_deepruotv2

pfi.flow.solvers.external._deepruotv2

DeepRUOTv2 external solver.

DeepRUOTv2_(dist, times, flow_model, growth_model=None, **solver_kwargs)

Fit DeepRUOTv2 from snapshot tensors.

Parameters:

• dist (list of torch.Tensor) –

  Snapshot tensors, each of shape (n_k, ndim).

• times (torch.Tensor of shape (n_snaps,)) –

  Snapshot times.

• flow_model (None) –

  Unused placeholder for API compatibility.

• growth_model (None, default: None ) –

  Unused placeholder for API compatibility.

• **solver_kwargs (dict, default: {} ) –

  DeepRUOTv2 solver options. Supported keys include exp_name and config.

Returns:

• flow ( Module ) –

  Fitted DeepRUOTv2 flow wrapper.

• growth ( Module or None ) –

  Fitted DeepRUOTv2 growth model.

• loss_hist ( ndarray ) –

  Training loss history.

to_deepruot_csv(X, csv_path)

Export time-augmented samples to DeepRUOTv2 CSV format.

Parameters:

• X (array-like of shape (n_samples, ndim + 1)) –

  Input matrix with time in the last column.

• csv_path (str or Path) –

  Output CSV path.

Returns:

• df ( DataFrame ) –

  Written dataframe with columns samples, x1, ..., xN where samples stores biological times shifted so the minimum time is 0.