n_storeys must honour storey_minimum, not just level: keys (programme-house storey_minimum:2, all rooms level:0 - was seeded 1 storey short; cq1)."}
 {"_type":"memory","key":"9o5-multi-use-leaves-is-path-a-superposition","value":"9o5 multi-use leaves is path (a) — superposition as SEARCH RELAXATION that COLLAPSES to specific usage at the end, NOT path (b) loose-fit/no-collapse. Bruno's intent: codes with SIMILAR leaf requirements form an interchangeable equivalence class; during evolution the solver doesn't commit which leaf serves which specific usage (smoother landscape, no fighting over exact leaf usage); at the end the layout is CONDENSED to specific usages by brute-forcing the in-class assignment (3 interchangeable usages over 3 leaves = 3! = 6 combinations to check, pick best). 'Derive automatically' compatibility = requirement-similarity grouping. This reverses the issue's stated 'path b preferred' note."}
 {"_type":"memory","key":"cli-tool-style-prefer-python-m-homemaker-module","value":"CLI tool style: prefer python -m homemaker.module --parameters pattern, installable via pip install -e . with pyproject.toml entry_points. Not standalone bin/ scripts."}
 {"_type":"memory","key":"unfold-strategy-for-shared-leaves-homemaker-py-8iv","value":"Unfold strategy for shared leaves (homemaker-py-8iv, resolved 2026-07-16): use the BALANCED GRID (operators._grow_balanced/_size_subtree_equal), NOT circulation-aware slicing. Slicing a shared leaf perpendicular to its access edge so every child touches the corridor was implemented + A/B-tested and LOST decisively (150k-eval warm-start polish from evolved-3M: slice 41 fails/3.5e-14 vs grid 25 fails/2.4e-09, grid ahead at every milestone). Reason: k rooms all touching one wall are intrinsically thin slices; that geometric debt (proportion/long/width) is unfixable without topology change, while the grid's squarer children let local search re-route access cheaply via level_retype/place_missing/level_fix. Lesson: at the sharing-\u003eno-sharing transition, prioritise squarer children and leave access to local search; do not reintroduce slicing in Schedule B (kpu)."}
 {"_type":"memory","key":"adjacency-in-binary-slicing-tree-is-structural-not","value":"Adjacency in binary slicing tree is structural, not geometric: the inner-loop NM cannot fix topological adjacency failures. Two paths exist: (1) tree-sibling adjacency — a node is adjacent to its sibling in the tree; (2) cross-zone geometric adjacency — leaves from different subtrees that happen to share a boundary. Staircase/adjacency fails require a topology mutation that changes which nodes are siblings or which zones touch. This was proved empirically on programme-house: staircase fail from rot=0 layout could not be fixed by NM but was fixed by level_retype creating a two-C topology (2026-06-14/15)."}
 {"_type":"memory","key":"ld2-13-6-interior-o-seed-diagnostic-all","value":"ld2/§13.6 interior-O seed diagnostic: ALL crinkliness fails in the constructed bal+share seed are UNDER-exposed (crink\u003c0.62, landlocked rooms with no facade + no uncovered-O neighbour) — zero over-exposed sliver fails. So the erc crinkliness residual is genuine under-daylighting, validating the interior light-well premise. Default outside_divisor=6 was too sparse (null: harbor 147-\u003e142, crinkliness even rose). odiv=3 is the seed-optimal joint setting: harbor seed fails 147-\u003e129 (-18), maple 219-\u003e206 (-14), landlocked fails drop, at cost of more leaves (harbor +4, maple +8). Because it ADDS leaves it carries the §13.4 wash-out risk; A/B to convergence pending."}
 {"_type":"memory","key":"run-to-run-reproducibility-in-homemaker-layout-serial","value":"Run-to-run reproducibility in homemaker-layout: serial search (workers=1) is byte-for-byte deterministic; parallel (workers\u003e1) is now deterministic too AFTER fixing driver._run_batch to admit futures in submission order (was as_completed/completion order, bug xcy). Reproducibility holds only for a FIXED worker count — serial vs parallel differ because children-per-iteration is 1 vs n_workers (different batch granularity), which is expected, not a bug. The constructive seeder was NEVER nondeterministic: _assign_adjacency_aware has unique idx tiebreaks; comparing topologies with Python builtin hash() of the signature STRING is invalid (PYTHONHASHSEED salts str hashing per process) — use a stable hash (sha1) or genome.signature equality."}
 {"_type":"memory","key":"homemaker-py-3l6-fix-leaf-sharing-evolve-runs","value":"homemaker-py-3l6 fix: leaf-sharing evolve runs now auto-finish before write via driver.polish_finish — unfold_shared_leaves() then a warm-started leaf_sharing=False polish search (--polish-budget, default budget//2). Makes the written .dom honest under canonical homemaker-fitness (internal==canonical when leaf_sharing off). Interrupt path forces polish_budget=0 (unfold+rescore only). This is yaa's unfold-then-polish, made automatic; Schedule B annealing is still kpu."}
 {"_type":"memory","key":"homemaker-py-pythonpath-set-pythonpath-home-bruno-src","value":"homemaker-layout PYTHONPATH: package installed as 'homemaker-layout' via pip install -e . so 'import homemaker_layout' works from anywhere without PYTHONPATH. For running tests use 'python -m pytest' from project root /home/bruno/src/homemaker-layout (pyproject.toml adds src/ automatically). Never try pip show homemaker — that's the old homemaker-addon conflict."}
