arXiv subjects

Joshua Robinson

Publications and source records attributed to Joshua Robinson.

At least 19 recordsLinked to original sources

arXiv2025

Remote epitaxial frustration

Remote epitaxy relaxes the constraints of conventional epitaxy, to enable low defect density, chemically abrupt heterostructures and exfoliation of single crystalline membranes. However, definitive evidence for a true remote mechanism remains elusive because most experiments can be explained by alternative mechanism that are macroscopically indistinguishable from true remote epitaxy. Using GdAuGe films grown on graphene/SiC (0001), we present two signatures that cannot be explained by the leading alternatives to the remote mechanism: (1) a few atomic layer thick disordered interlayer at the GdAuGe/graphene interface and (2) a $30\degree$ rotated epitaxial relationship between the GdAuGe film and the SiC substrate. Density functional theory calculations indicate these signatures arise from remote epitaxial \textit{frustration}, a competition amongst epitaxy to the remotely screened substrate, to graphene, and to the graphene-induced interfacial reconstruction. Tuning the amplitudes and periodicities of these competing potentials provides new opportunities to intentionally disrupt long-range order.

cond-mat.mtrl-sci↗

arXiv2025

Blazar PKS 0446+11 -- Neutrino connection study using a lepto-hadronic model

We present a multi-wavelength study of a blazar PKS 0446+11, motivated by its spatial association with the neutrino event IC240105A detected by the IceCube Neutrino Observatory on 2024 January 5. The source is located 0.4 degrees from the best-fit neutrino direction and satisfies selection criteria for VLBI-selected, radio-bright AGN that have been identified as highly probable neutrino associations. PKS 0446+11 exhibited a major gamma-ray flare in November 2023, reaching approximately 18x its 4FGL-DR4 catalog average. Around the neutrino epoch, PKS 0446+11 remained in an elevated state, with the gamma-ray flux more than six times above its catalog level, the X-ray flux an order of magnitude above the archival measurements, and the optical-UV emission also enhanced. We used Fermi-LAT, Swift-XRT/UVOT, and archival multi-wavelength data to construct multi-wavelength light curves and spectral energy distributions (SEDs). SED modeling shows that the emission is best described by a leptonic scenario, with synchrotron emission at low energies and external Compton scattering of broad-line region and dusty torus photons dominating the X-ray - gamma-ray output. A lepto-hadronic model fails to adequately reproduce the observed SED, although hadronic cascades can broadly account for the X-ray and gamma-ray spectral coverage at lower flux levels. We compute the expected neutrino flux for the hadronic scenario and compare it to the IceCube 90% upper limit. Our results highlight the importance of continued multi-wavelength and neutrino monitoring to better understand the physical conditions under which this blazar may serve as neutrino source.

astro-ph.HE↗

arXiv2025

Robust Super-Moir\'e in Large Angle Single-Twist Bilayers

Forming long wavelength moir\'e superlattices (MSL) at small-angle twist van der Waals (vdW) bilayers has been a key approach to creating moir\'e flat bands. The small-angle twist, however, leads to strong lattice reconstruction, causing domain walls and moir\'e disorders, which pose considerable challenges in engineering such platforms. At large twist angles, the rigid lattices render a more robust, but shorter wavelength MSL, making it difficult to engineer flat bands. Here, we depict a novel approach to tailoring robust super-moir\'e (SM) structures that combines the advantages of both small-twist and large-twist transition metal dichalcogenides (TMDs) bilayers using only a single twist angle near a commensurate angle. Structurally, we unveil the spontaneous formation of a periodic arrangement of three inequivalent commensurate moir\'e (CM) stacking, where the angle deviation from the commensurate angle can tune the periodicity. Electronically, we reveal a large set of van Hove singularities (VHSs) that indicate strong band hybridization, leading to flat bands near the valence band maximum. Our study paves the way for a new platform of robust SM bilayers with structural rigidity and controllable wavelength, extending the investigation of the interplay among band topology, quantum geometry, and moir\'e superconductivity to the large twist angle regime.

