Algorithms for ntuple production

[jbsauvan]

Building blocks

• Energy compression
  • Floating point 8 bits (4+4) with rounding
  • Floating point 7 bits (4+3) with rounding
• Concentrator
  • Threshold 2mipT
  • SuperTC 2x2
• Backend stage-1
  • Topological 2D clustering with size constraint
  • Dummy
• Backend stage-2
  • Old 3D iterative clustering
  • HistoMax with variable dR
    • No seed threshold, default dR
    • Seed threshold 10 mipT, default dR
    • Seed threshold 20 mipT, default dR
    • Seed threshold 10 mipT, default dR x 1.25
    • Seed threshold 10 mipT, default dR x 1.5
    • Seed threshold 10 mipT, default dR x 2

Chains

• Reference: FP8 + Old 2D and 3D clusterings
• FP8 + Threshold + All HistoMax flavours (6)
• FP8 + SuperTC + All HistoMax flavours (6)
• FP7 + Threshold + default HistoMax

[jbueghly]

I think it would also make sense to include Threshold 0mipT, and possibly also Threshold 2mipT + SuperTC 2x2. That allows for more comparisons between trigger cells and super trigger cells.

[jbsauvan]

and possibly also Threshold 2mipT + SuperTC 2x2

Do you mean applying a threshold of 2mipT on TCs before forming the SuperTCs?

[jbueghly]

Yes, that's what I meant.

[jbsauvan]

It is currently not an option that is implemented, but we can eventually add it to the SuperTC todo list

[jbsauvan]

For the TC 0 mipT threshold, I think we could do it, but only with the default clustering to reduce CPU time