Skip to content

Deterministic co-evolving ecosystem core and CLI - #1

Closed
mahim37 wants to merge 2 commits into
mainfrom
revamp/m1-core
Closed

mahim37 wants to merge 2 commits into
mainfrom
revamp/m1-core

Conversation

@mahim37

@mahim37 mahim37 commented Sep 5, 2026 •

Copy link
Copy Markdown
Owner

Superseded by #3.

…e 1)

Adds crates/rustywings-core and crates/rustywings-cli alongside the legacy
tutorial-derived libs/ and www/, which stay building and deployed until the
new frontend replaces them in milestone 2.

Core
- Open-ended evolution: energy, reproduction thresholds, death by starvation
  and age. No generations, no fitness function.
- Two trophic levels (sparrows eat seeds, hawks eat sparrows) with logistic
  plant regrowth around drifting patches.
- Genome = feed-forward brain weights + heritable body traits (FOV angle,
  sight range, top speed, size), each with an energy cost.
- Bit-identical across native and wasm: owned xoshiro256++ RNG, libm for
  transcendentals, no hash maps or threads in a step. Golden checksum
  pinned; CI runs it on Linux, macOS and Windows.
- Struct-of-arrays agents for zero-copy export to WebGL; counting-sort
  spatial hash rebuilt per tick; postcard snapshots that resume bit-for-bit.
- Validated Config with dotted-path errors instead of panics.
- Standardised arena: scores a genome alone in a fixed-seed world so
  "it learns" is a measured ratio, not a claim.

CLI (rustywings): run (CSV/JSONL stats, snapshots), bench, verify, arena, config.

Defaults were tuned with a 22-variant sweep. Over 60k ticks on seeds 1-4:
~450 sparrows, ~85 hawks, no rescue immigration, sparrow FOV evolving to
~300 deg and hawk FOV to ~120 deg. Arena: evolved sparrows eat 3-6.5x more
than random ones after 30k ticks. Native bench: 8.4 ms/tick at ~4k birds.

Also: GitHub Actions CI (fmt, clippy, tests, wasm check, cross-OS golden,
arena learning test), docs/ARCHITECTURE.md, five decision records, the UI
mockups the frontend direction was chosen from, Cargo.lock committed,
CLAUDE.md rewritten for the new layout.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LTmV279xygsXfALmMmQeeL
@vercel

vercel Bot commented Sep 5, 2026 •

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated
rusty-wings Ready Ready Preview Sep 5, 2026 10:01pm UTC

Vercel selects its build runtime from `engines.node`. The project was
configured in April 2025 and its runtime has since reached end of life,
which fails preview deployments. The legacy app stays deployed until the
new frontend (milestone 2) replaces it, so it needs a supported runtime.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LTmV279xygsXfALmMmQeeL
@mahim37 mahim37 closed this Sep 6, 2026
@mahim37
mahim37 deleted the revamp/m1-core branch September 6, 2026 10:23
@mahim37 mahim37 changed the title Rebuild milestone 1: deterministic co-evolving ecosystem core + CLI Deterministic co-evolving ecosystem core and CLI Sep 6, 2026
@mahim37

mahim37 commented Sep 6, 2026

Copy link
Copy Markdown
Owner Author

Superseded by #3: the branch was renamed to core-sim and the history rewritten into self-contained commits.

This branch was successfully deployed

1 active deployment
Preview — 581dd8d0 Deployed Sep 5, 2026 by vercel[bot]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant