In the documentation for the UnscentedKalmanPredictor, it states that the default parameter for kappa is 3-Ns, where Ns is the state dimension. However, the class actually uses a default value of kappa=0. In the gauss2sigma function, kappa will be set to 3-Ns if None is passed to the kappa argument, but when an UnscentedKalmanPredictor is initialised, it will default to 0.
We should either change the documentation to suggest that kappa=0 is the default in the UnscentedKalmanPredictor, or change the default value to None so that we utilise the kappa=3-Ns in the gauss2sigma function.
The default value of kappa=3-Ns is also mentioned in the unscented Kalman filter tutorial.
In the documentation for the UnscentedKalmanPredictor, it states that the default parameter for kappa is 3-Ns, where Ns is the state dimension. However, the class actually uses a default value of kappa=0. In the gauss2sigma function, kappa will be set to 3-Ns if None is passed to the kappa argument, but when an UnscentedKalmanPredictor is initialised, it will default to 0.
We should either change the documentation to suggest that kappa=0 is the default in the UnscentedKalmanPredictor, or change the default value to None so that we utilise the kappa=3-Ns in the gauss2sigma function.
The default value of kappa=3-Ns is also mentioned in the unscented Kalman filter tutorial.