The Physics Undergraduate Colloquia is a programme of research talks given by undergraduate students to undergraduate students.

Undergraduate Colloquia Lectures

2024-25 Lectures

All talks will be followed by lunch in the Blackett Level 2 Foyer

Date: 9 October 2024 at 12:00
Location: Blackett Lecture Theatre 1
Talk 1 Title: Investigation of Magnetic Pitch Angle and Toroidal Current Density on the MAST-U tokamak
Speaker: Juan Rajagopal
Abstract: The plasma current density and magnetic pitch angle in the MAST-U spherical tokamak have been investigated for different energy confinement categories.  The results provide crucial evidence that the transport barrier formation begins in the outer plasma edge region up to 50 ms before the neutral recycling signal drops due to reduced ion flux to the machine targets.  The impact of these results on our understanding of turbulence suppression and plasma self-organisation will be outlined. Plans for further investigation will also be discussed.

Talk 2 Title: Speed Enhancements of the Adaptive Random Fourier Features With Metropolis Sampling Algorithm
Speaker: Wadoud Charbak
Abstract: In this presentation, we introduce an advanced implementation of a random feature-based neural network training algorithm, adaptive random Fourier Features (ARFF).   We achieved significant speed improvements by utilising JAX, a high-performance machine learning library, and NVIDIA's CUDA for parallel GPU computations. We compare the computational time of ARFF training by considering a supervised learning problem in the following settings: a Standard NumPy-based version, a JAX version with GPU execution but no Just in time  (JIT) compilation, a JAX-based version with partial JIT compilation, a JAX-based version with full JIT compilation. Our approach significantly reduces computational time while maintaining high accuracy. We demonstrate the efficiency gains through comparative analyses and discuss potential applications in large-scale machine learning tasks.

Date: 16 October 2024 at 12:00
Location: Blackett Lecture Theatre 1
Talk 1 Title: Radiation asymmetries: EMC3-EIRENE Modelling of the Wendelstein 7-X edge plasma
Speaker: Amy Thornton
Abstract: The concept of the Stellarator has existed as long as the Tokamak, but due to engineering challenges, is only more recently undergoing a renaissance. Wendelstein 7-X is the world's largest Stellarator and has been in operation since 2015. Modelling of the W7-X edge plasma using EMC3-EIRENE, a 3D Monte-Carlo plasma fluid code, based on Braginskii’s fluid equations, (EMC3), and a kinetic neutral model, (EIRENE) [1], is performed for the investigation of radiation asymmetries arising during nitrogen impurity seeding, with implications for the power loss due to impurity radiation and impurity transport itself.

[1] Y. Feng, F. Sardei, and J. Kisslinger, “3d fluid modelling of the edge plasma by means of a monte carlo technique,” Journal of Nuclear Materials, vol. 266-269, pp. 812–818, 1999. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0022311598008447

Talk 2 Title:
Speaker:
Abstract:

Date: 13 November 2024 at 12:00
Location: Blackett Lecture Theatre 1
Talk 1 Title: Bridging Quantum Error Correcting Codes and the Bose-Marletto-Vedral Gravitational Entanglement Proposal
Speaker: Arush Garg
Abstract: The 2017 Bose-Marletto-Vedral (BMV) experiment proposed a novel approach to probe the quantum nature of gravity by its ability to induce entanglement between two spin-1/2 masses. Recently, this protocol has been generalised to spin-j systems. Simultaneously, advancements in quantum information science have led to the development of various physical qubit implementations and sophisticated error correction techniques. This project attempted to map concepts from quantum computing to the BMV proposal. We show that, under certain limits, the experiment is simply the action of a logical gate and begin exploring the use of quantum error correction to mitigate the impact of decoherence.

Talk 2 Title: Attempts to infer mass outflow rate in the circumgalactic medium from the absorption spectrum of a distant quasar
Speaker: Chi-Yu Huang
Abstract: The galactic mass outflow rate in the circumgalactic medium (CGM) is a useful measure for studying a galaxy and its evolution. For starburst galaxies, the common procedure to infer the outflow rate is to measure the column density and velocity from the absorption spectrum of starlight coming from within the galaxy . However, for other galaxies where the CGM is instead probed by the absorption spectrum of light from a distant quasar, the above described procedure is problematic. Nonetheless, my project looks at galaxies in the latter scenario. We attempt to infer the outflow rate by measuring the column density of many simulated absorption lines of random light rays going through the CGM of a galaxy built with RAMSES. In particular, we focus on H I, C IV, Mg II, and O VI absorption lines in the range of 800Å to 1600Å, similar to the range of the HST-COS. In this presentation I will present the different behaviour of the galactic mass outflow rates depending on 1. the way we define the outflow rate,  2. the inclination of the galaxy, and 3. the geometry of the galaxy. We find that the correlation between the outflow rate and column density for O VI and H I is consistently higher than C IV and Mg II, suggesting that the former two may be a better tracer for the outflow rate.

Date: 15 January 2025 at 12:00
Location: Blackett Lecture Theatre 1
Talk Title: Use of an Autoencoder for SETI related Anomaly Detection in Green Bank Telescope Data
Speaker: Kai Mulcock
Abstract: Anomaly detection methods are applied to Green Bank Telescope data to aid in the search for astronomical phenomena and potential techno signatures. A deep autoencoder is trained on background and injected signals, and the latent space is explored for anomalous classes based on their reconstruction loss. Broadband complex signals injected as anomalous classes are successfully retrieved in a proof-of-concept test.

Date: 19 February 2025 at 12:00
Location: Blackett Lecture Theatre 1
Talk 1 Title: Simulating the magnetic fields in the early stages of laser-plasma interactions
Speaker: Elias Fink
Abstract: In inertial confinement fusion, plasmas are generated by irradiating a target with a strong laser field. Understanding the non-uniform magnetic fields of the laser-target and laser-plasma interaction is essential to mitigating instabilities in the plasma and increasing the yield of the fusion reactor. One way of measuring the magnetic field is proton radiography which determines the integrated magnetic field along the proton path from the beam divergence caused by the Lorentz force. The issue is that this only allows for a 2D image as not enough information is present for a full 3D reconstruction of the magnetic field geometry. The aim of the project was to simulate the magnetic fields generated by the Biermann Battery effect and visualise what proton radiography through this field should look like. Matching the resulting data to the experimental proton radiography data allows for a better understanding of the magnitude and location of plasma instabilities in the early stages of the laser pulse.