29 lines
1.5 KiB
Markdown
29 lines
1.5 KiB
Markdown
# ELAC-Plan: Edge-Latency Aware Cross-Venue Execution Planner
|
|
|
|
ELAC-Plan is an open-source platform prototype for generating cross-venue execution plans locally near exchanges. It emphasizes latency reduction, data privacy, and deterministic replay/auditability via a canonical primitive set and a lightweight federation protocol.
|
|
|
|
This repository implements a production-ready MVP skeleton in Python with a FastAPI API, in-memory storage, and a minimal LocalProblem/SharedVariables/PlanDelta model suite. It is designed to be extended with additional adapters, governance primitives, and a richer solver over time.
|
|
|
|
Key primitives (canonical IR):
|
|
- LocalProblem (Objects): per-asset, per-venue optimization task definitions
|
|
- SharedVariables (Morphisms): privacy-bounded summarized signals
|
|
- DualVariables: cross-venue coupling signals
|
|
- PlanDelta: incremental plan updates with metadata and cryptographic tags
|
|
- AuditLog: governance and provenance blocks
|
|
|
|
Architecture overview
|
|
- API: FastAPI app exposing /problems and /status (and delta reconciliation in future)
|
|
- In-memory storage with deterministic replay hooks
|
|
- Lightweight solver: generates PlanDelta from LocalProblem
|
|
- Adapters: translate market feeds and broker APIs into the ELAC canonical representation
|
|
|
|
How to run
|
|
- Install: python3 -m pip install -e .
|
|
- Run API (example): uvicorn elac_plan.api.app:app --reload --port 8000
|
|
- Tests: bash test.sh
|
|
|
|
Contributing
|
|
- See AGENTS.md for contribution rules and how to run tests locally.
|
|
|
|
This is an early MVP; feedback and contributions are welcome.
|