PyKF.KalmanFiltering.LinearFilter

Package PyKF :: Module KalmanFiltering :: Class LinearFilter

[show private | hide private]

Class LinearFilter

Linear Kalman filter, using Ricker recruitment as an example.
Algorithms based on Peterman et al. (2000), Haykin (2001)
    
Measurement equation:
    y[t] = H*x[t] + v[t]
    
Process equation:
    x[t+1] = F*x[t] + w[t]
    
This algorithm also (optionally) updates the measurement variance
according to West and Harrison (1997), pp52-57.

Method Summary
	`__init__(self, system_coefficients, measurement_coefficients, process_covariance, measurement_covariance)`
	`filter(self, x_hat, P_hat, y)` Performs Kalman filtering on estimated state (parameter) and covariance, based on an observation.
	`measurement(self, x_, P_, y)` Update estimates of state and covariance:...
	`process(self, x, P)` Propagates system from time t to t+1:...

Method Details

filter(self, x_hat, P_hat, y)

Performs Kalman filtering on estimated state (parameter) and covariance, based on an observation.

measurement(self, x_, P_, y)

Update estimates of state and covariance:
    
    x[t] = x[t]_ + G*(y[t] - H*x[t]_)
    
    P[t] = (I - G*H)*P[t]_

process(self, x, P)

Propagates system from time t to t+1:
    
    x[t+1] = F*x[t]
    
    P[t+1] = F*P[t]*t(F) + V[t]

Generated by Epydoc 2.1 on Tue Mar 22 13:16:54 2005

http://epydoc.sf.net