Browse Articles

A deep reinforcement learning account of competition and integration of visual and goal vector signals for spatial navigation

Abstract Integrating different sources of information is essential to successful spatial navigation. For instance, animals often rely on a combination of visual impressions, self-motion, olfaction, and other signals to navigate towards a goal. This is especially important when navigating in uncertain environments, where switching from one source of information to another or integrating multiple sources of information may be required to make navigation decisions. We propose a deep reinforcement learning model of the interaction of visual and goal-vector signals based on reinforcement learning and use it to study behavior and spatial representations. We show that the nature and degree of signal noise strongly influences how the signals drive behavior and spatial representations. Our model also demonstrates that the ability to navigate using each information source independently, in addition to integrating them, is crucial to successfully navigating in uncertain environments. Counterintuitively, our model shows that when one of the signals is removed, navigation may be improved if the remaining signal is reliable and sufficient to navigate. However, this improvement comes at the expense of robustness. Our modeling results demonstrate that combining redundant sources of information in biological systems is far more complex than suggested by sensor fusion in the engineering context.

Light harvesting and carrier separation in N-doped graphene clusters/Si heterojunctions for photodetection and spatial image recognition

Two-dimensional (2D) graphene photodetectors have limited light absorption, short carrier lifetimes, and a dependence on external bias, which hinder sensitivity and power efficiency. This study presents a three-dimensional (3D) heterojunction of nitrogen (N)-doped graphene clusters, featuring 82% porosity, constructed on a silicon (Si) substrate. The 3D structure enhances light–matter interaction and trapping through localized optical field confinement in concave areas. Scanning kelvin probe microscopy and conductive atomic force microscopy results reveal that N doping enhances band alignment, boosts the built-in electric field, and improves carrier transport. The device enables broadband self-powered detection from 440 to 1550 nm without bias. At 1550 nm, it has a responsivity of 24.5 A/W and specific detectivity of 1.2 × 1012 Jones (under −3 V bias and 40 mW/cm2), and a low noise power spectral density of 3 × 10−26 A2/Hz. It features a −3 dB bandwidth of 1.5 kHz and rise/fall times of 173/189 μs, maintaining stable performance for 4 months. By combining spatial photocurrent mapping with a deep learning model, it achieves over 97% recognition accuracy for various images. This work presents a strategy for graphene optoelectronic devices featuring broadband, high-sensitivity, self-powered, and intelligent sensing.

Synthesis and characterization of polycitric acid-functionalized MXene nanofluid for enhanced oil recovery

Single-sided magnetic-field generation based on transformation optics with a multilayer coil structure

We propose a transformation-optics-based method for generating a single-sided time-varying magnetic field. In this method, an inhomogeneous anisotropic medium exhibiting a spatially asymmetric response is constructed and effectively emulated by a distribution of current density, which can independently generate a single-sided time-varying magnetic field. Guided by this approach, we design a single-port-fed multilayer structure consisting of stacked variable-pitch planar spiral coils interconnected in series, eliminating the need for high-permeability materials or Halbach-array-based configurations. A six-layer prototype is fabricated and experimentally characterized at 1 MHz. The measured magnetic flux density on the working side closely matches simulations, while the field on the opposite side is suppressed to near zero, validating the feasibility of the proposed method and demonstrating its potential for active metamaterials with simplified architectures and low magnetic-field leakage, which do not rely on high-permeability magnetic materials.

Decoupling of expansive and structural growth induced by β-aminobutyric acid (BABA) is associated with metabolic reconfiguration in linseed (Linum usitatissimum L.)

