Cluster Finder

Cleaning

We remove strips for which conditions :

Signal > 3 * rms
Noise > 0

are not satisfied
I ran over Production2 files and the next plots are showimg the size of clusters for ladders on p and n side.


fig 1 : p-side ladders	fig 2 : n-side ladders

Details ladder by ladder are here : - side P
and side N

Side N

The peaks seen for side N are more or less du to ladder 8 (fig a).
In this ladder, the peaks are not due to wafers at this edge (fig b)


fig a : n-side ladders	fig b : ladder 8n-side

Removing wafers at the edge

Here I show again the distribution of the size of the clusters but with removig wafers at the edge of all ladders, eg all wafers 1 and 16 are removed.


fig c : p-side ladders	fig d : n-side ladders

So we see that :

for ladders P it reduces the size of the clusters, even without the cleaning
Then the cleaning for these ladders (side P) decrease again the distribution
For ladders on N-side, the peaks for size = 35 , size = 90 are still here but I showed before that they are coming from ladder 8.
But the one at size = 65 has disappear.

Vi Nham also did the same but with higher statistics. : - here

In both side, this cleaning decreases the size of clusters.

Difference

I also plot the difference of the number of clusters with and w/o cleaning.

Figure 3

We observe that with the cleaning :

the population of clusters with 1 strip with cleaning is bigger than w/o.
the population for the rest of the clusters (size>1) is decreasing, as shown in the previous plots where we see that the size of clusters is smaller with the cleaning.

Impact on hits type

Figure 4

The hit type is characterising the number of clusters in each hit : the more this number is higher the more the hit density per wafer is higher.
It seems the cleaning decreases the number of this ambiguous hits.

Ratio matched clusters /all clusters

no cleaning


fig 5 : p-side ladders	fig 6 : n-side ladders

with cleaning


fig 7 : p-side ladders	fig 8 : n-side ladders

Efficiency

We change our definition of the efficiency.
It is now defined as the ratio of track with (TPC+)SSD+SVT / (TPC+)SSD+SVT + (TPC+)SVT
Basically we are looking at hits associated to the tracks and check if for tracks with at least 2 svt hits we have (we expect) an ssd hits
Cuts remain the same.
The 2 plots show this efficiency as a function of &eta


fig 9 : no cleaning : Eff = 0.616563 +/- 0.041429	fig 10 : with cleaning : Eff = 0.616879 +/- 0.042790

I also dump the values in txt files : - no cleaning and - with cleaning

Jonathan Bouchet

Last modified: Thu Jul 5 14:57:10 CEST 2007