On the treatment of energy loss in track fitting (Q1299726)

In track reconstruction with Kalman filter, both measurement and process noise are implicitly assumed to be Gaussian. However, what concerns the energy loss of electrons in matter, this is not always true and the loss is usually described by strongly non-Gaussian distribution. In such a case the Kalman filter is not necessarily the optimal filter, even if the model is perfectly linear. Therefore, in this short paper the authors examine two nonlinear estimators by means of a simulation experiment, i.e. the Gaussian-sum filter and the Metropolis-Hastings algorithm. It is shown, based on a simulation experiment, that in the studied case the momentum resolution of electrons can be improved by using a nonlinear estimator.