? 2 M. Tiemens Hit Creation Cluster 3 M. Tiemens Topology of - - PowerPoint PPT Presentation

▶
2 m tiemens hit creation cluster 3 m tiemens topology of
SMART_READER_LITE
LIVE PREVIEW

? 2 M. Tiemens Hit Creation Cluster 3 M. Tiemens Topology of - - PowerPoint PPT Presentation

Nov 26, 2022 389 likes •693 views

1 M. Tiemens The Latest Developments on the Online Cluster Finding Algorithms ? 2 M. Tiemens Hit Creation Cluster 3 M. Tiemens Topology of the Data Stream t Datastream: , defined by time difference between Two

slide-1
SLIDE 1

The Latest Developments on the Online Cluster Finding Algorithms

  • M. Tiemens

1

?

slide-2
SLIDE 2

Hit Creation

  • M. Tiemens

2

Cluster

slide-3
SLIDE 3

Topology of the Data Stream

  • M. Tiemens

3

Datastream: ∆τ ∆t Two time-scales: ∆ , τ defined by time difference between consecutive hits ∆t, the time difference between any pair of hits

slide-4
SLIDE 4

Topology of the Data Stream

  • M. Tiemens

4

Datastream: ∆τ ∆t Hit belonging to the same cluster 4D position needed to disentangle hits because t(E), timestamps can be spread out drastically

slide-5
SLIDE 5

Scenarios

  • M. Tiemens

5

No gain when using ∆t Make ∆t large to make sure no hits are excluded Spatially fully separated clusters Cluster Hit

slide-6
SLIDE 6

Scenarios

  • M. Tiemens

6

Use ∆t to separate May not help within events Spatially

  • verlapping clusters

Cluster

slide-7
SLIDE 7

Scenarios

  • M. Tiemens

7

Use ∆t to separate Spatially

  • verlapping clusters

Cluster

slide-8
SLIDE 8

Time is Important!

  • M. Tiemens

8

However, there are problems:

slide-9
SLIDE 9

Time is Important!

  • M. Tiemens

9

The evil-doer:

slide-10
SLIDE 10

Time is Important!

  • M. Tiemens

10

The evil-doer: The prize to pay: No more pile-up recovery

slide-11
SLIDE 11

Time is Important!

  • M. Tiemens

11

The prize to pay: No more pile-up recovery Possible solution: Shorten length of waveforms HOWEVER Severely compromises reconstruction effeciency! WHY? Unknown... Some bug in the tasks?

slide-12
SLIDE 12

'Default' Cluster Finding (PndEmcMakeCluster)

  • M. Tiemens

12

Datastream: … ∆τ Hit Hit Is neighbour? Yes No Hit Is neighbour? Yes No …

slide-13
SLIDE 13

'Default' Cluster Finding (PndEmcMakeCluster)

  • M. Tiemens

13

Datastream: … ∆τ Clusters

Grow clusters from “seeds”

Issues:

  • For each new hit, check if neighbour to cluster ≡ check if neighbour to

member hits

  • Large number of loops → clock cycles on a computing chip → latency
slide-14
SLIDE 14

'Default' Cluster Finding (PndEmcMakeCluster)

  • M. Tiemens

14

Datastream: … ∆τ

Grow clusters from “seeds”

Get 4-Momenta: Clusters IP EMC r

slide-15
SLIDE 15

Online Cluster Finding (PndEmcMakeClusterOnline)

  • M. Tiemens

15

Datastream: … ∆τ

Build hit neighbour relations, make clusters

Clusters

slide-16
SLIDE 16

Distributed Cluster Finding (PndEmcDistributedClustering)

  • M. Tiemens

16

Datastream: … ∆τ Preclusters

Assign to virtual Data Concentrators, build hit neighbour relations, make preclusters

DC1 DC2

slide-17
SLIDE 17

Distributed Cluster Finding (PndEmcDistributedClustering)

  • M. Tiemens

17

Datastream: … ∆τ Preclusters

Assign to virtual Data Concentrators, build hit neighbour relations, make preclusters

DC1 DC2

Clusters

Build precluster neighbour relations, make clusters

slide-18
SLIDE 18

2-Step Cluster Finding (PndEmcMakePreclusters)

  • M. Tiemens

18

Datastream: ∆τ … Preclusters

Assign to DCs, build hit neighbour relations, make preclusters

DC1 DC2

slide-19
SLIDE 19

2-Step Cluster Finding (PndEmcMakePreclusters)

  • M. Tiemens

19

Datastream: ∆τ … Preclusters

Assign to DCs, build hit neighbour relations, make preclusters

DC1 DC2

…

Repeat for all timebunches, build precluster datastream

Precluster datastream

slide-20
SLIDE 20

2-Step Cluster Finding (PndEmcMergePreclusters)

  • M. Tiemens

20

Datastream: ∆τ … Preclusters

Assign to DCs, build hit neighbour relations, make preclusters

DC1 DC2

…

Repeat for all timebunches, build precluster datastream

Precluster datastream ∆τ

slide-21
SLIDE 21

2-Step Cluster Finding (PndEmcMergePreclusters)

  • M. Tiemens

21

Datastream: ∆τ … Preclusters

Assign to DCs, build hit neighbour relations, make preclusters

DC1 DC2

…

Repeat for all timebunches, build precluster datastream

Precluster datastream ∆τ Clusters

Build precluster neighbour relations, make clusters

slide-22
SLIDE 22

Testing the Algorithms

  • M. Tiemens

22

slide-23
SLIDE 23

A More Challenging Channel

  • M. Tiemens

23

slide-24
SLIDE 24
  • M. Tiemens

24

Finding Optima for ∆τ and ∆t

Integrate If (Integral ≈ #events): Lower bound = ∆τ ∆t?

Low rates <2 MHz

slide-25
SLIDE 25
  • M. Tiemens

25

Finding Optima for ∆τ and ∆t

Low rates <2 MHz

slide-26
SLIDE 26
  • M. Tiemens

26

Finding Optima for ∆τ and ∆t

High rate: 20 MHz

slide-27
SLIDE 27
  • M. Tiemens

27

Finding Optima for ∆τ and ∆t

High rate: 20 MHz

Integrating doesn’t work because of event mixing and pile-up

slide-28
SLIDE 28
  • M. Tiemens

28

Finding optima for ∆τ and ∆t

High rate: 20 MHz

slide-29
SLIDE 29
  • M. Tiemens

29

Conclusions for ∆τ and ∆t

∆τ can be determined from time-difference spectra ∆t should be larger than ∆τ, but not too much larger, + 25 ns Exception: high rate, but ∆t is still ∆τ + 25 ns