 {"_type":"memory","key":"island-model-psk-14-is-a-null-priming","value":"Island model (psk, §14) is a NULL: priming a population from N converged independent elites + crossover-heavy migration does not beat best-of-N at equal total budget (maple island 124 vs control 116). The child_probe instrument shows WHY: area-matched crossover across independently-converged elites almost never synthesizes (1-3 of ~64 children beat the better parent, max drop 2-5) because the slicing encoding is non-canonical (9gp), so splices are disruptive not combinatorial. Search-machinery null #3 after graded-objective and niching/restarts; residual stays geometry/shape-bound."}
 {"_type":"memory","key":"collapse-global-94g-and-any-label-usage-optimisation","value":"collapse_global (94g) and any label/usage optimisation CANNOT fix geometry-intrinsic fails. The harbor-house 15-fail best layout contains long-thin cells that are useless whatever room usage is assigned — their width/proportion/crinkliness fails are shape-bound, not label slack. Two consequences: (1) do not over-claim collapse gains — only ~2-3 of that layout's fails are reclaimable relabel slack, the rest are geometry- or building-level bound; (2) the threshold objective must not be tuned to 'pass' a degenerate cell via a permissive room type — a metric-pass on a physically useless space is gaming, not a fix. Real remedies for these are geometry/topology search (cell shape) and circulation placement, filed separately, not the collapse."}
 {"_type":"memory","key":"correction-to-urb-fitness-bug-memory-bruno-2026","value":"CORRECTION to urb-fitness-bug memory (Bruno, 2026-06-12): 'C' is NOT a 'covered' type — Is_Covered is a geometric predicate (indoor space above). Urb's generic types are canonically UPPERCASE: C=circulation, O=outside, S=sahn (get_space_types qw/C O S/; corpus is 100% uppercase, never 'c'/'o' leaves). The mixed-case designs that fired the latent ratio_type first-match bug were created by homemaker's own operator type pool emitting lowercase 'c'/'o' — fixed: driver/operators now emit uppercase generics only, and class checks use t[0].lower() in 'cos'. The Urb class-sum patch stays as defensive hardening (zero impact on canonical designs). Native port (3y7/gnw): treat type classes case-insensitively, generics canonically uppercase."}
 {"_type":"memory","key":"user-preference-bruno-this-is-a-fedora-system","value":"User preference (Bruno): this is a Fedora system — NEVER install Python packages via pip without asking first; always ask whether to install the rpm via dnf (e.g. python3-cma) before considering pip. Applies to any dependency additions."}
 {"_type":"memory","key":"warm-x0-initialization-bug-pattern-when-a-topology","value":"warm_x0 initialization bug pattern: when a topology operator explicitly sets division ratios on a newly-created node (e.g. compound_fix sets node.division=[0.25,0.25] for t3), parent.ratios has no entry for that node (it was a leaf). warm_x0 defaults it to 0.5, corrupting the inner loop's starting point and making the operator invisible to lex comparison. Fix: only propagate child ratios for nodes where the parent node was NOT already divided; stale hidden nodes revealed by structural mutations (swap flipping b.below) must NOT contribute their pre-writeback values. See driver.py lines 259-267 (fixed 2026-06-14)."}
 {"_type":"memory","key":"collapse-global-s-jacobi-adjacency-relaxation-homemaker-py","value":"collapse_global's Jacobi adjacency relaxation (homemaker-py-94g) is a synchronous per-round linear-assignment re-solve, which can 2-cycle indefinitely between two labellings that each satisfy ZERO adjacency requirements even though a permutation satisfying ALL of them exists -- proven on a minimal 4-cell chain (p1-q1-p2-q2, two disjoint adjacency pairs p1\u003c-\u003ep2/q1\u003c-\u003eq2) in test_two_opt_polish_escapes_jacobi_plateau. homemaker-py-9wi added Fitness._two_opt_adjacency_polish: a same-level pairwise-swap local search run after the Jacobi fixpoint, gated behind collapse_global(local_search=True) (default off, exposed as homemaker-collapse --local-search). Monotone by construction (a swap is kept only if it strictly increases total reward). Empirically on the 11 harbor-house evolved-*.dom/3m.dom/materialised-3M.dom layouts: 10 matched Jacobi-only exactly, 0 regressed, and evolved-anneal-3M.dom improved 21-\u003e19 fails (fixed a genuine mutual da1\u003c-\u003ek1 adjacency miss the Jacobi loop couldn't reach)."}
-{"_type":"memory","key":"never-use-corpus-filenames-candidate-001-dom-candidate","value":"Never use corpus filenames (candidate-001.dom, candidate-002.dom, generated.dom, init.dom, etc.) as --output targets when running experiments. These are test fixtures. Always write experimental outputs to scratch/ or a timestamped path. Lesson from 2026-06-14: warm-start runs overwrote candidate-001/002.dom and broke graph tests."}
+{"_type":"memory","key":"experiment-harness-gotcha-the-leaf-sharing-relaxed-objective","value":"Experiment harness gotcha: the leaf-sharing RELAXED objective (§13.3) is injected ONLY by monkeypatching fitness.load_config in the parent process (run_staged_search.py / probe scripts). This is parent-process-only and does NOT propagate into ProcessPoolExecutor workers (n_workers\u003e1), which re-import fitness fresh and score under the STRICT on-disk patterns.config -\u003e r.n_fails MISMATCH (worker strict vs parent relaxed re-score). ALL §13.x floor runs were therefore SERIAL. Any future PARALLEL leaf-sharing experiment will silently mis-score until leaf_sharing lives on disk/CLI (tracked: homemaker-py-x3b). The parallel driver itself is correct; both paths score via load_config(programme_dir)."}
 {"_type":"memory","key":"programme-house-optimisation-result-2026-06-14-15","value":"Programme-house optimisation result (2026-06-14/15): best achievable is 1 fail (l1 wrong level, score ~0.005). 0 fails is geometrically impossible: l1 (min 27m²) must occupy ll (~23m²) at level 0, which eliminates the t3-adj-C provider; dividing ll into lll(l1)+llr(C) gives llr proportion ~6:1 (fails). Python memetic optimizer achieves 1 fail in 50k evals vs Perl optimiser's 2-3 fails. Winning topology: TWO C nodes at level 0 — ll(C) for t3-adj-C via geometric contact, rl(C) for staircase via tree-sibling adjacency to rrr(O). Best .dom: scratch/from-warmstart-fixed.dom and scratch/from-compound3-fixed.dom."}
 {"_type":"memory","key":"deceptive-valleys-in-topology-search-when-every-single","value":"Deceptive valleys in topology search: when every single-step mutation from a target state passes through a high-fail intermediary (e.g. level_fix displaces a room into 5+ new fails), a compound operator that atomically applies two coordinated changes can escape. Design compound operators to land on the low-fail state directly, bypassing the deceptive gradient. Programme-house example: level_compound_fix atomically moves the level-constrained room AND re-inserts the displaced room adjacent to C in one step (operators.py, 2026-06-14)."}
 {"_type":"memory","key":"strategy-decision-2026-06-12-bruno-occlusion-daylight","value":"Strategy decision 2026-06-12 (Bruno): occlusion/daylight is ORTHOGONAL to building a scalable optimiser. Disable it in Urb (env flag, homemaker-py-gp2) rather than port it; native fitness uses simple crinkliness (illumination factor = 1); rebuild occlusion in Python only after optimisation is fully native (homemaker-py-2g5, now P4). Consequence: all scores change when the flag flips — re-baseline corpus/.score, DESIGN \\$4.5 gains, gate bars at one clean boundary AFTER homemaker-py-1p0 closes; Phase-2 urb-evolve benchmark must run with the same flag."}
 {"_type":"memory","key":"urb-oracle-nondeterminism-urb-fitness-pl-output-varies","value":"Urb oracle nondeterminism: urb-fitness.pl output varies run-to-run from Perl hash-order randomisation — .fails line ORDER shuffles (compare sorted, use oracle.Score.fail_lines) and the score float can flip by ~1 ULP (compare with math.isclose rel_tol=1e-12, never ==). Not a batching artifact; affects single runs too. Matters for the Phase 3 native-fitness parity gate (homemaker-py-uxz)."}
--- a/DESIGN.md
+++ b/DESIGN.md
@ -4005,3 +4005,89 @@ faster in wall-clock/eval terms than flat lex at the SAME budget (this A/B
 measured fail composition at fixed budget, not convergence speed); an
 apples-to-apples "evals to 0 hard fails" race is a natural follow-up once
 `2g7.1`/`2g7.2` ground truth lands.
 ### 37.2 `homemaker-py-2g7.4` shape-curve DP prototype — measured 2026-08-02, ACCEPTANCE: PASS
 **What was built.** `experiments/shapecurve_spike.py` + `experiments/
 validate_shapecurve.py`: an Otten/Stockmeyer-style shape-curve DP answering
 "does some equal-offset ratio assignment clear the size/width/proportion
 FAIL_THRESHOLD for every leaf" in one bottom-up pass, for a frozen topology on
 harbor-house-l0. Each leaf's feasible (width, height) region is bounded by an
 area hyperbola, a min-width line, and an aspect-ratio wedge — closed-form
 FAIL_THRESHOLD inversions of `quality_size`/`quality_width`/
 `quality_proportion` (`leaf_constraints`, verified against the real Gaussian
 formulas by construction, not reimplemented magic numbers: same `conf`/
 `get_space_params` lookups `fitness.py` uses, including the "any type code
 starting with 'c' or 's'/'o' hits the circulation/outside branch, not its own
 programme params" quirk — confirmed this is existing product behaviour, not a
 bug, by reading `get_space_params`/`quality_size` together). Regions compose
 bottom-up through the slicing tree: a node's cut is either a "width-split"
 (children share height, widths sum) or "height-split" (heights sum), measured
 once from the actual baseline geometry (`_orientation`) rather than derived
 symbolically from `rotation` — robust to any rotation convention. Composition
 is done on a shared log-spaced grid (interval-sum + a numpy-vectorised
 inversion, `_invert`); leaf curves themselves are exact closed forms, so all
 discretisation error is confined to internal-node composition. A top-down
 `realise()` back-substitution converts a feasible root point into actual
 `division` ratios, so the DP's output is a real, scoreable `.dom` tree, not
 just a yes/no.
 **Explicit scope (per the plan's own caveats).** Only size/width/proportion
 is modelled — crinkliness/adjacency/access/level connectivity are graph
 terms, out of scope by design. Every quad is approximated by its axis-aligned
 bounding box (exact only for a true rectangle). `leaf_sharing`/`co_type`
 target-adjustment is not modelled (harbor-house-l0's programme doesn't
 exercise either).
