5.3. pypfilt.sweep¶
The pypfilt.sweep
module iterates over forecast scenarios defined in
any number of scenario files.
5.3.1. Forecast scenarios¶
-
pypfilt.sweep.
forecasts
(configs, load_obs=True)¶ Iterate over the forecast settings for each scenario.
Parameters: - configs (Union[pypfilt.config.Config, Sequence[pypfilt.config.Config]]) – The forecasting configuration(s).
- load_obs (bool) – Whether to load observations from disk.
Return type: Iterator[Forecasts]
-
class
pypfilt.sweep.
Forecasts
¶ Forecast settings.
Parameters: - scenario_id (str) – The scenario identifier.
- scenario_name (str) – A descriptive name for the scenario.
- params (dict) – The simulation parameters.
- all_observations (List[dict]) – All of the available observations.
- List[Sequence[dict]] – observation_streams: The observations, separated into streams for each observation model.
- file_descriptor (str) – A descriptor to identify files associated with these forecasts.
- config_strings (List[str]) – The contents of the configuration files in the forecasting sweep.
-
scenario_id
¶ Alias for field number 0
-
scenario_name
¶ Alias for field number 1
-
params
¶ Alias for field number 2
-
observation_models
¶ Alias for field number 3
-
all_observations
¶ Alias for field number 4
-
observation_streams
¶ Alias for field number 5
-
file_descriptor
¶ Alias for field number 6
-
config_strings
¶ Alias for field number 7
5.3.2. Forecasting in parallel¶
The pypfilt.sweep
module also provides support for running forecasts
across multiple Python processes:
-
pypfilt.sweep.
forecasts_mp
(configs)¶ Iterate over the forecast settings for each scenario, yielding tuples that can be serialised and sent to other Python processes. Each process can then retrieve the associated
Forecasts
value by callingget_forecasts_mp()
.Parameters: configs (Union[pypfilt.config.Config, Sequence[pypfilt.config.Config]]) – The forecasting configuration(s).
-
pypfilt.sweep.
get_forecasts_mp
(mp_tuple, load_obs=True)¶ Return the forecast settings for a tuple yielded by
forecasts_mp()
.Parameters: - mp_tuple – A value returned by
forecasts_mp()
. - load_obs (bool) – Whether to load observations from disk.
Return type: - mp_tuple – A value returned by