arXiv subjects

Jack B. Muir

Publications and source records attributed to Jack B. Muir.

9 recordsLinked to original sources

arXiv2026

MABLE: Masked Autoencoding with Bi-Lipschitz Decoding for Embeddings and Graph Metric Learning

We propose MABLE (Masked Autoencoding with Bi-Lipschitz Decoding for Embeddings and Graph Metric Learning), a self-supervised framework for learning node and graph embeddings from large, heterogeneous graphs, demonstrated here on geospatial mineral-exploration data. MABLE combines masked reconstruction with fixed cosine-similarity losses that align matched augmented views while keeping unpaired embeddings well spread. A bi-Lipschitz feature decoder ties a low-dimensional reconstruction component of each node embedding to feature similarity, while matched-node consistency shapes the remaining context used by graph pooling. Lipschitz-controlled pooling helps stabilize graph-level representations under perturbations of retained node embeddings, while augmentation alignment trains robustness to masking, node dropping, and sampling variation. Across local copper and regional Arabian Shield studies, MABLE embeddings provide complementary downstream signal and produce coherent embedding-derived layers for hypothesis generation without learned discriminators or hard-negative selection.

cs.LG↗

arXiv2025

Superradiant Organic Light-Emitting Diodes

Organic light-emitting diodes (OLEDs) are central to modern display technologies and are promising candidates for low-cost energy-efficient lighting. Their performance is determined by numerous, intricate fabrication parameters, but not least by the number of emissive molecules N, which provide sites for electron-hole recombination and photon generation in the diode host matrix. Counterintuitively, larger concentrations of emitters do not always lead to brighter or more efficient OLEDs due to concentration quenching of luminescence meaning that rates of radiative electron-hole recombination can become severely reduced, negatively impacting charge-to-photon conversion efficiency. In this work we trigger steady-state superradiant light emission from a series of Fabry-P\'erot microcavity OLEDs by scaling the operating voltage of each device with emitter concentration. We demonstrate a collective enhancement in the luminance of a microcavity OLED that scales super-extensively when compared to no-cavity controls fabricated in the same run. Triggering quantum correlations between emitters via the confined cavity field allows devices with fewer emitters to match or even exceed the brightness of control OLEDs even when driven by lower voltages. Moreover, our devices show significant narrowing of their emission spectra, offering purer colours at low applied voltages. Leveraging collective effects in microcavity OLEDs provides a new approach to enable brighter, more efficient devices paving the way for next-generation displays and lighting that do not compromise performance for operational efficiency or device lifetime.

quant-ph↗

arXiv2025

Experimental demonstration of a scalable room-temperature quantum battery

Harnessing quantum phenomena in energy storage systems offers an opportunity to introduce a new generation of batteries with quantum-enhanced performance. Until now, the quantum battery has largely remained a theoretical concept, with little progress towards experimental realisation, due to the challenges in quantum coherent control. Here, we experimentally demonstrate a scalable room-temperature quantum battery with a multi-layered organic-microcavity design. We show that it exhibits superextensive charging, metastabilisation of stored energy, and generates superextensive electrical power, the latter an unpredicted phenomenon. The combination of these properties in a single device is the first demonstration of the full cycle of a quantum battery, laying the framework for future designs.

quant-ph↗

arXiv2024

Imaging topological polar structures in marginally twisted 2D semiconductors

Moire superlattices formed in van der Waals heterostructures due to twisting, lattice mismatch and strain present an opportunity for creating novel metamaterials with unique properties not present in the individual layers themselves. Ferroelectricity for example, arises due to broken inversion symmetry in twisted and strained bilayers of 2D semiconductors with stacking domains of alternating out-of-plane polarization. However, understanding the individual contributions of twist and strain to the formation of topological polar nanostructures remains to be established and has proven to be experimentally challenging. Inversion symmetry breaking has been predicted to give rise to an in-plane component of polarization along the domain walls, leading to the formation of a network of topologically non-trivial merons (half-skyrmions) that are Bloch-type for twisted and Neel-type for strained systems. Here we utilise angle-resolved, high-resolution vector piezoresponse force microscopy (PFM) to spatially resolve polarization components and topological polar nanostructures in marginally twisted bilayer WSe2, and provide experimental proof for the existence of topologically non-trivial meron/antimeron structures. We observe both Bloch-type and Neel-type merons, allowing us to differentiate between moire superlattices formed due to twist or heterogeneous strain. This first demonstration of non-trivial real-space topology in a twisted van der Waals heterostructure opens pathways for exploring the connection between twist and topology in engineered nano-devices.

cond-mat.mes-hall↗

arXiv2023

Bayesian eikonal tomography using Gaussian processes

Eikonal tomography has become a popular methodology for deriving phase velocity maps from surface wave phase delay measurements. Its high efficiency makes it popular for handling datasets deriving from large-N arrays, in particular in the ambient-noise tomography setting. However, the results of eikonal tomography are crucially dependent on the way in which phase delay measurements are interpolated, a point which has not been thoroughly investigated. In this work, I provide a rigorous formulation for eikonal tomography using Gaussian processes (GPs) to interpolate phase delay measurements, including uncertainties. GPs allow the posterior phase delay gradient to be analytically derived. From the phase delay gradient, an excellent approximate solution for phase velocities can be obtained using the saddlepoint method. The result is a fully Bayesian result for phase velocities of surface waves, incorporating the nonlinear wavefront bending inherent in eikonal tomography, with no sampling required. The results of this analysis imply that the uncertainties reported for eikonal tomography are often underestimated.