cond-mat.mtrl-sci↗

arXiv2025

Humanity's Last Exam

Benchmarks are important tools for tracking the rapid advancements in large language model (LLM) capabilities. However, benchmarks are not keeping pace in difficulty: LLMs now achieve over 90\% accuracy on popular benchmarks like MMLU, limiting informed measurement of state-of-the-art LLM capabilities. In response, we introduce Humanity's Last Exam (HLE), a multi-modal benchmark at the frontier of human knowledge, designed to be the final closed-ended academic benchmark of its kind with broad subject coverage. HLE consists of 2,500 questions across dozens of subjects, including mathematics, humanities, and the natural sciences. HLE is developed globally by subject-matter experts and consists of multiple-choice and short-answer questions suitable for automated grading. Each question has a known solution that is unambiguous and easily verifiable, but cannot be quickly answered via internet retrieval. State-of-the-art LLMs demonstrate low accuracy and calibration on HLE, highlighting a significant gap between current LLM capabilities and the expert human frontier on closed-ended academic questions. To inform research and policymaking upon a clear understanding of model capabilities, we publicly release HLE at https://lastexam.ai.

cs.LG↗

arXiv2025

Charge to spin conversion in atomically thin bismuth

We report charge to spin conversion in a hybrid heterostructure comprised of atomically thin bismuth (Bi) confined between a silicon carbide (SiC) substrate and epitaxial graphene (EG). We confirm composition, dimensionality, and a 96.5 \% intercalation coverage using X-ray photolectron spectroscopy, scanning transmission microscopy, low energy electron diffraction, and Raman spectroscopy. Electrical transport measurements show signs of weak antilocalization in the heterostructure, consistent with spin-orbit coupling in this hybrid heterostructure. Spin torque ferromagnetic resonance measurements in permalloy/EG/2D-Bi heterostructures probe charge-to-spin conversion and revealing that an in plane polarization of the spin current, perpendicular to the charge current. The ratio of the in-plane to out-of-plane torque is 3.75 times higher than in hydrogenated graphene control samples.

cond-mat.mes-hall↗

arXiv2024

A Comprehensive Hadronic Code Comparison for Active Galactic Nuclei

We perform the first dedicated comparison of five hadronic codes (AM$^3$, ATHE$\nu$A, B13, LeHa-Paris, and LeHaMoC) that have been extensively used in modeling of the spectral energy distribution (SED) of jetted active galactic nuclei. The purpose of this comparison is to identify the sources of systematic errors (e.g., implementation method of proton-photon interactions) and to quantify the expected dispersion in numerical SED models computed with the five codes. The outputs from the codes are first tested in synchrotron self-Compton scenarios that are the simplest blazar emission models used in the literature. We then compare the injection rates and spectra of secondary particles produced in pure hadronic cases with monoenergetic and power-law protons interacting on black-body and power-law photon fields. We finally compare the photon SEDs and the neutrino spectra for realistic proton-synchrotron and leptohadronic blazar models. We find that the codes are in excellent agreement with respect to the spectral shape of the photons and neutrinos. There is a remaining spread in the overall normalization that we quantify, at its maximum, at the level of $\pm 40\%$. This value should be used as an additional, conservative, systematic uncertainty term when comparing numerical simulations and observations.

astro-ph.HE↗

arXiv2024

Neutrino detection rates from lepto-hadronic model simulations of bright blazar flares

There is mounting evidence that blazars are the sources of part of the very-high-energy astrophysical neutrino flux detected by IceCube. In particular, there have been several spatial and temporal coincidences of individual IceCube neutrino events with flaring blazars, the most prominent of them being IceCube-170922A, coincident with a multi-wavelength flare of TXS~0506+056. Motivated by this, we used the time-dependent lepto-hadronic code OneHaLe to model the spectral energy distributions and light curves of a sample of bright $\gamma$-ray flares of blazars detected by Fermi-LAT, for which Kreter et al. (2020) provided calorimetric estimates of the expected neutrino detection rates. Flares were modelled with temporal changes of the proton injection spectra. Our analysis shows that the calorimetric approach overestimates the increase in neutrino production by a factor of typically $\sim 10$ if the $\gamma$-ray emission is dominated by proton-synchrotron radiation.

