CASBI.inference

Module Contents

Classes

CustomDataset_halo	Custom dataset class for the training and validation datasets.
CustomDataset_subhalo	Custom dataset class for the training and validation datasets.

Functions

train_inference

Train a ltu-ili ensable model on (x, theta) couples. The training data is split into training and validation data. The model is saved in the output_dir directory.

API

class CASBI.inference.CustomDataset_halo(observation, parameters, device)

Bases: torch.utils.data.Dataset

Custom dataset class for the training and validation datasets.

observationtorch.Tensor

The observation data.

parameterstorch.Tensor

The parameter data.

Initialization

__len__()

__getitem__(idx)

class CASBI.inference.CustomDataset_subhalo(observation, parameters, device)

Bases: torch.utils.data.Dataset

Custom dataset class for the training and validation datasets.

observationtorch.Tensor

The observation data.

parameterstorch.Tensor

The parameter data.

Initialization

__len__()

__getitem__(idx)

CASBI.inference.train_inference(x: torch.Tensor, theta: torch.Tensor, validation_fraction: float = 0.2, output_dir: str = './', device: str = 'cuda', N_nets=4, hidden_feature: int = 100, num_transforms: int = 20, model: str = 'nsf', embedding_net: str = ConvNet_halo(output_dim=32), custom_dataset: torch.utils.data.Dataset = CustomDataset_halo, custom_dataloader: torch.utils.data.DataLoader = torch.utils.data.DataLoader, minimum_theta: list = [3.5, -2.0], maximum_theta: list = [10, 1.15], batch_size: int = 2048, learning_rate: float = 1e-05, stop_after_epochs: int = 20, norm_x=False, norm_theta=True)

Train a ltu-ili ensable model on (x, theta) couples. The training data is split into training and validation data. The model is saved in the output_dir directory.

xtorch.Tensor

The observation data.

thetatorch.Tensor

The parameter data.

validation_fractionfloat, optional

The fraction of the data to use for validation, by default 0.2

output_dirstr, optional

The directory where the model is saved, by default ‘./’

devicestr, optional

The device to use, by default ‘cuda’

N_netsint, optional

The number of nets to ensable, by default 4

hidden_featureint, optional

The number of hidden features of the model, by default 100

num_transformsint, optional

The number of transforms of the model, by default 10

modelstr, optional

The model to use, by default ‘nsf’

embedding_netstr, optional

The embedding network to use, by default ‘CNN’. The CNN.py file should be in the same directory as the training script.

output_dimint, optional

The output dimension of the embedding network, by default 32

minimum_thetalist, optional

The minimum value of the theta parameters, by default [3.5, -2.]

maximum_thetalist, optional

The maximum value of the theta parameters, by default [10, 1.15]

batch_sizeint, optional

The batch size, by default 1024

learning_ratefloat, optional

The learning rate, by default 0.00001

stop_after_epochsint, optional

The number of epochs to train, by default 20

posterior_ensembleili.posterior.PosteriorEnsemble

The posterior ensemble model.

summariesdict

The summaries of the training process.