BID

1
BID Data Formats

• G. Watts (UW)

2
d0correct Future of BID
TMB
Algorithm must be able to read from TMB
1
d0root allows us to do that.
Generate Certified Results
2
d0correct
SVT, JLIP, CSIP, SLT all run in d0root Certified
on several formats
Write Results Back to Chunk Format
3
Chunks

• Cant do this
• Tracking object usage through d0root and back
  again is mostly implemented
• Sometimes we generate a new vertex
• No reason this cant be done by RECO vertexing
• SVT Kalman fitter makes new tracks. ?
• What do we write back now?

We could write a very simple chunk with a y/n
answer in it
3
Time Scale Problem
We are one of the higher level objects
To certify and tune we need the CSG skims already
root-tupled!
d0correct Ready (w/o BID)
DZERO Makes Rootuples!
BID Tunes and Certifies
New Version of d0correct
Maybe this isnt so bad from the analyzer point
of view
Most people run the BID algorithms in their root
tuple makers
But without large samples we cant certify
4
Data Format
This isnt a 1 page talk
I have come to like the top_tree
It is to the TMBTree what the TMBTree is to the
TMB
but missing d0root!
Any final root format will have to accommodate
group specific items
As part of the generic objects!!!
For example
Is this jet tagged with a SVT Medium? What are
the JES lq, MC, hf, etc., correction factors?
Also, all groups will have to involved in this
decision/design!