Papers

Multicomponent vesicles suspended in viscoelastic fluids are crucial for understanding a variety of physiological processes. In this work, we develop a continuum surface force (CSF) phase-field model to investigate the hydrodynamics of inextensible multicomponent vesicles in viscoelastic fluid flows with inertial forces. Our model couples a fluid field comprising both Newtonian and Oldroyd-B fluids, a surface concentration field representing the multicomponent distribution on the vesicle membrane, and a phase-field variable governing the membrane evolution. The viscoelasticity effect of extra stress is well incorporated into the full Navier-Stokes equations in the fluid field. The surface concentration field is determined by Cahn-Hilliard equations, while the membrane evolution is governed by a nonlinear advection-diffusion equation. The membrane is coupled to the surrounding fluid through the continuum surface force (CSF) framework. To ensure stable numerical solutions of the highly nonlinear multi-field model, we employ a residual-based variational multiscale (RBVMS) method for the Navier-Stokes equations, a Streamline-Upwind Petrov-Galerkin (SUPG) method for the Oldroyd-B equations, and a standard Galerkin finite element framework for the remaining equations. The system of PDEs is solved using an implicit, monolithic scheme based on the generalized-$\alpha$ time integration method. To enhance spatial accuracy, we employ isogeometric analysis (IGA). We present a series of two-dimensional numerical examples in shear and Poiseuille flows to elucidate the influence of membrane composition and fluid viscoelasticity on the hydrodynamics of multicomponent vesicles.

Physics-Informed Neural Networks with hard constraints (HC-PINNs) are increasingly favored for their ability to strictly enforce boundary conditions via a trial function ansatz $\tilde{u} = A + B \cdot N$, yet the theoretical mechanisms governing their training dynamics have remained unexplored. Unlike soft-constrained formulations where boundary terms act as additive penalties, this work reveals that the boundary function $B$ introduces a multiplicative spatial modulation that fundamentally alters the learning landscape. A rigorous Neural Tangent Kernel (NTK) framework for HC-PINNs is established, deriving the explicit kernel composition law. This relationship demonstrates that the boundary function $B(\vec{x})$ functions as a spectral filter, reshaping the eigenspectrum of the neural network's native kernel. Through spectral analysis, the effective rank of the residual kernel is identified as a deterministic predictor of training convergence, superior to classical condition numbers. It is shown that widely used boundary functions can inadvertently induce spectral collapse, leading to optimization stagnation despite exact boundary satisfaction. Validated across multi-dimensional benchmarks, this framework transforms the design of boundary functions from a heuristic choice into a principled spectral optimization problem, providing a solid theoretical foundation for geometric hard constraints in scientific machine learning.

Two different non-destructive techniques of characterization were coupled to study the mechanical behavior of materials inside a battery during operation: measurement of the deformation of a battery casing by strain gauge and acoustic emission due to the release of mechanical stress inside the battery materials. Experiments were conducted on a commercial Ni-Zn AA-type battery and, particularly, phenomena occurring during overcharge were considered. Beyond the example of this study on a Ni-Zn battery, these two techniques offer complementary information that can be very useful for the monitoring and management of different type of batteries.

The energy-transfer efficiency of the natural photosynthesis system seems to be perfectly optimized during the evolution for millions of years. However, how to enhance the efficiency in the artificial light-harvesting systems is still unclear. In this paper, we investigate the energy-transfer process in the photosystem I (PSI). When there is no effective coupling between the outer antenna (OA) and the reaction center (RC), the two light-harvesting networks are disconnected and thus the energy transfer is inefficient. In order to repair these disconnected networks, we introduce a bridge with three sites between them. We find that by modulating the level structure of the 3-site bridge to be resonant, the energy transfer via the dark state will be enhanced and even outperform the original PSI. Our discoveries may shed light on the designing mechanism of artificial light-harvesting systems.

This paper presents a test for wide-sense stationarity (WSS) based on the geometry of the covariance function. We estimate local patches of the covariance surface and then check whether the directional derivative in the $(1,1,0)$ direction is zero on each patch. The method only requires the covariance function to be locally smooth and does not assume stationarity in advance. It can be applied to general stochastic dynamical systems and provides a time-resolved view. We apply the test method to an SDOF system and to a stochastic Duffing oscillator. These examples show that the method is numerically stable and can detect departures from WSS in practice.