astro-ph.HE↗

arXiv2024

LocateBench: Evaluating the Locating Ability of Vision Language Models

The ability to locate an object in an image according to natural language instructions is crucial for many real-world applications. In this work we propose LocateBench, a high-quality benchmark dedicated to evaluating this ability. We experiment with multiple prompting approaches, and measure the accuracy of several large vision language models. We find that even the accuracy of the strongest model, GPT-4o, lags behind human accuracy by more than 10%.

cs.CV↗

arXiv2024

Giant and Tunable Bosonic Quantum Interference Induced by Two-Dimensional Metals

Harnessing quantum interference among bosons provides significant opportunities as bosons often carry longer coherence time than fermions. As an example of quantum interference, Fano resonance involving phonons or photons describes the coupling between discrete and continuous states, signified by an asymmetric spectral lineshape. Utilizing photon-based Fano resonance, molecule sensing with ultra-high sensitivity and ultrafast optical switching has been realized. However, phonon-based Fano resonance, which would expand the application space to a vaster regime, has been less exploited because of the weak coupling between discrete phonons with continuous states such as electronic continuum. In this work, we report the discovery of giant phonon-based Fano resonance in a graphene/2D Ag/SiC heterostructure. The Fano asymmetry, being proportional to the coupling strength, exceeds prior reports by two orders of magnitude. This Fano asymmetry arises from simultaneous frequency and lifetime matching between discrete and continuous phonons of SiC. The introduction of 2D Ag layers restructures SiC at the interface and facilitates resonant scattering to further enhance the Fano asymmetry, which is not achievable with conventional Ag thin films. With these unique properties, we demonstrated that the phonon-based Fano resonance can be used for ultrasensitive molecule detection at the single-molecule level. Our work highlights strong Fano resonance in the phononic system, opening avenues for engineering quantum interference based on bosons. Further, our findings provide opportunities for advancing phonon-related applications, including biochemical sensing, quantum transduction, and superconductor-based quantum computing.

cond-mat.mtrl-sci↗

arXiv2024

RelBench: A Benchmark for Deep Learning on Relational Databases

We present RelBench, a public benchmark for solving predictive tasks over relational databases with graph neural networks. RelBench provides databases and tasks spanning diverse domains and scales, and is intended to be a foundational infrastructure for future research. We use RelBench to conduct the first comprehensive study of Relational Deep Learning (RDL) (Fey et al., 2024), which combines graph neural network predictive models with (deep) tabular models that extract initial entity-level representations from raw tables. End-to-end learned RDL models fully exploit the predictive signal encoded in primary-foreign key links, marking a significant shift away from the dominant paradigm of manual feature engineering combined with tabular models. To thoroughly evaluate RDL against this prior gold-standard, we conduct an in-depth user study where an experienced data scientist manually engineers features for each task. In this study, RDL learns better models whilst reducing human work needed by more than an order of magnitude. This demonstrates the power of deep learning for solving predictive tasks over relational databases, opening up many new research opportunities enabled by RelBench.

cs.LG↗

arXiv2023

Relational Deep Learning: Graph Representation Learning on Relational Databases

Much of the world's most valued data is stored in relational databases and data warehouses, where the data is organized into many tables connected by primary-foreign key relations. However, building machine learning models using this data is both challenging and time consuming. The core problem is that no machine learning method is capable of learning on multiple tables interconnected by primary-foreign key relations. Current methods can only learn from a single table, so the data must first be manually joined and aggregated into a single training table, the process known as feature engineering. Feature engineering is slow, error prone and leads to suboptimal models. Here we introduce an end-to-end deep representation learning approach to directly learn on data laid out across multiple tables. We name our approach Relational Deep Learning (RDL). The core idea is to view relational databases as a temporal, heterogeneous graph, with a node for each row in each table, and edges specified by primary-foreign key links. Message Passing Graph Neural Networks can then automatically learn across the graph to extract representations that leverage all input data, without any manual feature engineering. Relational Deep Learning leads to more accurate models that can be built much faster. To facilitate research in this area, we develop RelBench, a set of benchmark datasets and an implementation of Relational Deep Learning. The data covers a wide spectrum, from discussions on Stack Exchange to book reviews on the Amazon Product Catalog. Overall, we define a new research area that generalizes graph machine learning and broadens its applicability to a wide set of AI use cases.

