The Fitterless Time Spread (FTS) Cut

This is a cut based on the time spread of hit PMTs within an event. The main idea is that events due to Cerenkov light have a relatively narrow time distribution compared to some types of non-physics events (e.g. flashers) and this becomes a potentially useful handle on tagging these undesirable events. Here, the absolute time differences of all pairs of hit tubes in close proximity to each other in the detector are calculated and the median time difference in the event is used as a cut statistic.

The cut algorithm is simply as follows, where words in bold/capitals are quantities that currently can be varied by the user and the corresponding word in the FTS_CUT part of the FLTH titles bank is shown in parentheses. We loop over all pairs $(i,j)$ of hit tubes (with $i \not= j$) in an event and record the absolute value of the time difference between tubes ( $\mid t_i - t_j \mid \equiv \Delta t$) for a valid pair, taking care not to double count for $(i,j)$ and $(j,i)$. A valid pair is a pair of hit PMTs which satisfy the following criteria:

• Neither channel has failed the timing calibration. This assumes the CAL processor has been run prior with full ECA + PCA calibration.
• The two PMTs are separated by a linear distance less than DIST_THRESH (word 3).
• $\Delta t$ for the pair is less than DT_THRESH (word 2).
• The two tubes are not adjacent channels nor are they both part of the same channel cluster. The criterion only applies if CLUSTER_MODE (word 1) has a value of 2.

There is a hardwired maximum of 10000 valid pairs to keep the cut from getting bogged down on high NHIT events which have an enormous number of unique pairs. The median $\Delta t$ for all valid pairs in the event is then calculated using the SNOMAN routine median.for. If the median $\Delta t$ is greater than MEDIAN_CUT (word 4) AND there are at least COUNT_THRESH (word 5) valid pairs in the event, the cut returns 'false', otherwise 'true' is returned.

The cut parameters previously described which can be adjusted are contained in the titles file filter.dat, the relevant part of which is shown below:

*DO   FLTH  11  -i(30I 1I 3F 1I)           #.  FLTH: Fitterless time spread cut
#.
19750101        0 20380517 03331900   6*0   #.   Standard Database Header
19750101        0 20380517 03331900  16*0
#.
1                            #. Cluster flag (1=keep cluster/adjacent pairs,
                             #.               2=ignore cluster/adjacent pairs)
1.0e8                        #. Only consider PMT pairs with an absolute time 
                             #.  difference less than this many nanoseconds
200.                         #. Only consider PMT pairs with tubes less than 
                             #.  this distance (in cm) from each other
10.                          #. Cut value (in ns) for median delta_t
5                            #. Require at least this many valid pairs to
                             #.  calculate median delta_t