This is a guest post from Ted Kolberg, who is a fellow graduate student from Notre Dame here at CERN. He has worked on the CMS ECAL for the last 4 years.
Physics object commissioning at CMS: Photons
As we start to take high-energy data with CMS, one might ask: how do we
know that the so-called “physics objects” that we reconstruct (photons,
electrons, muons, jets, taus, bs, etc…) are working properly? It’s
one matter to determine if the detector is functioning properly
(e.g. voltages, temperatures, and electronics) but it’s something
else to know if the reconstructed physics objects are working the way they
should. Commissioning these objects is a task that CMS physicists will
spend quite a bit of time on in the coming months. Using my favorite particle,
the photon, I’ll show you a few examples of the types of studies we will
do to check that our physics algorithms are correct.
One of the first areas we will have to check are the so-called isolation
and identification variables. Isolation is a measure of how many other
particles are near the particle in question in the detector. It is
defined by summing up the calorimeter and/or tracker energy in a cone
around the particle candidate. At a hadron collider like the LHC, isolationis
very important. In the case of photons,they are copiously produced
inside of jets. Not only do charged particles have a chance
to radiate photons, they also are produced as the result of
meson decays (primarily of neutral pions and eta mesons). However, these
photons are almost impossible to identify as photons because of the many
other particles nearby, and it is more theoretically interesting to look
at the ones that are isolated. Even a photon which comes from
the primary interaction might still have other particles nearby from the
underlying event (the remnants of the colliding proton), multiple
parton interactions from the same proton collision, or from pile-up
(multiple collisions occuring in the same bunch crossing). These effects
can be difficult to estimate or calculate ahead of time, so we will have
to compare the observed values of the isolation sums with our Monte Carlo
models and tune them to agree before we can proceed to studying the isolated
photons themselves.
Identification is a related but separate issue. Whereas isolation looks at
what’s around the photon, the identification step looks at the characteristics
of the photon candidate itself. We look at the shape of the photon’s energy
deposition in the electromagnetic calorimeter, whether or not the photon
“converted” into an electron-positron pair while crossing the material
in front of the calorimeter, and if there is any hadronic energy deposited
behind the candidate. All these variables give us a clue as to if the
candidate is a real photon, or a jet which just happens to look sort of like
one. The isolation described above can also be a clue as to whether it’s a
real photon or not. The unfortunate reality is that the selection is never
perfect. Sometimes, a real photon might fail your identification (this
introduces what we call “inefficiency” of the selection) or a jet might
sneak through (we call the rate of this happening the “fake rate”). For
whatever identification cuts we choose, we have to measure how efficient the
selection is and what the fake rate is. Ideally we would want 100% efficiency
and 0% fake rate, but the normal situation is that the fake rate goes up
as the efficiency goes up. There are various ways of optimizing this trade-off
and we will study this topic a lot in the coming months.
We will also check the “kinematics” of our photons: how energetic are they,
and where do they tend to fall in the detector? We have some idea of what this
should look like from the theoretical predictions and the measurements of
previous experiments, but since nobody has ever run an experiment as this high
of an energy before, we have to remain alert for any surprises. Thankfully,
basic physical concepts can help us to cross-check these distributions.
For example, everything we know about high-energy physics tells us that any
physical distribution should be symmetric around the beam pipe (e.g. there
should be the same number of particles to the top, bottom, left, and
right of the detector). Any departure from this rule in out observations
means that there is probably some kind of detector effect manifesting
itself in the data (hardware problems in a section of the detector). Just
like the distributions should be symmetric around the beam, they should
also be symmetric to the front and back of the detector. Since the LHC
collides identical protons of identical energy, the front and back of the
events should, on average, look the same. We have to make all these
basic sanity checks before moving on to more interesting measurements.
There are lots of other things we will look at to check how well photons are
working, but these are some basic concepts that also apply to most other
particles we are interested in reconstructing at CMS. Hopefully it gives
you a flavor of how we’ll be spending our time in the coming weeks and
months!
Ted
Be the first to like.
