Estimators¶
This page summarizes the estimator functions exposed by pysib.
All estimators return:
where theta is the estimated parameter vector and model is a dictionary with polynomial arrays A, B, C, D, and F.
ARX estimators¶
pysib.arx¶
Least-squares estimator for the ARX structure.
Parameters:
u: input signal.y: output signal.na: number of estimatedAparameters.nb: number of estimatedBparameters.nz: input delay.
Returns theta = [a1, ..., a_na, b1, ..., b_nb].
pysib.iv¶
Instrumental-variable estimator for the ARX structure using two repeated experiments.
Parameters:
u: input signal, shared by both experiments.y1: output signal from the first experiment.y2: output signal from the second experiment.na: number of estimatedAparameters.nb: number of estimatedBparameters.nz: input delay.
Returns theta = [a1, ..., a_na, b1, ..., b_nb].
pysib.correlation¶
ARX estimator based on correlation error minimization.
Parameters:
u: input signal.y: output signal.na: number of estimatedAparameters.nb: number of estimatedBparameters.nz: input delay.M: number of correlation lags.z: optional external instrument. If omitted,uis used.
Returns theta = [a1, ..., a_na, b1, ..., b_nb].
Output Error estimators¶
pysib.sm¶
Stieglitz-McBride estimator for the Output Error structure.
Parameters:
u: input signal.y: output signal.nb: number of estimatedBparameters.nf: number of estimatedFparameters.nz: input delay.
Returns theta = [b1, ..., b_nb, f1, ..., f_nf].
pysib.oe¶
Prediction-error estimator for the Output Error structure using the C/LAPACK nonlinear optimizer.
Parameters:
u: input signal.y: output signal.nb: number of estimatedBparameters.nf: number of estimatedFparameters.nz: input delay.
Returns theta = [b1, ..., b_nb, f1, ..., f_nf].
pysib.oe_filtered¶
Filtered continuation variant of the Output Error estimator.
Parameters and return layout are the same as pysib.oe.
ARMAX estimators¶
pysib.armax¶
Prediction-error estimator for the ARMAX structure using the C/LAPACK nonlinear optimizer.
Parameters:
u: input signal.y: output signal.na: number of estimatedAparameters.nb: number of estimatedBparameters.nc: number of estimatedCparameters.nz: input delay.
Returns theta = [a1, ..., a_na, b1, ..., b_nb, c1, ..., c_nc].
pysib.armax_filtered¶
Filtered continuation variant of the ARMAX estimator.
Parameters and return layout are the same as pysib.armax.
Box-Jenkins estimators¶
pysib.bj¶
Prediction-error estimator for the Box-Jenkins structure using the C/LAPACK nonlinear optimizer.
Parameters:
u: input signal.y: output signal.nb: number of estimatedBparameters.nc: number of estimatedCparameters.nd: number of estimatedDparameters.nf: number of estimatedFparameters.nz: input delay.
Returns theta = [b1, ..., b_nb, c1, ..., c_nc, d1, ..., d_nd, f1, ..., f_nf].
pysib.bj_filtered¶
Filtered continuation variant of the Box-Jenkins estimator.
Parameters and return layout are the same as pysib.bj.