In this work, we study the detection of quantum phase transitions (QPTs) in coherent Ising machines (CIMs) through a spectral mapping between the one-dimensional XY spin model and a network of degenerate optical parametric oscillators (DOPOs). This exact correspondence reveals that the DOPO spectrum faithfully encodes the critical behavior of anisotropic and isotropic XY chains, while a strongly anisotropic regime provides a controlled near-transverse-field-Ising realization, with the energy gap closing precisely at the critical point. The ground-state energy density and its derivatives are analyzed, and the resulting singularities in magnetic susceptibility accurately pinpoint the quantum critical points of the corresponding spin model. These results show that CIMs not only serve as powerful platforms for solving combinatorial optimization problems but also provide a versatile tool for probing universal quantum critical phenomena, bridging quantum-spin models and photonic quantum systems.

H2O is a unique substance with exceptional thermal properties arising from the subtle interplay between its electronic, phononic, and structural degrees of freedom. Of particular interest in H2O are the negative thermal expansion (NTE) phenomena, observed in its solid phase (ice) at low temperature, and in its liquid phase (water) near the freezing temperature. Furthermore, ice and water exhibit the abnormal volume isotope effect (VIE), where volume expansions occur when replacing H with its heavier isotope, deuterium (D). In order to capture more conceptual and intuitive understanding of intriguing NTE and VIE phenomena in ice and water, we have explored isotope effects in their NTE and melting properties by employing a type of Born-Oppenheimer-approximation approach and the Lindemann criterion. Our findings demonstrate that unusual isotope effects in these phenomena stem from competition between zero-point-energy phonons, thermal phonons, and the hydrogen bonding in H2O. All these components originate from nuclear quantum mechanical (QM) processes, revealing that QM physics plays a crucial role in the seemingly classical ice/water systems.

An elongated plasma column has been successfully generated and sustained in a linear plasma device using a hollow cathode discharge in the presence of an axisymmetric magnetic field. The confinement of cold energetic electrons produced near the hollow cathode plays a crucial role in guiding the plasma along the device axis. Experimental diagnostics reveal a high concentration of energetic electrons in the peripheral region near the source, which progressively converge toward the axis at a downstream location approximately 3.0 meters from the cathode. The length of the plasma column exhibits an inverse relationship with the electron-neutral collision frequency, indicating the significance of collisional damping in the propagation of energetic electrons. These observations are further supported by fluid simulations performed using COMSOL Multiphysics, which qualitatively reproduce the experimental trends. The results are consistent with a theoretical model previously proposed by the authors, reinforcing the understanding of energetic electron behaviour in magnetically guided plasma columns.

We demonstrate anisotropic photostriction in two-dimensional orthorhombic semiconductors using time-dependent density functional theory. By tracing the dynamics of photoexcited carriers, we establish a quantitative link between carrier density and lattice deformation in layered black phosphorus and germanium selenides. The structural response exhibits significant anisotropy, featuring lattice expansion along the armchair direction and contraction along the zigzag direction, which is attributed to the interplay between charge redistribution and intrinsic lattice anisotropy. Both the magnitude and orientation of the photostrictive strains can be tuned by photodoping densities, enabling precise control over the photoinduced response. Notably, the photoinduced strains significantly increase carrier recombination lifetimes by suppressing nonradiative recombination, primarily due to the enlarged bandgap and weakened nonadiabatic coupling. These results provide microscopic insight into the origin of anisotropic photostriction in low-dimensional systems and lay the groundwork for light-controllable, directionally sensitive optomechanical devices at the atomic scale.

We present an analytical framework for the emergence of nonclassical radiation in strongly laser-driven quantum systems, with a focus on high-order harmonic generation (HHG). Starting from a Pauli-Fierz description, we employ a parametric factorization of the coupled light-matter wavefunction that reduces the dynamics to coupled equations for a field-driven electronic state and a quantized light mode. Within this framework, we identify a simple and predictive mechanism for nonclassicality: it originates from the nonlinear dependence of the electronic dipole response on the light-mode coordinate. An approximately constant dipole yields coherent radiation, a linear dependence produces squeezing, and higher-order nonlinearities give rise to Wigner-function negativity. We illustrate this mechanism for atomic and molecular model systems and analyze its scaling in multi-emitter configurations, indicating routes toward high-photon-number nonclassical radiation in HHG. Our results provide a transparent connection between strong-field dynamics and quantum-optical properties of emitted light, offering a basis for engineering nonclassical states in intense laser-matter interactions.

Due to the prolonged MRI encoding process, respiratory motion can cause undesired artifacts and image blurring, degrading image quality and limiting clinical applications in abdominal and pulmonary imaging. In this work, we develop a two-stage respiratory motion-resolved radial MR image reconstruction pipeline using an interpretable deep unrolled network (MoraNet), enabling high-quality imaging under free-breathing conditions. Firstly, low-resolution images are reconstructed from the central region of successive golden-angle radial k-space to extract respiratory motion signals. The binned k-space data based on the respiratory signal are then used to reconstruct the motion-resolved high-resolution image for each motion state. The MoraNet applies nonuniform fast Fourier transform (NUFFT) to operate radial encoding and convolutional neural network (CNN) modules to conduct image regularizations. The MoraNet was trained on retrospectively acquired lung MRI images for both fully sampled and undersampled acquisitions. The performance of the proposed method was evaluated on digital CT/MRI breathing XCAT (CoMBAT) phantom data, QUASAR motion phantom data acquired from a 1.0T MRI scanner and volunteer chest data acquired from a 1.5T MRI scanner. The MoraNet pipeline was compared with motion-averaged reconstruction and a conventional compressed sensing (CS)-based method in terms of SSIM, RMSE and computation time. Simulation and experimental results demonstrated that the proposed network could provide accurate respiratory signal estimation and enable effective motion correction. Compared with the CS method, the MoraNet preserved better structural details with lower RMSE and higher SSIM values at acceleration factor of 4, and meanwhile took ten-fold faster inference time.

Biological membranes are known to form various structural motifs, from lipid bilayers to tubular filaments and networks facilitating e.g. adhesion and cell-cell communication. To understand the biophysical processes underpinning lipid-lipid interactions in these systems, synthetic membrane models are crucial. Here, we demonstrate the formation of tubular networks from supported lipid membranes of controlled lipid composition on glass. We quantify tube radii using quantitative differential interference contrast (qDIC) and propose a biophysical mechanism for the formation of these structures, regulated by surface tension and lipid exchange with connected supported membranes. Two lipid types are investigated, namely DOPC and DC15PC, exhibiting a liquid disordered and a solid ordered phase at room temperature, respectively. Tube formation is studied versus temperature, revealing bilamellar layers retracting and folding into tubes upon DC15PC lipids transitioning from liquid to solid phase, which is explained by lipid transfer from bilamellar to unilamellar layers. This study introduces a novel model system for bilayer tubes, allowing to elucidate the biophysics of lipid-lipid interactions governing lipid membrane reshaping into tubular structures, important for our understanding of biological membrane filaments.

There has been debate for over 70-years about whether active skeletal muscle is dynamically stable at lengths greater than its optimal length. The stability of computational muscle models is a critical issue, as it directly affects our ability to simulate muscle deformation across different operating lengths, especially at lengths where muscles are known to remain functional despite model-predicted instabilities. In this study, we revisit the question of dynamical stability of ODE-based models of skeletal muscle. In particular, we investigate whether activation-independent tissue properties can provide stability to contractions along the dip region of the total force-length curve. First, using a combination of analytical tools (eigenvalue analysis and non-dimensionalization) and numerical simulations, we confirm that traditional Hill-type muscle models can display divergent dynamics in this region. Then, we propose a stabilized version of a 1D Hill-type muscle model that incorporates the 3D nature of skeletal muscle deformation. This results in a completely convex force-length relationship that can bring robustness to numerical simulations, while preserving the computational efficiency of 1D models. Our findings suggest that activation-independent intrinsic mechanical properties of muscle are sufficient to stabilize contractions even in the dip region, offering new insight into how muscles maintain functional integrity during active stretch.

