Hi HN!
I got completely sidetracked during a patent law lecture & ended up building physics simulations of Leonardo da Vinci’s 500-year-old mechanical designs with the new GPT-5 Codex.
I built an open repository that reconstructs Leonardo’s machines as digital twins. Each invention links to folio-level provenance (original codex manuscripts), includes uncertainty quantification for key parameters, and ships with FMEA safety notes.
The repo has runnable Jupyter notebooks for:
– Ornithopter: Flight dynamics with modern material upgrades
– Parachute: Terminal velocity analysis (preliminary results ~7 m/s)
– Self-propelled cart: Range/energy studies with spring mechanics
– Aerial screw: Lift generation vs power requirements
– Mechanical odometer: Measurement accuracy analysis
All simulations include parameter sensitivity studies and link back to original Renaissance manuscripts. Everything’s open source with comprehensive documentation.
I’d love feedback on: (1) modeling assumptions and unit choices, (2) where reconstructions diverge from museum builds, and (3) which inventions to prioritize next.
Repo: https://github.com/Shannon-Labs/davinci-codex
Comments URL: https://news.ycombinator.com/item?id=45275352
Points: 1
# Comments: 0
Source: github.com
