Table of Context
Object Oriented Programming Analysis and Design books generally
emphasize the need for a problem domain specific glossary.
The purpose of such a glossary is to make fuzzy concepts more concrete
and ensure that all participants are speaking the same language when
they use the same words. Specificity can help to eliminate the confusion
that arises when two people use the same term with subtly different
meanings or nuances.
To that end, this glossary is expected to be an evolving document
with additions as new terms come into use and individual terms have
their meanings more narrowly focussed. The discussion section at the
end is intended as a place where justifications for particulars
can be made without cluttering the glossary itself.
- active detector
- A component of the overall detector that gives measurable
- A generic term for a group of objects. Typically these
are further classified as: sets (no duplicates),
bags (duplicates), and sequences (duplicates, ordered).
Any collection object should at minimum support the
- return the number of elements in the collection;
for bag or sequence duplicates are counted as separate items.
- loop over the items in the collection visiting each item once;
for a sequence the ordering is significant.
- coordinate system
proposal: we continue to support
both a global (detector-level) coordinate system as well
as a local coordinate system for individual detector elements
(e.g. strips (logical vs. physical?)).
question: do we continue to use
cm as our basic unit of length?
question: do we continue to use
the tpos (transverse position) variable. Is this
useful? Is it well defined. A related question is the definition
of the rotations X,Y,U,V. Perhaps, U and V should be redefined
to straddle the X rather than the Y orientation.
- digitization or digit
- An unit of data corresponding to one electronic channel's response.
For the MINOS detector this is expected to consists of
two parts: an ADC value representing the photoelectron
count, and a timestamp. As the channels are expected
to be self triggering, there needn't be a digit for every
channel. The timestamp may consist of more than one
part, e.g., the output of a fast clock may be combined
with a coarse bucket number to produce a unique time.
(cf. hit and discussion
on the structure of digits below).
- This is a term fraught with ambiguity;
see discussion below.
- flat file
- flux file
- An source of particles, generated by an external program,
that serves as input to the simulation. These simulate
a variety of physics sources.
- neutrino flux
- NuMI beam simulation
- muon flux
- comic rays
- rock interactions (CC neutrino events)
question: what about correlated
muons from atmospheric showers ... are these fluxes or
- A hit, following GEANT parlance, is
a collection of unknowable truth information recorded
while tracking particles in the simulation. The hit
information relates to a single particle in a single
sensitive (or active) volume. Multiple hits in the
same volume will be collected together and used to
compute the digitization after the tracking is complete.
- An unfortunately overused term; overloading of this term
has lead to ambiguity.
- A short sequence of passive and active detector planes
that repeats itself an integral number of times to
form a supermodule.
This usage drove the origin of the term supermodule,
which has achieved common usage even if module in
this context has not.
[used by the simulation code gminos]
- A unit steel plane and active detector plane.
[used by the TDR, page 3-15]
- A light-tight package of 20 or 28 scintillator strips
that make up a active detector plane. These come in
a variety of shapes (boxes and trapazoids) and sizes.
[used by the scintillator group, cf. TDR page 5-1]
- A logical module consists of those strips
that map, due to clear fiber routing in the MUX boxes,
to different pixels. This term is nonsensical for
some optical summing schemes.
[used by the scintillator group, cf. TDR page 5-5]
- overlay event
- Overlay or pileup events are the results of multiple interactions
(beam neutrinos, through-going muons associated with the beam,
cosmic ray muons and neutrino interactions) occurring in
the same TimeSlice.
- passive plane
- A component of the overall detector that serves solely
as target material and absorber.
- The input to a uniquely identifiable electronics channel.
question: is this to be exactly
a channel, or should cross-talk impose a distinction between
- sensitive volume
- The sensitive volume of an active detector is that region
that can return a measurable signal based on the energy
deposition of a particle traversing the volume.
- Standard Reconstruction
- A length of scintillating plastic.
question: what about
"logical" vs. "physical" strips,
i.e., does the term refer to aligned strips that
are in two parts due to the break imposed by the coil, or
is each strip a single unbroken piece of plastic.
- A sequence of passive and active detector planes that
constitute a logical unit.
proposal: The baseline MINOS detectors
each have two (see discussion below).
