bruno/homemaker-layout
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
Python rewrite of the Urb/Homemaker stack
189 commits 2 branches 0 tags 25 MiB
Find a file
Bruno Postle 0ec404c73d Spike: torch-autodiff inner-loop ratio optimisation (homemaker-py-2ax) — negative 
Build a torch-differentiable local proxy for the ratio-to-fitness path (exact
port of geometry.py's coordinate recursion + the 5 continuous per-leaf quality
factors, with discrete/structural facts frozen from a real fitness.py
snapshot and the 0.5^n cliff relaxed to a sigmoid) and compare Adam ascent
against nm_search on frozen topologies from programme-house and harbor-house.

Result: ~30-35x slower per unit of search progress than nm_search at both
6 DOF and 36 DOF (per-op torch tensor dispatch overhead with no batching
opportunity, plus snapshot/resnapshot cost on par with a full oracle eval),
and no better quality at matched budget. A step-size sensitivity check
confirmed the flagged 0.5^n cliff risk is real, but autodiff doesn't make the
gradient direction any cheaper to obtain here. Not recommended; kept as
reference only, not wired into innerloop.py. Full writeup in DESIGN.md §34.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01R8agJBT2ZpmF3ErW7wi2wY
2026-08-01 10:26:43 +01:00
.beads Spike: torch-autodiff inner-loop ratio optimisation (homemaker-py-2ax) — negative 2026-08-01 10:26:43 +01:00
.claude Scaffold homemaker-py with validated geometry port 2026-06-10 20:50:20 +01:00
examples homemaker-py-1s3: multi-use leaves as permanent design goal (§26 path b) 2026-07-31 00:16:12 +01:00
experiments Spike: torch-autodiff inner-loop ratio optimisation (homemaker-py-2ax) — negative 2026-08-01 10:26:43 +01:00
src/homemaker_layout homemaker-py-1s3: land precision-weighted shape combination for multi-use leaves 2026-07-31 10:20:49 +01:00
tests homemaker-py-1s3: land precision-weighted shape combination for multi-use leaves 2026-07-31 10:20:49 +01:00
.gitignore bd init: initialize beads issue tracking 2026-06-11 23:27:11 +01:00
AGENTS.md bd init: initialize beads issue tracking 2026-06-11 23:27:11 +01:00
CLAUDE.md docs: DESIGN.md §26/§27 — backfill 9o5/xi7/b3v (type superposition) and mi7 (bubble-diagram signal) 2026-07-26 23:16:21 +01:00
DESIGN.md Spike: torch-autodiff inner-loop ratio optimisation (homemaker-py-2ax) — negative 2026-08-01 10:26:43 +01:00
pyproject.toml 94g: public-access pin + keep-better wrapper + CLI/finish-hook wiring 2026-07-18 10:29:44 +01:00
README.md update 2026-07-31 09:57:45 +01:00

README.md

homemaker-layout

Programme-driven building-layout search over slicing trees. A clean-room Python successor to the Perl Urb project, intended to eventually be 100% Python.

Why a rewrite

Urb represents a building as a binary slicing tree where room sizes are derived top-down from division ratios. That makes room area an emergent property of every cut above it, which:

  • gives the genome low locality (a cut near the root rescales every descendant),
  • makes target room sizes nearly impossible to hit, so the gaussian size penalty dominates fitness, and
  • defeats crossover (transplanted subtrees lose their proportions).

homemaker inverts this: leaves carry target dimensions from the programme and division ratios are solved bottom-up for a fixed topology. The evolutionary search then only explores topology + types + adjacency.

Phase plan

  1. Solver experiment: port Urb's geometry, re-solve ratios from programme targets, score the result against the original via the Perl oracle. ✓
  2. Native Python fitness (retire the Perl oracle). ✓
  3. Memetic search: canonical slicing genome + high-locality operators + Nelder-Mead inner loop. ✓
  4. Penalty reshaping: lexicographic (-n_fails, fitness) outer-search comparison. ✓
  5. Representation upgrade: canonical slicing encoding + bottom-up shape feasibility, scaled to larger programmes. ✓
  6. Search-quality experiments (current): a long running series of opt-in levers tried against the harbor-house, health-centre, and programme-house example corpora — leaf-sharing, finish-time cell→room collapse, ruin-and-recreate LNS, 2-opt polish, multi-use/co-located leaves, adjacency-graph and bubble-diagram fitness signals, and more. Most of these are negative/null results kept as opt-in flags or reference code rather than defaults. See DESIGN.md §11 onward for the full, numbered experiment log with methodology and results for each.

Layout

  • src/homemaker_layout/dom.py — read/write Urb .dom YAML into a Node tree.
  • src/homemaker_layout/geometry.py — faithful port of Urb's top-down geometry.
  • src/homemaker_layout/programme.py — parse patterns.config space requirements.
  • src/homemaker_layout/solver.py — bottom-up ratio solve (scipy).
  • src/homemaker_layout/fitness.py — native Python fitness evaluator.
  • src/homemaker_layout/fitness_cmd.py — homemaker-fitness CLI (drop-in for urb-fitness.pl).
  • src/homemaker_layout/collapse_cmd.py — homemaker-collapse CLI: finish-time global cell→room relabel of a .dom.
  • src/homemaker_layout/graph.py — leaf-adjacency graph for programme-driven checks.
  • src/homemaker_layout/genome.py — topology genome: base-floor tree + per-storey deltas.
  • src/homemaker_layout/operators.py — high-locality mutation and subtree crossover.
  • src/homemaker_layout/innerloop.py — ratio optimisation inner loop (Nelder-Mead / CMA-ES).
  • src/homemaker_layout/driver.py — memetic search outer loop.
  • src/homemaker_layout/evolve.py — homemaker-evolve CLI entry point.
  • src/homemaker_layout/oracle.py — legacy Perl shim, kept for cross-validation only.
  • src/homemaker_layout/bubble.py — 3D bubble-diagram adjacency fitness-signal prototype (DESIGN.md §27); validated null, not wired into fitness.py — reference only.