I will show how the machine lattice can be constructed from TBT data
with the joint Kalman filter (JKF). The JKF algorithm finds a fit
to the TBT data so that the beta functions can be constructed and
lattice errors can be identified. Another feature of JKF is that
noise from the BPMs, pulse to pulse variations of the injected beam,
error tolerances of the results can all be taken into account. As a
demonstration of JKF, I will show how the JKF can reconstruct the
lattice and identify the source of the error when it is applied to
TBT (turn by turn) data generated from a toy FODO lattice with a
quad strength error.