Sinergym output

When a simulation is running, it creates a directory named Eplus-env-<env_name>-res<num_simulation>. The modeling module is responsible for managing the directory structure generated during a simulation.

The contents of this root output directory include the results of the simulation, and consists of the following:

Output directories and files.

  • Eplus-env-sub_run<num_episode> directories record the results of each simulation episode. The number of directories retained depends on the value specified by the max_ep_data_store_num parameter (see Maximum episode data stored in Sinergym output). Within these directories, the structure is consistent and follows the same format, including:

    • A copy of the environment.epJSON used during the simulation episode. This does not need to match the original one, as the simulation can be modified to accommodate specific user-defined settings when defining the Gymnasium environment.

    • A copy of the weather.epw used during the simulation episode. This file will not be identical to the original when using weather variability.

    • monitor/: this directory contains information about the agent-environment interactions during the episode, timestep by timestep. It is only created when the environment has been wrapped with LoggerWrapper and CSVLogger (see Logger Wrappers). Several CSV files are included:

      • observations.csv. This file contains the observations of the environment at each timestep. The header includes the observation variables names.

      • agent_actions.csv. Contains the actions taken by the agent at each timestep. The header includes the action variables names.

      • simulated_actions.csv. Contains the actions executed during the simulation at each timestep. This values do not have to be the same as the agent actions. For example, if actions are normalized. The header includes the action variables names.

      • rewards.csv. This file contains the rewards obtained by the agent at each timestep.

      • infos.csv. This file contains the info dictionary returned by the environment at each timestep. The header corresponds the dictionary keys. Some of these keys can be ignored depending on the CSVLogger configuration.

      • normalized_observations.csv. It is only generated when the environment is wrapped with NormalizeObservation (see NormalizeObservation). The structure is similar to observations.csv, but the values are normalized.

      • custom_metrics.csv. This file is created when the logger wrapper includes custom metrics. The header includes the custom metric names.

    • output/. This directory contains EnergyPlus simulation output data. To learn more about these files, visit the EnergyPlus Output File List.

  • progress.csv. This file contains information about general simulation results. Each row contains episode information registering relevant data such as mean power consumption, rewards or comfort penalties. This file is only available when the environment has been wrapped with a LoggerWrapper and CSVLogger (see Logger Wrappers for more information). The structure of this file is defined by the LoggerWrapper class.

  • data_available.txt. It is generated when the EnergyPlus API initializes all callbacks and handlers for the simulation. In this file, you can find all the available components of the building model, such as actuators, schedulers, meters, variables, internal variables, etc.

Warning

Some component lists, such as Output:Variable, may not fully appear in data_available.txt because they need to be declared in the building model first. To view all variables or meters specifically, you should check the EnergyPlus output file. If you specify a valid variable in the environment, Sinergym will automatically add the Output:Variable element to the building model before the simulation begins.