 **Validation** (`experiments/validate_shapecurve.py`, harbor-house-l0, 200
 `driver.random_topology` topologies, 2-14 leaves, seed 12345): compared
 against NM search **minimising shape-fail count directly** (`ShapeFailEvaluator`,
 budget 100), not `innerloop.optimise`'s full aggregate objective — the first
 version of this harness used the full objective and found spurious
 "disagreements" where the DP's own realised point independently verified at
 **zero** shape fails but NM's full-objective search had wandered away from it,
 because on a topology missing most of its programme, the 0.5^n missing-space
 penalty swamps the objective and NM has no pressure to preserve
 shape-feasibility specifically. Minimising shape-fail count alone is the
 correct apples-to-apples comparison against what the DP claims to solve.
 | metric | result | target |
 |---|---|---|
 | agreement | 198/200 = **99.0%** | >= 95% |
 | false positives (DP feasible, NM can't reach 0) | 2 | — |
 | false negatives (DP infeasible, NM reaches 0 anyway) | **0** | — |
 | speedup (grid_n=150, vs 100-eval NM) | **93.6x** | >= 50x |
 | speedup (grid_n=300) | 42.7x | — |
 | plot-level bbox area error | measured **+7.5%** overestimate | quantified |
 Zero false negatives across 200 topologies: the DP never wrongly rejects a
 topology NM finds feasible — the safe direction for a pre-filter (worst case
 it fails to prune, never wrongly prunes a viable topology). grid_n=150 vs 300
 gave **identical** agreement (99.0%, the same 2 mismatches) at 2.2x the
 speedup — internal-node grid resolution has headroom below 300 with no
 measured accuracy cost on this benchmark; `_invert`'s pure-Python O(N²)
 double loop was ~70% of DP wall-clock before vectorising with numpy
 (profiled: 170ms → 40ms/topology at grid_n=300 from that change alone).
 **Approximation error, root-caused.** Both false positives were traced to the
 bounding-box approximation, not a DP logic bug: the DP's own realised point
 for both cases had one leaf whose bbox-approximated area (e.g. 29.54 m²,
 comfortably inside `[27.12, 52.88]`) was a **real skewed quad** whose true
 `geometry.area` (26.90 m²) fell just *below* the true lower bound — a bbox
 overestimate of the same ~8-12% magnitude as the plot-level +7.5% figure
 above (harbor-house-l0's plot is a near-rectangular trapezoid, not a true
 rectangle). Every mismatch occurred within one bbox-error-width of a boundary
 — exactly the failure mode the plan's caveat predicted ("equal-offset
 skew-quad geometry means DP areas are approximate — measure the approximation
 error on real plots first").
 **ACCEPTANCE: PASS** — all three criteria cleared (agreement, speedup,
 quantified approximation error). **Not done in this session** (follow-on,
 new bead needed before this can replace `operators.predicted_shape_fails` in
 `driver.py`'s real pre-filter path): wiring the DP into `driver._evaluate`/
 `innerloop.optimise` as an actual pre-filter + NM warm-start, multi-storey
 (`below`-link) support, `leaf_sharing`/`co_type` modelling, and a true
 skew-quad (non-bbox) leaf region to remove the measured approximation-error
 source rather than just quantify it. `experiments/shapecurve_spike.py` is
 kept as a reference/prototype (the §34 `autodiff_spike.py` precedent), not
 wired into `innerloop.py`.
--- a/experiments/shapecurve_spike.py
+++ b/experiments/shapecurve_spike.py
@ -0,0 +1,373 @@
 """Spike (homemaker-py-2g7.4): Otten/Stockmeyer shape-curve DP vs nm_search.
 Motivation (DESIGN.md §37, plan point 2): the inner loop answers "does some
 equal-offset ratio assignment clear the size/width/proportion FAIL_THRESHOLD
 for every leaf" by an 80-200 eval Nelder-Mead search per topology. The
 classic slicing-floorplan result answers the size/width/proportion family of
 this question EXACTLY in one bottom-up pass: each leaf's feasible (width,
 height) region is bounded by an area hyperbola (``quality_size``), a min-width
 line (``quality_width``), and an aspect-ratio wedge (``quality_proportion``) --
 all three are FAIL_THRESHOLD-inversions of the Gaussian/clipped-Gaussian
 factors in ``fitness.py`` (see ``leaf_constraints`` below). These per-leaf
 regions compose bottom-up through the slicing tree: a "width-split" node
 (children share height, widths sum) or "height-split" node (children share
 width, heights sum) -- see ``_orientation``.
 Approximations made explicit (the plan's caveats, DESIGN.md §37 point 2):
  * Every quad (leaf or internal) is approximated by its axis-aligned
    bounding-box (w, h) -- exact only for a true rectangle; harbor-house-l0's
    plot is a near-rectangular trapezoid (DESIGN.md says "harbor plot is a
    near-rect quad"), so this is the intended first target, not a general
    solution for skew quads.
  * A node's cut orientation (does it split width or height?) is measured
    once from the ACTUAL geometry at ratio=0.5 baseline, not derived from
    ``rotation`` symbolically -- robust to any rotation convention, but a
    property of the *frozen topology*, computed once, not re-derived by the
    DP itself.
