Tutorial: using NeurEco command line interface for a Discrete Dynamic problem

NeurEcoRNN is the executable used for building, evaluating and exporting Dynamic models (Discrete Dynamic). The executable can be called directly from a terminal / powershell only after a full installation (the portable version doesn’t allow this option). To call the executable, run the following command:

neurecoRNN

which will output:

Running NeurEco Dynamic version 4.01.1154.0 compiled with MSVC v1928  on Dec  5 2022 @ 15:55:33
usage: neurecoRNN [-h] [command <parameters>]

Entry point for neurecoRNN network building and evaluation.

Commands:
build <configurationFilename>
        build a neurecoRNN network from a given input solution/excitation set.

evaluate <configurationFilename>
        evaluate a neurecoRNN network from a given input solution/excitation set.

exportFMU <serialized network full path> <FMU model full path> <ORed platform flag (windows=1, linux=2)>
        export a serialized network as an FMU file.


Optional arguments:
-h, --help   show this message and exit

The following section uses the test case Temperature forecasting. This test case is delivered with the NeurEco installation package.

To build a Discrete Dynamic model using the executable:

• Create a configuration file .conf for build, here called build_configuration_file.conf (see Build NeurEco Discrete Dynamic model with the command line interface). For the test case Temperature forecasting, the configuration file for build looks as follows:

{"neurecoRNN_build":
        {
            "exc_filenames": ["./x_first_year.npy"],
            "output_filenames": ["./y_first_year.npy"],
            "validation_exc_filenames": [],
            "validation_output_filenames": [],
            "test_exc_filenames": ["./x_second_year.npy"],
            "test_output_filenames": ["./y_second_year.npy"],
            "write_model_to": "./TemperatureForecasting.ernn",
            "write_model_output_to_directory": "",
            "checkpoint_address": "./TemperatureForecasting.checkpoint",
            "resume": false,
            "settings": {
                "valid_percentage": 30,
                "min_hidden_state": 1,
                "max_hidden_state": 0,
                "steady_state_exc": [],
                "steady_state_out": [],
                "input_normalization": {
                    "shift_type": "mean",
                    "scale_type": "l2",
                    "normalize_per_feature": true},
                "output_normalization": {
                    "shift_type": "mean",
                    "scale_type": "l2",
                    "normalize_per_feature": true}
                    }
        },
}

• Place this configuration file in the same directory as the data of the test case (x_first_year.npy, x_second_year.npy, y_first_year.npy, y_second_year.npy), otherwise adjust the relative paths to the data files in the configuration file

• To launch the build, run the following command in the terminal (opened in the data directory, otherwise adjust the relative path to the configuration file):

neurecoRNN build ./build_configuration_file.conf

• The build starts automatically:

00h00m00s info > Running NeurEco Dynamic version 4.01.1154.0 compiled with MSVC v1928  on Dec  5 2022 @ 15:55:33
00h00m00s warning > Configuration file has no member 'write_model_output_to_directory', using default value '.'
00h00m00s info > Reading Dataset...

To evaluate a Discrete Dynamic model using the executable:

• Create a configuration .conf file for evaluation, here called eval_configuration_file.conf (see Evaluate NeurEco Discrete Dynamic model with the command line interface). For the test case Temperature forecasting, the configuration file for evaluation looks, for example, as follows:

{
         "neurecoRNN_evaluate": {
     "exc_filenames": ["./x_second_year.npy"],
     "init_output_filenames": [],
     "init_exc_filenames": [],
     "ernn_filename": "./TemperatureForecasting.ernn",
     "write_model_output_to_directory": "./EvaluationResults"
 }
     }

• Place this configuration file in the same directory as the data of the test case (x_second_year.npy), otherwise adjust the relative paths to the data files in the configuration file

• To launch the evaluation, run the following command in the terminal (opened in the data directory, otherwise adjust the relative path to the configuration file):

neurecoRNN evaluate ./eval_configuration_file.conf

• The model is evaluated on the testing data in “./x_second_year.npy”, and the results are saved in the directory created by NeurEco: “./EvaluationResults”.

To export a Discrete Dynamic model to the FMU format using the executable (embed license is required):

• Run the following command:

neurecoRNN exportFMU ./TemperatureForecasting.ernn ./TemperatureForecasting.fmu