physics.geo-ph↗

arXiv2023

A Deep Gaussian Process Model for Seismicity Background Rates

The spatio-temporal properties of seismicity give us incisive insight into the stress state evolution and fault structures of the crust. Empirical models based on self-exciting point-processes continue to provide an important tool for analyzing seismicity, given the epistemic uncertainty associated with physical models. In particular, the epidemic-type aftershock sequence (ETAS) model acts as a reference model for studying seismicity catalogs. The traditional ETAS model uses simple parametric definitions for the background rate of triggering-independent seismicity. This reduces the effectiveness of the basic ETAS model in modelling the temporally complex seismicity patterns seen in seismic swarms that are dominated by aseismic tectonic processes such as fluid injection rather than aftershock triggering. In order to robustly capture time-varying seismicity rates, we introduce a deep Gaussian process formulation for the background rate as an extension to ETAS. Gaussian processes (GPs) are a robust non-parametric model for function spaces with covariance structure. By conditioning the lengthscale structure of a GP with another GP, we have a deep-GP: a probabilistic, hierarchical model that automatically tunes its structure to match data constraints. We show how the deep-GP-ETAS model can be efficiently sampled by making use of a Metropolis-within-Gibbs scheme, taking advantage of the branching process formulation of ETAS and a stochastic partial differential equation (SPDE) approximation for Matérn GPs. We illustrate our method using synthetic examples, and show that the deep-GP-ETAS model successfully captures multiscale temporal behavior in the background forcing rate of seismicity. We then apply the results to two real-data catalogues: the Ridgecrest, CA July 5 2019 Mw 7.1 event catalogue and the 2016--2019 Cahuilla, CA earthquake swarm.

physics.geo-ph↗

arXiv2023

Effects of Floquet Engineering on the Coherent Exciton Dynamics in Monolayer WS$_2$

Coherent optical manipulation of electronic bandstructures via Floquet Engineering is a promising means to control quantum systems on an ultrafast timescale. However, the ultrafast switching on/off of the driving field comes with questions regarding the limits of validity of the Floquet formalism, which is defined for an infinite periodic drive, and to what extent the transient changes can be driven adibatically. Experimentally addressing these questions has been difficult, in large part due to the absence of an established technique to measure coherent dynamics through the duration of the pulse. Here, using multidimensional coherent spectroscopy we explicitly excite, control, and probe a coherent superposition of excitons in the $K$ and $K^\prime$ valleys in monolayer WS$_2$. With a circularly polarized, red-detuned, pump pulse, the degeneracy of the $K$ and $K^\prime$ excitons can be lifted and the phase of the coherence rotated. We demonstrate phase rotations during the 100 fs driving pulse that exceed $π$, and show that this can be described by a combination of the AC-Stark shift of excitons in one valley and Bloch-Siegert shift of excitons in the opposite valley. Despite showing a smooth evolution of the phase that directly follows the intensity envelope of the pump pulse, the process is not perfectly adiabatic. By measuring the magnitude of the macroscopic coherence as it evolves before, during, and after the pump pulse we show that there is additional decoherence caused by power broadening in the presence of the pump. This non-adiabaticity may be a problem for many applications, such as manipulating q-bits in quantum information processing, however these measurements also suggest ways such effects can be minimised or eliminated.

cond-mat.mes-hall↗

arXiv2022

HypoSVI: Hypocenter inversion with Stein variational inference and Physics Informed Neural Networks

We introduce a scheme for probabilistic hypocenter inversion with Stein variational inference. Our approach uses a differentiable forward model in the form of a physics informed neural network, which we train to solve the Eikonal equation. This allows for rapid approximation of the posterior by iteratively optimizing a collection of particles against a kernelized Stein discrepancy. We show that the method is well-equipped to handle highly multimodal posterior distributions, which are common in hypocentral inverse problems. A suite of experiments is performed to examine the influence of the various hyperparameters. Once trained, the method is valid for any seismic network geometry within the study area without the need to build travel time tables. We show that the computational demands scale efficiently with the number of differential times, making it ideal for large-N sensing technologies like Distributed Acoustic Sensing. The techniques outlined in this manuscript have considerable implications beyond just ray-tracing procedures, with the work flow applicable to other fields with computationally expensive inversion procedures such as full waveform inversion.

physics.geo-ph↗

arXiv2022

Exciton-polaron interactions in monolayer WS$_2$

Interactions between quasiparticles are of fundamental importance and ultimately determine the macroscopic properties of quantum matter. A famous example is the phenomenon of superconductivity, which arises from attractive electron-electron interactions that are mediated by phonons or even other more exotic fluctuations in the material. Here we introduce mobile exciton impurities into a two-dimensional electron gas and investigate the interactions between the resulting Fermi polaron quasiparticles. We employ multi-dimensional coherent spectroscopy on monolayer WS$_2$, which provides an ideal platform for determining the nature of polaron-polaron interactions due to the underlying trion fine structure and the valley specific optical selection rules. At low electron doping densities, we find that the dominant interactions are between polaron states that are dressed by the same Fermi sea. In the absence of bound polaron pairs (bipolarons), we show using a minimal microscopic model that these interactions originate from a phase-space filling effect, where excitons compete for the same electrons. We furthermore reveal the existence of a bipolaron bound state with remarkably large binding energy, involving excitons in different valleys cooperatively bound to the same electron. Our work lays the foundation for probing and understanding strong electron correlation effects in two-dimensional layered structures such as moiré superlattices.

cond-mat.mes-hall↗