pyacs.lib.robustestimators.dikin

L1-norm estimation using Dikin’s method.

pyacs.lib.robustestimators.dikin.Dikin(A, y, W, eps=0.003)[source]

L1-norm estimation using Dikin’s method (linear program y = Ax + e).

Parameters:

A (numpy.ndarray) – Design matrix.
y (numpy.ndarray) – Observation vector.
W (numpy.ndarray, optional) – Diagonal weight matrix for observables. Can be None.
eps (float, optional) – Stopping criterion for iteration. Default is 3e-3.

Returns:

x (numpy.ndarray) – Estimated parameters.
e (numpy.ndarray) – Residuals.

Notes

Reference: A. Khodabandeh and A. R. Amiri-Simkooei, “Recursive Algorithm for L1 Norm Estimation in Linear Models”, J. Surveying Engineering ASCE (2011). doi:10.1061/(ASCE)SU.1943-5428.0000031. Translated from Matlab by J.-M. Nocquet (2011).