To help the discussion of building electron library and maybe another
fast simulation code, I did some more tests with electron singles and
ep elastics. In both cases, electrons were generated randomly from
5-75 degree uniformly (I don't have CELEG events yet, will have later).
Speed:
A. electron singles
1. with all default cuts and all default settings 3.7 sec/event
2. NOSEC 'OTHE' 1.6 sec/event
3. NOSEC 'EC1' 'EC' 'OTHE' 0.5 sec/event
4. NOSEC 'EC1' 'EC' 2.5 sec/event
B. ep elastic (uniform angular distribution, no cross-ection weight yet)
1. with all default cuts and settings 4.7 sec/event
2. NOSEC 'OTHE' 2.0 sec/event
3. NOSEC 'EC1' 'EC' 'OTHE' 0.6 sec/event
4. NOSEC 'EC1' 'EC' 2.9 sec/event
It appears tracking electron takes most of the time (with Will's CELEG
file the speed on my machine is about 5 sec/event). Building an electron
library will speed up by a factor of 3-5.
I assume someone is still building the shower library for EC, once this is done,
the current gsim can do 2 events/sec.
Can we have an estimate of projected speed after building the electron library
from available numbers?
I don't know how this number would compare to the modified SDA simulator.
I looked at the electron percentage with or without NOSEC 'OTHE',
there is no loss of fraction of electrons (this needs to be further tested).
There is consistent ~1 degree shift for the re-constructed electron angle. I
tested with input of theta_e 20 30 40 degree and got back 19, 29, 39 degree.
There were shifts in W when I looked at them with KJOO.
I don't have a full understanding yet because JKOO claims there shouldn't be
according to Burin's corrected results. So much for now.
Jianguo
--
***************************************************************
* Jianguo Zhao Telephone: (617)258-5438 *
* Room 26-402C, MIT Fax: (617)258-5440 *
* Cambridge, MA 02139 E-mail: jzhao@mitlns.mit.edu *
***************************************************************