Generates batch_size Brownian Bridge trajectories.

Generates batch_size Brownian motion trajectories, extracts sub-trajectories.

Convert a Tensor with dimensions [batch, series * time steps, ...] to one with dimensions [batch, series, time steps, ...]

Standardize column names and perform necessary data transformations.

Replace the dataframe's datetime index with a continuous daily index. For each unique date in the original index, a new date

Build a dictionary structure of checkpoints from specified dataset directories. Args: base_log_dir: Base path containing dataset-spe

Parameters ---------- ts_iterator Iterator over time series forecast_iterator Iterator over

Parameters ---------- all_agg_metrics dictionary with aggregate metrics of individual dimensions all_met

Check the total memory and occupied memory for GPU

input : the forecasts from forecasts = list(forecast_it) metric func in "total_variation",

Filter and extract rows from csv_file based on config and hypothesis count.

Recursively convert string values in a dictionary (or list) to numeric values (int or float) if possible.

cosine schedule as proposed in https://openreview.net/forum?id=-NEXDKk8gZ

Count FLOPs for the prediction process

Create and return the network used for training (i.e., computing the loss). Returns ------- pl.LightningModu

Create and return a predictor object. Parameters ---------- transformation Transformation to be applied

Create a data loader for training purposes. Parameters ---------- data Dataset from which to create the

Create and return the network used for training (i.e., computing the loss). Returns ------- nn.Module

Add a transformation that replaces NaN in the input data with zeros, and mention whether the data was a NaN or not in another field.

Create and return the transformation needed for training and inference. Returns ------- Transformation T

Create a data loader for validation purposes. Parameters ---------- data Dataset from which to create th

Undo the normalization done in the normalize() function. Parameters: ----------- norm_value: Tensor [batch, series,

Parameters ---------- time_series forecast Returns ------- np.ndarray time seri

Parameters ---------- time_series forecast Returns ------- np.ndarray time seri

compute the forecasts from the BaseForecastModel

Compute the embedding for each series and time step. Parameters: ----------- encoded: Tensor [batch, series, time st

Compute the cumulative density function of a marginal conditioned using the given context, for the given value of x. Also returns the

Transform the given value according to the given parameters, but does not compute the derivative of the transformation. param

Compute the cumulative density function of a marginal conditioned using the given context, for the given value of x. Parameters:

Generates complete trajectories of an AR(p) process and extracts sub-trajectories. :param batch_size: Number of sub-trajectories

Recursively find all checkpoint files in a folder. Args: folder_path: Path to the folder to search in Returns: List of

Instantiates callbacks from config. :param callbacks_cfg: A DictConfig object containing callback configurations. :return: A list of instanti

Instantiates loggers from config. :param logger_cfg: A DictConfig object containing logger configurations. :return: A list of instantiated lo

NOTE: Added for compatability with Alex' notebook "ground_truth_copula.ipynb" demonstrating the 2-dimensional example This function

Controls which config parts are saved by Lightning loggers. Additionally saves: - Number of model parameters :param object_dict: A d

Compute a sum in logarithm space: log(exp(a) + exp(b)) Properly handle values which exponential cannot be represented in float32.

Find the exact file path in the directory matching the given path pattern.

Main function to run the checkpoint extraction script. Command line arguments: --log_dir: Base path containing dataset-specific log

nb simu : the number of time you want to compute the models time nb_hyp: the number of hypothesis

Parse a folder name to extract its components. Args: folder_name: String in format: 'YYYY-MM-DD_HH-MM-SS_dataset_model_n

Plot the history and forecasts with the mean trajectory. Also save the data in a pkl file.

Plot forecasts from either timeMCL or TimeGrad models. Args: ... (existing args) ... is_mcl: Whether the forecasts come from

Plot a single method's results in one column

Plot different methods side by side Args: dataset: str, name of the dataset methods: list of str, names of methods to plot (e