2026-03-23 | | Total: 15
This work presents a variational physics-informed deep learning framework for phase-field modelling of brittle crack propagation in anisotropic media. Previous Deep Ritz Method (DRM) approaches have focused on second-order, isotropic phase-field fracture formulations. In contrast, the present work introduces, for the first time within a variational deep learning setting, a family of higher-order anisotropic phase-field models through a generalised crack density functional. The resulting fracture problem is solved by minimising the total energy using the DRM. The trial space is enriched with higher-order B-spline basis functions to represent higher-order gradients accurately and stably, thereby eliminating the need for conventional automatic differentiation. The methodology is assessed for isotropic, cubic, and orthotropic fracture surface energy densities. Numerical examples demonstrate direction-dependent crack growth in anisotropic cases, highlighting the capability of the method to accurately capture this behaviour.
Accurately calculating time delays between signals is pivotal in many modern physics applications. One approach to estimating these delays is computing the cross-spectrum in the time-frequency domain. Linear time-frequency representations, such as the continuous wavelet transform (CWT), are widely used to construct these cross-spectra. However, it is well known that the frequency resolution is inherently limited by the localized nature of the convolving wavelet. Moreover, the functional form of the CWT cross-spectrum is not a proper correlation measure and typically requires post-processing smoothing. Conversely, quadratic representations achieve joint time-frequency resolution approaching the Gabor-Heisenberg limit while also providing an adequate measure of similarity between the signals. Motivated by these advantages, we propose a time-delay estimation method based on the Wigner-Ville Distribution (WVD). Considering nonstationary signals arising from two typical wave-physics scenarios, we show that the WVD yields more accurate time-delay estimates with lower uncertainty, particularly in the most energetic frequency bands.
Machine-learning electronic Hamiltonians achieve orders-of-magnitude speedups over density-functional theory, yet current models omit long-range Coulomb interactions that govern physics in polar crystals and heterostructures. We derive closed-form long-range Hamiltonian matrix elements in a nonorthogonal atomic-orbital basis through variational decomposition of the electrostatic energy, deriving a variationally consistent mapping from the electron density matrix to effective atomic charges. We implement this framework in HamGNN-LR, a dual-channel architecture combining E(3)-equivariant message passing with reciprocal-space Ewald summation. Benchmarks demonstrate that physics-based long-range corrections are essential: purely data-driven attention mechanisms fail to capture macroscopic electrostatic potentials. Benchmarks on polar ZnO slabs, CdSe/ZnS heterostructures, and GaN/AlN superlattices show two- to threefold error reductions and robust transferability to systems far beyond training sizes, eliminating the characteristic staircase artifacts that plague short-range models in the presence of built-in electric fields.
Quantitative high-resolution transmission electron microscopy (HRTEM) provides an indispensable means to understand the structure-property relationships of a material in atomic dimensions. Successful quantification requires reliable retrieval of essential atomic structural information despite artifacts arising from unwanted but practically unavoidable imaging imperfections. Experimental observation carried out in tandem with model-based iterative image simulation shows vast applications in quantitative structural and chemical determination of objects spanning zero to three dimensions [Prog. Mater. Sci. 133, 101037, 2023]. However, the large number of parameters involved in the simulations make the current multi-step, user-guided iterative approach highly time consuming, thereby restricting its application primarily to small sample areas and to experienced users. In this work, we implement and apply a physics-informed Bayesian optimization (BO) framework to advance HRTEM quantification towards full automation and large-field-of-view analysis. Unlike conventional optimization approaches, our method adopts a stepwise strategy that fully leverages the strength of BO in handling high-dimensional parameters, while its probabilistic engine rigorously and efficiently refines the parameter space to enable rapid quantification. Using a BaTiO3 single crystal that contains heavy, medium and light elements as a model system, we demonstrate that the three-dimensional crystal structure can be determined from a single HRTEM image with a three to four order-of-magnitude improvement in time efficiency. This approach thus opens new avenues for fast and automated image quantification over larger sample volumes and, potentially, in the time domain.
We introduce a distribution-free lattice Boltzmann formulation for general compartmental reaction--diffusion systems arising in mathematical epidemiology. The proposed scheme, termed a single-step simplified lattice Boltzmann method (SSLBM), evolves directly macroscopic compartment densities, eliminating the need for particle distribution functions and explicit streaming operations. This yields a compact and computationally efficient framework while retaining the kinetic consistency of lattice Boltzmann methodologies. The approach is applied to a SEIRD (Susceptible-Exposed-Infected-Recovered-Deceased) reaction-diffusion model as a representative case. The resulting discrete evolution equations are derived and shown to recover the target macroscopic dynamics. The method is systematically validated against a fourth-order finite difference reference solution and compared with a standard BGK lattice Boltzmann formulation. Numerical results demonstrate that the SSLBM consistently improves accuracy across all compartments and norms. The error reduction is robust with respect to both the basic reproduction number and diffusion strength, typically ranging between factors of approximately two and five depending on the regime. In particular, the method shows enhanced control of localised errors in regimes characterised by strong nonlinear coupling and steep spatial gradients. Our findings indicate that the proposed formulation provides an accurate and efficient alternative to classical lattice Boltzmann approaches for reaction-diffusion systems, with particular advantages in stiff and nonlinear epidemic dynamics.
Purpose: To present a fully open-source framework for quasi-real-time streaming and cloud-based processing of low-field (LF) MRI data, addressing the growing computational demands of advanced reconstruction and post-processing pipelines in portable and affordable MRI systems. Methods: The proposed framework integrates open-source scanner control software with a network-enabled streaming architecture, allowing for raw data to be transmitted directly from the MRI console to remote compute resources. Cloud-based processing modules support image reconstruction and advanced post-processing, including computationally intensive physics- and learning-based methods, while maintaining compatibility with low-cost on-device control hardware. Results: The system enables continuous acquisition-to-reconstruction workflows in LF-MRI without requiring specialized high-performance console architectures. Selected example applications include deep-learning-based denoising, field-induced distortion correction, and non-Cartesian image reconstruction. Experimental demonstrations show reliable streaming performance. Conclusions: Open-source streaming and cloud-processing provide an effective pathway to overcome the computational limitations of embedded LF-MRI consoles. By decoupling acquisition hardware from intensive reconstruction workloads, the proposed framework supports scalable deployment of advanced algorithms while preserving the affordability and portability that motivate LF-MRI.
We study the coupling mechanism of surface acoustic waves (SAW) with spin waves (SW) using micromagnetic analysis. The SAW magnetoacoustic excitation field is fully implemented, i.e., all strain and lattice-rotation terms are included. A realistic CoFeB film with a weak in-plane uniaxial anisotropy is considered. We investigate the conditions for efficient SAW--SW coupling, with particular emphasis on the case where the SAW propagates parallel to the external magnetic field, a configuration of special interest for magnonic applications. Remarkably, we find that the anisotropy orientation serves as a knob to tune the parallel resonant interaction. Overall, this work provides a unified and practical picture of SAW--SW coupling in thin magnetized films.
Hydroxyapatite (HA) on a magnesium (Mg) surface is studied using density functional theory, to help understand the effect of HA coating and alloying in the surfaces of Mg-based biodegradable implants. We determine the adsorption energies and structural changes of a single layer of HA on pure Mg(0001) and on sparsely calcium (Ca) or zinc (Zn) doped Mg(0001) and find that both Zn and Ca doping improves the adsorption, except in a few positions of HA relative to the dopant position. All adsorption configurations, whether with pure or doped Mg surfaces, show deformation of the surface and HA layer. For Ca doping, we found that for a certain adsorption configuration, the dopant Ca atom moves out of the Mg surface and into the HA layer, leaving behind a Mg vacancy in the top layer of the Mg surface. Plots of electron density changes show that electrons accumulate around the Ca dopant and the neighboring Mg atoms, while in Zn doping this is less pronounced. Overall, our results demonstrate that the dopant choice and relative position of HA influence the interaction between HA and Mg-surfaces, and affect both adsorption energies and atomic and electronic structures.
