Skip to main content

ECS boids simulation — free tier: Vec3 math + 3-force flocking

Project description

Boidswarm (zombiebubble-boids)

An advanced, high-performance Entity Component System (ECS) flocking simulation.

While traditional "Boids" algorithms simulate flocking using 3 basic rules (separation, alignment, cohesion), boidswarm introduces an advanced Graph-Laplacian Leader Election system. This allows your flock to intelligently pass messages, nominate leaders, and gracefully recover when leaders get stuck or destroyed.

Installation

pip install boidswarm

Features (No Advanced Math Required!)

Boids Animation

If you've never used Boids before, think of them as AI agents that group together like flocks of birds or schools of fish. This library makes managing them incredibly robust:

1. The "Ripple Effect" (Signal Propagation)

In basic flocking, a boid only reacts to its immediate neighbors. If a boid at the front spots the player or a target, the back of the flock won't know and might break formation.

  • How it works: Information (like a "pursuit signal") diffuses through the flock like heat spreading through metal.
  • How to use it: You control this with the alpha parameter.
    • alpha = 0.1 to 0.3: Highly recommended. The signal spreads fluidly and realistically.
    • alpha = 0.0: No spreading (traditional, dumb boids).
    • alpha = 1.0: Instant hive-mind transmission.

2. Auto-Healing Leader Election

Navigating complex environments usually causes flocks to get stuck in corners.

  • How it works: Beneath the hood, the system dynamically classifies every boid into 7 roles (e.g., Follower, Sub-Leader, Leader, Prime).
  • How to use it: It works automatically! If your current leader drops out of the simulation (becomes a "Phantom Dead") or its speed drops too low (becomes "Phantom Stuck"), the system immediately promotes a Sub-Leader to take its place. The flock never permanently breaks down.

3. Customizable Flock Personalities (Gram Matrix)

You can tune the underlying scoring engine to fit the specific needs of your game or simulation without rewriting any core logic:

  • Resilience-First: Best for massive, chaotic, or combat-heavy flocks. It guarantees the flock stays together even if multiple leaders are lost simultaneously.
  • Expressiveness-First: Best for small, tight-knit flocks where you want highly nuanced, synchronized, and agile movements.

Command Line Usage

After installation, the zombiebubble command is available with three subcommands:

zombiebubble COMMAND [options]

Commands:
  run     Run a headless boids simulation
  check   Statically analyse a Python file for risky Swarm() call sites
  info    Show library version and build information

zombiebubble run — headless simulation

zombiebubble run [--boids N] [--steps S] [--dt DT] [--print-every K] [--no-intro]
Option Default Description
--boids N 20 Number of boids to simulate
--steps S 200 Total simulation steps
--dt DT 0.016 Time delta per step (seconds)
--print-every K 40 Print a metrics row every K steps
--no-intro Skip the Boids explanation text

Examples:

# Quick default run (20 boids, 200 steps)
zombiebubble run

# Large flock, long run, silent intro
zombiebubble run --boids 200 --steps 1000 --no-intro

# Slow-motion with fine-grained output
zombiebubble run --boids 30 --dt 0.008 --print-every 10

Understanding the output columns:

Column Meaning
step / time Simulation frame and elapsed simulated time
spread Average distance from the flock centroid (metres). Shrinking = cohesion working
avg speed Mean boid speed (m/s)
status Human-readable flock state

After the run, the tool prints the initial vs. final spread and confirms whether the Cohesion rule pulled the flock together.