cs.LG↗

arXiv2023

Expressive Sign Equivariant Networks for Spectral Geometric Learning

Recent work has shown the utility of developing machine learning models that respect the structure and symmetries of eigenvectors. These works promote sign invariance, since for any eigenvector v the negation -v is also an eigenvector. However, we show that sign invariance is theoretically limited for tasks such as building orthogonally equivariant models and learning node positional encodings for link prediction in graphs. In this work, we demonstrate the benefits of sign equivariance for these tasks. To obtain these benefits, we develop novel sign equivariant neural network architectures. Our models are based on a new analytic characterization of sign equivariant polynomials and thus inherit provable expressiveness properties. Controlled synthetic experiments show that our networks can achieve the theoretically predicted benefits of sign equivariant models. Code is available at https://github.com/cptq/Sign-Equivariant-Nets.

cs.LG↗

arXiv2023

On Retrieval Augmentation and the Limitations of Language Model Training

Augmenting a language model (LM) with $k$-nearest neighbors ($k$NN) retrieval on its training data alone can decrease its perplexity, though the underlying reasons for this remain elusive. In this work, we rule out one previously posited possibility -- the "softmax bottleneck." We then create a new dataset to evaluate LM generalization ability in the setting where training data contains additional information that is not causally relevant. This task is challenging even for GPT-3.5 Turbo. We show that, for both GPT-2 and Mistral 7B, $k$NN retrieval augmentation consistently improves performance in this setting. Finally, to make $k$NN retrieval more accessible, we propose using a multi-layer perceptron model that maps datastore keys to values as a drop-in replacement for traditional retrieval. This reduces storage costs by over 25x.

cs.CL↗

arXiv2023

Dynamic STEM-EELS for single atom and defect measurement during electron beam transformations

On- and off-axis electron energy loss spectroscopy (EELS) is a powerful method for probing local electronic structure on single atom level. However, many materials undergo electron-beam induced transformation during the scanning transmission electron microscopy (STEM) and spectroscopy, the problem particularly acute for off-axis EELS signals. Here, we propose and operationalize the rapid object detection and action system (RODAS) for dynamic exploration of the structure-property relationships in STEM-EELS. In this approach, the electron beam is used to induce dynamic transformations creating new defect types at sufficiently small rates and avoiding complete material destruction. The deep convolutional neural networks trained via the ensemble learning iterative training (ELIT) approach are used to identify the defects as they form and perform EELS measurements only at specific defect types. Overall, in this case the EEL spectra are collected only at predefined objects of interest, avoiding measurements on the ideal regions or holes. We note that this approach can be extended to identify new defect classes as they appear, allowing for efficient collection of structure-property relationship data via balanced sampling over defect types.

cond-mat.mtrl-sci↗

arXiv2023

On the Stability of Expressive Positional Encodings for Graphs

Designing effective positional encodings for graphs is key to building powerful graph transformers and enhancing message-passing graph neural networks. Although widespread, using Laplacian eigenvectors as positional encodings faces two fundamental challenges: (1) \emph{Non-uniqueness}: there are many different eigendecompositions of the same Laplacian, and (2) \emph{Instability}: small perturbations to the Laplacian could result in completely different eigenspaces, leading to unpredictable changes in positional encoding. Despite many attempts to address non-uniqueness, most methods overlook stability, leading to poor generalization on unseen graph structures. We identify the cause of instability to be a ``hard partition'' of eigenspaces. Hence, we introduce Stable and Expressive Positional Encodings (SPE), an architecture for processing eigenvectors that uses eigenvalues to ``softly partition'' eigenspaces. SPE is the first architecture that is (1) provably stable, and (2) universally expressive for basis invariant functions whilst respecting all symmetries of eigenvectors. Besides guaranteed stability, we prove that SPE is at least as expressive as existing methods, and highly capable of counting graph structures. Finally, we evaluate the effectiveness of our method on molecular property prediction, and out-of-distribution generalization tasks, finding improved generalization compared to existing positional encoding methods. Our code is available at \url{https://github.com/Graph-COM/SPE}.

