We have had discussions here at Sitges about including backgrounds in our LCD simulation system. While it is fresh in my mind I wanted to summarize it and distribute it for further comment especially since I won´t be back for another two weeks.

In addition to simulating "signal" processes (eg ee-> ZZ, etc) we have at present two kinds of backgrounds to simulate. One is the two photon (minijet) process using the BBHAD program and the other is beamsthralung using Guinea Pig. The general plan is to create generator level output files for each type of background in stdhep format. Then run those through the detector simulation program creating simulation level output. Then at the reconstruction/analysis stage the backgrounds will be overlain on the signal processes.

More detail is required here. First, define some terms. The beam consists of separate bunch trains each consisting of order 100 bunches. Because of  angled crossing each bunch will only interact with one bunch from the opposite bunch train. Call this a bunch crossing. The two photon process is conceptually like any signal process. On any bunch crossing a single interaction has some relatively low probability to take place. The beamsthralung is quite different. On each bunch crossing  beamsthralung interactions will produce approximately 50K ee pairs. Most of these will have very low transverse momentum and will never leave the beampipe due to the detector magnetic field. The order 5K pairs that will interact will be saved in the generator level output file.

It is possible for different detector components to have different read out characteristics  regarding timing. One component may integrate its readout over the entire bunch train and another may read out at the bunch by bunch level. To allow for these different readout possibilities when events are overlain at the reconstruction level each hit will include a timestamp/bunch bin ID tag. Then a general overlay scheme would involve overlaying a number of events which could be selected by the user at reconstruction/analysis time. 

For example, if the analysis is sensitive to forward backgrounds one might overlay a signal event, one or more minijet events and an entire bunch train of beamsthralung events. The signal event might be assigned to bunch bin 53, two minijets assigned to bins 22 and 56 and one beamsthralung event to each bin. 

Suppose the forward calorimeters had fast readout and could discriminate at the level of 10 bunch bins but the forward tracking could only discriminate at the bunch train level. One could include all the overlay data in the tracking reconstruction but only include cal information from the ten bins surrounding the signal event. This is just a madeup example to be concrete about the ideas. However, it seems likely that some of the trigger systems will be sensitive to the sub-bunch train timing level.

This is the picture I have of what we need to implement. Advice? Corrections? Comments?

-----------------------------------

Gary Bower  ([log in to unmask])
Stanford Linear Accelerator Center
Stanford University
650/926-2460