Dynamically scalable agent-based simulation framework, written in Rust
4 days ago
- #simulation
- #scalability
- #distributed-systems
- Dynamically scalable virtual worlds framework for large agent-based simulations.
- Potential applications include games and business intelligence.
- Supports billions of entities and thousands of connected clients.
- Under development with a roadmap towards alpha 0.2.0 release.
- Inspired by ECS pattern and actor model.
- Entities composed of components for flexible data storage and fast migration.
- Logic provided via behaviors or external services with synchronization mechanisms.
- Cluster workers perform system-wide queries and serve data to subscribers.
- Generic runtime allows dynamic cluster enlargement and simulation model changes.
- Actors can deploy nested simulations and exhibit complex behaviors.
- Command-line tool (bigworlds-cli) for creating and interacting with simulations.
- Examples and documentation available for getting started.
- Aims to support socio-economic and ecological simulations on off-the-shelf hardware.
- Potential for AI/LLM training in near-real-world scenarios.
- Trade-offs in data retrieval times and entity size due to distributed storage.
- No strong atomicity guarantees for transactions.