cs.LG↗

arXiv2023

Atomistic Control in Molecular Beam Epitaxy Growth of Intrinsic Magnetic Topological Insulator MnBi2Te4

Intrinsic magnetic topological insulators have emerged as a promising platform to study the interplay between topological surface states and ferromagnetism. This unique interplay can give rise to a variety of exotic quantum phenomena, including the quantum anomalous Hall effect and axion insulating states. Here, utilizing molecular beam epitaxy (MBE), we present a comprehensive study of the growth of high-quality MnBi2Te4 thin films on Si (111), epitaxial graphene, and highly ordered pyrolytic graphite substrates. By combining a suite of in-situ characterization techniques, we obtain critical insights into the atomic-level control of MnBi2Te4 epitaxial growth. First, we extract the free energy landscape for the epitaxial relationship as a function of the in-plane angular distribution. Then, by employing an optimized layer-by-layer growth, we determine the chemical potential and Dirac point of the thin film at different thicknesses. Overall, these results establish a foundation for understanding the growth dynamics of MnBi2Te4 and pave the way for the future applications of MBE in emerging topological quantum materials.

cond-mat.mtrl-sci↗

arXiv2023

Structuring Representation Geometry with Rotationally Equivariant Contrastive Learning

Self-supervised learning converts raw perceptual data such as images to a compact space where simple Euclidean distances measure meaningful variations in data. In this paper, we extend this formulation by adding additional geometric structure to the embedding space by enforcing transformations of input space to correspond to simple (i.e., linear) transformations of embedding space. Specifically, in the contrastive learning setting, we introduce an equivariance objective and theoretically prove that its minima forces augmentations on input space to correspond to rotations on the spherical embedding space. We show that merely combining our equivariant loss with a non-collapse term results in non-trivial representations, without requiring invariance to data augmentations. Optimal performance is achieved by also encouraging approximate invariance, where input augmentations correspond to small rotations. Our method, CARE: Contrastive Augmentation-induced Rotational Equivariance, leads to improved performance on downstream tasks, and ensures sensitivity in embedding space to important variations in data (e.g., color) that standard contrastive methods do not achieve. Code is available at https://github.com/Sharut/CARE.

cs.LG↗

arXiv2023

A deep learning approach to using wearable seismocardiography (SCG) for diagnosing aortic valve stenosis and predicting aortic hemodynamics obtained by 4D flow MRI

In this paper, we explored the use of deep learning for the prediction of aortic flow metrics obtained using 4D flow MRI using wearable seismocardiography (SCG) devices. 4D flow MRI provides a comprehensive assessment of cardiovascular hemodynamics, but it is costly and time-consuming. We hypothesized that deep learning could be used to identify pathological changes in blood flow, such as elevated peak systolic velocity Vmax in patients with heart valve diseases, from SCG signals. We also investigated the ability of this deep learning technique to differentiate between patients diagnosed with aortic valve stenosis (AS), non-AS patients with a bicuspid aortic valve (BAV), non-AS patients with a mechanical aortic valve (MAV), and healthy subjects with a normal tricuspid aortic valve (TAV). In a study of 77 subjects who underwent same-day 4D flow MRI and SCG, we found that the Vmax values obtained using deep learning and SCGs were in good agreement with those obtained by 4D flow MRI. Additionally, subjects with TAV, BAV, MAV, and AS could be classified with ROC-AUC values of 92%, 95%, 81%, and 83%, respectively. This suggests that SCG obtained using low-cost wearable electronics may be used as a supplement to 4D flow MRI exams or as a screening tool for aortic valve disease.

cs.LG↗