High-power and efficient operation of Ho:Lu2O3 ceramic laser at 2.1  <b> <i>μ</i> </b> m spectral gain region

We report on the power scaling and high-efficiency operation of a Ho:Lu2O3 ceramic laser at 2.1 μm in-band pumped by a high-power, narrow-linewidth Tm-doped fiber laser at 1942 nm. The 0.5 at. % Ho3+-doped Ho:Lu2O3 ceramic was fabricated in-house through a vacuum sintering plus hot isostatic pressing (HIPing) process, showing a relatively low scattering loss of &amp;lt;2.1% cm−1. A record output power of 10.2 W at 2121 nm was achieved with 45 W of absorbed pump power, corresponding to a slope efficiency of 27.5%. The beam quality factors (M2) at the maximum output were measured to be 1.34 and 1.28 in the x- and y-directions, respectively. The laser power and efficiency were enhanced by two orders of magnitude and nearly 30-fold compared to the previous results, which confirms the significant power scaling potential of in-band pumped Ho:Lu2O3 ceramic lasers for hundred-watt-level operation upon improved quality of the ceramic with a more homogeneous and denser microstructure.

Structural brain alterations in frailty among end-stage renal disease patients

Phthalocyanine-based molecular memristor consisting of a tin-bridged heterojunction

Molecular memristive devices are attractive for nanoscale memory and logic because their resistance can be modulated and retained through electrically addressable molecular states. Herein, we design a phthalocyanine-based molecular memristor constructed with a Cu/SnPc/graphene heterojunction. The SnPc molecule possesses two stable structural conformations, namely Sn-up and Sn-down orientations, which are reversible and switchable via pulsed current stimulation. In the Sn-up conformation, the electronic properties of the heterojunction are primarily governed by interlayer van der Waals interactions. In contrast, the Sn ion in the Sn-down conformation serves as a conductive bridge between the Cu electrode and phthalocyanine layer, which markedly diminishes the interlayer tunneling barrier. Cross-plane charge transport calculations demonstrate that the Sn-down configuration delivers a much higher current density compared with the Sn-up counterpart. These results suggest that a SnPc-based vertical heterojunction may provide a useful molecular building block for future memristive devices, while device-level neuromorphic or in-memory computing functions remain to be demonstrated.

Lattice-based post-quantum image encryption with NDFA-driven spin-matrix transformation

Abstract The increasing reliance on visual data transmission across distributed and cloud-based environments requires image encryption frameworks that provide both strong cryptographic guarantees and practical computational efficiency. Conventional image encryption schemes frequently rely on heuristic transformations whose security is primarily evaluated through statistical analyses, often lacking formal resistance against emerging post-quantum threats. A layered post-quantum image encryption framework is presented, integrating lattice-based key establishment, deterministic image scrambling, and authenticated encryption within a unified architecture. A shared secret is established using the sntrup761 key encapsulation mechanism, providing IND-CCA security under lattice-based assumptions. The derived session key is expanded using HKDF into independent keys for permutation scheduling and encryption, enabling separation between cryptographic security and image transformation layers. A nondeterministic finite-state scheduling mechanism drives spin-matrix-based scrambling to enhance spatial diffusion, while authenticated encryption with associated data ensures confidentiality and integrity through strict verification prior to image recovery. Security analysis shows that post-quantum security comes from the lattice-based key encapsulation mechanism rather than the image-transformation layer, and demonstrates robustness against adaptive adversarial models, while computational evaluation indicates linear scalability with image size. Statistical and robustness analyses further confirm strong diffusion properties and stable scrambling-layer behavior under noise and occlusion conditions. The proposed framework bridges standardized post-quantum cryptography with lightweight image-oriented transformations, providing a structured and practical approach for secure visual data transmission.

Tissue Heterogeneity‐Driven Parallel Acquisition for High‐Coverage MS/MS Imaging of Lipids

ABSTRACT Elucidating tissue architecture and function necessitates lipid analysis that is both spatially comprehensive and achieved at high resolution while maintaining spatial context. Conventional mass spectrometry imaging (MSI) techniques typically perform MS/MS analysis sequentially for individual lipids, creating a significant trade‐off between spatial resolution and analytical depth. To circumvent this limitation, we introduce a deconvolution‐based structure‐specific spatial lipidomics (DeconS 2 L) method for coupling with per‐pixel and broadband MS/MS sampling. DeconS 2 L capitalizes on the spatial and compositional heterogeneity inherent to all biological tissues for large‐scale lipid annotation and lipidome imaging. To facilitate rapid imaging throughput and improved sample usage, lipids per pixel are co‐fragmented to generate convolved MS/MS spectra. DeconS 2 L analysis of ∼4000 pixels in a mouse cerebellum tissue yielded 100 annotated lipids and allowed multiplexed MS/MS imaging of the tissue lipidome. Furthermore, applying DeconS 2 L to human hepatocellular carcinoma (HCC) tissues effectively resolved isobaric interferences that obscured tumor margins in conventional imaging. It successfully revealed the tumor‐specific enrichment of odd‐chain lipid PC 33:1 and distinct spatial heterogeneity in triglyceride saturation, correlating with metabolic reprogramming in HCC. DeconS 2 L represents a versatile methodology that effectively integrates molecular annotation with spatial lipidomics, demonstrating significant potential for in‐depth biomarker discovery in clinical pathology.

Ge content compromise to the performance optimization of GexAsySez OTS materials for selector applications

Germanium is a critical dopant in ovonic threshold switching (OTS) selectors, yet how its concentration influences the performance and failure mechanisms of Ge–As–Se-based OTS materials remains poorly understood. In this study, Ge–As–Se OTS devices with varying Ge concentrations are fabricated and characterized. Increasing the Ge content is found to reduce the threshold voltage, albeit at the expense of prolonged switching time and increased voltage drift. The leakage current initially decreases with higher Ge content but subsequently rises due to the formation of conductive Ge–Ge bonds. Microstructural analysis of the devices after 109 cycles reveals that void formation and interlayer diffusion are the key failure mechanisms. An optimal Ge content window of 28%–44% is identified, which balances low leakage current, high endurance, and acceptable switching speed. These findings elucidate the role of Ge doping and provide a theoretical foundation for the design of novel high-performance OTS materials.

A cross-sectional analysis of the associations between dry eye disease and depression

Abstract Dry Eye Disease (DED) is a common condition worldwide that is often associated with mental health conditions such as depression. The study evaluates the associations between dry eye disease and depression in the Hong Kong population, based on a large population-based cohort. A total of 614 participants were enrolled, and 1228 eyes were evaluated. The Ocular Surface Disease Index (OSDI) questionnaire was administered to assess subjective dry eye symptoms. Lipid layer thickness (LLT) was measured using the IDRA ocular surface analyzer, while tear meniscus height (TMH), first non-invasive tear break-up time (NITBUT), average NITBUT, Meiboscore, and redness were assessed with the Keratograph 5 M. The osmolarity was measured by the TearLab. Dry eye and non-dry eye were categorized based on the Tear Film and Ocular Surface Society Dry Eye Workshop II (TFOS DEWS II) diagnostic criteria. Depression classification was based on participant self-report, together with concurrent antidepressant use. For statistical analysis, the Kolmogorov–Smirnov test was used to examine normality. Fisher’s exact test was applied to categorical variables, and the Mann–Whitney U test was used for continuous variables in two-group comparisons. A Generalized Estimating Equation (GEE) model was constructed to explore associations between dry eye parameters and depression. Among the 614 participants recruited from the general Hong Kong population, 596 were classified as not having depression (1192 eyes; median age, 46 years), and 18 were classified as having depression (36 eyes; median age, 51 years). The OSDI score was higher among participants with depression than among participants without depression [20 (9–36) vs. 14 (6–23), P  = 0.004]. However, other parameters showed no significant differences between the two groups. In the multivariate model, the OSDI score was significantly associated with depression, showing an odds ratio (OR) of 1.03 (95% CI: 1.01–1.06; P  = 0.015). Furthermore, the severe subjective dry eye (OSDI &gt; 32) was significantly associated with depression, with an OR of 3.00 (95% CI: 1.14–7.86; P  = 0.026). Severe DED symptom score was significantly associated with depression in the population of Hong Kong, highlighting the clinical relevance of the relationship between DED and mental health.

