Hi Miles,
Thanks for your report on your studies. If you can make it to any of our
meetings, we can discuss this further. Also, if you have any responses
from what I write in the minutes, you can send them to
clas_gsim@cebaf.gov. We're very interested in what you can tell us.
Best regards,
- Will
--------------02E8844A572A4E979DCCD85B
Content-Type: message/rfc822
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Received: from smtp3.erols.com (smtp3.erols.com [207.172.3.236]) by jlabh4.jlab.org with ESMTP (8.8.6 (PHNE_14041)/8.7.1) id MAA21779 for <brooksw@cebaf.gov>; Tue, 24 Mar 1998 12:14:49 -0500 (EST)
Received: from resnet (207-172-108-12.s69.as2.wlm.erols.com [207.172.108.12])
by smtp3.erols.com (8.8.8/8.8.5) with SMTP id MAA21639
for <brooksw@cebaf.gov>; Tue, 24 Mar 1998 12:14:33 -0500 (EST)
Message-Id: <3.0.5.32.19980324121606.009454d0@pop.erols.com>
X-Sender: macon@pop.erols.com
X-Mailer: QUALCOMM Windows Eudora Pro Version 3.0.5 (32)
Date: Tue, 24 Mar 1998 12:16:06 -0500
To: brooksw@cebaf.gov
From: "Miles Cary Macon Johnston, III" <macon@erols.com>
Subject: about the GSIM project group
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Hello, my name is Miles Johnston, and I am currently doing a senior project
under Herbert Funsten, and it deals with speeding up the GSIM program. I
have only a limited amount of time left this semester, and it seems this
will be my last (hopefully). I was looking mainly at the energy cuts on
the eletron calorimeter, and whether or not the default cuts may be made at
higher energies to help speed the program up. At the moment, it seems that
setting cuts will not help much. I have taken the original defaults
(which, as far as I can tell are 1 MeV) up to 100MeV and looked at the
speed increase and the data out. Supposedly, the graph (scatterplot) or
energy on the inner and outer calorimeters should have a band which could
be fit by a straight line. In the case of 100MeV, this band all but falls
apart. And the time per event in this case is dramatically reduced. I
have tlked wiht Maurik Holtrop about the times per event, and might need to
do a little recalculation, since he said that the GSIM output file time is
biased. I will check on that. But, as of this point in looking into the
different energy cuts, the electron and gamma cuts make the difference.
Which is what could be expected in the calorimeter.
So at this point in time, I feel that it is safe to say that the default
energy cuts in the ffread.in file used in GSIM cannot be altered to produce
a significant speed increase per event. I quote Maurik:
>The GSIM code uses GeV throughout. This makes sense, it
>means that 1.e-4 cuts are equivalent to 0.1 MeV, or 100 KeV, which is
about the lowest scale >we'd be interested in. Setting the cuts to 10 MeV
(1.e-2GeV) would mean that the granularity of the >EC code gets too large.
However,at that threshold a lot of tracks are cut, hence the speed increase.
Yes, it seems that lots of tracks are cut, and the granularity does
increase significantly. I was not in contact with Maurik or yourself, or
anyone strictly involved with the GSIM program from the beginning, and due
to several reroutes and setbacks with disk space, etc., was unable to
really get much further than the cuts. I would have liked to get more into
the problem, since I now know more about the program, but as time would
have it, must finish the project by April 20th or so. I have a
presentation to make sometime around then. Any comments you might have
would be greatly appreciated. If things come up, the worst case is that I
continue to look into the problem and try to help out and learn even more,
and present in the Fall, but I feel that if I were to do so, the project
would suddenly become more than jsut a senior project, and would lead to
many other directions.
I told him that:
>I have heard that there is a team looking at the GSIM speed increases, and
>at the moment they are looking into the farming out of GSIM to the parallel
>processing route. It seems that since GSIM is very adaptable to the
>parallel route that that will significantly speed things up. Beyond that,
>I am not sure.
and of my mention of a possible parallel route for GSIM:
>Which group is this ? Could you tell me ?
>It would be very helpful (especially for the GSIM Focus Group) to have a
>more complete report of your study on GSIM, with some details of exactly
>what you did. You could send that to the clas_gsim@cebaf.gov mailing list
>(or send it to me and I'll forward it.)
>We also have weekly teleconferences on Thursdays at 1:30pm. If you would
>like you could also make a presentation there.
Well, there is a brief synopsis of what I have done, and what I hope to get
into a nice presentable form at some point soon. Please, let me know if
you have any suggestions, or if you have anything to add. I have been told
that I should give a copy of my paper to the GSIM project, and possibly
present to them at one of the meetings. I do have class during the meeting
times, but would be willing to get together and share what I know, and my
progress with the team if it would help, and I think it might help me out
as well. Oh well, sorry for the long synopsis, but I felt that I should
contact you about this.
Sincerely,
Miles Johnston
The Lord above made liquor for temptation, to see if man would turn away
from sin.
The Lord above made liquor for temptation, but, wit a lill' bit o' luck,
wit a lill' bit o' luck,
When temptation comes, you'll give right in!
--------------02E8844A572A4E979DCCD85B--