When we look at the results produced by HDGeant from simulating a set of events that were generated by a Monte Carlo program, for example genr8, it lists for the coordinates of the event vertex 0,0,0. Yet according to the documentation for the geometry specification, the origin lies upstream of the target. Shouldn't the simulation produce events that start inside the target?

The answer to your question is found in the following comment which I cut and paste from the head of the hdgeant source file hddmInput.c.

 * Usage Notes:
 * 1)   Most Monte Carlo generators do not care where the vertex is placed
 *      inside the target, and specify only the final-state particles'
 *      momenta.  In this case the vertex position has to be randomized by
 *      the simulation within the beam/target overlap volume.  If the vertex
 *      position from the generator is (0,0,0) then the simulation vertex is
 *      generated uniformly inside the cylinder specified by TARGET_LENGTH,
 *      BEAM_DIAMETER, and TARGET_CENTER defined below.
 */

#define TARGET_LENGTH 30
#define BEAM_DIAMETER 0.5
#define TARGET_CENTER 65

Do you need a new "truth" tag in the hitView group that reports the actual vertex? If you think that would be useful then I can add it, although I would warn you against relying on it since for real data that information will be missing.

When we look at the results produced by HDGeant from simulated photoreactions, we see what we many of us think are surprising distributions for the energy deposition of charged pions in the forward calorimeter. The response of lead-glass to charged pions should be minimal unless they interact, and even then only a small fraction of the energy should be contained inside the glass. For an example of this, look at Fig. 1 at the right. Is there something wrong with the way the simulation is treating pions in the lead glass? Is this simulation realistic enough to use as a basis for the trigger design?

Figure 1: In this simulation, the charge-exchange reaction γp ­> n X(1600) is simulated where the decay of the X is into phase-space π⁺π⁺π^-. The charged pions are produced preferentially in the forward direction and produce light in the forward calorimeter. The histogram shows the total energy deposited in the lead-glass per event, which includes the sum from all three charged tracks if they are within the acceptance of the LGD. The energy of the incident photon is 9 GeV.

I know a good way to find out: look at some real data. The data from E852 at Brookhaven contain a large number of charged pions. Their momentum is measured in the dipole magnet and by looking for clusters around the impact point on the face of the LGD I can measure the pions energy deposition fraction. I do the following:

1. I select data where the trigger had no LGD requirement
2. I find charged tracks in the spectrometer
3. I swim those tracks to the plane of the LGD and demand the projected point be in the useful area of the LGD
4. I find the nearest cluster to the projected impact point
5. If that cluster is within 20 cm of the impact point I assume it is associated with the pion.
6. If no cluster is found, an LGD energy of zero is used otherwise, the observed energy in the LGD is used.
7. I plot up E_lgd/E_pion as shown in Fig. 2.

Figure 2: The ratio E_lgd/E_pion for charged pion tracks that intersect the LGD in AGS experiment E852. I find:

Dr. Teige thanks for the information. I have made a plot from our data for just a pi+. I've moved all of the non interacting pi+'s off of the graph. The result is shown in Fig. 3.

Figure 3: Energy deposited in the LGD by a single pi+, plotted as a fraction of the incident pi+ energy, for the reaction X(1600) -> n pi+ pi- pi+. The data is for a single track. The graphs for the other pi- and pi+ are basically the same.

Total Events:	9851
Non Inter:	2170	22%
.1 - .2	2154	22%
.2 - .3	1008	10%
> .8	75	0.7%

There seems to be quite a bit of difference between what we're getting and what Dr. Teige is getting.

The agreement actually looks rather good to me. I get 16% between 0.1 and 0.2, MC gets 22%. I get 9% between .2 and .3, MC gets 10%. The fraction that do not interact (when measured from data) depends on software thresholds, this may be a source of disagreement. The shape of the energy fraction distribution also seems to agree but a numerical comparison would be nice.

I've got a couple of questions. We've developed a couple of ideas about settings for a trigger. One cut involves the total energy in the barrel and forward cal as well as the whether the energy is greater in the barrel. The other cut involves the energy in the forward cal and whether or not there are any tracks in the time of flight.

1. Is there some way of changing the probability of a particles interacting without having to re-compile or do we need to modify geant? (If we need to modify geant can you point me to a place can make changes?)
2. It's basically the same, but I need to know if there is a way that I can modify the amount of hadronic energy being deposited? (Can you point me to a starting point where I can make changes?)

Before we start kludging the simulation, we need to see if it is correct. When Scott says that the comparison looks good to him, he is not claiming that the plots are identical. As you have pointed out they are not. But they are not plots of exactly the same thing. His data were collected with a 18GeV pion beam, different cuts and maybe clusterization in the LGD. His point is simply that from his data he thinks that the interaction rates in the simulation look reasonable. If you would like a more quantitative comparison then we will need to simulate Scott's conditions, and analyze the data with his analyzer.

Some questions for you, Scott. Can you give more details about how your plot was produced? Your pions were tracked through a magnetic field and projected onto the LGD, right? Did you do a clusterization of the LGD hits and then take the closest cluster to the pion impact? What cuts did you make on that association? What were your clusterization parameters?