Status: under active development — no stable release yet.
ATLAS simulates electricity market mechanisms across the full trading horizon. It models the sequential decisions of market participants — from order formulation ahead of the auction to market clearing and portfolio optimisation — using a modular, configurable architecture.
Each module is an independent computation unit that reads from a shared dataset and produces structured changes applied by the orchestrator. Modules are chained into workflows, where the output of each step feeds into the next.
Modules are grouped into market chains, each representing a full simulation cycle.
Day-Ahead — available
| Module | Description |
|---|---|
| Day-Ahead Orders | Generates market orders for all equipment types based on asset characteristics and cost structure. |
| Market Clearing | Determines market equilibrium by matching supply and demand across interconnected areas, under ATC or flow-based network constraints. |
| Portfolio Optimisation | Optimises energy asset portfolios (thermal, hydro, storage, solar, wind, load) to maximise profit under market conditions. |
Intraday — available
| Module | Description |
|---|---|
| Intraday Price Forecast | Forecasts intraday prices per market area from the deviation between the latest load, wind and solar forecasts and the day-ahead baseline, using a price sensitivity ratio and price caps. |
| Portfolio Optimisation | Re-optimises the portfolio against the forecast prices, starting from the day-ahead cleared position. |
| Intraday Orders | Formulates intraday orders (thermal, hydro, storage, solar, wind, load, non-dispatchable) from the gap between the optimised schedule and the current engagement. |
| Market Clearing | Clears the intraday market over the remaining horizon. |
| Portfolio Optimisation | Final re-optimisation against the intraday cleared position. |
Both chains reuse the same Market Clearing and Portfolio Optimisation modules — only the parameters and the position they hold in the workflow differ.
- Python 3.13+, managed with uv
- OR-Tools for optimisation, Pydantic for data models, Polars for data processing
- Typer for the CLI, Ruff for linting, mypy for type checking, pytest for tests
Install uv, then:
git clone https://github.com/rte-france/ATLAS.git && cd ATLAS && uv sync
from atlas import AtlasDataset, ControlBlock, MarketArea, Node, Portfolio, Thermal, Timeseries cb = ControlBlock(name="cb_fr") area = MarketArea(name="fr", control_block=cb) node = Node(name="node_fr", control_block=cb, market_area=area) portfolio = Portfolio(name="gen_fr", control_block=cb, market_area=area) variable_cost = Timeseries.from_index( start_date="2024-01-01 00:00:00", frequency="1h", end_date="2024-01-02 00:00:00", default_value=45.0, ) nuclear = Thermal( name="fr_nuclear", node=node, portfolio=portfolio, installed_capacity=1584.0, variable_cost=variable_cost, ) dataset = AtlasDataset( control_block=[cb], market_area=[area], node=[node], portfolio=[portfolio], thermal=[nuclear], ) dataset.to_directory("./data/input/")
# parameters.yaml temporal: start_date: "2024年01月01日 00:00:00" end_date: "2024年01月02日 00:00:00" execution_date: "2023年12月31日 12:00:00" timestep: "PT1H" solver: solver_name: "SCIP" load_price: 3000
atlas module list # See available modules
atlas module run DayAheadOrders -p parameters.yaml -d ./data/input/Or from Python:
from atlas import AtlasDataset, ModuleRun result = ModuleRun( module="DayAheadOrders", dataset="./data/input/", parameters="parameters.yaml", ).run()
# workflow.yaml name: day-ahead dataset_path: ./data/input/ output_dataset_path: ./data/output/ output_dir: ./results/ steps: - module: DayAheadOrders parameters_path: ./parameters/day_ahead_orders.yml - module: MarketClearing parameters_path: ./parameters/market_clearing.yml - module: PortfolioOptimisation parameters_path: ./parameters/portfolio_optimisation.yml
atlas workflow list workflow.yaml # List all workflow steps
atlas workflow run workflow.yamlThe intraday chain starts from a price forecast and re-optimises the portfolio twice — once against the forecast prices, once against the intraday cleared position. The same module can appear several times in a workflow with different parameters.
# intraday_workflow.yaml name: intraday dataset_path: ./data/input/ output_dataset_path: ./data/output/ output_dir: ./results/ steps: - module: IntradayPriceForecast parameters_path: ./parameters/intraday_price_forecast.yml - module: PortfolioOptimisation parameters_path: ./parameters/portfolio_optimisation_1.yml - module: IntradayOrders parameters_path: ./parameters/intraday_orders.yml - module: MarketClearing parameters_path: ./parameters/market_clearing.yml - module: PortfolioOptimisation parameters_path: ./parameters/portfolio_optimisation_2.yml
atlas workflow run intraday_workflow.yaml
A runnable example is available in the test dataset:
uv run atlas workflow run tests/dataset/parameters/intraday/workflow.yml
Full documentation: rte-atlas.readthedocs.io
See CONTRIBUTING.
See CHANGELOG.
See AUTHORS.