Next: Robust Estimation---Motivation---III
Up: 5: Goodness of
Previous: 5.6Robust Estimation--Motivation--I
- Consider an example from physics where one observes mass
distribution with a Breit Wigner for a resonance, say,
plus ``small'' (usually large!) contributions to the probability
distribution (e.g., background which must be estimated as
polynomial/phase space) which must be parameterized in an ad hoc fashion.
- It is hard to evaluate distortion due to an incorrect
parameterization of these unknown contributions.
- Obviously, the correct choice of method will depend on relative
size of signal (with known functional form), and background.
Geoffrey Fox, Northeast Parallel Architectures Center at Syracuse University, gcf@npac.syr.edu