Regulation of vacancy formation in h-BN via carbon doping toward improved resistive switching

Hexagonal boron nitride (h-BN) exhibits promising potential in resistive switching (RS) applications, yet the performance variability remains a major obstacle, primarily due to the uncontrollable nature of intrinsic defects. Carbon doping can offer a potential means to regulate defect types and their spatial distribution. However, the microscopic mechanisms by which carbon dopants regulate defects in h-BN remain unclear. In this work, the modulation of the RS behavior in h-BN-based resistive random-access memory (RRAM) by carbon dopants is studied using first-principles calculations. The results indicate that carbon substituted for boron (CB) not only promotes the formation of boron vacancies (VB), but its interaction with vacancies further facilitates vacancy clustering around the dopant. Notably, interlayer interactions depend on defect type, with VB in adjacent layers exhibiting mutual repulsion, whereas nitrogen vacancies (VN) attract and stabilize VB. Climbing image nudged elastic band calculations demonstrate that RS does not rely on VB migration but is mediated by boron–nitrogen divacancy (VBN) enabled conductive bridge formation (−1q/−2q charge states) and breaking (0q/+1q charge states). Nonequilibrium Green's function transport calculations verify a pronounced enhancement of the current upon conductive bridge formation. This work provides fundamental theoretical insights for defect engineering in the development of h-BN-based RRAM.

Normal inflammatory marker levels do not exclude metabolic disturbances in young male team sport athletes

Deep learning-based prediction of highly efficient optical probes for two-photon fluorescence imaging and photodynamic therapy of tumor

This study reports an innovative intelligent screening strategy based on a directed message-passing neural network (DMPNN) to predict two key physical properties of corrole molecule probes involving the Q_band absorption and singlet–triplet intersystem crossing rate (kISC). Feature engineering and heterogeneous model fusion are performed, including the concatenation of molecular fingerprints, integration with additional algorithms, incorporation of physical constraints into molecular descriptors, and introduction of attention mechanisms. The DMPNN model alone achieves excellent predictive performance for Q_band absorption. The combination of the DMPNN with specific Morgan fingerprints yields an R2 of up to 0.926 for kISC prediction. Interpretable analyses of node- and bond-level features reveal the primary structural information related to the metal center and its N4 coordination environment, the conjugated macrocyclic framework, and the presence of lone pair electrons. The model screens gallium-centered corrole derivatives with peripheral pentafluorophenyl substitutions to identify promising candidates. According to two-photon fluorescence lifetime imaging, the probes enable effective discrimination between normal and cancer cells. Under red-light irradiation, the probe efficiently generates reactive oxygen species and effectively eliminates cancer cells both in vitro and in vivo. The intelligent screening strategy provides a new paradigm for developing high-performance optical theranostic probes with strong potential for clinical translation.

Sickle cell disease premarital screening utilization and associated factors among youth corps members in Osun State, Nigeria

Diverse ratio of electronic and internal-strain piezoelectricity in half-Heusler semiconductor for the design of multifunctional piezoelectric devices

Half-Heusler (HH) semiconductors are promising for piezoelectric applications owing to their non-centrosymmetric structures and thermal stability, but the corresponding piezoelectric mechanisms and atomic-scale tunability remain elusive. This work addresses this gap by investigating non-centrosymmetric HH compounds via atomic substitution. Using first-principles calculations and lattice dynamics analysis, we quantify electronic and internal-strain piezoelectric contributions for 47 HH compounds with the F4¯3m space group, which exhibit remarkable diversity in the ratio of electronic and internal-strain piezoelectric contributions. Critically, atom substitution enables a smooth transition between electronic- and ionic-dominated piezoelectricity, a unique advantage over traditional single-mechanism piezoelectric materials. This tunability provides a unified platform, where the ratio of the two piezoelectric components becomes a key design parameter, bridging atomic-scale structure manipulation and macroscale functionality. Moreover, shear strain engineering demonstrates that piezoelectric performance of most HH materials decreases under strain, yet exceptions (e.g., SbVRu) highlight the potential of mode modulation for performance enhancement. This work not only uncovers the physical origin of HH piezoelectricity but also establishes a framework for strain-controlled multifunctional devices.