  * Leaf curves are EXACT closed forms (hyperbola/line/wedge intersection --
    no discretisation error). Internal-node composition is done on a shared
    discretised grid (log-spaced) with linear interpolation -- this is where
    approximation error enters, and is quantified in ``validate.py``.
  * ``leaf_sharing``/``co_type`` (multi-use leaves) target-adjustment is NOT
    modelled -- ``leaf_constraints`` uses each leaf's own type's base params
    only. harbor-house-l0's programme does not exercise these, so this is a
    scoping simplification, not a validated-safe omission for programmes that
    do.
 Only the size/width/proportion family is modelled -- crinkliness, access,
 adjacency, level/vertical connectivity are graph/topology terms, not per-leaf
 shape, and are explicitly out of scope (DESIGN.md §37 point 2 caveat).
 """
 from __future__ import annotations
 import math
 import warnings
 from dataclasses import dataclass
 import numpy as np
 from homemaker_layout import dom as dom_mod
 from homemaker_layout import geometry
 # sqrt(2*ln(10)): FAIL_THRESHOLD=0.1 inversion of a unit-height Gaussian,
 # gaussian(x,1,target,sigma) >= 0.1  <=>  |x-target| <= K*sigma.
 _K = math.sqrt(2.0 * math.log(10.0))
 Interval = tuple[float, float] | None  # None = infeasible
 def _interval_add(a: Interval, b: Interval) -> Interval:
    if a is None or b is None:
        return None
    return (a[0] + b[0], a[1] + b[1])
 # --------------------------------------------------------------------------- #
 # Per-leaf feasible region (exact closed form; FAIL_THRESHOLD inversion of
 # fitness.py's quality_size/quality_width/quality_proportion).
 # --------------------------------------------------------------------------- #
@dataclass
 class LeafBounds:
    amin: float
    amax: float
    wmin: float
    rmax: float  # max aspect ratio (>= 1)
    def h_range(self, w: float) -> Interval:
        if w < self.wmin - 1e-12:
            return None
        lo = self.wmin
        if self.amin > 0:
            lo = max(lo, self.amin / w)
        lo = max(lo, w / self.rmax)
        hi = w * self.rmax
        if self.amax < math.inf:
            hi = min(hi, self.amax / w)
        if lo > hi + 1e-12:
            return None
        return (lo, hi)
    def w_range(self, h: float) -> Interval:
        # symmetric in (w, h) -- same box+hyperbola+wedge shape.
        return self.h_range(h)
    def range_grid(self, grid: np.ndarray) -> list[Interval]:
        """Vectorised ``h_range``/``w_range`` (symmetric) over a whole grid."""
        lo = np.maximum(self.wmin, grid / self.rmax)
        if self.amin > 0:
            lo = np.maximum(lo, self.amin / grid)
        hi = grid * self.rmax
        if self.amax < math.inf:
            hi = np.minimum(hi, self.amax / grid)
        feasible = (grid >= self.wmin - 1e-12) & (lo <= hi + 1e-12)
        return [(float(lo[i]), float(hi[i])) if feasible[i] else None for i in range(len(grid))]
 def leaf_constraints(fit, leaf: dom_mod.Node) -> LeafBounds:
    """FAIL_THRESHOLD-inverted (amin, amax, wmin, rmax) for one leaf.
    Mirrors the branching of ``Fitness.quality_size``/``quality_width``/
    ``quality_proportion`` (fitness.py) but returns the (target, sigma)-derived
    hard bounds instead of evaluating a Gaussian against actual geometry.
    Ignores leaf-sharing/co_type target adjustment (see module docstring).
    """
    t0 = leaf.type[0].lower() if leaf.type else ""
    # --- size -> (amin, amax) ---
    if t0 in ("o", "s"):
        amin, amax = 0.0, math.inf
    else:
        params = fit.conf("size_circulation") if t0 == "c" else fit.get_space_params(leaf.type, "size")
        target, sigma = params[0], params[1]
        # NB: quality_size's ``target > 0`` gate governs only the leaf-sharing/
        # co_type k-scaling of (target, sigma) (not modelled here, see module
        # docstring) -- the underlying gaussian(area, target, sigma) test
        # always applies, including target==0 (e.g. size_circulation's [0.0,
        # 14.0] default: a real one-sided "as small as possible" constraint,
        # not "unconstrained").
        amin, amax = max(0.0, target - _K * sigma), target + _K * sigma
    # --- width -> wmin ---
    if (
        t0 in ("o", "s")
        and not dom_mod.is_covered(leaf)
        and not dom_mod.is_supported(leaf)
        and dom_mod.level_of(leaf)
    ):
        wmin = 0.0
    else:
        if t0 in ("o", "s"):
            params = fit.conf("width_outside")
        elif t0 == "c":
            params = fit.conf("width_circulation")
        else:
            params = fit.get_space_params(leaf.type, "width")
        target, sigma = params[0], params[1]
        wmin = max(0.0, target - _K * sigma)
    # --- proportion -> rmax ---
    if t0 in ("o", "s"):
        params = fit.conf("proportion_outside")
    elif t0 == "c":
        params = fit.conf("proportion_circulation")
    else:
        params = fit.get_space_params(leaf.type, "proportion")
    target, sigma = params[0], params[1]
    rmax = max(1.0 + 1e-9, target + _K * sigma)
    return LeafBounds(amin=amin, amax=amax, wmin=wmin, rmax=rmax)
 # --------------------------------------------------------------------------- #
 # Bounding-box geometry + cut-orientation detection (rectangular approximation)
 # --------------------------------------------------------------------------- #
 def _bbox(n: dom_mod.Node) -> tuple[float, float]:
    """Axis-aligned bounding-box (w, h) of a quad's 4 corners."""
    xs = [geometry.coordinate(n, i)[0] for i in range(4)]
    ys = [geometry.coordinate(n, i)[1] for i in range(4)]
    return (max(xs) - min(xs), max(ys) - min(ys))
 def _orientation(node: dom_mod.Node) -> str:
    """'w' (width-split, children share height) or 'h' (height-split),
    measured from the actual baseline geometry -- see module docstring."""
    bw, bh = _bbox(node)
    lw, lh = _bbox(node.left)
    rw, rh = _bbox(node.right)
    err_w = abs((lw + rw) - bw)
    err_h = abs((lh + rh) - bh)
    return "w" if err_w <= err_h else "h"
 def annotate_orientations(level_root: dom_mod.Node) -> dict[int, str]:
    """Baseline-geometry orientation per internal node, keyed by id(node).
    Sets every free branch's division to [0.5, 0.5] on the LIVE tree (matching
    the inner loop's cold-start convention), clears the geometry cache, then
    measures. Caller must re-clear the cache afterwards if it goes on to use
    different ratios (the DP itself never reads real coordinates again after
    this call -- only the plot bbox, computed separately).
    """
    from homemaker_layout import solver
    for b in solver._branches(level_root):
        if b.below is None or not b.below.divided:
            b.division = [0.5, 0.5]
    geometry.clear_cache()
    orientations: dict[int, str] = {}
    def _walk(n: dom_mod.Node) -> None:
        if not n.divided:
            return
        orientations[id(n)] = _orientation(n)
        _walk(n.left)
        _walk(n.right)
    _walk(level_root)
    return orientations
 # --------------------------------------------------------------------------- #
 # The DP itself
 # --------------------------------------------------------------------------- #
@dataclass
 class Curve:
    """A node's feasible region, both ways: w_of_h[i] is the feasible w-range
    at h=grid[i]; h_of_w[j] is the feasible h-range at w=grid[j]. Same shared
    grid at every node, so composition needs no cross-node interpolation."""
    w_of_h: list[Interval]
    h_of_w: list[Interval]
 def _interp_range(grid: np.ndarray, arr: list[Interval], x: float) -> Interval:
    if x <= grid[0]:
        return arr[0]
    if x >= grid[-1]:
        return arr[-1]
    j = int(np.searchsorted(grid, x)) - 1
    j = max(0, min(j, len(grid) - 2))
    a, b = arr[j], arr[j + 1]
    if a is None or b is None:
        return a if x - grid[j] < grid[j + 1] - x else b
    t = (x - grid[j]) / (grid[j + 1] - grid[j])
    return (a[0] * (1 - t) + b[0] * t, a[1] * (1 - t) + b[1] * t)
 def _invert(grid: np.ndarray, arr: list[Interval]) -> list[Interval]:
    """Given arr[i] = feasible cross-range at grid[i], return the inverse:
    inv[j] = {y : arr's cross-range at y contains grid[j]}, assumed contiguous
    in y (true for the monotonic hyperbola/line/wedge-composed regions this
    DP produces). O(N^2) but numpy-vectorised (the naive Python double loop
    was ~70% of total DP wall-clock, profiled on harbor-house-l0)."""
    lo_arr = np.array([r[0] if r is not None else np.nan for r in arr])
    hi_arr = np.array([r[1] if r is not None else np.nan for r in arr])
    # mask[i, j]: does grid[i]'s range contain grid[j]?
    mask = (lo_arr[:, None] - 1e-9 <= grid[None, :]) & (hi_arr[:, None] + 1e-9 >= grid[None, :])
    grid_masked = np.where(mask, grid[:, None], np.nan)
    any_feasible = mask.any(axis=0)
    with np.errstate(invalid="ignore"), warnings.catch_warnings():
        warnings.simplefilter("ignore", category=RuntimeWarning)
        inv_lo = np.where(any_feasible, np.nanmin(grid_masked, axis=0), np.nan)
        inv_hi = np.where(any_feasible, np.nanmax(grid_masked, axis=0), np.nan)
    return [None if np.isnan(lo) else (float(lo), float(hi)) for lo, hi in zip(inv_lo, inv_hi)]
 def make_grid(wmax: float, n: int = 400, wmin: float = 0.1) -> np.ndarray:
    return np.geomspace(wmin, wmax, n)
@dataclass
 class Feasibility:
    feasible: bool
    h_range_at_w: Interval
    w_range_at_h: Interval
 def check_feasible(root_curve: Curve, grid: np.ndarray, w_plot: float, h_plot: float) -> Feasibility:
    hr = _interp_range(grid, root_curve.h_of_w, w_plot)
    wr = _interp_range(grid, root_curve.w_of_h, h_plot)
    ok_h = hr is not None and hr[0] - 1e-6 <= h_plot <= hr[1] + 1e-6
    ok_w = wr is not None and wr[0] - 1e-6 <= w_plot <= wr[1] + 1e-6
    return Feasibility(feasible=bool(ok_h or ok_w), h_range_at_w=hr, w_range_at_h=wr)
 # --------------------------------------------------------------------------- #
 # Top-down back-substitution: realise one feasible point as division ratios.
 # --------------------------------------------------------------------------- #
 def realise(
    node: dom_mod.Node,
    curves: dict[int, tuple[Curve, Curve]],
    orientations: dict[int, str],
    grid: np.ndarray,
    w: float,
    h: float,
 ) -> None:
    """Write ``division`` on every free branch under ``node`` so its subtree
    realises the (w, h) target, given each descendant's precomputed curves.
