crlb#
- blochsim.optim.crlb()[source]#
Cramer-Rao lower bound on the variance of each parameter.
- Parameters:
jacobian (torch.Tensor) –
(..., parameters, samples), derivative of the signal with respect to each parameter. Real and imaginary parts count as separate measurements.noise_variance (float, default=1.0) – Variance of the noise per real measurement, in signal units squared.
singular ({"raise", "infinite"}, default="raise") – Behaviour where the parameters are not jointly identifiable: raise, or return an infinite variance at those entries.
- Returns:
(..., parameters), diagonal of the inverse Fisher matrix.- Return type:
torch.Tensor
- Raises:
ValueError – If the Jacobian has no parameter axis, or
singularis invalid.torch.linalg.LinAlgError – If the Fisher matrix is singular and
singular="raise".