Modeling ultrafast laser excitation of fused silica with a hybrid Lorentz–Drude dielectric response and density-dependent two-temperature model

Abstract Ultrashort laser excitation of dielectric materials is governed by a complex interplay between nonlinear photoexcitation, transient optical response, and energy transfer to the lattice. A consistent physical description of these processes across different pulse durations remains challenging, as purely free-electron-based optical models and classical thermal approaches do not adequately capture the strongly dynamic and non-equilibrium nature of the excited electron system. In this work, a coupled modeling framework is developed for amorphous fused silica (SiO 2 ) that directly links the transient optical response to a density-dependent two-temperature description of the subsequent thermal evolution. The optical properties are described using a hybrid Lorentz–Drude formalism, enabling a consistent representation of bound, localized, and free electronic states during excitation. The resulting energy deposition is coupled to a modified thermal model incorporating electron-density-dependent material parameters derived from first-principles calculations. The model is evaluated by comparison with experimentally measured ablation geometries and single-pulse ablation thresholds for pulse durations in the femtosecond and picosecond range. The threshold analysis indicates that experimentally measurable material removal can occur below the model-internal phase-explosion criterion, particularly for longer pulse durations. In the femtosecond regime, both ablation depth and diameter are reproduced within a limited deviation range. In the picosecond regime, the ablation depth remains consistent with experimental observations, while systematic deviations in the lateral extent become apparent. At longer pulse durations, the experimentally observed structures are increasingly influenced by melt-mediated material redistribution, which is not captured by the present model and leads to pronounced deviations in the ablation geometry. These results demonstrate that the proposed framework provides an effective physical description of ultrafast laser–matter interaction in non-equilibrium and transition regimes, while also defining its limitations in thermally dominated regimes where hydrodynamic effects become significant.

Late‐Stage Functionalization of Peptides on the Solid Phase

ABSTRACT Chemical modification is a means of tailoring the structure and function of peptides. This is of particular interest in the field of peptide drug development. Methods that enable straightforward modification of peptides with non‐canonical residues are therefore in high demand. On‐resin late‐stage functionalization is a powerful, cost‐ and time‐efficient tool for modifying peptides. It also allows for high‐throughput experimentation, making it well suited to streamlining the development of peptide therapeutics. In this minireview, we provide an overview of the chemical approaches for on‐resin late‐stage peptide functionalization , most of which have been developed over the past decade. These reactions range from nucleophilic substitution and carbonyl chemistry to metal catalysis and photocatalysis. Many of the presented approaches have been shown to be compatible with short and long peptides, and to modulate their bioactivity. Impressive examples of peptide library synthesis in high‐throughput formats demonstrate the enormous potential of late‐stage peptide functionalization for the development of peptide therapeutics in the years to come.

Regrown n++-GaN ohmic contacts to high-Al content AlxGa1−xN (x &amp;gt; 0.6) digital alloy channels using AlN/GaN short-period superlattice

In this paper, we report a novel ohmic contact formation scheme for extreme bandgap AlxGa1−xN (x &amp;gt; 0.6) channel high electron mobility transistors with undoped barrier layers. Our approach consists of using a new low-temperature pulsed metal-organic chemical vapor deposition doping scheme for the n++-GaN regrown contacts and an AlxGa1−xN digital alloy (DA) channel layer comprising short-period superlattices of AlN and GaN. Pulsed growth and doping yield a sheet resistivity that is a factor of 3–5 lower than that of conventional doped n++-GaN layers grown under identical conditions. Moreover, the regrown n++-GaN layer has no hetero-barrier with the GaN layers of the DA channel, which leads to linear ohmic contacts and a record low contact resistance Rc ∼ 6.5 Ω-mm to the Al0.62Ga0.38N DA channel layer of a HEMT with an AlN barrier layer.