Note: If you pass a very small --boids value, the tool will print a warning to stderr explaining which internal thresholds are at risk. These warnings do not stop the simulation.


zombiebubble check — static analysis (gcc-style)

Analyses a Python source file for Swarm(N) call sites where N falls below the kernel thresholds required for reliable leader election and full flock expressiveness. Useful in CI pipelines.

zombiebubble check FILE [--error]
Option Description
FILE Python source file to analyse
--error Treat warnings as errors (exit code 2 instead of 1)

Exit codes:

Code Meaning
0 No issues found
1 One or more warnings emitted
2 Warnings emitted and --error was passed

Example output:

$ zombiebubble check my_game.py
error[BW001]: 2 boids < 3 (min-distance d): error-correction structure collapsed
  --> my_game.py:12:13
    |
 12 |     swarm = Swarm(2)
    |             ~~~~~~~~
    |
    = help: the kernel corrects t=⌊(d−1)/2⌋=1 simultaneous leader loss but needs ≥ d=3 boids to form the correction graph; add at least 1 more

my_game.py: 1 error(s)

Understanding Diagnostics (The Bird Analogy)

The simulation relies on advanced group dynamics. If your flock is too small, the system emits rustc-style diagnostics to let you know why the flock might behave poorly. You don't need a math degree to fix them — just think of them as real birds!

Code Level What it means in plain English
BW001 Error The Lonely Birds (N < 3): A flock of 1 or 2 birds isn't really a flock. If one gets lost or stuck, the other is completely alone. You need at least 3 birds so that if one drops out, the remaining two can still navigate together.
BW002 Warn The Understaffed Committee (N < 5): To make complex, agile turns, the flock divides decision-making into 5 independent "thoughts" (degrees of freedom). With fewer than 5 birds, they simply don't have enough brainpower to perform fancy, nuanced maneuvers.
BW003 Warn Missing the Alpha (N < 7): A healthy flock organizes itself into a 7-tier social hierarchy. With fewer than 7 birds, the absolute top rank (the "prime-leader") remains empty. The flock will fly, but without a strong central guide.
BW004 Note The Missing Partner (Exactly 6 birds): This is a mathematical quirk! With exactly 6 birds, an Alpha is elected, but the 7th rank (the "lone scout" or isolated counterbalance) is empty. The leader feels unsupported. Adding just 1 more bird (to exactly 7) fixes their social dynamic entirely.
BW005 Note The Awkward Tie (Exactly 4 birds): 4 birds hit an invisible resonance boundary in how the algorithm calculates distances. They get slightly confused about who is leading whom. Adding 1 bird steps off this awkward boundary.
(note) Note The Safest Journey (N ≥ 9): If you want an ultra-resilient flock that can survive 2 leaders getting lost simultaneously without any hesitation, aim for at least 9 birds.

Use in CI (GitHub Actions example):

- name: Check boid counts
  run: zombiebubble check src/my_sim.py --error

zombiebubble info — build info

zombiebubble info

Prints the installed package version, Python version, and OS/architecture.


License

Licensed under GPL-3.0-only.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

boidswarm-0.1.8.tar.gz (293.9 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

boidswarm-0.1.8-py3-none-any.whl (294.7 kB view details)

Uploaded Python 3

File details

Details for the file boidswarm-0.1.8.tar.gz.

File metadata

  • Download URL: boidswarm-0.1.8.tar.gz
  • Upload date:
  • Size: 293.9 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.5

File hashes

Hashes for boidswarm-0.1.8.tar.gz
Algorithm Hash digest
SHA256 a178c378b1667b8ad3cee8e25afaf9af3986e8df3c960ce74044e64249a5dcb8
MD5 2c7c65058f7ff5de49d23d9e6e4a6616
BLAKE2b-256 09746d8834c92b68ccff474ef40bcdea33983251cc497835bc98b9558eebc54e

See more details on using hashes here.

File details

Details for the file boidswarm-0.1.8-py3-none-any.whl.

File metadata

  • Download URL: boidswarm-0.1.8-py3-none-any.whl
  • Upload date:
  • Size: 294.7 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.5

File hashes

Hashes for boidswarm-0.1.8-py3-none-any.whl
Algorithm Hash digest
SHA256 fa3c6db3196ef4084613fdd6149cdaa485396a977d1b295c74c13c36117c0e5f
MD5 e5069e3c0a9e19510782e964fe242a1c
BLAKE2b-256 3d28815c72950a79b827faf5545905d4f2ee22a3a5b6f4ef0c367d7694c7f64e

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page