Adversarial vulnerability in vision and hallucination in large language models are conventionally viewed as separate problems, each addressed with modality-specific patches. This study first reveals that they share a common geometric origin: the input and its loss gradient are conjugate observables subject to an irreducible uncertainty bound. Formalizing a Neural Uncertainty Principle (NUP) under a loss-induced state, we find that in near-bound regimes, further compression must be accompanied by increased sensitivity dispersion (adversarial fragility), while weak prompt-gradient coupling leaves generation under-constrained (hallucination). Crucially, this bound is modulated by an input-gradient correlation channel, captured by a specifically designed single-backward probe. In vision, masking highly coupled components improves robustness without costly adversarial training; in language, the same prefill-stage probe detects hallucination risk before generating any answer tokens. NUP thus turns two seemingly separate failure taxonomies into a shared uncertainty-budget view and provides a principled lens for reliability analysis. Guided by this NUP theory, we propose ConjMask (masking high-contribution input components) and LogitReg (logit-side regularization) to improve robustness without adversarial training, and use the probe as a decoding-free risk signal for LLMs, enabling hallucination detection and prompt selection. NUP thus provides a unified, practical framework for diagnosing and mitigating boundary anomalies across perception and generation tasks.
Energy renormalizations of resident carriers and excitons are studied theoretically, and compared with recent experiments of electrostatically-doped WSe$_2$ monolayers. The calculated energy renormalization of resident carriers, subjected to strong out-of-plane magnetic field, reveals the importance of dynamical screening in transition metal dichalcogenides. The energy renormalization of tightly bound excitons is analyzed through the exchange interaction between the electron (or hole) component of the exciton and resident carriers that share the same spin and valley quantum numbers. Our theory explains the weak energy shift of excitonic resonances despite the strong energy renormalization of resident carriers. We identify the dependence of the energy renormalization on the envelope function of a tightly-bound exciton, showing that unlike free electron-hole pairs, this energy renormalization is not the added renormalizations of a resident electron and resident hole.
In this work, we investigate observable signatures of a magnetically charged Anti-de Sitter black hole in string-inspired Euler-Heisenberg theory. We analyze photon trajectories, the photon sphere, and the resulting black hole shadow. We derive the photon sphere and shadow radii and show that both deviate from the Schwarzschild and Schwarzschild-AdS cases. In particular, the radii decrease monotonically as the magnetic charge parameter $Q_m$ increases, indicating that magnetic charge modifies light propagation near the black hole. We also study neutral and charged particle motion and compute the corresponding epicyclic frequencies. Using the effective potential method, we obtain the specific energy and angular momentum for stable circular orbits and determine the innermost stable circular orbit (ISCO). The presence of $Q_m$ shifts the ISCO radius and alters the orbital structure. The radial, vertical, and orbital frequencies show clear deviations from the Schwarzschild case. Finally, we confront the model with twin-peak quasi-periodic oscillation (QPO) data from stellar-mass, intermediate-mass, and supermassive black hole candidates. A two-dimensional Delta chi-square analysis in the ($r$, $Q_m$) space shows that the best fit corresponds to $Q_m=0$, although finite values remain allowed within confidence levels. At the 1 sigma level, we obtain an upper bound $Q_m/M$ less than about 0.2. These results indicate that while magnetic charge produces measurable theoretical deviations, current QPO data place only moderate constraints on its magnitude.