- Proposal: a term for a collection
of digitizations within a time window (marked by a beginning
and ending timestamp). In the simulation there should also
be a mechanism for relating this window to the individual
events that contributed to the list of digitizations, even
those events that originated before the start of the TimeSlice
but generated digits within the window due to propagation times.
- A datum associated with each digitization that uniquely
identifies the time at which the information was recorded.
The structure of this item must be fleshed out in conjunction
with the DAQ software group, but it is expected to have
two components: a finely segmented clock tick and a coarse
- This is a term, like event, fraught with ambiguity;
see discussion below.
This might seem to be a silly question, but actually it is quite subtle
on several levels. Different definitions can come into play if one
is talking about the simulation or the reconstruction levels.
The intuitive definition involves assigning to an event the properties
of a single physics interaction. Furthermore one normally does not
differentiate between the "hard" interaction (say the
initial neutrino interaction) and the subsequent interactions of the
daughters products (the shower development). Thus an "event"
in such a usage would involve a particular neutrino flavor and 4-vector,
the interaction (NC vs. CC), and resultant kinematics (outgoing lepton
and fragmentation products).
While a hit can obviously be associated
with a single unique event, the same is not true of a
digit when one considers pileup or overlapping
interactions. Similarly, are digits due to noise (photodetector,
radioactivity, cosmic background) part of an event or not?
Proposal: adopt and use the term
event to mean a physics interaction, and designate a separate
distinct term for reference to a collection of digits within
a time window. Such digits may or may not be logically related via
a single event. During reconstruction one can then assign individual
digits from a collection, say TimeSlice to an object,
say RecoEvent. The digits in the data store are not owned
by either TimeSlice nor RecoEvent which thus allows
them to be used by multiple versions of each.
hits vs. digits
This bugaboo is a pet peeve of Robert's. When doing simulations one must
clearly distinguish between the information recorded while tracking particles
(hit) and the detector response
(digit). The first contains information
that is simply unknowable in the real world, while the second may be the
result of multiple particles in the same
This choice of terminology follows that used by GEANT.
Sloppiness in common usage (saying hits when one means
digits) is usually decipherable, but will occasionally lead
to temporary confusion. It would be better if all involved would
attempt to stick to precise usage. Alternatively, if there is a strong
upswelling for the use of hits as the term for data that
is delivered out of the detector, then we must find an alternative
term for the truth information.
What is an digit?
We need some discussion on what constitutes a digit.
The GMINOS concept of a digit as realized by
the FLSDigit structure. FLSDigit associates two ADC counts
and two TDC values with each logical strip.
This makes the assumption that the event is either demultiplexed
or that ghosts digits occupy each of the 7+7 alternative strips
(assuming non-zero values on each side).
Proposal: It is recommended that we
define the base digit as the output of a single electronics channel,
e.g. PixelDigit. These would be the data coming from the
DAQ system. Tentative assignment to the physical strips via a
demultiplexing algorithm can be collected into StripDigit
objects, which would not have ghosts. In reference to the
"What is an event?" discussion above, one model might have
TimeSlice collections contain PixelDigits, while
RecoEvent collections contain StripDigit objects.
What is a track?
Tracks, like the term event, has subtly different meaning
depending on the context. The issue is further confused by the
question of assigning digits (not hits) to
tracks; like events a digit may truly belong to more than one track.
This implies that any "track object" should not own
the digits that make it up.
....needs more discussion
As written, collections are a general term.
When it is relevant one should use a more specific term (set, bag, sequence)
for the particular collection type under discussion. It is expected
that in the MINOS software most collections will be either sets or
sequences. The choice has implications on what one can expect out
of the iterator.
supermodule(s) in the near detector
The term supermodule has achieved
a general usage and understanding in the context of the
baseline MINOS far detector.
In the near detector there remains an ambiguity.
One can view the near detector in three different ways.
- A single unit. There is only one coil, thus only one supermodule.
- As four parts. This distinguishes between
identically constructed regions based on their functional use.
- veto region
- target region
- hadron shower region
- muon spectrometer
- As two parts. This only distingishes between the forward
region and the downstream spectrometer.
Proposal: It is recommended that we adopt
the third usage, with the near detector having two supermodules.
This allows flexibility in designating where the "target"
region starts and ends, while retaining the distinction between
the forward region and the spectrometer.