Sparsity and memory constraints interact with training sequence to bias learning of associative maps.

This paper introduces a hybrid computational framework that integrates the boundary element method (BEM) with physics-informed neural networks (PINNs) to address inhomogeneous elasticity problems in thin-walled structures. In the proposed framework, PINNs are employed to approximate the particular solution corresponding to the inhomogeneous terms, and the original inhomogeneous problem is reformulated into an equivalent homogeneous problem, thereby retaining the boundary-only discretization advantage and avoiding the need for costly domain meshing. Furthermore, a nonlinear sinh transformation is incorporated to regularize nearly singular integrals by mapping them into a transformed coordinate system, thereby effectively smoothing the integrand. The synergy of BEM, PINNs, and the sinh transformation results in an efficient and highly accurate computational framework for analyzing elasticity in thin-walled structures.

Patient-specific fluid-structure interaction modeling of cerebral aneurysm: influence of wall compliance, tissue prestress, and blood rheology.

Real-time route optimization in smart cities via Bidirectional A* algorithm

Different outcomes of multicomponent reactions between 4-fluorobenzoic acid and 2-amino-5-nitro/chloro pyridines: Insights through structural, computational and energy framework analysis

The increasing sophistication and frequency of cyber threats have rendered conventional protection strategies inadequate. Artificial Intelligence (AI) is becoming central to modern cybersecurity, strengthening capabilities in vulnerability assessment, malware detection, phishing prevention, intrusion detection, and deception technologies. Simultaneously, quantum computing introduces both challenges to classical cryptography and opportunities for new forms of quantum-enhanced defenses. This review integrates advances in AI, quantum methods, and ethical governance to provide an integrated perspective on the future of secure digital systems. It evaluates state-of-the-art AI models, including explainable frameworks and quantum-inspired approaches, such as Quantum Convolutional Neural Networks and Quantum Support Vector Machines, along with recent progress in post-quantum cryptography. Ethical concerns, particularly bias, transparency, privacy, and accountability, are examined as essential foundations for trustworthy cybersecurity design in system-on-chip and embedded AI environments. In addition to technical developments, this study considers regulatory frameworks, governance structures, and societal expectations, highlighting the need for responsible and adaptive approaches. A comparative SWOT analysis outlines the strengths, limitations, and areas for cross-domain integration. Finally, a roadmap of future research directions is presented, aligning AI-driven defenses, quantum resilience, and ethical safeguards into flexible and reliable cybersecurity architectures. By linking the technological, ethical, and policy dimensions, this review offers a consolidated foundation to guide the evolution of cybersecurity in a globally connected era.

This study investigates the thermo-fluidic dynamics of a magnetised hybrid Ag-TiO₂/EG-water nanofluid over a bi-directionally stretching/shrinking surface, a configuration relevant to advanced thermal management in biomedical, aerospace, and electronics cooling. To address the limitations of traditional numerical approaches in multi-parameter optimization, novel Artificial Neural Network (ANN) optimized with the Levenberg–Marquardt algorithm (LMA) is developed. The model incorporates realistic physical effects, including temperature-dependent viscosity and thermal conductivity, surface suction, and Joule heating. A high-fidelity dataset was generated by solving the transformed governing equations using MATLAB's bvp4c solver, covering the parameter ranges: magnetic parameter (1 ≤ M ≤ 5), mixed convection (−1 ≤ λ ≤ 6), variable viscosity (0.1 ≤ a ≤ 1.5), thermal conductivity (0.1 ≤ b ≤ 0.5), stretching ratio(0.1 ≤ ε ≤1), suction parameter (−1 ≤ S ≤ 1), and nanoparticle volume fraction (0.01 ≤ ϕ ≤ 0.1). The velocity and temperature data, varying with M , ε , λ , a , and b , were divided into 70% training, 15% validation, and 15% testing sets for ANN-LMA modelling. The framework achieved absolute 10 −3 –10 −9 and MSE between 10 −10 –10 −7 , demonstrating high predictive accuracy. Results reveal that the magnetic field enhances vertical velocity in shrinking flows but reduces it in stretching flows, while horizontal velocity is suppressed in both cases. Temperature rises with magnetic and mixed convection effects, and variable conductivity causes hybrid nanofluids to exhibit up to 45% higher thermal elevation than mono nanofluids. Notably, a 10% Ag + TiO₂ mixture enhances heat transfer by 45%, compared to 18.6% for 10% Ag alone. The novelty of this work lies in its integrated AI-driven framework that accurately captures coupled multiphysics interactions, providing a rapid and reliable predictive tool for the design of advanced thermal-MHD systems. • A high-fidelity computational framework is developed for MHD hybrid nanofluid flow over a bi-directional deformable surface. • Variable viscosity, thermal conductivity, Joule heating, suction, and mixed convection effects are incorporated for realistic modelling. • A data-assisted ANN–LMA model accurately predicts thermo-fluidic behaviour with absolute errors as low as 10 −9 . • Hybrid Ag–TiO₂ nanofluid exhibits significantly enhanced thermal performance compared to conventional nanofluids. • Heat transfer is improved by 18.6% with Ag nanoparticles and by 45% using Ag–TiO₂ hybrid nanoparticles.

