Opis raspodjele kvarkova i gluona u protonu pomoću gausijanskih procesa

master's thesis

• Title (english): Description of distribution of quarks and gluons in proton using Gaussian processes
• Author: Andrija Radočaj
• Mentor: Krešimir Kumerički (mentor)
• Committee member: Krešimir Kumerički (predsjednik povjerenstva)
• Committee member: Sanjin Benić (član povjerenstva)
• Committee member: Petar Žugec (član povjerenstva)
• Committee member: Davor Horvatić (član povjerenstva)
• Committee member: Ivica Friščić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Science; (Department of Physics); Zagreb
• Defense date and country: 2024-09-24, Croatia
• Scientific / art field, discipline and subdiscipline: NATURAL SCIENCES; Physics

Abstract

Partonske distribucijske funkcije protona opisuju raspodjelu impulsa njegovih sastavnica te stoga nose važnu informaciju o njegovoj kvarkovsko-gluonskoj dinamici. S obzirom na to da su fundamentalno neperturbativna veličina kvantne kromodinamike, određuju se posredno prilagodbom modela na eksperimentalne podatke. Diferencijalni udarni presjeci dubokih neelastičnih raspršenja ovise upravo o njima te se mogu smisliti razne metode ekstrakcije tih funkcija iz eksperimentalnih podataka različitih... More procesa. U ovom radu ispituje se metoda strojnog učenja za ekstrakciju spomenutih distribucijskih funkcija njihovim modeliranjem kao gausijanskih procesa te Monte Carlo dropout metoda neuronskih mreža za usporedbu. U svrhu ispitivanja metode uvedena su određena nefizikalna pojednostavljenja. Također, s istim ciljem zadane su nefizikalne partonske distribucijske funkcije koje služe za simulaciju podataka diferencijalnih udarnih presjeka nefizikalnih procesa. Distribucijske funkcije ekstrahirane iz simuliranih podataka mogu se tada usporediti s onim zadanim kako bi se na taj način mogla procijeniti uspješnost metode kao i potencijalno zaključiti nešto o njenoj primjenjivosti na stvarne eksperimentalne podatke. Također, određeni zaključci doneseni su usporedbom rezultata metode gausijanskih procesa s rezultatima Monte Carlo dropout metode neuronskih mreža. Less

Abstract (english)

Parton distribution functions of proton describe the distribution of momenta of its components and thus provide an important information about its quark-gluon dynamics. Considering they are fundamentally non-perturbative quantity of quantum chromodynamics, they are determined indirectly by fitting the model to experimental data. Differential cross sections of deep inelastic scatterings depend on those functions and various methods of their extraction can be devised from experimental data of... More different processes. This thesis examines the machine learning method for extraction of the mentioned distribution functions by modeling them as Gaussian processes and the Monte Carlo dropout method of neural networks for comparison. For testing purposes of the method certain non-physical simplifications were introduced. Likewise, with the same goal there are given non-physical parton distribution functions which serve to simulate differential cross section data of different nonphysical processes. Distribution functions extracted from the simulated data can then be compared with the given ones in order to evaluate the efficacy of the method and potentially conclude something on its applicability to real experimental data. Also, certain conclusions were drawn by comparing the results of the Gaussian processes method with the results from Monte Carlo dropout method of neural networks. Less

• Keywords: duboko neelastično raspršenje; partonske distribucijske funkcije; strojno učenje; Gausijanski procesi; neuronske mreže; Monte Carlo dropout
• Keywords (english): deep inelastic scattering; parton distribution functions; machine learning; Gaussian processes; neural networks; Monte Carlo dropout
• Language: croatian
• URN:NBN: urn:nbn:hr:217:474705
• Study programme: Title: The University Integrated Undergraduate and Graduate Programme in Physics; Research Physics; specializations in: Research Physics; Course: Research Physics; Study programme type: university; Study level: integrated undergraduate and graduate; Academic / professional title: sveučilišni magistar fizike (sveučilišni magistar fizike)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2025-01-31 11:49:23