Discussion on content and format of L3 monitor statistics
---------------------------------------------------------


Example L3 monitor output for a filter script containing two filters
(in addition to the Mark&Pass filter):

Run Number: 149344 L2 Trig bit: 10 L2 Trig name: dummy14 L1 Trig bit: 10
L3 Script bit:  14 L3 Script name: 2TAU_EM10_2CJT5
 # called = 35449 # passed = 8316 # force_unbiased = 532 # prescaled = 0
L3 Filter name:  t15s1_mp50
 # called = 35449 # passed = 35449
L3 Filter name:  t15s2_Tau
 # called = 35449 # passed = 18483
L3 Filter name:  t15s3_Ele
 # called = 18816 # passed = 7917


Because of the presence of Mark&Pass events there are a number of
features that sometimes cause confusion.  For example:

  - Script # passed     (= 8316)   .ne.  Last filter  # passed (= 7917)
  - 2nd script # passed (= 18483)  .ne.  3rd script # called   (= 18816) 


 

Proposal from Moacyr and Terry
------------------------------


a) Script statistics:
  - Add number of "true passes" (i.e.,   # passed - # force_unbiased)
  - Implement option to express "passes" as fraction (%) of # called. 

b) Filter statistics:
Would like to have three options for how the  # called and # passed 
should be calculated:
  1) count total number (as is done now)
  2) count only "true" calls and passes (i.e., ignore force_unbiased) 
  3) use only Mark&Pass events,
  + implement option to express "passes" as fraction (%) of # called. 

  We have to figure out exactly how to implement this.