Difference between revisions of "Private:Student Projects"

From GlueXWiki
Jump to: navigation, search
(Created page with "===Project 1=== <div class="toccolours mw-collapsible mw-collapsed"> Student Project 1       (Click "Expand" to the right for more details -->): <di...")
 
(→‎Project 6)
 
(23 intermediate revisions by 2 users not shown)
Line 2: Line 2:
  
 
<div class="toccolours mw-collapsible mw-collapsed">
 
<div class="toccolours mw-collapsible mw-collapsed">
−
Student Project 1 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp; (Click "Expand" to the right for more details -->):
+
PDF for dark scalar boson &nbsp;&nbsp;&nbsp;&nbsp;&nbsp; (Click "Expand" to the right for more details -->):
  
 
<div class="mw-collapsible-content">
 
<div class="mw-collapsible-content">
  
−
# Make sure you have created a BlueJeans account via your JLab CUE account using this link:
+
* Determine the Probability Density Function (PDF) of the dark scalar gauge boson, S, for the process gamma proton -> eta  + proton, eta -> S + pi0, S -> gamma gamma. Since the hypothetical S mass is unknown, the PDF has to be determined for all masses between ~ a few MeV/c^2 and ~ 420 MeV/c^2 scannable by JEF. This can be done in 3 steps:
−
#* [http://jlab.bluejeans.com http://jlab.bluejeans.com]  (You should only need to do this once)
+
# Determine PDF for fixed masses from histograms I will provide
−
#:<br>
+
# Parametrize the PDF vs mass and possibly vs incident photon-beam energy
−
# '''Meeting ID: 859833522 '''
+
# Toy Monte Carlo simulation study to show that the model is not introducing biases for our observables#:<br>
−
#* (you may need to type this in, depending how you connect)
+
  
 +
</div>
 +
[https://halldweb.jlab.org/DocDB/0055/005500/001/Notes_from_Jared%281%29.pdf Jared G. Richards report]
 
</div>
 
</div>
  
 +
===Project 2===
 +
 +
<div class="toccolours mw-collapsible mw-collapsed">
 +
Island Algorithm &nbsp;&nbsp;&nbsp;&nbsp;&nbsp; (Click "Expand" to the right for more details -->):
 +
 +
<div class="mw-collapsible-content">
 +
 +
* Use Simon's and Igal's work to test the GAMS Island algorithm. Steps:
 +
# Grab the farm version (also exists on a GitHub branch: https://github.com/JeffersonLab/halld_recon/tree/FCALIsland)
 +
# Use Igal's environment and build scripts.
 +
# Use factory flag/tag to change clustering algorithms on the command line (or in config file): DEFTAG:DFCALCluster=Island
 +
# Merge/convert with Jon's code.
 +
# Run photon guns through MCWRAPPER, to see single photon, and two-photon hits. 
 +
# Check upstream conversions in TOF/DIRC.
 +
# Check average no of reconstructed clusters.  Extract 'resolving power' of FCAL-I and FCAL-II eventually.
 +
# Stitch boundary between insert and outer blocks.
 +
# Vaildate with data?  Use omega skims for FCAL-II / COMCAL is subtends too small of an angle
 +
<br>
 +
 +
</div>
 +
</div>
 +
 +
===Project 3===
 +
 +
<div class="toccolours mw-collapsible mw-collapsed">
 +
Background suppression via MVA (eta -> pi^0 gamma gamma) &nbsp;&nbsp;&nbsp;&nbsp;&nbsp; (Click "Expand" to the right for more details -->):
 +
 +
<div class="mw-collapsible-content">
 +
Study of the process gamma + proton -> eta + proton [1], eta -> pi^0 gamma gamma , and pi^0 -> gamma gamma; and in particular background suppression via multivariate analysis (MVA) using the Fisher discriminant [2] method
 +
in the TMVA [3] package.
 +
 +
[1]G.Liping et al, Precision tests of fundamental physics with η\etaη and η′\eta^\primeη′ mesons, [arXiv:2007.00664]
 +
 +
[2]R. A. Fisher, The use of multiple measurements in taxonomic problems, Annals Eugen.7(1936) 179.
 +
 +
[3]H. Voss, A. Hocker, J. Stelzer and F. Tegenfeldt,TMVA - Toolkit for Multivariate DataAnalysis, PoS ACAT (2007) 040 [arXiv:physics/0703039]
 +
<br>
 +
 +
</div>
 +
[https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=5336 Chase W. Peterson talk & report]
 +
</div>
 +
 +
===Project 4===
 +
 +
<div class="toccolours mw-collapsible mw-collapsed">
 +
Background suppression via MVA (eta -> pi^0 e^+ e^-) &nbsp;&nbsp;&nbsp;&nbsp;&nbsp; (Click "Expand" to the right for more details -->):
 +
 +
<div class="mw-collapsible-content">
 +
Study of the process gamma + proton -> eta + proton [1], eta -> pi^0 e^+ e^- , and pi^0 -> gamma gamma; and in particular background suppression via multivariate analysis (MVA) using the Fisher discriminant [2] method
 +
in the TMVA [3] package.
 +
 +
[1]G.Liping et al, Precision tests of fundamental physics with η\etaη and η′\eta^\primeη′ mesons, [arXiv:2007.00664]
 +
 +
[2]R. A. Fisher, The use of multiple measurements in taxonomic problems, Annals Eugen.7(1936) 179.
 +
 +
[3]H. Voss, A. Hocker, J. Stelzer and F. Tegenfeldt,TMVA - Toolkit for Multivariate DataAnalysis, PoS ACAT (2007) 040 [arXiv:physics/0703039]
 +
<br>
 +
 +
</div>
 +
</div>
 +
 +
===Project 5===
 +
 +
<div class="toccolours mw-collapsible mw-collapsed">
 +
Classification methods optimizable for unknown cross-sections and particle masses. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp; (Click "Expand" to the right for more details -->):
 +
 +
<div class="mw-collapsible-content">
 +
Search for dark particles needs powerful classification methods, with the capacity to be optimized for unknown cross-sections and particle masses, and also be able to account for systematic effects ie a fully differentiable analysis framework.
 +
<br>
 +
</div>
 +
</div>
 +
 +
===Project 6===
 +
<div class="toccolours mw-collapsible mw-collapsed">
 +
Create a PID classification model to distinguish e/pi/K/p
 +
 +
<div class="mw-collapsible-content">
 +
Create a classification model within the Jlab ML workflow for Particle IDentification with a likelihood loss function and determine PID systematic error with the ensemble method.
 +
<br>
 +
</div>
 
</div>
 
</div>

Latest revision as of 12:09, 2 February 2024

Project 1

PDF for dark scalar boson       (Click "Expand" to the right for more details -->):

  • Determine the Probability Density Function (PDF) of the dark scalar gauge boson, S, for the process gamma proton -> eta + proton, eta -> S + pi0, S -> gamma gamma. Since the hypothetical S mass is unknown, the PDF has to be determined for all masses between ~ a few MeV/c^2 and ~ 420 MeV/c^2 scannable by JEF. This can be done in 3 steps:
  1. Determine PDF for fixed masses from histograms I will provide
  2. Parametrize the PDF vs mass and possibly vs incident photon-beam energy
  3. Toy Monte Carlo simulation study to show that the model is not introducing biases for our observables#:

Jared G. Richards report

Project 2

Island Algorithm       (Click "Expand" to the right for more details -->):

  • Use Simon's and Igal's work to test the GAMS Island algorithm. Steps:
  1. Grab the farm version (also exists on a GitHub branch: https://github.com/JeffersonLab/halld_recon/tree/FCALIsland)
  2. Use Igal's environment and build scripts.
  3. Use factory flag/tag to change clustering algorithms on the command line (or in config file): DEFTAG:DFCALCluster=Island
  4. Merge/convert with Jon's code.
  5. Run photon guns through MCWRAPPER, to see single photon, and two-photon hits.
  6. Check upstream conversions in TOF/DIRC.
  7. Check average no of reconstructed clusters. Extract 'resolving power' of FCAL-I and FCAL-II eventually.
  8. Stitch boundary between insert and outer blocks.
  9. Vaildate with data? Use omega skims for FCAL-II / COMCAL is subtends too small of an angle


Project 3

Background suppression via MVA (eta -> pi^0 gamma gamma)       (Click "Expand" to the right for more details -->): 

Study of the process gamma + proton -> eta + proton [1], eta -> pi^0 gamma gamma , and pi^0 -> gamma gamma; and in particular background suppression via multivariate analysis (MVA) using the Fisher discriminant [2] method
in the TMVA [3] package.

[1]G.Liping et al, Precision tests of fundamental physics with η\etaη and η′\eta^\primeη′ mesons, [arXiv:2007.00664]

[2]R. A. Fisher, The use of multiple measurements in taxonomic problems, Annals Eugen.7(1936) 179.

[3]H. Voss, A. Hocker, J. Stelzer and F. Tegenfeldt,TMVA - Toolkit for Multivariate DataAnalysis, PoS ACAT (2007) 040 [arXiv:physics/0703039]

Chase W. Peterson talk & report

Project 4

Background suppression via MVA (eta -> pi^0 e^+ e^-)       (Click "Expand" to the right for more details -->): 

Study of the process gamma + proton -> eta + proton [1], eta -> pi^0 e^+ e^- , and pi^0 -> gamma gamma; and in particular background suppression via multivariate analysis (MVA) using the Fisher discriminant [2] method
in the TMVA [3] package.

[1]G.Liping et al, Precision tests of fundamental physics with η\etaη and η′\eta^\primeη′ mesons, [arXiv:2007.00664]

[2]R. A. Fisher, The use of multiple measurements in taxonomic problems, Annals Eugen.7(1936) 179.

[3]H. Voss, A. Hocker, J. Stelzer and F. Tegenfeldt,TMVA - Toolkit for Multivariate DataAnalysis, PoS ACAT (2007) 040 [arXiv:physics/0703039]

Project 5

Classification methods optimizable for unknown cross-sections and particle masses.       (Click "Expand" to the right for more details -->):

Search for dark particles needs powerful classification methods, with the capacity to be optimized for unknown cross-sections and particle masses, and also be able to account for systematic effects ie a fully differentiable analysis framework.

Project 6

Create a PID classification model to distinguish e/pi/K/p

Create a classification model within the Jlab ML workflow for Particle IDentification with a likelihood loss function and determine PID systematic error with the ensemble method.

Retrieved from "https://halldweb.jlab.org/wiki/index.php?title=Private:Student_Projects&oldid=124555"