Generalized Least Squares Fit: the Sequel

For the bias-free pT reconstruction of tracks in the HARP TPC, especially of helical tracks with small pT, and for optimum resolution and charge separation, our `Generalized Least Squares Fit' is the best algorithm currently available. We present an efficient implementation of this fit of a helical track, based on the TOPAZ parametrization. Apart from the fast convergence in typically three iterations, our fit allows for change of charge sign in the search for minimum Chi2, which permits optimum charge separation capability. For typical HARP TPC coordinate errors, our fit achieves a 12% better pT resolution than obtained with the original `Chernov-Ososkov' circle fit, and by a factor of two smaller charge misidentification probability at pT= 15 GeV/c.