io

class irsa.io.ExpPredDataset(exp, exp_labels, pred, pred_labels)

Bases: Dataset

Create pairs for M experiment spectra and N predicted spectra. (M x N pairs)

Parameters:

• exp (array) – Array of experiment values.

• exp_labels (array) – Array of experiment labels.

• pred (array) – Array of predicted values.

• pred_labels (array) – Array of predicted labels.

_create_pair(exp_idx, pred_idx)

Create data pairing for a single experimental instance paired with a predicted instance.

class irsa.io.PairedExpPredDataset(exp, exp_labels, pred, pred_labels, deterministic=False)

Bases: Dataset

Create true positive & true negative pairs of experiment and predicted spectra.

Parameters:

• exp (array) – Array of experiment values.

• exp_labels (array) – Array of experiment labels.

• pred (array) – Array of predicted values.

• pred_labels (array) – Array of predicted labels.

• deterministic (bool) – True for np.random shuffling only once at initialization. False for np.random shuffling each iteration.

_get_different()

Return different experimental and predicted data, label=0. As this class is larger, indices will be selected randomly.

_get_different_deterministic(idx)

Return same-labeled spectra, label=1.

_get_same(idx)

Return same-labeled spectra, label=1.

irsa.io.load_experimental(path)

Load experimental spectra from path using JCAMP.

Parameters:

path (str) – Path to DX file.

Returns:

• freq (array) – Array of frequencies.

• intensity (array) – Array of intensities.

irsa.io.load_model(path, model=<class 'irsa.networks.PairedNeuralNet'>, **kwargs)

Load model from path.

Parameters:

• path (str) – Path to saved model state dict.

• model (Module) – Model class to initialize.

• kwargs – Keyword arguments for model initialization.

Returns:

Initialized model.

Return type:

Module

irsa.io.load_predicted(path)

Load NWChem-predicted spectra from path using ISiCLE.

Parameters:

path (str) – Path to ISiCLE file.

Returns:

• freq (array) – Array of frequencies.

• intensity (array) – Array of intensities.

irsa.io.preprocess_predicted(freq, intensity, exp_freq, sigma=2)

Process predicted spectra to broaden, normalize, and map to experimental frequencies.

Parameters:

• freq (array) – Array of predicted frequencies.

• intensity (array) – Array of predicted intensities.

• exp_freq (array) – Array of experimental frequencies.

• sigma (int) – Sigma value to broaden by Gaussian function.

Returns:

• freq (array) – Array of experimental frequencies.

• intensity (array) – Array of processed intensities at experimental frequencies.