Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Machine learning approaches towards tuning ALICE TRD simulations
In this work an exploration of the discrepancies existing between real and simulated data pertaining to the ALICE Transition Radiation Detector is carried out as a motivation to tune the necessary parameters in the ALICE Online-Offline simulation software (O2 ). After such exploration a single param...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | Eng |
| Published: |
Department of Statistical Sciences
2024
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| Summary: | In this work an exploration of the discrepancies existing between real and simulated data pertaining to the ALICE Transition Radiation Detector is carried out as a motivation to tune the necessary parameters in the ALICE Online-Offline simulation software (O2 ). After such exploration a single parameter namely the Xe gas gain is subjected to modification. A machine learning approach is taken with the use of deep learning discrimination mechanisms namely artificial neural networks and convolutional neural networks to quantify the effect that our tuning has on the improvement of the simulation results and their conformation to the real data. The correspondence of the optimal values suggested by deep learning approaches is investigated with pulse height spectrometry. It is shown that the optimal parameters suggested by our deep learning models through inference of their performance metrics are not clear and in agreement with that suggested by naive pulse height inspections. |
|---|