Astronomy, often perceived as a distant or luxury science, holds immense potential as a driver for sustainable local socio-economic development. This paper explores how astronomy can create tangible benefits for communities through education, tourism, technology transfer, and capacity building. Using case studies from South Africa, Chile, Indonesia, and India, we demonstrate how astronomical facilities and initiatives have stimulated local economies, generated employment, supported small enterprises, and enhanced STEM participation, while simultaneously inspiring a sense of shared global heritage. The analysis identifies both successes and challenges, including unequal benefit distribution, limited local ownership, and sustainability gaps once external funding ends. Building on these lessons, we propose a practical framework/guidelines for designing, implementing, and evaluating astronomy-based community initiatives, rooted in participatory engagement and aligned with the UN Sustainable Development Goals (SDGs). This paper positions astronomy as a catalyst for inclusive growth, demonstrating that investment in the cosmos can translate into grounded, measurable benefits for people and places on Earth.

Zero-dimensional cardiovascular models provide a computationally efficient framework for studying global hemodynamic behavior, yet the influence of model complexity on parameter sensitivity remains insufficiently understood. This work investigates two lumped-parameter cardiovascular models, a simplified single-ventricle configuration and a detailed four-chamber representation, to examine how physiological parameter sensitivities vary with model structure. Time-varying elastance functions are used to represent cardiac dynamics, and global sensitivity analysis is performed using Sobol and Morris methods to quantify the impact of key physiological parameters, including venous return, myocardial contractility, total peripheral resistance, and arterial compliance. The results demonstrate that sensitivity rankings differ substantially between the two models, highlighting the role of model granularity and parameter interactions in shaping cardiovascular responses. These findings support sensitivity-driven model reduction and provide a foundation for scalable, non-invasive cardiovascular simulation frameworks.

Polymers-macromolecular systems composed of repeating chemical units-constitute the molecular foundation of living organisms, while their synthetic counterparts drive transformative advances across medicine, consumer products, and energy technologies. While machine learning (ML) models have been trained on millions of quantum chemical atomistic simulations for materials and/or small molecular structures to enable efficient, accurate, and transferable predictions of chemical properties, polymers have largely not been included in prior datasets due to the computational expense of high quality electronic structure calculations on representative polymeric structures. Here, we address this shortcoming with the creation of the Open Polymers 2026 (OPoly26) dataset, which contains more than 6.57 million density functional theory (DFT) calculations on up to 360 atom clusters derived from polymeric systems, comprising over 1.2 billion total atoms. OPoly26 captures the chemical diversity that makes polymers intrinsically tunable and versatile materials, encompassing variations in monomer composition, degree of polymerization, chain architectures, and solvation environments. We show that augmenting ML model training with the OPoly26 dataset improves model performance for polymer prediction tasks. We also publicly release the OPoly26 dataset to help further the development of ML models for polymers, and more broadly, strive towards universal atomistic models.

The simulation of turbulent flows remains a central challenge, as even our most powerful computers cannot resolve the finest scales of motion in many flows of practical interest. As a result, the effects of unresolved scales on large eddies must be modelled via closures and coarse-graining procedures. Large-eddy simulation (LES) traditionally coarse-grains Navier-Stokes equations using Smagorinsky's effective viscosity model. This has the merit of simplicity but fails to account for strong non-equilibrium effects, as they typically arise in most flows in the vicinity of solid walls, the reason being that the notion of eddy viscosity assumes scale separation between small and large eddies, an assumption that fails for high-Reynolds flows far from equilibrium. The lattice Boltzmann method (LBM) offers an alternative by coarse-graining at the kinetic level, potentially capturing non-equilibrium effects beyond reach of hydrodynamic closures. This paper addresses whether LBM-Smagorinsky LES of forced homogeneous isotropic turbulence (FHIT) exhibits kinetic behavior. We test whether the turbulent Knudsen number $K_t$, measuring scale separation, reaches order one (kinetic regime) or remains asymptotically small (hydrodynamic regime). Using reference DNS ($800^3$) and iso-Reynolds LES ($100^3$) at $Re = 2 \times 10^4$, we quantify $K_t$ via spatial maps, temporal statistics, energy spectra, and higher-order moments. Results show $K_t \sim O(10^{-3})$, strictly positive without negative excursions, with spectra and flatness following canonical LES behavior. We conclude that despite its kinetic formulation, LBM-Smagorinsky LES operates in the hydrodynamic regime, with small FHIT eddies remaining in local equilibrium with larger ones, validating Smagorinsky viscosity and confirming that LBM-LES functions as conventional hydrodynamic LES while preserving LBM efficiency and locality.

