Changes for page Antonio Romero Vidal and julio Novoa Fernández

Summary

• Page properties (2 modified, 0 added, 0 removed)

Details

Page properties

Title

...	...	@@ -1,1 +1,1 @@
1		-Xabier Cid Vidal
	1	+Antonio Romero Vidal and julio Novoa Fernández

Content

...	...	@@ -1,13 +1,28 @@
1		-=== Generative Models for Fast Simulation of a Neutron Tomography Detector ===
2		-==== Supervisor: María Pereira Martínez
	1	+=== Machine Learning Tools for Precision Tests of the Standard Model at LHCb ===
	2	+==== Supervisor: Julio Novoa Fernández
3	3	====
4	4
5		-Neutron tomography is an innovative technique for the detailed analysis of dense materials such as metals and alloys. Thanks to the high penetration capability of neutrons, it is possible to detect internal manufacturing defects and, in certain cases, determine the chemical composition of a sample without destroying it.
	5	+Leptons are a fundamental class in the Standard Model (SM) of particle physics. According to this model, there is no apparent difference between leptons apart from their masses - this is known as Lepton Flavour Universality (LFU). However, recent measurements of LFU observables show significant disagreement with respect to SM predictions, which could suggest new physics
	6	+beyond SM.
6	6
7		-To achieve high precision in neutron tomography, optimizing the components of the tomograph is essential. One of the main challenges lies in the simulation and optimization of the primary detector, a process that requires significant computational resources. To address this, the use of generative machine learning models offers a groundbreaking approach: these models can greatly accelerate simulations when trained to mimic the detector’s response under various experimental conditions, drastically reducing computational cost.
	8	+To reduce their uncertainty and solve the current situation, these LFU observables need to be measured with great precision. In particular, the experimental data has to be carefully processed to separate “signal” and “background” events. This selection includes Machine Learning (ML) tools that improve the efficiency of this procedure.
8	8
9		-The aim of this project is to explore the potential of generative models in the simulation and optimization of the neutron tomography system. In this context, the summer student will:
	10	+This project will introduce the student to the use of ML in high-energy physics. In particular, the
	11	+student will:
10	10
11		-* Gain an understanding of how the neutron tomography system and its main detector work.
12		-* Become familiar with fundamental machine learning concepts and implement a generative neural network to replicate the detector’s response under different configurations.
13		-* Develop programming skills by working with Python and commonly used machine learning libraries such as scikit-learn, TensorFlow and PyTorch, with the potential to explore GPU-accelerated training techniques.
	13	+* Analyse real data from LHCb, one of the main experiments of the Large Hadron Collider at CERN (the world’s largest and most powerful particle accelerator).
	14	+* Learn how ML is used for measuring LFU observables with better accuracy, by implementing tools such as neural networks, decision trees and more.
	15	+* Train and optimise their own ML models with different Python environments, which they could apply to other physical phenomena and beyond.
	16	+
	17	+----
	18	+
	19	+=== P1 - ML study for B->D* 3Pi decays ===
	20	+
	21	+This decay is used as the normalization channel for the semileptonic ratio R(D*), one of the LFU observables that is studied at LHCb. In this particular case, the D* meson decays into a pion and a D meson, which then decays into a kaon and three pions.
	22	+
	23	+----
	24	+
	25	+=== P2 - ML study for B->D 3Pi decays ===
	26	+
	27	+A related LFU observable is the semileptonic ratio of branching fractions R(D). One possible choice for the normalization channel is B->D 3Pi, where D decays into two pions and a kaon.
	28	+