Recent advances in nanoscale fabrication enable atomic-scale manipulation of two-dimensional (2D) materials by introducing engineered pores and perforations. This provides new opportunities to tailor functional properties of 2D materials for applications such as selective ion transport, desalination membranes, and molecular filtration. Despite this progress, the auxetic mechanical behavior of perforated 2D materials has received little attention. In this work, large-scale reactive molecular dynamics (MD) simulations, validated against experimental measurements and first-principles calculations, are employed to investigate the mechanical response of perforated monolayer titanium-based MXene metamaterials. Architectures containing rectangular perforations with straight ligaments and sinusoidally curved ligaments are systematically examined under uniaxial tension and compression over a range of geometric parameters and temperatures, from the onset of deformation to fracture. The results demonstrate that MXene metamaterials exhibit a tunable negative Poisson's ratio (NPR), which can be controlled through the perforation geometry and surface termination. Atomistic stress analysis reveals alternating in-plane shear stresses at the junctions that induce rotational deformation of the ligaments. This rotating-junction mechanism is coupled with out-of-plane deflections arising from the low bending rigidity of atomically thin materials, producing complex three-dimensional deformations. Comparison with graphene metamaterials indicates that the perforation geometry governs qualitative auxetic trends, whereas intrinsic material properties determine quantitative responses. These findings identify MXenes as a versatile candidate for the design of tunable 2D mechanical metamaterials and provide atomistic insight into the interplay between geometry, bending rigidity, and auxetic deformation mechanisms.
The discovery of constitutive laws for complex materials has historically faced a dichotomy between high-fidelity data-driven approaches, which demand prohibitive full-field experimental data, and traditional engineering fitting, which often yields numerically unstable models outside calibration regimes. In this work, we propose an Engineering-Oriented Symbolic Regression (EO-SR) framework that bridges this gap by leveraging Large Language Models (LLMs) as "Physics-Informed Agents." Unlike unconstrained symbolic regression, our framework utilizes an LLM Agent to zero-shot synthesize executable physical constraints -- specifically thermodynamic consistency and frame indifference -- transforming the search process from mathematical curve-fitting into a physics-governed discovery engine. We validate this approach on the hyperelastic modeling of rubber-like materials using standard Treloar datasets. The framework autonomously identifies a novel hybrid constitutive law that combines a Mooney-Rivlin linear base with a rational locking term. This discovered model not only achieves high predictive accuracy across multi-axial deformation modes (including zero-shot prediction of pure shear) but also guarantees unconditional convexity. Finite element validation demonstrates that while industry-standard models (e.g., Ogden N=3) fail due to numerical singularities under severe transverse compression, the EO-SR-discovered model maintains robust convergence. This study establishes a generalized, low-barrier pathway for discovering simulation-ready constitutive closures that satisfy both data accuracy and rigorous physical laws.
We study the symmetry of the coupling between surface acoustic waves and ferromagnetic resonance in a thin magnetic film of CoFeB deposited on top of a piezoelectric Z-cut LiNbO3 substrate. We vary the orientation of the applied magnetic field with respect to the wavevector of the surface acoustic wave. Experiments indicate an unexpected 2-fold symmetry of the absorption of the SAW energy by the magnetic film. We discuss whether this symmetry can arise from the magnetoelastic torque of the longitudinal strain and the magnetic susceptibility of ferromagnetic resonance. We find that one origin of the 2-fold symmetry can be the weak in-plane uniaxial anisotropy present within the magnetic film. This phenomena adds to the previously identified other source of 2-fold symmetry but shall persist for ultrathin films when the dipolar interactions cease to contribute to the anisotropy of the slope of the spin wave dispersion relation.
We study the frequency non-reciprocity of the spin waves in symmetric CoFeB/Ru/CoFeB synthetic antiferromagnets stacks set in the scissors state by in-plane applied fields. Using a combination of Brillouin Light Scattering and propagating spin wave spectroscopy experiments, we show that the acoustical spin waves in synthetic antiferromagnets possess a unique feature if their wavevector is parallel to the applied field: the frequency non-reciprocity due to layer-to-layer dipolar interactions can be so large that the acoustical spin waves transfer energy in a unidirectional manner for a wide and bipolar interval of wavevectors. Analytical modeling and full micromagnetic calculations are conducted to account for the dispersion relations of the optical and acoustical spin waves for arbitrary field orientations. Our formalism provides a simple and direct method to understand and design devices harnessing unidirectional propagation of spin waves in synthetic antiferromagnets.