Turnip (Brassica rapa) is a native species of the Mediterranean area, spread from northwest France to south Algeria. In this study, dormancy and germination traits were assessed for 61 wild Brassica rapa populations collected across the Mediterranean region. Seed dormancy is a key factor influencing germination and seedling establishment. Three dormancy-breaking methods were compared: gibberellic acid, scarification and cold plasma. The efficiency and selectivity were evaluated through germination ability, time to 10% germination (T10), mean germination time and greenhouse emergence. Five days after imbibition, germination was only 18% for the untreated seeds but 60% for the plasma-treated seeds. Germination also began 24 hours earlier and mean germination time was reduced across most populations. However, there was a limited effect on seedling emergence, which remained around 55% for both untreated and treated samples. Comparative analysis indicates that cold plasma was more effective in alleviating embryo dormancy. In addition, histological and scanning electron microscopy showed that the seed coat differed according to the geographical origin of the populations, with a deeper dormancy in seeds from Sicilian populations.

Viruses display striking diversity in structure, transmission mode, immune interaction, and evolutionary behavior. Despite this diversity, viral strategies are not unconstrained. Here we present a unifying framework that treats viral evolution as a problem of constrained optimization governed by physical decay, immune pressure, mutation robustness, and transmission architecture. We model virions as multi-component physical systems subject to irreversible environmental failure and viruses as replicators operating under immune-driven selection and mutation-selection balance. Within this framework, major viral transmission strategies arise as necessary solutions rather than taxonomic accidents. Environmentally transmitted and airborne viruses are predicted to be structurally simple, chemically stable, and reliant on replication volume rather than immune suppression. Structurally complex viruses tolerate rapid environmental decay by encoding immune-modulatory machinery, latency, or persistent replication, at the cost of reduced mutation robustness. Temperature-dependent seasonality emerges naturally from the thermally activated nature of viral decay, without invoking host behavior.

What happens to a unified conscious field when its physical implementation straddles a black hole event horizon? This paper addresses that question for integration-based theories, including Integrated Information Theory, Global Workspace Theory, and Predictive Processing. These views share a structural commitment: unity requires a single strongly connected component (SCC) in an effective causal graph over a finite integration window $\tau$. Using the standard black hole causal structure, I show that no SCC can span an event horizon. Any theory that ties unity to strong connectivity must therefore accept that a single conscious field cannot remain numerically identical and unified across such a configuration. From the perspective of the theories themselves, the outcome is bifurcation: each causally connected subsystem continues to satisfy the very structural criteria the theory declared necessary for a unified field. On any such view, the number and boundaries of unified conscious fields are therefore fixed not by the substrate alone but by the conjunction of its internal architecture with the relativistic causal structure of the spacetime it occupies, a dependency that ordinary spacetime conceals by supplying it so abundantly, but which an event horizon abruptly withdraws.

We propose a new analytical potential function to model proton transfer in the adenine-thymine base pair and develop a non-adiabatic quantum mechanical framework to calculate genetic mutation probabilities. This potential has been used to calculate the probability of mutation in a non-adiabatic process. The results of the new model have been shown to be consistent with the findings of other researchers.

The discovery of novel odorant molecules is key for the fragrance and flavor industries, yet efficiently navigating the vast chemical space to identify structures with desirable olfactory properties remains a significant challenge. Generative artificial intelligence offers a promising approach for de novo molecular design but typically requires large sets of molecules to learn from. To address this problem, we present a framework combining a variational autoencoder (VAE) with a quantitative structure-activity relationship (QSAR) model to generate novel odorants from limited training sets of odor molecules. The self-supervised learning capabilities of the VAE allow it to learn SMILES grammar from ChEMBL database, while its training objective is augmented with a loss term derived from an external QSAR model to structure the latent representation according to odor probability. While the VAE demonstrated high internal consistency in learning the QSAR supervision signal, validation against an external, unseen ground truth dataset (Unique Good Scents) confirms the model generates syntactically valid structures (100% validity achieved via rejection sampling) and 93.9% unique structures. The latent space is effectively structured by odor likelihood, evidenced by a Frechet ChemNet Distance (FCD) of ~6.96 between generated molecules and known odorants, compared to ~21.6 for the ChEMBL baseline. An ablation of the property head shows that its contribution lies in the density of retrievable structures near an odorant seed – a 4.3-fold increase in distinct novel compounds per seed – rather than in the odor-likeness of any individual compound. Bemis-Murcko analysis shows 2,499 of 6,765 ring-containing candidates, spanning 1,429 distinct core frameworks, absent from all training sets, while a further 294 molecules...

