annular.market.tulipa.utils

Attributes

logger

Functions

strip_timezone(→ pandas.Series | pandas.DatetimeIndex)	Convert an input to pd.Series or pd.Index and strip timezone.
create_tulipa_input_data(→ None)	Creates the Tulipa input from the given arguments.
write_inputs_to_db(→ None)	Write input data to DuckDB tables.
read_outputs_from_db(→ tuple[pandas.DataFrame, ...)	Read output data from DuckDB tables.
convert_data_for_tulipa(→ tuple[pandas.DataFrame, ...)	Convert data so that Tulipa can use them.

Module Contents

annular.market.tulipa.utils.logger

annular.market.tulipa.utils.strip_timezone(timestamp: pandas.Series | pandas.DatetimeIndex | list) → pandas.Series | pandas.DatetimeIndex

Convert an input to pd.Series or pd.Index and strip timezone.

Parameters:

timestamp – The input to process.

annular.market.tulipa.utils.create_tulipa_input_data(generator_configs: dict[str, dict], timeseries_data: pandas.DataFrame, db_file: str | None = None, csv_folder: str | pathlib.Path | None = None) → None

Creates the Tulipa input from the given arguments.

Populates a duckdb database with the input files for Tulipa. This creates one producer asset per generator in generator_configs plus one consumer asset named bid_manager.

The profiles for the generators is stored in the timeseries_data.

Parameters:

• generator_configs – dict of configurations defining of all generators.

• timeseries_data – DataFrame containing all timeseries_data for generators.

• db_file – A path to duckdb database. By default, an in-memory database is used and not persisted, in which case csv_folder should be specified.

• csv_folder – If given, all tables are saved as csv files in this location.

Notes

The intended use of this function is to transform supply-side input data for an annular model to a Tulipa model. The bid_manager asset is created because we attach bids later to an (arbitrary) existing consumer asset, but the supply-side data do not have any demand assets.

annular.market.tulipa.utils.write_inputs_to_db(connection: duckdb.DuckDBPyConnection, bids: pandas.DataFrame, timeseries_data: pandas.DataFrame, window_df: pandas.DataFrame) → None

Write input data to DuckDB tables.

Parameters:

• connection – Open DuckDB connection.

• bids – DataFrame with bids.

• timeseries_data – DataFrame with timestamp and timestep columns.

• window_df – DataFrame with timestamp and timestep columns for the window.

annular.market.tulipa.utils.read_outputs_from_db(connection: duckdb.DuckDBPyConnection, snapshots: pandas.DatetimeIndex, bid_idx_cols: list[str], bids: pandas.DataFrame) → tuple[pandas.DataFrame, pandas.DataFrame]

Read output data from DuckDB tables.

Parameters:

• connection – Open DuckDB connection

• snapshots – DatetimeIndex to filter market_price rows (the bidding window)

• bid_idx_cols – Column names to use for setting the index

• bids – Dataframe with with submitted bids.

Returns:

Tuple of (market_price, scheduled_bids) DataFrames. scheduled_bids timestamps are localized back to UTC.

annular.market.tulipa.utils.convert_data_for_tulipa(bids: pandas.DataFrame, timeseries_data: pandas.DataFrame, snapshots: pandas.DatetimeIndex, bidding_window: pandas.DatetimeIndex) → tuple[pandas.DataFrame, pandas.DataFrame, pandas.DatetimeIndex, pandas.DataFrame]

Convert data so that Tulipa can use them.

In particular - Strip timezone from inputs to prepare for database storage - Reset all indexes - Create 1-based time index that is used in Tulipa - Denote sense column with strings “demand” and “supply” instead of numerics

To avoid mutation, input dataframes are copied.

Parameters:

• bids – MultiIndex DataFrame with a ‘timestamp’ level.

• timeseries_data – DataFrame with DatetimeIndex.

• snapshots – Optimization window as a DatetimeIndex of timesteps.

• bidding_window – timestamps for bidding window.

Returns:

Tuple of (bids, timeseries_data, snapshots, and window_df). window_df is a correspondence between pd.Datetime timestamps in Annular and integer timesteps in Tulipa, subset to the optimization window.