bigworlds is a novel framework for assembling and running large simulation models.

It tries to solve the problem of dynamic yet efficient distribution of discrete simulation elements over many machines.

Goals

Provide a batteries-included solution for creating large virtual worlds with millions of persistent entities. Provide a way for creating multiverse systems made up of model-compatible worlds.

Allow building massive multiplayer experiences with thousands of concurrent players at relatively high degree of fidelity. Allow operating them at a fraction of the cost required by currently available closed-source systems.

Make assembling and running large-scale socio-economic simulations feasible on off-the-shelf hardware. Make model definition process scalable, allow multiple ways for importing data and defining logic.

Design

Focus is on discrete entity-based models, with capability for driving logic based on both global events and regular time cycles.

Entities are generally composed of components defining the data they store. Components also potentially define the logic that should be applied to given entities.

Storage is optimized for entity-level retrieval and transfer between nodes.

The runtime is kept as generic as possible, allowing for wide variety of logic and compute configurations. This includes dynamic cluster enlargement and changes to the simulation model at runtime.

Actors with complex sets of behaviors can be constructed and deployed into the simulation world at any point, including by other actors at runtime.

Interesting recursion patterns are possible, with actors being able to start up bigworlds simulations themselves.