TorchBicop.from_data

classmethod TorchBicop.from_data(u, controls=None, *, cache_integrals=True, device=None, dtype=torch.float64)

Fits a bicop on pseudo-observations and wraps in a TorchBicop.

Dispatches on controls.method ("tll": pure-torch Transformed Local Likelihood, matching the Bicop TLL fit to machine precision).

Parameters:

undarray or Tensor, shape (n, 2), dtype float: Pseudo-observations.
controlsFitControlsTorchBicop or None, default=None: Fit-time controls. None defaults to TLL with grid_size=30 on the normal-spaced grid.
cache_integralsbool, default=True: See __init__.
devicetorch.device or None, default=None: See __init__.
dtypetorch.dtype, default=torch.float64: See __init__.

Returns: