Papers

Regulators in China, the United States, and the European Union now provide low-altitude airspace access as priced, time-windowed corridor authorizations, booked in advance and forfeited if unused. We ask how much capacity a UAV logistics planner should reserve on each corridor–time unit before demand is realized, to maximize expected profit net of reservation cost. Reserved capacity cannot be transferred across corridors or time windows and is consumed jointly along time-respecting paths, so reservations are coupled through the network in ways that models with exogenous airspace capacity cannot capture. We formulate a two-stage stochastic program whose recourse selects and routes accepted requests on a time-expanded network, prove its arc-based and path-packing forms equivalent, and solve it by Benders decomposition with column-generated subproblems. The decomposition operates on the LP relaxation, and all reported reservation and routing decisions are recovered as integer plans. Computational experiments achieve single-digit LP-Benders gaps on moderate-sized networks and extend to much larger instances through a truncated-path approach. A Shenzhen case study shows reservations concentrating on structurally central corridors, with demand level and reservation price having more influence on the quantity of capacity reserved than the selection of corridors.

Bio-based alternatives for conventional rigid foams have proven to be good substituents owing to their enhanced sustainability and competitive performance. However, because their manufacturing processes are complex and destructive testing is often impractical, this study investigates whether microstructural features can be correlated with mechanical properties in lignin-containing rigid polyurethane (PU) foams using machine-learning approaches. Various types and percentages of lignin-based polyols were investigated as a partial polyol replacement. Scanning electron microscopy (SEM) images and corresponding mechanical compression data were used to train a custom state-of-art dual-head convolutional neural network (CNN) targeting the specific prediction of density, specific compression modulus, specific yield strength, and specific compression strength. The CNN was optimized with a weighted multi-output loss function, achieving strong predictive performance R2 values ranging from 0.850 to 0.91 and correlation coefficients above 0.92 while maintaining mean absolute error percentages below 9%. This proves the trained network capability to predict and capture morphological features governing load bearing responses. On the other hand, Grad-CAM visualization revealed that the network focused its predictions on physically meaningful microstructural regions such as cell walls and strut junctions, which confirm that the proposed network can be classified as an interpretable, non-destructive and data-driven framework for predicting and understanding bio-based PU foams mechanical behavior, hence reducing the inconvenience caused by time consuming manufacturing and destructive testing.

While existing benchmarks have made substantial progress in evaluating LLMs across STEM domains, financial reasoning over structured data remains comparatively less explored. We introduce V-FiLLM, a framework that generates financial reasoning benchmarks from executable computation trees grounded in real tables, yielding items whose answers are correct by construction. Trees are evaluated symbolically to obtain ground truth and rendered into natural-language questions, removing any model from the labeling loop, so items can be generated at arbitrary scale without annotation cost and without inheriting a generator's error rate. V-FiLLM exposes four independently controllable axes of difficulty including computation depth, expression breadth, financial concept complexity, and context size. By evaluating on open-source models, we find that accuracy falls up to 51% as reasoning depth increases, and up to 47% points under adversarial numerical perturbations, highlighting remaining challenges in robust financial reasoning over tables. We further show that lightweight LoRA fine-tuning on verified chain-of-thought traces improves accuracy from 81.1% to 85.6% on held-out problems and outperforms the base model by 5% points on FinQA (Chen et al., 2022a), s), suggesting that targeted, low-cost adaptation is a promising direction for compositional reasoning in financial QA.

Causal discovery at large spatiotemporal scale is difficult: variables are physically embedded, candidate interactions grow quadratically with system size, and causal structure changes with system state. We introduce GeoDCD, a geometry-aware neural framework that uses spatial coordinates to initialize a learnable hierarchy and converts a trained nonlinear predictor into time-varying directed graphs through input-output Jacobian sensitivity analysis. On chaotic Lorenz-96 dynamics, GeoDCD attains an F1 score of 0.99, reducing structural Hamming distance by 36.8% relative to the strongest neural baseline, and still leads flat baselines when coordinates are uninformative. Runtime scales approximately linearly over the evaluated range, enabling discovery on a 10,512-node sea-level-pressure grid. Applied to observations, GeoDCD identifies circulation-consistent gateways, resolves El Nino/La Nina-dependent reorganization, and separates energy-to-traffic from traffic-to-energy influence in coupled electric-vehicle and road systems. Edges are neural-Granger sensitivities rather than interventional effects, positioning GeoDCD for mechanistic hypothesis generation where interventions are unavailable.

Gene regulatory networks (GRNs) are fundamental in understanding cellular dynamics and underlying mechanisms during development and disease. Although scRNA-seq technologies have enabled the collection of vast numbers of gene expression profiles at single-cell resolution, inferring GRNs from scRNA-seq data remains a significant challenge due to high dimensionality and dropout. Recently, FlowGRN has shown promising results in reconstructing cell trajectories and inferring GRNs by applying conditional flow matching (CFM) to learn the cell dynamics. However, FlowGRN still faces limitations in the temporal coherence of reconstructed dynamics and relies on human inspection, which hinders downstream applications and reproducibility. In this paper, we propose FlowGRN+, an improved version of FlowGRN that integrates spline fitting into the CFM framework to generate more stable reference trajectories for training, thereby improving the temporal coherence of the learned dynamics. To address overshooting in spline fitting, we further introduce a projection scheme that projects spline tangents onto the local tangent space of the data manifold. We evaluate FlowGRN+ on the BEELINE benchmark and show improved trajectory smoothness with a competitive GRN inference performance. FlowGRN+ provides a practical framework for reconstructing cell trajectories and inferring GRNs from scRNA-seq data, and the insights from this work may also be useful for other CFM-based models of cellular dynamics.

Volatility control converts risk estimates into portfolio exposure, yet existing approaches often rely on a fixed volatility estimator or a pre-defined control rule that may not adapt to changing market conditions. We propose VolRouter, a modular framework that formulates volatility control as state-conditioned routing over estimator-controller pairs. VolRouter first summarizes market conditions into a control-relevant state profile and then performs routing through three stages: state inference, switch review, and pair selection. The Router can be implemented using rule-based, learnable, or LLM-based decision modules, while portfolio actions remain generated by predefined control policies. We evaluate VolRouter across S&P 500, Multi-Asset, Bitcoin, and USDT volatility-control settings. VolRouter achieves the highest Sharpe ratio in three of four settings. On S&P 500, it improves Sharpe from 0.952 for RV + Naive Scaling to 1.222 while reducing maximum drawdown from 15.10% to 12.58% and daily CVaR from 1.76% to 1.32%. On Multi-Asset, it improves Sharpe from 1.498 to 1.540 and reduces CVaR from 1.56% to 1.18%. Bitcoin shows similar improvements in risk-adjusted performance, while USDT provides a boundary case where simpler state-aware selectors remain competitive. Ablation and sensitivity analyses show that the improvement comes from relative policy evaluation and selective persistent switching rather than simply expanding the policy library. These results suggest that volatility control can be viewed as a policy-selection problem when risk management requirements vary across market states.

Thermo-mechanical compliant devices are commonly designed with small-strain linear elasticity and temperature-independent material properties, even though they might operate hundreds of kelvin above ambient where both assumptions are questionable. In this work, we quantify the effect and cost of each assumption in multi-material topology optimization of thermally actuated compliant devices. To this end, we introduce a physics-informed, simultaneous analysis-and-design framework with (i) a finite-strain quadratic-Hencky (logarithmic-strain) constitutive model whose isotropic thermal eigenstrain admits an exact additive split in log-strain space, and (ii) temperature-dependent conductivity, thermal expansion, and elastic moduli for a titanium–copper–steel material system. We optimize a thermal actuator and a thermal gripper at three design temperatures under both a baseline model and the full physics, subject to mass and manufacturability constraints. Every converged design is re-evaluated by verified nonlinear finite element solvers in the full factorial of constitutive law and property model. The comparison between the two factors reveals that the constitutive law is the decisive modeling choice: These devices work as linkages where linear kinematics mistakes rotation for compressive strain; its error therefore grows with the design temperature and concentrates on the very layouts that exploit rotation best. Because a linear optimizer also steers away from the rotation-rich mechanisms that would expose this bias, the model can deceptively appear trustworthy when validated against its own designs. Designing with the full physics yields consistently stronger and more temperature-robust devices at a modest increase in design-time cost.

A thin shell carries load through its shape and, at finite deflections, its stiffness changes with the load itself, so a shape optimum found with a linear model can be far from optimal. We present an automated framework that embeds the fully geometrically nonlinear shell response in the shape-optimisation loop. The forward model, a five-parameter nonlinear Naghdi shell stabilised against locking by partial selective reduced integration, operates directly on a discrete (faceted) triangulation with a numerically recovered director field – dispensing with the exact mid-surface parameterisation of isogeometric approaches, a chart that ceases to exist once the geometry itself is the design variable. Implemented in Firedrake, the model generates its residual, consistent tangent and adjoint automatically; shape derivatives, computed by algorithmic differentiation through the full load-continuation solve, drive the Fireshape/ROL trust-region optimiser. The forward solver reproduces the Sze/Abaqus benchmark for a clamped semi-cylindrical shell under a point load, capturing the progressive stiffening that a geometrically linear model cannot reproduce. The optimisation is validated against the COMSOL benchmark, a sheet-metal bracket under bending: the framework develops the same off-mid-plane corrugation mechanism and attains an 87% reduction of elastic strain energy within the prescribed displacement budget, matching the benchmark's magnitude and area change. Applied to the curved semi-cylinder, it forms a smooth stiffening crease that reduces the shell's average deflection under load by 78%.

The Woods-Saxon mean-field approach offers a compact description of bound single-particle motion in finite nuclei, while physics-informed neural networks (PINNs) provide a differentiable route to the inverse problem of recovering potential parameters from sparse spectral data. We develop a probabilistic physics-informed framework in which a WaveNet represents the separated single-particle wavefunction and a ParamNet maps selected spectra, nuclear descriptors, and quantum numbers to a learned distribution over six global Woods-Saxon parameters. The Hamiltonian includes the central Woods-Saxon, proton Coulomb, and spin-orbit terms; training enforces spectral energy consistency, Schr"odinger-equation residuals, boundary conditions, normalization, orthogonality, spin-orbit splitting constraints, and latent regularization. The distribution mean serves as a selection-free parameter estimate, validated against an independent finite-difference radial solver. Synthetic closure tests with the Seminole and Wahlborn parameterizations recover all six parameters with sub-percent relative errors and reproduce reference spectra with mean absolute deviations of (0.0109) and (0.0131~\mathrm{MeV}). For experimental spectra with the Wahlborn form, the estimator reduces the all-state mean absolute error from (1.0783) to (0.8303~\mathrm{MeV}); with the Seminole form it attains (0.8068~\mathrm{MeV}), close to the (0.7969~\mathrm{MeV}) from Seminole reference parameters, using only (42) experimental levels – about (51%) fewer than the Seminole calibration. These results show that sparse, structured single-particle spectra can constrain global Woods-Saxon interactions within a differentiable framework addressing both the forward eigenvalue problem and inverse parameter identification, while the learned output spread offers a model-derived, qualitative measure of parameter stiffness.

Operator learning is a rapidly advancing area of computational science. It is particularly well suited to problems where a partial differential equation (PDE) must be solved repeatedly under varying physical configurations. Most existing architectures represent the solution operator in a fixed basis. While this assumption is well aligned with global structures, it is less suitable for phenomena governed by local interactions in physical space. We explore an alternative perspective motivated by the observation that the continuum limit of coupled oscillator systems can describe a broad class of PDEs. Building on this idea, we introduce the Kuramoto Neural Operator (KNO), which represents the solution through the evolution of a latent field of interacting oscillators. Across a diverse collection of PDE benchmarks, KNO achieves strong predictive performance, with improvements over competing approaches. Our experimental evaluation also includes an extensive ablation study that quantifies the contribution of each architectural component incorporated into KNO. Furthermore, we show that the model's prediction error is closely linked to the collective dynamics of the latent oscillators. It varies systematically with their degree of synchronization, providing insights into the underlying mechanisms.

Large language models often produce plausible numerical claims from financial filings while using the wrong reporting period, unit, line item, or formula. Verifying such claims requires more than rechecking arithmetic: a verifier must ground the relevant facts, establish the authorized calculation, and determine whether the candidate value follows from both. We propose VeriFin, a neurosymbolic verification framework for numerical financial question answering that grounds operands in filed XBRL facts, derives calculations from the question, filing linkbases, or documented metric definitions, and checks claims using Z3. When a claim is inconsistent, solver-derived unsatisfiable cores identify the conflicting facts, formula, and candidate value, enabling targeted repair. We evaluate VeriFin on XBRLFiling, a new 600-question benchmark constructed from 10-K filings of 28 U.S. companies, and on FinanceBench. On fixed candidate pools shared by all verification methods, VeriFin accepts none of the incorrect claims, whereas the baselines accept 6 to 92 incorrect claims on XBRLFiling's 600 claims and 4 to 21 incorrect claims on FinanceBench's 67 claims. The zero false-accept result persists across multiple answer-generator models. Solver-derived feedback further improves repair, achieving up to 69.9% recovery among true catches. These results show that source-grounded symbolic verification can provide a reliable and auditable acceptance control for LLM-generated financial claims.

Inferring gene regulatory networks (GRNs) from single-cell RNA sequencing (scRNA-seq) data offers insights into cellular behavior, but is complicated by the lack of temporal information and the prevalence of dropout noise. To address these challenges, we present FlowGRN, a method that integrates conditional flow matching and score matching for robust trajectory reconstruction with dynGENIE3 for scalable GRN inference. FlowGRN incorporates a novel cell similarity measure that is resilient to dropout effects in high-dimensional scRNA-seq data. Evaluation on the BEELINE benchmark demonstrates that FlowGRN achieves state-of-the-art performance on both synthetic and experimental datasets. Ablation studies validate the importance of both the dropout-robust similarity measure and the trajectory reconstruction step, highlighting FlowGRN's ability to accurately model dynamic regulatory relationships.

Accurate air quality forecasting is essential for public health and urban environmental management, but remains challenging because pollutant channels differ in periodicity and distribution drift, while their concentration trajectories contain both multi-scale dependencies and rapid changes. Recent methods have improved spatial dependency learning and meteorological covariate modeling. However, pollutant channels are still passed through the same normalization rule and temporal backbone, using a shared latent representation for channel-specific distributions and changes at different rates. To address this limitation, we propose AirFlow, a pollutant-aware dual-stream framework that operates on station multivariate observations without additional graph propagation or predefined signal decomposition. Specifically, AirFlow designs two novel blocks: (1) a statistic-guided normalization routing mechanism that selects a normalization path for each pollutant according to its 24-hour autocorrelation and distribution drift; and (2) a hierarchical dual-stream state model that combines multi-scale state space propagation with learnable response coefficients, where gated bidirectional cross-attention exchanges information and adaptively fuses the resulting representations. Experiments on real-world data from multiple cities show that AirFlow achieves the best performance in 34 of 36 metrics comparisons, with reductions of up to 11.11% root mean square error over the state-of-the-art baseline. AirFlow also requires only 0.0483M parameters and 0.0215G FLOPs, achieving high forecasting accuracy with low computational overhead.

Large language models can write parametric CAD programs from a natural-language description (text-to-CAD generation), but a single sample is often wrong. Increasing test-time compute by sampling multiple candidates only helps if a good candidate can be identified, yet no ground-truth model is available at generation time. Existing systems often require a separate verifier, such as a vision-language judge, to select among candidates. We investigate whether the candidate pool itself provides enough signal for effective selection and a verifier-free alternative. We introduce 3D CAD consensus selection, hereafter consensus selection: sample $N$ parametric CAD programs, compile them to 3D models, and return the candidate that agrees most with the rest of the pool. The method is training-free and compatible with existing CAD agents. We investigate geometric and topological notions of agreement, each of which improves its corresponding evaluation metric. On the exact candidate pools of a state-of-the-art CAD generation method, geometric consensus improves all three geometric metrics over the method's verifier, while topological consensus matches it on topology. Across every tested LLM and prompt variant, geometric consensus also improves geometric accuracy over random selection from the same pool, reducing Chamfer distance by $1-10\%$.

In battery manufacturing, the low-dew-point air conditioning of dry rooms is among the largest energy consumers, and a physics-based digital twin offers insight for operating-point optimization beyond the installed monitoring points. Building one and calibrating it to field data each demand distinct expertise, which limits industrial uptake. We present a framework that generates a dynamic digital twin of an HVAC system from a declarative topology specification, concise enough to draft from a natural-language plant description, compiled against a purpose-built physical component library with wiring, solver, and telemetry synthesized automatically. The models carry equipment-level physics: the desiccant wheel couples heat and mass transfer through an interchangeable sorption-isotherm component, so an undisclosed commercial sorbent is calibrated as an effective isotherm rather than asserted as a material. For experimental validation we built an industrial-grade, ten-component low-dew-point system whose commercial desiccant-wheel unit holds a chamber near -40 {\deg}C frost point, and operated it in both dehumidification and bypass regimes. Generation reached a runnable model fifteen times faster than expert manual construction, and a single parameter set, fitted only to three closed-loop humidity nodes, predicts the bypass regime within 0.1 {\deg}C, the reactivation-heater power within 5%, and measured input-step responses. Identifiability analysis shows why this is prediction, not fitting: ordinary operating points constrain only one parameter combination, and the deep-dry equilibrium level of the recirculating loop supplies the missing signal. The framework shortens the path from plant description to measurement-validated twin; its criteria-based calibration is a step toward twins calibrated, not only constructed, automatically.

Each quasi-static load step of phase-field fracture is a bound-constrained minimization of a nonconvex, coupled displacement-damage energy under an irreversibility bound on the damage. Monolithic Newton stalls once the nonlinearity localizes at the advancing crack front, and staggered (alternate-minimization) schemes converge slowly there. We present an on-demand nonlinear-elimination preconditioned trust-region Newton method: an energy Steihaug-Toint trust region, a primal-dual active set for irreversibility, and a bound-constrained field-split sweep that eliminates an algebraically-identified "hard set" spanning both fields before each step. The elimination is applied on demand – triggered by the coupled Newton's own stalling and otherwise skipped – so the method reduces to monolithic Newton at no surcharge where the step is already healthy. We find the robustness to come from the energy trust region: with that globalization fixed, monolithic Newton already completes every loading history without cutbacks, where residual-merit Newton death-spirals, alternate minimization stalls, and the full-field sweep loses robustness. Against that well-globalized baseline, the on-demand elimination cuts outer nonlinear iterations by 19-25% (brittle) and 17% (ductile), with always-on elimination reaching 26-28% and about $39\%$ at the ductile nucleation step. Measured machine-independently, as a full-mesh-equivalent assembly-work proxy rather than wall-clock, it is competitive with – not faster than – monolithic Newton (within about 10%), whereas an always-on sweep adds up to 30%. Nonlinear elimination is thus an iteration-reduction mechanism whose overhead the on-demand gate bounds, with no demonstrated total-work advantage over well-globalized monolithic Newton.

An annular liquid jet may enclose a volume of gas and, when surface tension dominates over inertia, close on the symmetry axis at a finite distance. In the one-dimensional membrane model, closure provides an algebraic relation among the state variables at the free end. Earlier domain-adaptive schemes differentiate this relation to obtain an evolution equation for the domain length. We instead absorb the endpoint closure identically in the dependent variable and discretise the reduced system by Chebyshev collocation. Convergence is geometric, published steady data are recovered, and the unsteady calculations are checked by componentwise off-grid residuals, temporal refinement, and an independent upwind finite-difference discretisation. For jets of finite thickness, the convergence length decreases with nozzle thickness when the Froude number, Weber number, nozzle angle, and pressure coefficient are held fixed and the terminal zero remains transversal; this is a parameter-specific model result rather than a universal experimental thickness law. The forced response has a pressure-coefficient maximum near a Strouhal number of 0.093, whereas the convergence-length gain increases monotonically over the globally admissible frequency interval studied. Most importantly, a global positivity check changes the interpretation of the first finite-time event. Under body-force forcing, the mean radius first reaches zero at an interior point while the tracked endpoint remains transversal. The interior contact is quadratic, and the independent finite-difference scheme converges to the same first-contact time. After this event, the tracked endpoint is no longer the first physical closure point, so the single-interval annular-jet representation is no longer globally admissible.

The carbon footprint of any deployed Large Language Model (LLM) accumulates during inference, where repeated use of the model substantially exceeds the one-time cost of fine-tuning. Yet most efficiency interventions target either pre-training scale or post-hoc compression. We ask whether folding a calibrated, differentiable energy surrogate into the fine-tuning objective can produce inference behavior that gains task accuracy at zero or near-zero carbon cost, a break-even configuration. We propose a joint loss mechanism with a per-model carbon-emission parameter, a linear surrogate over parameter norm, FLOP proxy, and a memory proxy, fit from on-hardware energy profiling. We fine-tune three architecturally distinct families: Gemma-2 2B, Llama-3.1 8B, and Qwen-2.5 14B, and evaluate inference F1 and CO$_2$ emissions on three MMLU subjects: abstract algebra, philosophy, and formal logic. We discover from several outcomes that the carbon term behaves as either harmful interference or beneficial regularization depending on the task structure. We position calibrated carbon-aware fine-tuning as a lightweight, drop-in regularizer with a non-empty but model and task-dependent break-even region. This is an ongoing work, and we will release our codebase soon.

The scaling-up of large language models (LLMs) necessitates computing systems to have multi-processor-chip architectures, elevating the importance of chip-to-chip (C2C) communication. However, designing efficient C2C hardware architectures for LLM workloads faces three key challenges: generating realistic LLM-specific C2C traffic, accurately simulating hardware-level communication at scale, and efficiently exploring the exponentially large C2C design space. We propose C2C-Explorer, an adaptive Bayesian DSE framework that integrates a LLM-workload-driven traffic generator, a scalable interconnect simulator (switch/full-mesh, up to 512 chips), and a metric-guided evaluator into a workload-to-hardware optimization pipeline, enabling systematic C2C architectural co-design under realistic LLM workloads. Validated against FPGA-based C2C prototypes, the C2C simulator achieves 2.46-8.23% end-to-end timing error across diverse traffic patterns. Its hybrid cycle and event model further accelerates large-scale simulation by up to 7.8$\times$ over a pure cycle-accurate baseline. Applied to a 32-XPU DeepSeek-R1-671B inference workload, C2C-Explorer identifies configurations that improve goodput by 44.1% and reduce memory by 98.4%. C2C-Explorer is open-source and available at https://github.com/Selinaee/C2C-Explorer.

