2 The CDF Experiment at Fermilab Contents - Harvard University ...

Section 2: Run II Discovery Physics 65
2.4 Run II and Beyond
The prospect of a much larger and improved data set raises hopes for the discovery of physics
beyond the Standard Model. The Harvard group has set its sights on discovery physics and
on measurements of CP violation in the b system. We describe our plans and achievements
in these two areas below.
2.4.1 Discovery Physics
There are two approaches to searching for New Physics which are sometimes called \datadriven"
and \theory-driven." They are somewhat opposed philosophically, and each has it
advantages and disadvantages.
The idea behind data-driven searches is that a well-dened event sample thought to be
explained by known processes should be studied thoroughly with an eye open for important
discrepancies between observation and expectation. For example, a valid measurement of
the cross section for tt production, in principle a well-understood SM process, could turn
out higher than expected. One would conclude that the selected event sample would contain
events from new physics processes. If the measurement were below expectation, then it could
indicate unusual top decays to states not satisfying the given selection criteria. If pursued
seriously and with alacrity, almost any measurement of SM processes could open a window
to an unexpected discovery.
The weakness of data-driven searches is the danger that uctuations are innocently massaged
into signals. This weakness is avoided in the theory-driven approach, in which a signal
is proposed and modeled, allowing a priori selection criteria to be designed using simulations
before the data are queried. In principle the criteria are unbiased with respect to features of
the data, so when applied to the real data, a signicant excess of events is either observed, or
not. In practice the diculty of modeling some of the backgrounds often precludes nalizing
the selection criteria before the data are looked at. More importantly, the criteria designed
for a signal suggested by a popular theory may fail to ferret out a real signal quite dierent
from what that theory predicts. In other words, there is a danger of looking very seriously
and systematically for the wrong signals.
We do not wish to adopt a dogma for Run II. We believe that the the job of the experimenter
is to study the data and to see whether a given interesting sample can be fully
understood. At the same time, current theoretical ideas about extensions of the SM are