Accurate prediction of electron temperature ($T_{\rm e}$) is critical for non-equilibrium plasma applications ranging from hypersonic flight to plasma-assisted combustion. We recently proposed a thermodynamically consistent model for vibrational-electron heating [Phys. Fluids 37, 096141 (2025)] that enforces the convergence of $T_{\rm e}$ to the vibrational temperature ($T_{\rm v}$) at equilibrium. However, the original derivation was restricted to single-quantum transitions, limiting its validity to low-temperature regimes ($T_{\rm e} \lesssim 1.5$ eV). In this Letter, we generalize the model to include multi-quantum overtone transitions, extending its applicability to high-energy regimes. We demonstrate that previous models neglecting hot-band transitions incur a systematic heating error of $\exp(-\theta_{\rm v}/T_{\rm v})$, where $\theta_{\rm v}$ is the characteristic vibrational temperature. This error exceeds 40% when $T_{\rm v}$ is greater than $\theta_{\rm v}$, effectively preventing thermal relaxation. To correct this, we derive a formulation where the total heating rate is a summation of channel-specific cooling rates $Q_{\rm e-v}^{(m)}$, each associated with a quantum jump $m$, scaled by a thermodynamic factor $\exp(m\theta_{\rm v}/T_{\rm e}-m\theta_{\rm v}/T_{\rm v})$. This generalized model preserves thermodynamic consistency by ensuring zero net energy transfer at equilibrium.

Exponential divided differences arise in numerical linear algebra, matrix-function evaluation, and quantum Monte Carlo simulations, where they serve as kernel weights for time evolution and observable estimation. Efficient and numerically stable evaluation of high-order exponential divided differences for dynamically evolving node sets remains a significant computational challenge. We present a Chebyshev-polynomial-based algorithm that addresses this problem by combining the Chebyshev-Bessel expansion of the exponential function with a direct recurrence for Chebyshev divided differences. The method achieves a computational cost of ${\cal O}(qN)$, where $q$ is the divided-difference order and $N$ is the Chebyshev truncation length. We show that $N$ scales linearly with the spectral width through the decay of modified Bessel coefficients, while the dependence on $q$ enters only through structural polynomial constraints. We further develop an incremental update scheme for dynamic node sets that enables the insertion or removal of a single node in ${\cal O}(N)$ time when the affine mapping interval is held fixed. A full C++ reference implementation of the algorithms described in this work is publicly available.

We construct the first physics-informed neural-network (PINN) surrogates for relativistic magnetohydrodynamics (RMHD) using a hybrid PDE and data-driven workflow. Instead of training for the conservative form of the equations, we work with Jacobians or PDE characteristics directly in terms of primitive variables. We further add to the trainable system the divergence-free condition, without the need of cleaning modes. Using a novel MUON optimizer implementation, we show that a baseline PINN trained on early-time snapshots can extrapolate RMHD dynamics in one and two spatial dimensions, and that posterior residual-guided networks can systematically reduce PDE violations.

Partial differential equations (PDEs) govern a wide range of physical systems, and recent multimodal foundation models have shown promise for learning PDE solution operators across diverse equation families. However, existing multi-operator learning approaches are data-hungry and neglect physics during training. Here, we propose a physics-informed multimodal foundation model (PI-MFM) framework that directly enforces governing equations during pretraining and adaptation. PI-MFM takes symbolic representations of PDEs as the input, and automatically assembles PDE residual losses from the input expression via a vectorized derivative computation. These designs enable any PDE-encoding multimodal foundation model to be trained or adapted with unified physics-informed objectives across equation families. On a benchmark of 13 parametric one-dimensional time-dependent PDE families, PI-MFM consistently outperforms purely data-driven counterparts, especially with sparse labeled spatiotemporal points, partially observed time domains, or few labeled function pairs. Physics losses further improve robustness against noise, and simple strategies such as resampling collocation points substantially improve accuracy. We also analyze the accuracy, precision, and computational cost of automatic differentiation and finite differences for derivative computation within PI-MFM. Finally, we demonstrate zero-shot physics-informed fine-tuning to unseen PDE families: starting from a physics-informed pretrained model, adapting using only PDE residuals and initial/boundary conditions, without any labeled solution data, rapidly reduces test errors to around 1% and clearly outperforms physics-only training from scratch. These results show that PI-MFM provides a practical and scalable path toward data-efficient, transferable PDE solvers.

Direct-view LED displays are widely adopted in large-format applications due to their high luminance and reliability. However, visual comfort and accurate performance evaluation remain challenging due to the complex interaction between pixel luminance, human visual perception, and measurement artifacts. In this work, we introduce a novel 2D imaging luminance meter that replicates key optical parameters of the human eye, including entrance pupil size and angular resolution, to assess perceived pixel luminance. We report comprehensive measurements across various visual field angles and distances and establish a refined luminance metric that aligns with foveal vision standards (1/120 degree). Furthermore, a new method to quantify and mitigate stray light effects significantly improves measurement precision by reducing luminance overestimation from 7% to 2%. Our findings provide a foundation for optimizing LED display design for perceptual comfort and advancing standardization in pixel luminance evaluation.

Extreme El Niño events, such as occurred in 1997–1998, can induce severe weather on a global scale, with significant socioeconomic impacts that motivate efforts to understand them better. However, extreme El Niño events are rare, and even in a very long direct numerical simulation (DNS) occur too infrequently for robust statistical characterization. This study seeks to generate extreme El Niño event model data at a lower cost, while preserving statistical fidelity, using a rare event sampling technique, which preferentially devotes computational resources toward extreme events by generating a large, branched ensemble of interrelated trajectories through successive targeted perturbations. We specifically use the “trying-early adaptive multi-level splitting” (TEAMS) algorithm, which is well-suited for El Niño's relative timescales of predictability and event duration. We apply TEAMS to the Zebiak-Cane model, an intermediate-complexity ENSO model for which it is feasible to run a long DNS (500000 years) for validation. We compare extreme El Niño event return time estimates from TEAMS to those from the long DNS to assess TEAMS' accuracy and efficiency. We find that TEAMS accurately reproduces the return time estimates of the DNS at about one fifth the computational cost. Therefore, TEAMS is an efficient approach to study rare ENSO events that can be plausibly applied to full-complexity climate models.

The aim of this study was to investigate the impact of ambient air plasma treatment on both sunflower seed germination and early steps of seedling development under water stress. Dry seeds were exposed to a cold atmospheric plasma (CAP) generated in a dielectric barrier device where excited molecular nitrogen and ozone were detected by optical emission spectroscopy and mass spectrometry respectively. Interestingly, we explain the crucial role of gap's accuracy when treating seeds with CAP, especially to improve interaction between seeds and plasma and therefore ensure an efficient treatment. CAP significantly improved the germination rates of seeds of dormant and non-dormant sunflower genotypes under water stress, demonstrating its efficiency in alleviating seed dormancy and in improving germination under suboptimal conditions. Furthermore, greenhouse experiments demonstrated that plasma treatment also stimulated seedling development under water stress conditions. These findings highlight the potential of CAP treatment as an effective approach to promote the whole process of emergence of crop species.

Density functional theory (DFT) and machine learning potentials (MLPs) are essential for predicting and understanding materials properties, yet preparing, executing, and analyzing these simulations typically requires extensive scripting, multi-step procedures, and significant high-performance computing (HPC) expertise. These challenges hinder reproducibility and slow down discovery. Here, we introduce Masgent, an AI-assisted materials simulation agent that unifies structure manipulation, automated VASP input generation, DFT workflow construction and analysis, fast MLP-based simulations, and lightweight machine learning (ML) utilities within a single platform. Powered by large language models (LLMs), Masgent enables researchers to perform complex simulation tasks through natural-language interaction, eliminating most manual scripting and reducing setup time from hours to seconds. By standardizing protocols and integrating advanced simulation and data-driven tools, Masgent lowers the barrier to performing state-of-the-art computational methodologies, enabling faster hypothesis testing, pre-screening, and exploratory research for both new and experienced practitioners.