Fourier neural operators (FNOs) provide efficient nonlocal spectral learning, but varying geometries and independently chosen discretizations remain difficult to accommodate. We introduce the ambient-domain extension Fourier neural operator (ADEx-FNO), a deterministic framework that incorporates geometry without modifying the defining Fourier-operator layers. Each physical domain is embedded in a fixed ambient hypercube and represented by a signed distance function. Inputs and solution fields are deterministically extended to the ambient domain, transferred to a common, potentially nonuniform rectilinear latent grid, processed by the FNO, then interpolated to an independently chosen target discretization and restricted to the physical domain. All geometry-transfer operations lie outside the optimization procedure and require no trainable graph, point-cloud, deformation, or geometry-decoding modules. ADEx-FNO achieves relative l2 errors of 0.32%-0.77% on held-out smooth-domain nonlinear Poisson and advection-reaction-diffusion problems in 2D and 3D, and is also evaluated on unseen nonsmooth geometries. A single ADEx-FNO inference is then used to initialize conventional CFD solvers. For all 29 converged 2D and 3D RANS cases, pseudo-time iterations decrease, with mean reductions of 44.17% and 43.03%, respectively, with comparable gains across three mesh resolutions. URANS cases reduce post-window physical-time advances by 18.52%-27.51%. In transfer from 2D URANS training data to DNS at different Mach and Reynolds numbers, the bootstrap interval decreases by 23.47%-48.21%, depending on the target statistic. In all CFD tests, ADEx-FNO provides only the initial field; the governing-equation solver controls the subsequent solution, while physical or statistical consistency is assessed separately from computational savings.

Spiking neural networks (SNNs) offer a promising pathway to energy-efficient AI and brain-inspired computing. However, their widespread adoption is hindered by a lack of fast, accessible, and versatile simulation frameworks. In this paper, we introduce SuperNeuroMAT, an open-source, scalable, and highly efficient Python-based SNN simulator. We devise a novel matrix-based approach to model the leaky integrate-and-fire (LIF) neuron dynamics and natively support dense and sparse execution modes. This enables fast simulation of approximately 10,000 neurons in dense mode and 100,000 neurons in sparse mode on standard laptops and desktops without requiring specialized hardware. We demonstrate that SuperNeuroMAT consistently outperforms four established SNN simulators—NEST, Brian2, BindsNET, and snnTorch—on two performance metrics (execution speed and peak resident memory) and across various network sizes and connection probabilities. Furthermore, we demonstrate SuperNeuroMAT's applicability across a diverse set of problems. SuperNeuroMAT can efficiently handle conventional machine learning benchmarks such as the Digits and citation network datasets as well as neuromorphic event-based vision tasks such as N-CARS and ASL-DVS. Moreover, it can be extended beyond machine learning workloads and facilitate general-purpose workloads. We validated this by implementing the neuromorphic shortest path algorithm and two arithmetic primitives (addition and multiplication). SuperNeuroMAT can be installed via the Python Package Index (PyPI), thereby lowering the barrier to entry into the field of neuromorphic computing and accelerating the broader development of neuromorphic algorithms.

This paper presents, for the first time in power systems literature to our knowledge, analytical tools to explain the training performance of machine learning surrogate models for power system dynamics. Power system simulations are increasingly challenged by stiff and multi-timescale dynamics arising from converter-interfaced resources and fast control loops. Machine learning surrogates emerge as promising tools to handle this complexity and accelerate dynamic simulations. However, their performance remains difficult to interpret, which limits their adoption. Building on the small-signal eigenvalue analysis in power systems, this paper uses the Neural Tangent Kernel (NTK) method. NTK delivers a modal interpretation of the learning performance, identifying error modes that decay rapidly versus others that converge slowly. This connection explains how physical stiffness and timescale separation in power system dynamic models appear as optimization stiffness during Neural Network (NN) training. Based on this analysis, we develop adaptive loss-weighting strategies to improve and explain why structure-aware neural architectures, such as ActNet, perform better than vanilla NNs. We assess the proposed approach on physics-informed machine learning surrogate models of \acp{SM} and power electronic converters. The methods introduced in this paper can deliver the necessary analytical tools to interpret and improve the performance of machine learning surrogates, paving the way for the systematic, physics-aware design of NN architectures and training strategies. By moving beyond trial-and-error development, these tools reveal training dynamics and failure modes, support more reliable design decisions, and strengthen confidence in machine-learning surrogates for engineering applications.

Purpose: The lamina cribrosa (LC) is considered the initial site of glaucomatous retinal ganglion cell (RGC) injury, and is also the region where unmyelinated RGC axons become myelinated. Here we sought to use finite element (FE) modeling to investigate how the configuration of the myelination transition zone (MTZ) influences the mechanical insult to RGC axons. Methods: A multiscale FE framework was developed to investigate the biomechanical effect of myelin distribution on IOP-induced axonal stress and strain at the MTZ. An anatomically based macro-scale FE eye model was used to compute LC deformations under 15 and 45 mmHg IOP. These deformations were then applied to micro-scale models of the posterior LC, consisting of axons, myelin sheaths, and surrounding matrix. Four distinct MTZ boundary configurations were simulated: one flat and three with random posterior offsets of 3, 6, or 9 {\mu}m, representing potential physiological variations. IOP-induced effective axonal strains and stresses were quantified across the different MTZ configurations. Results: Under IOP loading, axons exhibited longitudinal compression and transverse stretch, with marked effective stress and strain discontinuities at the myelin boundary. Across all models, the unmyelinated region exhibited higher effective stress and strain than the myelinated region, and this mechanical discontinuity increased with larger MTZ offsets. Conclusions: Glaucoma-associated demyelination has been previously suggested to precede RNFL thinning. Here we have shown that the MTZ configuration directly influences RGC axonal mechanics. Whether different MTZ profiles can initiate glaucomatous injury, whether demyelination accelerates disease progression, or whether both mechanisms contribute, remains to be determined.

We introduce StableEval Arena, a cost-aware benchmark framework for evaluating agentic AI systems on stablecoin peg-risk prediction. StableEval Arena evaluates LLM-backed agentic systems on diagnosing peg stress and forecasting deviations from the one-dollar peg over a hidden seven-day horizon, using leakage-safe historical replay with exchange price-volume data and market-context features. We report two complementary experiment blocks: a 120-case stress-enriched validation block and a 507-case natural-distribution full-arena evaluation block. Across six LLM-backed agent configurations and baselines, StableEval Arena measures prediction quality, calibrated-label behavior, structured-output reliability, latency, token consumption, and estimated inference cost. Rather than ranking agents by accuracy alone, the framework treats trustworthiness as a joint property of forecast quality, operational reliability, and computational cost. The results show a gap between protocol-following reliability and financial-risk reliability: agents reliably produce valid structured outputs at modest measured cost, but still miss most rare severe-stress and sustained-depeg cases. To support auditing and replication, we release the benchmark dataset on Hugging Face and the source code on GitHub.

Uplink power control in dense, heterogeneous Internet-of-Things (IoT) tiers is fundamentally limited by incomplete channel-state information (CSI) and mutual interference, while low size, weight, and power (SWaP) devices cannot afford the feedback and computation of conventional distributed schemes. This paper proposes EBGT, an epistemology-aided Bayesian game-theoretic framework for decentralized uplink power minimization in stochastically distributed IoT networks. Interfering users are modeled as spatially random through a Poisson point process (PPP), and each device reasons about its rivals through a two-layer belief hierarchy of inter-epistemic beliefs about opponents and intra-epistemic self-assessment, so that the transmit-power equilibrium is reached without repeated inter-node feedback. We derive a closed-form coverage-probability payoff via stochastic geometry and quantify belief convergence toward equilibrium using the Jensen–Shannon divergence (JSD) of the resulting power distributions. Monte-Carlo simulations validate the analytical coverage expressions and show that EBGT sustains the target coverage probability while reducing transmit power relative to fixed power control (FPC) and stochastic non-cooperative power control (SNCPC) baselines, particularly under stringent SINR and high-density regimes.

As artificial intelligence (AI) adoption accelerates across tech giants, AI-native startups, and non-technical organizations alike, a deceptively simple question remains hard to answer: is that spending efficient? AI consumption is priced by tokens, and costs vary by token type (input, output, reasoning) and model type, with usage ranging from a few hundred tokens for simple queries to over a million for multi-step agentic tasks. This variance makes cost comparison, both within and across organizations, difficult without a standardized framework. We introduce the Token Efficiency Index (TEI), a peer-benchmarked composite indicator that condenses token spend efficiency into a single 0-100 score. The TEI ingests an organization's AI usage data, independent of the underlying provider, and computes three direction-aware metrics: cache hit rate, cache amortization ratio, and premium model share. These are normalized to a common scale and aggregated via an equal weights composite and a Benefit-of-the-Doubt (BoD) Data Envelopment Analysis (DEA) model, with a robust order-m extension for sparse data. The result is a headline score, a peer percentile, and frontier-gap recommendations with estimated savings. Grounded in established methods from composite indicator and DEA literature, the TEI offers a transparent, interpretable approach to benchmarking AI token efficiency and identifying opportunities to optimize AI spend.

