rxmc.covariance.noise_term

rxmc.covariance.noise_term(parameter, log=True, basis=None, basis_params=(), support=None, coords=None) → Term

Unknown statistical noise diag((epsilon * basis)**2).

basis defaults to ones (constant noise); pass any basis(c, *basis_values) — e.g. exp_growth() with basis_params=(slope,) for noise growing along x.

This term is additive: a Constraint already adds each observation’s reported statistical diagonal, so the assembled covariance is diag(y_stat_err**2 + epsilon**2). To make the inferred noise replace the reported statistics, build the Observation with zero y_stat_err or pass include_statistical_term=False to the Constraint.