    ``curves[id(n)] = (left_curve, right_curve)`` for internal nodes."""
    if not node.divided:
        return
    cl, cr = curves[id(node)]
    orient = orientations[id(node)]
    if orient == "w":
        rl = _interp_range(grid, cl.w_of_h, h)
        rr = _interp_range(grid, cr.w_of_h, h)
        lo = max(rl[0], w - rr[1])
        hi = min(rl[1], w - rr[0])
        wl = min(max((lo + hi) / 2.0, rl[0]), rl[1])
        wl = min(max(wl, w - rr[1]), w - rr[0])
        wr = w - wl
        t = wl / w if w > 0 else 0.5
        node.division = [t, t]
        realise(node.left, curves, orientations, grid, wl, h)
        realise(node.right, curves, orientations, grid, wr, h)
    else:
        rl = _interp_range(grid, cl.h_of_w, w)
        rr = _interp_range(grid, cr.h_of_w, w)
        lo = max(rl[0], h - rr[1])
        hi = min(rl[1], h - rr[0])
        hl = min(max((lo + hi) / 2.0, rl[0]), rl[1])
        hl = min(max(hl, h - rr[1]), h - rr[0])
        hr = h - hl
        t = hl / h if h > 0 else 0.5
        node.division = [t, t]
        realise(node.left, curves, orientations, grid, w, hl)
        realise(node.right, curves, orientations, grid, w, hr)
 def build_curves_with_children(
    node: dom_mod.Node, fit, orientations: dict[int, str], grid: np.ndarray,
    out: dict[int, tuple[Curve, Curve]],
 ) -> Curve:
    """Like ``build_curves`` but also records each internal node's (left,
    right) curves in ``out`` for ``realise`` to consume."""
    if not node.divided:
        b = leaf_constraints(fit, node)
        w_of_h = h_of_w = b.range_grid(grid)
        return Curve(w_of_h=w_of_h, h_of_w=h_of_w)
    cl = build_curves_with_children(node.left, fit, orientations, grid, out)
    cr = build_curves_with_children(node.right, fit, orientations, grid, out)
    out[id(node)] = (cl, cr)
    orient = orientations[id(node)]
    if orient == "w":
        w_of_h = [_interval_add(cl.w_of_h[i], cr.w_of_h[i]) for i in range(len(grid))]
        h_of_w = _invert(grid, w_of_h)
    else:
        h_of_w = [_interval_add(cl.h_of_w[j], cr.h_of_w[j]) for j in range(len(grid))]
        w_of_h = _invert(grid, h_of_w)
    return Curve(w_of_h=w_of_h, h_of_w=h_of_w)
 def solve(level_root: dom_mod.Node, fit, grid_n: int = 150) -> tuple[bool, dict]:
    """End-to-end: orientation-annotate, compute plot bbox, build curves,
    check root feasibility, and (if feasible) write realising ratios in
    place. Returns (feasible, info) where info carries timing-relevant
    intermediates for the caller."""
    orientations = annotate_orientations(level_root)
    w_plot, h_plot = _bbox(level_root)
    grid = make_grid(max(w_plot, h_plot) * 1.2, n=grid_n)
    curves_by_node: dict[int, tuple[Curve, Curve]] = {}
    root_curve = build_curves_with_children(level_root, fit, orientations, grid, curves_by_node)
    feas = check_feasible(root_curve, grid, w_plot, h_plot)
    if feas.feasible:
        realise(level_root, curves_by_node, orientations, grid, w_plot, h_plot)
        geometry.clear_cache()
    return feas.feasible, {
        "w_plot": w_plot, "h_plot": h_plot, "orientations": orientations,
        "grid": grid, "h_range_at_w": feas.h_range_at_w, "w_range_at_h": feas.w_range_at_h,
    }
--- a/experiments/validate_shapecurve.py
+++ b/experiments/validate_shapecurve.py
@ -0,0 +1,152 @@
 """Validation harness for shapecurve_spike.py (homemaker-py-2g7.4).
 For N random harbor-house-l0 topologies: run the DP feasibility check and an
 NM search that directly MINIMISES the shape-fail count (size/width/
 proportion FAIL_THRESHOLD family only -- crinkliness/adjacency/level/etc are
 out of scope for this DP, see shapecurve_spike.py's module docstring), and
 compare verdicts.
 NB: the first version of this harness polished against innerloop.optimise's
 FULL aggregate fitness (missing-space/adjacency/etc fails included) and found
 many "false positives" -- but a directed check showed the DP's own realised
 ratios genuinely score 0 shape fails in those cases; NM's full-objective
 search had simply wandered away from that point, because on a topology
 missing most of its programme (a `random_topology`-grown tree rarely places
 all 10 codes), the 0.5^n missing-space penalty swamps the objective and NM
 has no gradient pressure to preserve shape-feasibility specifically. Scoring
 by shape-fail-count ALONE is the correct apples-to-apples comparison against
 what the DP claims to solve.
