HDFast questions and answers

[Elton Smith <elton@jlab.org>]

---------- Forwarded message ----------
Date: Mon, 21 Jan 2002 10:22:10 -0500
From: Richard Jones <Richard.T.Jones@uconn.edu>
Reply-To: jonesrt@uconnvm.uconn.edu
To: James Hubbard <jhubbard@MCS.UVAWISE.EDU>
Cc: hdgeant@zeus.phys.uconn.edu
Subject: Re: More questions

Hello James,

> 1.) We're now to the point where can do our own simulations.  Now we
> want to be able to do simulations for some other reactions. The 3 that
> we need to look at now are
>
> for background:
> gamma p -> delta+ -> p pi0
> gamma p -> delta+ -> n pi+
>
> for another signal:
> gamma p -> X(1600)+ n -> pi+pi-pi+n  (this is the same one as in the
> default HDFast code)
>
> Can you help us with this?

This sounds reasonable.  Why don't you go ahead with it, and continue to consult
with me as you run into puzzles or obstacles.

> 2.) In the analysis we're getting some events where there is only 1
> track in either the startCntr or the forwardTOF.  This seems to be
> happening where the angles are very far forward.
>
> We need to calculate the opening angle (or the minimum track angle
> picked up by the start counter.  What we need to know is the diameter,
> length, and interaction point in the start counter. We have looked in
> your geometry file but aren't sure how to interpret it.  Could you
> either give us the information or tell us how to dig it out of the
> geometry file?

The easiest way to look at the geometry is to plot it and use a ruler.  Geant
has a nice plotting package that I use to visualize the geometry.  To use it
you should build hdgeant++ (as in "make hdgeant++") and then start it.  It is
just like paw, except it includes geant commands.  If you prefer the Motif
interface, when it asks for your Workstation Type? reply with the letter "m",
otherwise you get the plain pawX11 interface.  To make a plot where you can
see the target region geometry, try

GEANT> satt cdc seen -2
GEANT> next; dcut hall x 1 5 10 0.1 0.1

The satt command is just there to tell Geant not to plot every straw in the
cdc, otherwise it takes a while.  To see what the contents of any region of
the plot are, use the "pick" command.  To measure distances between points
in real distance units, use the "measure" command.

> 3.)  I'm attaching some graphs of the analysis on the 10000 event file.
>     The main thing here is that the forward cal energy doesn't seem to
> be correct.  It's relatively flat whereas in the barrel cal it drops off
> as expected. We showed this analysis at our meeting today, and Elton was
> also slightly shocked at the "flatness" of the forward cal energy
> deposition.  You can see the minimum ionizing peak, but then the energy
> is fairly flat after that. Is it possible that GEANT is calculating the
> energy lost by the particle in interactions, but not taking into account
> that not all of that is seen in the lead glass?
>
> (You'll notice that there is a -1 for energy in the graphs for the
> forward and barrel cal. All that means is that there was nothing that
> happend in that cal.)

What you are looking at here is the response of the forward calorimeter to
incident charged pions, that is, to a hadronic shower.  The probability
that a high-energy pion passes through the Fcal without undergoing an
inelastic strong collision is only about 35%; that both pions leave only
MIP signals in the lead glass is about 10%.  When there is an inelastic
collision with a nucleus, it generally disperses energy through nucleon
emission and multi-pion production.  From your plot, it would appear that
in these cases on the average 50% of the energy is deposited via dE/dx in
the glass and 50% escapes (neutrons, penetrating pions, a few gammas).
This plot looks to me like the typical response of a leaky hadron calorimeter.
All this shows is that we have a poor hadron calorimeter.

What kind of structure did you expect to see?  The behavior in the barrel is
no different, except that there the incident energy distribution is much
softer.  I would expect that the lead/SciFi would be a better hadronic
calorimeter than the lead glass, but you would have to cast them in the same
geometry to be able to compare them.

Perhaps you are misinterpreting the peak at low energy from stopping protons
in the barrel as a MIP peak, whereas it is mostly a total absorption peak.

This simulation does fall short in the following sense, that it does not take
into account the Cerenkov nature of the lead-glass response.  Right now the
signal from the glass is proportional to the total ionization energy deposition
regardless of particle velocity.  As soon as we need a detailed response in
the lead glass to different particle types then we need to replace that with
real Cerenkov generation.  There is no reason why that has not been done
already, except that no one has done it.  I hope to get those responsible for
the subsystems to take responsibility for the response of their own detector
in the simulation.  Does that make sense?

Richard Jones