How WareMax works
WareMax is a Cargo workspace with a deterministic Rust simulation core and a Python Gymnasium bridge on top. Here is how a scenario becomes a byte-identical trajectory.
The pipeline
A scenario YAML is parsed and schema-validated by waremax-config, then
handed to the Rust DES core. The core advances a single event queue on one thread;
time moves only at typed discrete events. Routing on the graph topology is handled by
waremax-map, and the selected dispatching policy decides task allocation.
Every completed task is decomposed by waremax-analysis into the five delay
buckets.
The core can be driven two ways: the waremax CLI for simulations, sweeps,
A/B tests, and benchmarks; or the waremax-gym PyO3 bridge, which exposes
WaremaxAllocEnv as a Gymnasium environment for reinforcement learning.
Why it is deterministic
Reproducibility is enforced, not asserted. The core seeds a ChaCha8 RNG from a
u64, runs single-threaded per scenario, and applies canonical id-based
tie-breaking everywhere two choices are otherwise equal. The RL control loop adds a
crossbeam ping-pong handshake so the property survives a worker thread. The result:
the same seed and action sequence produce a
byte-identical trajectory.
The workspace
waremax-core DES kernel, event queue, IDs, SimTime, ChaCha8-seeded RNG
waremax-map Graph topology, shortest-path + congestion-aware routing, traffic
waremax-storage Racks, bins, SKUs, inventory replicas
waremax-entities Robot, Order, Task, Station, ChargingStation
waremax-policies Allocation, station assignment, batching, priority, traffic policies
waremax-config YAML / JSON scenario parsing and schema validation
waremax-metrics Event log, time-series, CSV/JSON export, HTML/PDF reports
waremax-analysis Delay attribution, critical-path analysis, bottlenecks, root-cause
waremax-rl RL control seam, Gym-style env, attribution & routed reward modes
waremax-gym PyO3 bindings, Python wrapper, training scripts
Ready to run it? Start with the quickstart, then write your own scenario YAML. WareMax is built by Skelf Research.