search for z’-boson signal using neural networks a.n.buryk and v.v.skalozub
TRANSCRIPT
![Page 1: Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub](https://reader036.vdocuments.us/reader036/viewer/2022072005/56649ceb5503460f949b61fd/html5/thumbnails/1.jpg)
Search for Z’-boson signalusing neural networks
A.N.Buryk and V.V.Skalozub
![Page 2: Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub](https://reader036.vdocuments.us/reader036/viewer/2022072005/56649ceb5503460f949b61fd/html5/thumbnails/2.jpg)
Many theories predict Z’-boson existance.
There are many methods to search it.
But the lack of experimental statisticks doesn’t allow to find it.
Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub
![Page 3: Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub](https://reader036.vdocuments.us/reader036/viewer/2022072005/56649ceb5503460f949b61fd/html5/thumbnails/3.jpg)
Starting point of this work was “Model-independent search for the Abelian Z’-boson in the Bha-Bha process”, made by A.V.Gulov and
V.V.Skalozub
jj
eff TL
O djdjddj
jSMjj
T
TTT
,,44
1
Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub
![Page 4: Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub](https://reader036.vdocuments.us/reader036/viewer/2022072005/56649ceb5503460f949b61fd/html5/thumbnails/4.jpg)
YYY~~~
2,1,
fBfYgi
fiBYgi
L DDew
fff
ew
RL
))(2
(|))(2
(|~~
~
,
,
2~~
~
,
avav ffff **
2;
2
~~~~,,
'
,,
'
YYv
YYa lLlRl
Z
lLlRl
Z
2
~Y
ala
YTa ff
~,3
Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub
![Page 5: Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub](https://reader036.vdocuments.us/reader036/viewer/2022072005/56649ceb5503460f949b61fd/html5/thumbnails/5.jpg)
YvYavava eeeeee ,,,,22
aY e2~
The Bha-Bha process was investigatedFor this process we have only one equation from the third formula:
The leading-order differential cross-section of the Bha-Bha process depends on next combinations of Z’-boson couplings:
Taking into account RG equations they are reduced to three parameters:
vava eeee,,22
Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub
![Page 6: Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub](https://reader036.vdocuments.us/reader036/viewer/2022072005/56649ceb5503460f949b61fd/html5/thumbnails/6.jpg)
Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub
mmg
mmg
Z
Ze
Z
Ze vvaaCosz 2
'
22
2
'
22
4;
4;
~~
...),(),(),(22 vazSFzSFzSF
dz
d
dz
d
dz
davav
SM
av
109109
42
42
av
![Page 7: Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub](https://reader036.vdocuments.us/reader036/viewer/2022072005/56649ceb5503460f949b61fd/html5/thumbnails/7.jpg)
Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub
z1 z2 dσ/dz error
189 … … … …
192 … … … …
… … … … …
207 … … … …
Common view of OPAL collaboration data
View of OPAL collaboration data after some preparations
z1 z2 Δ(dσ/dz) error
189 … … … …
192 … … … …
… … … … …
207 … … … …
![Page 8: Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub](https://reader036.vdocuments.us/reader036/viewer/2022072005/56649ceb5503460f949b61fd/html5/thumbnails/8.jpg)
Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub
On the diagram it looks like a band
1 2 3 4 5 6
![Page 9: Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub](https://reader036.vdocuments.us/reader036/viewer/2022072005/56649ceb5503460f949b61fd/html5/thumbnails/9.jpg)
Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub
Set of 2 neural networks was used.
Each network had its own discard criterion.
First criterion: discard that data, that correspond to very big v2
Second criterion: discard that data, for which RG equations are not applied
1 networkSignal: v2=0,0001 Antisignal: v2=0,01
2 network
Antisignal:
Signal: aY e2~
aY e30000~
![Page 10: Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub](https://reader036.vdocuments.us/reader036/viewer/2022072005/56649ceb5503460f949b61fd/html5/thumbnails/10.jpg)
Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub
Scheme of neural network structure
![Page 11: Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub](https://reader036.vdocuments.us/reader036/viewer/2022072005/56649ceb5503460f949b61fd/html5/thumbnails/11.jpg)
Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub
We used neural networks with 2 incoming neurons, 10 hidden neurons an 1 outgoing neurons.
After processing by first network we discard 2% of the width.After processing by the second network we discard 6,2% of bandwidth.
So, after all actions, the overall decrease is about 8,2%.
To construct and learn neural networks the MLPFit package was used.To use obtained neural networks to process the data, 2 programs were
written.
![Page 12: Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub](https://reader036.vdocuments.us/reader036/viewer/2022072005/56649ceb5503460f949b61fd/html5/thumbnails/12.jpg)
Search for Z’-boson signal using neural networks A.N.Buryk and V.V.Skalozub
1. A.V.Gulov and V.V.Skalozub “Model-independent search for the Abelian Z’boson in the Bhabha process”, arXiv: hep-ph/0408076.
2. MLPFit, http://home.cern.ch/~schwind/MLPfit.html
Bibliographical links: