Browse Articles

Learning continuum-level closures for control of interacting active particles

Active matter swarms—collectives of self-propelled particles that can self-assemble, ferry microscopic cargo, or endow materials with dynamic properties—remain hard to steer. In crowded systems, tracking or controlling individual agents becomes challenging, so strategies must operate on macroscopic fields like particle density. Yet predicting how density evolves is difficult because of inter-agent interactions. For model-based feedback control methods—such as Model Predictive Control (MPC)—fast, accurate, and differentiable models are crucial. Detailed agent-based simulations are too slow, necessitating coarse-grained continuum models. However, constructing accurate closures—approximations that express the effects of unresolved microscopic states (e.g., agent positions) on continuum dynamics in terms of the modeled continuum fields (e.g., density)—is challenging for active matter swarms. We present a learning-for-control framework that learns continuum closures from agent simulations, demonstrated with active Brownian particles under a controllable external field. Our Universal Differential Equation (UDE) framework represents the continuum as an advection–diffusion equation. A neural operator learns the advection term, providing closure relations for microscopic effects such as self-propulsion, interactions, and external-field responses. This UDE approach, embedding universal function approximators in differential equations, ensures adherence to physical laws (e.g., conservation) while learning complex dynamics directly from data. We embed this learned continuum model into MPC for precise agent-simulation control. We demonstrate this framework’s capabilities by dynamically exchanging particle densities between two groups and by simultaneously controlling particle density and mean flux to follow a prescribed sinusoidal profile. These results highlight the framework’s potential to control complex active-matter dynamics, foundational for programmable materials.

Rational Lithium Salt Selection Principle for Designing High-Entropy Electrolytes toward High-Performance Lithium Metal Batteries

6G conditioned spatiotemporal graph neural networks for real time traffic flow prediction

3D N and O co-doping hierarchical porous carbon with ultrahigh surface area as cathode material for high-performance zinc-ion hybrid capacitor

Aqueous zinc-ion hybrid capacitors (ZIHCs) have garnered significant attention due to their cost-effectiveness, safety, and high theoretical capacity. However, the use of carbon-based materials in ZIHCs faces challenges such as electrolyte ion and pore size mismatching, inadequate infiltration between electrolyte and electrode, and limited surface active sites and defects, all of which impede the device’s ability to achieve optimal energy density and electrochemical performance. To address these issues, a three-dimensional N and O co-doping hierarchical porous activated carbon (3DNOHC) with an ultrahigh specific surface area of 3477.69 m2 g−1 is synthesized through the direct calcination of nitrilotriacetic acid sodium salt precursor followed by a chemical activation process. Density functional theory calculations demonstrate that the N/O co-doping of activated carbon significantly enhances the ion adsorption/desorption capabilities on the surface of the materials, thereby improving their kinetic and electrochemical properties. The structural changes in zinc metal anodes and 3DNOHC cathodes during charging/discharging are investigated using ex situ XRD and ex situ Raman tests. Due to its abundant porous structure and active sites, the 3DNOHC-6 sample exhibits rapid ion transport and impressive electrochemical performance in ZIHCs. In particular, the 3DNOHC-6//Zn device demonstrates a high reversible capacity of 171/102 mA h g−1 at 0.2/10 A g−1 and an outstanding energy density of 137 Wh kg−1@160 W kg−1. Moreover, it exhibits excellent capacity retention of 80% at 5 A g−1 after 45 000 cycles. This study serves as a valuable reference for the development of activated carbon cathode materials for aqueous hybrid capacitors aiming for high energy/power density.

Extending the Distance Range in Double Electron–Electron Resonance Measurements of Transition Metal Clusters

Assessment of electrocardiography interpretation competency of Ethiopian medical interns: a multi-site study

How your brain chemistry rewards hard work

Encapsulation of fragmented cargo by virus coat proteins

The co-assembly of multiple nanoparticles (“fragmented cargo”) and virus coat proteins is very sensitive both to the size of the nanocolloids and the stoichiometric ratio of nanoparticles to coat proteins, as recent experiments demonstrate. In addition, in a head-to-head competition, larger nanoparticles turn out to be preferentially encapsulated. In order to rationalize these findings, we investigate a simple mass-action model in which we allow for the co-existence of free nanoparticles and coat proteins, complexes consisting of a nanocolloid bound to a coat protein, and fully formed capsids consisting of a fixed number of coat proteins and a variable number of nanoparticles. In qualitative agreement with the experimental findings, we find (i) that there is a relatively narrow range of concentrations of nanocolloids that allows for the formation of appreciable numbers of partially filled capsids, and (ii) that the number of nanocolloids adsorbed on the inner wall of the capsid shell is typically well below the maximum number that fits on the wall facing the lumen. We attribute this to the impact of entropy that offsets the increase in binding free energy gain, which for smaller particles tends to be weaker.

IR-VUV Photoionization Spectra of Hydrated BaOH Reveal Base Dissociation in Growing Water Clusters

Metacognitive ability is associated with reduced emotion suppression

Ab <i>initio</i> characterization of C2H4N2 isomers: Structures, electronic energies, spectroscopic parameters, and formation pathways

This work presents a comprehensive theoretical investigation of key isomers of C2H4N2 using state-of-the-art quantum chemical methods. The objective is to characterize their molecular structures, spectroscopic constants, and electronic energies and to elucidate plausible formation and destruction pathways, providing data critical for astrochemical and atmospheric detection. High-accuracy ab initio methods were employed, notably CCSD(T)-F12/cc-pVTZ-F12 for optimized geometries. Additional calculations were performed at the CCSD(T)/aug-cc-pVTZ, CCSD(T)/cc-pVTZ, MP2/aug-cc-pVTZ, and CIS levels. Intrinsic reaction coordinate calculations were performed at the B3LYP/6-31G(d,p) level to explore reaction pathways. The Zero-Point Energy (ZPE)-corrections were determined for all the isomers considered. Six low-energy C2H4N2 isomers were identified, all within 1 eV of the global minimum. Among them, methylcyanamide (MCA) exhibits the lowest relative energy (∼0.2 eV) and a significant electric dipole moment of 5.00 D, making it a strong candidate for detection in gas-phase environments. The rotational constants for MCA, computed at the level of CCSD(T)-F12/cc-pVTZ-F12, are Ae = 34 932.44 MHz, Be = 4995.31 MHz, and Ce = 4520.30 MHz. The V3 torsional barrier was found to be 631.19 cm−1. Centrifugal distortion constants were computed up to sextic order for all isomers. Formation pathways for MCA—such as CH3N + HCN → CH3NHCN—and related isomers were characterized. The combination of large dipole moments and distinct rotational signatures supports the detectability of MCA and related C2H4N2 isomers via radioastronomy, IR, and MW spectroscopy. Isomerization and reaction pathways involving radical-neutral and neutral-neutral processes were found to be key to their formation in gas-phase environments. These results offer a robust foundation for future observational and modeling efforts.

Synthesis of Phenyl-Substituted Poly(3-Hydroxybutyrates) with High and Tunable Glass Transition Temperatures via Sequential Catalytic Transformations

ADAT novel time-series-aware adaptive transformer architecture for sign language translation

Abstract Current sign language machine translation systems rely on recognizing hand movements, facial expressions, and body postures, and natural language processing, to convert signs into text. While recent approaches use Transformer architectures to model long-range dependencies via positional encoding, they lack accuracy in recognizing fine-grained, short-range temporal dependencies between gestures captured at high frame rates. Moreover, their quadratic attention complexity leads to inefficient training. To mitigate these issues, we introduce ADAT, an Adaptive Transformer architecture that combines convolutional feature extraction, log-sparse self-attention, and an adaptive gating mechanism to efficiently model both short- and long-range temporal dependencies in sign language sequences. We evaluate ADAT on three datasets: the benchmark RWTH-PHOENIX-Weather-2014 (PHOENIX14T), the ISL-CSLTR, and the newly introduced MedASL, a medical-domain American Sign Language corpus. In sign-to-gloss-to-text translation, ADAT outperforms the state-of-the-art baselines, improving BLEU-4 by at least 0.1% and reducing training time by an average of 21% across datasets. In sign-to-text translation, ADAT consistently surpasses transformer-based encoder-decoder baselines, achieving a minimum of 0.5% gains in BLEU-4 and an average training speedup of 21.8% across datasets. Compared to the encoder-only and decoder-only baselines in sign-to-text, ADAT is at least 0.7% more accurate, despite being up to 12.1% slower due to its dual-stream structure.

Nuclear–electronic orbital second-order coupled cluster for excited states

Excited-state methods within the nuclear–electronic orbital (NEO) framework have the potential to capture vibrational, electronic, and vibronic transitions in a single calculation. In the NEO approach, specified nuclei, typically protons, are treated quantum mechanically at the same level of theory as the electrons. Affordable excited-state NEO methods, such as time-dependent density functional theory, are limited to capturing the subset of excitations with single-excitation character, whereas existing methods that capture the full spectrum are limited in applicability due to their high computational cost. Herein, we introduce the excited-state variant of NEO coupled cluster with approximate second-order doubles (NEO-CC2) and its scaled-opposite-spin variant with electron–proton correlation scaling (NEO-SOS′-CC2). We benchmark this method for positronium hydride, where the electrons and positron are treated quantum mechanically, and find that NEO-CC2 deviates from exact results, but NEO-SOS′-CC2 can achieve near-quantitative accuracy by increasing the electron–positron correlation. Benchmarking NEO-CC2 and NEO-SOS′-CC2 on four different triatomic molecules with a quantum proton, we find that NEO-CC2 captures qualitatively correct vibrational features such as overtones and combination bands, as well as mixed electron–proton double excitations. Electron–proton correlation scaling that increases the excited-state correlation relative to the ground-state correlation improves the accuracy across all the molecular systems tested. Quantitative accuracy is not achieved due to a combination of finite basis set effects and incomplete description of excited-state electron–proton correlation. Nevertheless, NEO-SOS′-CC2 can describe single and mixed protonic and electronic excitations with accuracy approaching that of much more computationally intensive methods.

Large-Area Supramolecular Crystalline Thin Films of Polyoxometalates with Controlled 1-nm Pores Enabling Ultra-Selective Molecular Transport

Fabrication of anisotropic magnetic helical microswimmers utilizing Spirulina platensis templates and their integration with Janus PCL/Chitosan nanoparticles

libMobility: A Python library for hydrodynamics at the Smoluchowski level

Effective hydrodynamic modeling is crucial for accurately predicting fluid–particle interactions in diverse fields such as biophysics and materials science. Developing and implementing hydrodynamic algorithms is challenging due to the complexity of fluid dynamics, necessitating efficient management of large-scale computations and sophisticated boundary conditions. Furthermore, adapting these algorithms for use on massively parallel architectures such as GPUs adds an additional layer of complexity. This paper presents the libMobility software library, which offers a suite of CUDA-enabled solvers for simulating hydrodynamic interactions in particulate systems at the Rotne–Prager–Yamakawa level. The library facilitates precise simulations of particle displacements influenced by external forces and torques, including both the deterministic and stochastic components. Notable features of libMobility include its ability to handle linear and angular displacements, thermal fluctuations, and various domain geometries effectively. With an interface in Python, libMobility provides comprehensive tools for researchers in computational fluid dynamics and related fields to simulate particle mobility efficiently. This article details the technical architecture, functionality, and wide-ranging applications of libMobility. libMobility is available at https://github.com/stochasticHydroTools/libMobility.

Light-Driven Intraoctahedral Halide Isomerization in Two-Dimensional Mixed Halide Perovskites

Research on abnormal pressure of dark shale in the Tiemulike formation of the Yining Sag, Ili basin

Abstract The study of abnormal pressure in the dark shale of the Tiemulike Formation in the Yining Sag of the Ili Basin holds significant importance for its oil and gas exploration. Based on the Acoustic time difference data of shale, the overexposure development interval in the Tiemulike Formation was of the Yining Sag determined. The equilibrium depth method had been adopted to quantitatively calculate the excess pressure and pressure coefficient of dark shale in the Tiemulike Formation for the first time. The distribution map of abnormal excess pressure in single well profiles and well-tie comparison profile had been drawn, and the abnormal excess pressure characteristics of the shale had been fully Analyzed. The results show that the pressure coefficient of the dark shale in Tiemulike Formation is 1.21–1.95, belonging to the weak-high pressure to Ultra-high pressure layers. The excess pressure gradually increased from the top to the deeper layers in the Tiemulike Formation, with three abnormal pressure significantly increased layers. The abnormal excess pressure in Tiemulike Formation is about 5–10 MPa higher than that in the overlying Upper Permian Basiergan Formation, indicating high abnormal pressure characteristics. The horizontal distribution characteristics of abnormal excess pressure in Tiemulike Formation show that the abnormal excess pressure is mostly above 10 MPa, with a highest reaching 29.4 MPa. The Higher abnormal excess pressure has good continuity in the horizontal direction, several high abnormal fluid pressure compartments were formed in the central low-laying area of Yining Sag. The formation of abnormal excess pressure in the shale of the Tiemulike Formation in the Yining Sag was mainly controlled by the hydrocarbon generation expansion of dark shale, as well as the influence of clay mineral diagenesis.

Ground and excited state gradients with end-to-end differentiable semiempirical quantum chemistry

Accurate and efficient gradients of molecular energy with respect to nuclear degrees of freedom are essential for geometry optimization and molecular dynamics, including simulations that go beyond the Born–Oppenheimer regime. A common approach involves deriving analytical formulas for new electronic structure methods, which is often conceptually difficult and requires tedious coding. Here, we implement analytical, semi-numerical, and automatic differentiation (AD)-based gradient pathways for semiempirical Hamiltonian models in the PYSEQM software package, leveraging both graphics processing unit (GPU) and central processing unit (CPU) architectures. We further extend these capabilities to excited states calculated using the configuration interaction singles and time-dependent Hartree–Fock ansätze. We benchmark wall time, peak memory usage, and accuracy across three molecular families of varying chemical complexity, including systems of up to a thousand atoms. For ground-state simulations, analytical and AD gradients achieve near-identical GPU runtimes, while semi-numerical gradients are slower on GPU but remain competitive on CPU. For excited states, both analytical and custom AD approaches using implicit differentiation show similar performance and low memory requirements, whereas gradients with full AD are memory-limited. AD gradients match analytical ones in accuracy across all tested systems, aided by a quaternion-based diatomic frame rotation for two-center quantities that ensures smooth energy surfaces. Overall, automatic differentiation emerges as a practical alternative to analytical gradients in semiempirical quantum chemistry, offering high accuracy while allowing seamless integration in AI-driven workflows and popular packages, such as PyTorch and JAX. Our results provide actionable guidance for selecting optimal gradient strategies in large-scale ground- and excited-state molecular dynamics simulations.