 Usage: python experiments/validate_shapecurve.py [n_topologies] [nm_budget]
 """
 from __future__ import annotations
 import copy
 import sys
 import time
 import numpy as np
 from homemaker_layout import dom, driver, fitness as fit_mod, geometry, innerloop
 sys.path.insert(0, "experiments")
 import shapecurve_spike as sc  # noqa: E402
 PROGRAMME_DIR = "examples/harbor-house-l0"
 _SHAPE_SUFFIXES = (" size", " width", " proportion")
 class ShapeFailEvaluator(innerloop.NativeEvaluator):
    """Like NativeEvaluator, but ``evaluate`` scores -n_shape_fails (ties
    broken by the real fitness) so nm_search's greedy hill-climb directly
    minimises the shape-fail count instead of the full aggregate objective."""
    def evaluate(self, xs):
        results = []
        for x in xs:
            self.apply(x)
            root_copy = copy.deepcopy(self.root)
            score, fails = self._fit.score_with_fails(root_copy)
            n_shape = sum(1 for f in fails if f.endswith(_SHAPE_SUFFIXES))
            proxy_fitness = -n_shape + min(score, 0.999)  # tie-break, sub-1 so it never crosses a fail-count boundary
            results.append(innerloop._NativeScore(fitness=proxy_fitness, fail_lines=fails))
        self.n_evals += len(xs)
        self.n_oracle_calls += 1
        return results
 def main(n_topologies: int = 200, nm_budget: int = 100, grid_n: int = 150) -> None:
    seed_root = dom.load(f"{PROGRAMME_DIR}/init.dom")
    conf, cost = fit_mod.load_config(PROGRAMME_DIR)
    fit = fit_mod.Fitness(conf, cost)
    types = sorted(fit.spaces.keys())
    rng = np.random.default_rng(12345)
    n_agree = 0
    n_dp_feasible = 0
    n_nm_feasible = 0
    false_positive = 0  # DP says feasible, NM finds a shape fail
    false_negative = 0  # DP says infeasible, NM reaches 0 shape fails anyway
    dp_time_total = 0.0
    nm_time_total = 0.0
    rows = []
    i = 0
    attempts = 0
    while i < n_topologies:
        attempts += 1
        n_leaves = int(rng.integers(2, 15))
        seed = int(rng.integers(0, 2**31 - 1))
        trng = np.random.default_rng(seed)
        topo = driver.random_topology(seed_root, n_leaves, trng, types)
        dom._link(topo)
        lvl = dom.levels(topo)[0]
        if len(lvl.leaves()) < 2:
            continue  # undivided, nothing for the DP to do
        i += 1
        t0 = time.time()
        try:
            dp_ok, info = sc.solve(lvl, fit, grid_n=grid_n)
        except Exception as exc:  # noqa: BLE001 -- record and keep going
            dp_ok, info = None, {"error": repr(exc)}
        dp_time = time.time() - t0
        dp_time_total += dp_time
        t0 = time.time()
        topo_nm = copy.deepcopy(topo)
        geometry.clear_cache()
        with ShapeFailEvaluator(topo_nm, PROGRAMME_DIR) as ev:
            x0 = ev.x_current
            if len(x0) == 0:
                nm_shape_fails: list[str] = []
            else:
                r = innerloop.nm_search(ev, x0, budget=nm_budget)
                nm_shape_fails = [f for f in r.fail_lines if f.endswith(_SHAPE_SUFFIXES)]
        nm_time = time.time() - t0
        nm_time_total += nm_time
        nm_ok = len(nm_shape_fails) == 0
        if dp_ok is None:
            rows.append((i, n_leaves, seed, "ERROR", nm_ok, dp_time, nm_time, info.get("error")))
            continue
        if dp_ok:
            n_dp_feasible += 1
        if nm_ok:
            n_nm_feasible += 1
        agree = dp_ok == nm_ok
        if agree:
            n_agree += 1
        else:
            if dp_ok and not nm_ok:
                false_positive += 1
            else:
                false_negative += 1
        rows.append((i, n_leaves, seed, dp_ok, nm_ok, dp_time, nm_time, len(nm_shape_fails)))
    print(f"topologies: {n_topologies} (attempts {attempts})")
    print(f"agreement: {n_agree}/{n_topologies} = {100*n_agree/n_topologies:.1f}%")
    print(f"  false positive (DP feasible, NM shape-fails): {false_positive}")
    print(f"  false negative (DP infeasible, NM 0 shape-fails): {false_negative}")
    print(f"DP feasible: {n_dp_feasible}/{n_topologies}   NM 0-shape-fail: {n_nm_feasible}/{n_topologies}")
    print(f"DP total time: {dp_time_total:.2f}s  ({dp_time_total/n_topologies*1000:.1f} ms/topology)")
    print(f"NM total time: {nm_time_total:.2f}s  ({nm_time_total/n_topologies*1000:.1f} ms/topology)")
    print(f"speedup: {nm_time_total/dp_time_total:.1f}x")
    print("\nmismatches:")
    for row in rows:
        if row[3] != row[4] and row[3] != "ERROR":
            print(" ", row)
    print("\nerrors:")
    for row in rows:
        if row[3] == "ERROR":
            print(" ", row)
 if __name__ == "__main__":
    n = int(sys.argv[1]) if len(sys.argv) > 1 else 200
    budget = int(sys.argv[2]) if len(sys.argv) > 2 else 100
    grid_n = int(sys.argv[3]) if len(sys.argv) > 3 else 150
    main(n, budget, grid_n)