bruno/homemaker-layout
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
Python rewrite of the Urb/Homemaker stack
3 commits 1 branch 0 tags 5.4 MiB
Find a file
Bruno Postle d08d15e4d7 Full-fitness frozen-topology optimisation validates geometry inner loop 
Driving equal-offset cut ratios with Nelder-Mead against the REAL oracle
fitness (full objective, no proxy) improves all three test candidates with
zero new failures:

  2f45907 (best evolved)  0.012617 -> 0.015684  x1.24  (2->2 fails)
  candidate-002          0.007375 -> 0.012319  x1.67  (2->2 fails)
  c964435 (baseline)     0.003667 -> 0.005836  x1.59  (3->3 fails)

Headroom widens on weaker designs. The EA under-optimises geometry by
24-67% even on its best result. This validates a full-fitness geometry
inner loop (NOT the earlier area-proxy solver) and motivates a memetic
architecture: topology search outside, full-objective geometry optimise
inside, gated on a native Python fitness (oracle at ~3s/call is too slow).
2026-06-10 22:27:30 +01:00
.claude Scaffold homemaker-py with validated geometry port 2026-06-10 20:50:20 +01:00
experiments Full-fitness frozen-topology optimisation validates geometry inner loop 2026-06-10 22:27:30 +01:00
src/homemaker Add programme/solver/oracle + sizing experiments (negative result) 2026-06-10 21:49:31 +01:00
.gitignore Scaffold homemaker-py with validated geometry port 2026-06-10 20:50:20 +01:00
CLAUDE.md Scaffold homemaker-py with validated geometry port 2026-06-10 20:50:20 +01:00
pyproject.toml Scaffold homemaker-py with validated geometry port 2026-06-10 20:50:20 +01:00
README.md Scaffold homemaker-py with validated geometry port 2026-06-10 20:50:20 +01:00

README.md

homemaker-py

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 (current): 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. Canonical slicing encoding (normalized Polish expression) + memetic search.

Layout

  • src/homemaker/dom.py — read/write Urb .dom YAML into a Node tree.
  • src/homemaker/geometry.py — faithful port of Urb's top-down geometry.
  • src/homemaker/programme.py — parse patterns.config space requirements.
  • src/homemaker/solver.py — bottom-up ratio solve (scipy).
  • src/homemaker/oracle.py — Phase-1 scaffold: score a .dom via Urb's urb-fitness.pl.

The Perl oracle is the only throwaway component; everything else is permanent.