Physics-informed neural networks enable patient-specific tumor microenvironment modeling: Identifying parameters in combined immune therapy with integer- and fractional-order dynamics

An integrated experimental–computational framework for optimizing SLA 3D printing parameters via machine learning and genetic algorithms

Neuronal systems often preserve their characteristic functions and signalling patterns, also referred to as regimes , despite parametric uncertainties and variations. For neural models having uncertain parameters with a known probability distribution, probabilistic robustness analysis (PRA) allows us to understand and quantify under which uncertainty conditions a regime is preserved in expectation. We introduce a new computational framework for the efficient and systematic PRA of dynamical systems in neuroscience and we show its efficacy in analysing well-known neural models that exhibit multiple dynamical regimes: the Hindmarsh–Rose model for single neurons and the Jansen–Rit model for cortical columns. Given a model subject to parametric uncertainty, we employ generalised polynomial chaos to derive mean neural activity signals, which are then used to assess the amount of parametric uncertainty that the system can withstand while preserving the current regime, thereby quantifying the regime’s robustness to such uncertainty. To assess persistence of regimes, we propose new metrics, which we apply to recurrence plots obtained from the mean neural activity signals. The overall result is a novel, general computational methodology that combines recurrence plot analysis and systematic persistence analysis to assess how much the uncertain model parameters can vary, with respect to their nominal value, while preserving the nominal regimes in expectation. We summarise the PRA results through probabilistic regime preservation (PRP) plots, which capture the effect of parametric uncertainties on the robustness of dynamical regimes in the considered models. • We analyse probabilistic robustness of nonlinear systems with parametric uncertainty. • We quantify the preservation of dynamical regimes in expectation. • We efficiently compute mean system output via generalised polynomial chaos surrogate model. • We construct recurrence plots and detect loss of regime preservation via blob counts. • We assess the probabilistic robustness of regimes of key models in neuroscience

This work presents an efficient computational framework based on the J-A-φ formulation for the numerical modeling of high-temperature superconducting (HTS) cable-in-conduit conductors (CICCs). The formulation separates the magnetic vector and scalar potentials across conducting and non-conducting domains, significantly reducing computational cost without compromising accuracy. Three cable geometries configurations with varying levels of complexity and tape width reduction were analyzed under magnetization cycles to compute AC losses and evaluate computational performance. Two modeling strategies—thin strip and homogenized—were implemented using the J-A φ formulation and validated against a full T-A formulation taken as reference. Results show excellent agreement between formulations, with relative error coefficients R2 exceeding 0.99 in all cases, and computation time reductions reaching up to 57%. The critical current anisotropy of the HTS tapes was accurately captured using an empirical angular-dependent Ic (B,θ) model. The proposed method ology demonstrates high potential for accelerating the simulation of large-scale superconducting cable systems, especially in applications involving fusion magnets and high-field devices.

The design of solid rocket motors (SRMs) involves complex trade-offs between performance, structural integrity, and safety. This study presents an open-source computational framework that couples the openMotor simulation environment with a genetic algorithm (GA) to automate SRM geometry optimization. The framework enables exploration of multidimensional design spaces defined by user-specified constraints, targeting improved total impulse and thrust-time characteristics while maintaining structural safety margins. The algorithm evaluates simulation outputs using a weighted normalized penalty function. To validate the optimization results, two SRM prototypes—a baseline and an optimized design—were fabricated and tested on a horizontal static test stand. The optimized motor achieved an experimentally measured specific impulse corresponding to 89% of the simulated prediction. The results demonstrate that coupling open-source simulation with heuristic optimization produces realistic designs and reduces manual tuning. This framework establishes a foundation for adaptive optimization of SRMs using experimental feedback and open-source computational tools. • An open-source framework integrating openMotor with a genetic algorithm for automated SRM geometry optimization. • Demonstrated reduction of design–test cycles by coupling simulation and optimization with experimental calibration. • Experimental validation demonstrated agreement between optimized motor performance and simulation predictions.

Research on two-phase flow characteristics of single/dual-cell 18650 lithium-ion batteries under thermal runaway

Insilico engineering of transaminase variants for enhanced biocatalytic conversion of an ACE inhibitor precursor.

BLOOD: A fast, customizable, and patient-specific computational framework for assessing whole-body lymphocyte dose, survival, and replenishment after radiotherapy treatments.

MobiRes: An integrative pipeline for resistome risk prediction through mobilome profiling.

Partial differential equations (PDEs) are fundamental to modeling complex and nonlinear physical phenomena, but their numerical solution often requires significant computational resources, particularly when a large number of forward full solution evaluations are necessary, such as in design, optimization, sensitivity analysis, and uncertainty quantification. Recent advances in artificial intelligence – particularly operator learning – have enabled surrogate models that efficiently predict full-field PDE solutions; however, these models often struggle with accuracy and robustness when faced with highly nonlinear responses driven by sequential input functions. To address these challenges, we propose the Sequential Neural Operator Transformer (S-NOT), an architecture that combines gated recurrent units (GRUs) with the self-attention mechanism of transformers to address time-dependent, nonlinear PDEs. Unlike sequential-deep operator networks(S-DON), which use a dot product to merge encoded outputs from the branch and trunk sub-networks, S-NOT leverages attention to better capture intricate dependencies between sequential inputs and spatial query points. We benchmark S-NOT on three challenging datasets from real-world applications with plastic and thermo-viscoplastic highly nonlinear material responses: multiphysics steel solidification, a three dimensional (3D) lug specimen, and a dogbone specimen under temporal and path-dependent loadings. The results show that S-NOT yields prediction errors up to 4.5 times smaller than S-DON even for data outliers. Furthermore, S-NOT provides an acceleration of 4 orders-of-magnitude compared to traditional finite element method simulations, demonstrating its accuracy and robustness for drastically accelerating computational frameworks in scientific and engineering applications.

Computational methods for signal peptide prediction: From statistical models to deep learning.

Precision nutrition and food biomanufacturing for space missions: Toward intelligent and bioregenerative life-support systems.

TRIAG: Tri-reinforced infused generative agents for financial risk compliance