This paper proposes a Physics-Informed Neural Frequency Response Framework for learning and interpreting the frequency-domain behavior of semi-active shunted piezoelectric tuned mass dampers. The motivation is that the behavior of such systems is most naturally expressed through frequency response functions, while the governing electromechanical interactions depend strongly on hidden structural and shunt parameters. Conventional data-driven models can approximate these mappings, but they often lack physical consistency, require large training datasets, and provide limited interpretability. To address these limitations, the proposed framework combines a physics-based forward frequency-response model, a neural inverse learning module, and an explainability component. The forward model is used to generate synthetic complex-valued frequency-response data over a broad range of structural and shunt configurations while preserving the governing electromechanical behavior of the system. Based on synthetic frequency-response data, the neural inverse model is trained to estimate hidden parameters from spectral response signatures and is subsequently evaluated using independently measured experimental FRFs. This synthetic-to-experimental design enables fast parameter inference without solving a new optimization problem for each measured case. To improve robustness to realistic conditions, controlled noise is introduced only at the inverse-training stage, while the underlying physics model remains noise-free. In addition, the learned representation is analyzed through latent-space organization, sensitivity mapping, and reduced symbolic distillation in order to extract interpretable electromechanical response descriptors. The resulting framework provides a data-efficient and explainable ML approach for frequency-response-based identification and inverse tuning of STMD.

Molecular dynamics (MD) simulations are governed by explicit equations of motion, yet most neural approaches that accelerate or emulate MD rely on simulator-generated trajectories, forces, or energies for training. In this work we ask to what extent can physically meaningful molecular trajectories be recovered from the governing laws. We introduce the Differentiable Newtonian Molecular Solver (DINaMo), a physics-informed neural framework that represents molecular trajectories as differentiable functions of time and is trained exclusively through Newtonian dynamics, conservation laws, and analytic interaction potentials on a given equilibrated initial state. Unlike prior physics-informed MD formulations, DINaMo uses no simulator-generated trajectories, forces, velocities, or energies as supervisory targets. In Lennard–Jones argon systems, the learned trajectories reproduce short-time coordinate, energy, and structural observables, including in a larger and denser liquid-like setting where the radial distribution function is recovered. Although currently limited to short temporal horizons, the results indicate that physically meaningful molecular trajectories can emerge directly from physics-only supervision, supporting the feasibility of trajectory-unsupervised neural solvers for molecular dynamics.

Simulating complex fluid flows requires capturing full equilibrium distributions rather than just mean trajectories, yet high-fidelity solvers remain computationally prohibitive. Recent advances, such as Diffusion Graph Networks (DGNs), have combined diffusion models with graph neural networks to sample equilibrium states directly from unstructured meshes, enabling distributional accuracy even from short simulations. However, graph-based diffusion approaches suffer from hand-crafted architectural constraints, limited receptive fields in message passing, and costly multi-scale designs, which restrict scalability to larger and more complex domains. We propose Fluid-DiT, a Graph-Free Diffusion Transformer that replaces graph message passing with attention-based denoising, eliminating explicit graph design while preserving the ability to model distributions of chaotic flows. Our framework introduces a latent-space formulation that disentangles geometric fidelity from distributional learning, reducing high-frequency artifacts and accelerating sampling. By leveraging the transformer's global receptive field, Fluid-DiT naturally captures both local flow structures and long-range correlations without requiring hierarchical graph coarsening. On canonical benchmarks including laminar cylinder wakes, ellipse-flow systems, and turbulent 3D wing experiments, Fluid-DiT consistently outperforms graph-based diffusion baselines in both sample quality and distributional accuracy, achieving higher $R^2$ correlations and lower Wasserstein distances. Moreover, it generalizes robustly from short, incomplete trajectories to unseen Reynolds numbers and geometries, demonstrating strong scalability.

Improving the predictive capability of credit scoring models is always an active area of research in the financial sector. Recognising the impressive effectiveness of neural networks in different domains (such as computer vision and natural language processing), various neural networks have been tested to potentially improve loan default prediction on credit data. Nevertheless, a significant challenge emerges due to the predominantly tabular nature of credit data, which is not well-suited to the structure and strengths of neural networks, hindering their ability to surpass traditional machine learning models in credit scoring. To overcome the challenge, we propose a novel data transformation method called Tabular Image that converts tabular data into images to take advantage of the powerful two-dimensional convolutional neural networks that perform extremely well on images while mitigating the challenges tabular data poses to deep networks. The Tabular Image can convert tabular data into compact and resilient images compared with existing transformation methods by creatively embedding two crucial measures in credit scoring, the weight of evidence and information value, in the image. Applications to three credit scoring benchmark datasets suggest that simply training a two-dimensional convolutional neural network with Tabular Image can provide state-of-the-art predictive performance. In addition, the advantage of our proposed method's prediction is more evident in the large dataset. Our innovative approach raises the possibility of leveraging two-dimensional convolutional neural networks in credit scoring using a proper data representation method. Furthermore, a flexible framework is provided to suit various tabular datasets in other domains.