Min Jeong Kim has been working toward applying the cuts
Stepan defined to some recent GSIM runs. She ran into problems
with using the second and third moments, which were zero in
the 2.4gev_p_tgt_8.rzn file.
I tried to generate these by using the tcl switch anal_opt=2
defined by Stepan. This does not fill the ECRB, just the ECHB
which Stepan defined. Just changing this switch did not work.
After some experimentation, I found it was also necessary
to define the touch_id tcl variable (I set it to 2). Then these
moments are calculated and put in ECHB.
Now that we have a number of options for the reconstruction
code, we should do something about the interactions between
the parameters to avoid this kind of problem.
I looked at the ddl file for ECHB. There are several differences
between the names there and the ntuple names; some trivial (like
sec vs. sect for "sector") and some complicated (the sector and
layer are coded into one 32 bit word, as is the status word; these
are broken into a total of 4 16 bit words in the ntuple). This
is unfortunate from the point of view of uniform documentation.
I also tried to use the ECHB variables to look at the effect
of various cuts. I couldn't figure out a simple way to do this;
in ECRB I could plot Ein vs. Eout, and put cuts on this quantity.
In ECHB the inner data is distinguished from the outer by the
layer number, so there is no simple way to plot inner vs. outer.
When Stepan gets back we should make a final decision on the
format of the result bank. By now we have a reasonable amount
of operating experience. I would make the following proposal:
--------------------------------------------------------------
Have SEPARATE BANKS for inner information, outer information,
and combined information (Stepan's "whole" calorimeter). Encode
the "layers" as in ECRB from 1 to 6 only, for UVW(in) and UVW(out).
Keep the list of variable types which are in both banks, plus
the sector id. Use only a 16-bit status word (currently the upper
limit of the status word in both banks is set to 2**16 anyway).
--------------------------------------------------------------
Discussion:
There is not much difference between the ECHB and the ECRB.
The differences between ECRB and ECHB are that ECRB requires
early matching of inner and outer hits, and ECHB has the explicit
sector id. In the current implementation which includes the
"whole" calorimeter, there could not be much savings in space
with ECHB. Any space saving advantage would also occur by keeping
the information in three separate banks, plus would offer the
possibility of choosing which ones get written out, not presently
possible. It would also offer the possibility of trivially
plotting correlated variables connecting inner and outer
calorimeters, such as Einner vs. Eouter or X_in vs X_out. I think
potentially correlated variables should not have the same name
and be distinguished by other variables in the bank, such as
layer number, as is done in ECHB.
Any comments?
- Will