Peer Revenue Publish or profit

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Friday 4 September

Machine Learning

Representation Redundancy and Structural Complexity in Finite-Field Inversion

The representation chosen for a mathematical operation can affect both its algebraic form and its empirical learning difficulty. We study this phenomenon for inversion over \(\mathbb F_{2^n}\), with field elements expressed in varying ordered \(\mathbb F_2\)-bases. We prove that two ordered bases induce the same coordinate inversion map if and only if they belong to the same Galois orbit. Since every orbit has size \(n\), the correspondence between ordered bases and distinct inversion maps is exactly \(n\)-to-one. We then analyze three Boolean formulations of inversion. The reference formulation has algebraic degree \(n-1\) and joint ANF leap \(1\), the mixed representation formulation has degree \(2(n-1)\) and joint ANF leap \(2\), and the complete raw formulation has degree at most \(3(n-1)\) and joint ANF leap at least \(n\). Exhaustive computations agree with the theoretical results and bounds in the cases considered. Controlled experiments with multilayer perceptrons show the same ordering in learning difficulty, while Galois orbit redundancy provides only a limited generalization benefit under the tested conditions. These results show that exact redundancy among representations can coexist with changes in Boolean structure and learning behavior when the representation is exposed as part of the input.
Read the full abstract

Computation and Language

When Do Internal Probes Beat Reading the Answer? Miscalibrated Readouts and Behavior-Concealed Knowledge in Language Models

A 0.6B language model, asked to verify 1,200 logical conclusions (half valid, half corrupted by a single semantic edit), answers YES every time. Judged by behavior it discriminates nothing; linear probes on its hidden states read the correct verdict at 0.96 AUC, transferring to unseen logical structures and separating foils built from exactly the words of the true conclusion (0.90). We ask where the verdict is lost, and find the dominant failure is a single scalar. The verdict survives to the model's own output logits (margin AUC 0.89) along a well-aligned readout direction; a saturated decision threshold, offset by +4.6 sigma, erases it. The diagnosis generalizes: across 90 semantic-label configurations of a five-model, three-family factorial, behavioral accuracy collapses onto a single function of threshold offset (Spearman -0.93) while margin ranking moves far less. Across a 13x scale range, internal knowledge saturates while free-form behavior is non-monotone: an 8B model underperforms its 4B sibling through an answer-channel failure rather than the threshold; forced-choice accuracy is monotone. The diagnosis is actionable: a one-parameter correction, never fit on evaluated structures, repairs behavior from 50% to 81% (0.6B); calibrated margin decoding recovers 94% at 8B; few-shot prompting works the same way, recentering the threshold (+4.6 sigma to 0.0 sigma) while preserving ranking. Comparing probe to margin separates three regimes: concealed, miscalibrated, and undetected. On a maze task built so foils carry no surface cues, the audit correctly reports the third. In the standard generation setting, answer-surface features and heuristic labels reproduce published probing results without any internal access.
Read the full abstract

Materials Science

Metallic Organometallic and Semiconducting Covalent Phases of Free-Standing $\gamma$-Graphdiyne Molecular Wires

One-dimensional carbon allotropes combining sp and sp2 hybridizations provide a versatile platform for atomically precise quantum wires with tunable electronic, vibrational, and mechanical responses. Recent on-surface synthesis has realized the $\gamma$-graphdiyne molecular wire on Au(100) in two interconvertible phases sharing an identical aromatic backbone but distinct linkage chemistry: an organometallic intermediate with C-Au-C bridges and a fully covalent product with diacetylenic linkages. While experimental techniques characterize the on-surface periodicity and C$\equiv$C Raman signatures, the intrinsic properties of the free-standing wires remain inaccessible. Here, we address this gap using hybrid-functional first-principles calculations. The covalent phase is a semiconductor with a direct band gap of 2.02 eV, a polyynic bond-length alternation of 0.117 Å, and a chain periodicity reproducing experiment within -0.5%. Conversely, the organometallic phase exhibits partial cumulenization and 1D metallic behavior, with spin channels crossing the Fermi level. Both isolated polymers demonstrate dynamic stability with Young's moduli of 995 nN and 861 nN, respectively, where the C-Au-C bridge acts as a soft link absorbing 9.2% of axial deformation at +6% macroscopic strain. Species-resolved vibrational analysis identifies an effective-conjugation-coordinate band at 1408 cm$^{-1}$ for the covalent phase and Au-projected modes below 400 cm$^{-1}$ for the intermediate, offering diagnostic fingerprints for phase discrimination during thermal or chemical conversion.
Read the full abstract

All authors Antony G. L. Rodrigues, Guilherme S. L. Fabris, Bruno Ipaves, Fábio L. L. de Mendonça, Douglas S. Galvão, Marcelo L. Pereira Junior

Quantum Physics

Residual detuning in the laboratory-frame anti-Jaynes–Cummings model with a squeezed vacuum

The laboratory-frame anti-Jaynes–Cummings (AJC) interaction retains a residual detuning $2f\lambda$ that is absent from the rotating-frame model. We map two proposed remedies—a Kerr shift $\chi(\hat a^\dagger\hat a)^2$ and collective Dicke coupling of $N$ two-level emitters—for a squeezed vacuum on that ladder ($r=1$, $\langle n\rangle=\sinh^2 r\simeq 1.38$). All quoted contrasts are the amplitude of the first turning point of the atomic ground-state population, which coincides with the two-level formula $\mathcal{C}=1/[1+(2f+\chi)^2]$ at $r=0$ to $10^{-9}$. Three results follow. (i)~A single Kerr strength never restores unit contrast at $r=1$; the $r=0$ $n$-dependent shift $\chi(2n+1)$ cannot cancel $2f\lambda$ on every occupied Fock component. (ii)~The exact $r=1$ contrast at $\chi=0$ is not reproduced by an incoherent sum $\sum_n P_n(r)\,\mathcal{C}_n$ built from the $r=0$ two-level formula; pointwise deviations are several tenths. (iii)~Collective coupling raises the contrast systematically. At $f=5$ one finds $\mathcal{C}=0.140$ ($N=1$) and $\mathcal{C}=0.575$ ($N=8$), above the unsqueezed value $8/[8+(2f)^2]=0.074$. The $N=16$ point at this $f$ remains truncation-limited and is not quoted to three digits. The same $N\sim(2f)^2$ estimate for $\mathcal{C}=1/2$ places trapped-ion values $f\sim 10^{2}$–$10^{3}$ outside the present construction. The relevant platform is ultrastrong circuit QED with $f\sim 1$–$10$. The calculation is a numerical control landscape, not a new solvable limit.
Read the full abstract

Artificial Intelligence

Does the Selected Object Reach the Reader? Auditing Identity Handoffs in Grounded Language-Model Pipelines

Grounded language-model pipelines can be divided into three stages: selecting an object, retrieving passages for it, and using that evidence to answer. If the selected object must reach the reader, losing it breaks the handoff. Benchmark recall checks the dataset-linked object, which can differ. We audit 600 HybridQA questions across three selector families. On 1,463 resolvable records where the selected object matches the dataset-traced passage, exact key lookup and exact title matching return the object every time. With every ranked rule given the same decoded selected title, body-only BM25 omits it on 389 records (26.6%) at cutoff five, while hybrid retrieval with reranking omits it on 14 (1.0%). The two identities differ on 329 of 1,792 resolvable records. With original-question rankings, their top-five checks disagree on 106 records (5.9%). Frozen reader comparisons associate the aligned object's presence with 28.6 to 31.0 points higher exact match. In a deliberately selected 64-item cohort, removing that passage sharply lowers exact match, while removing a similar-length comparison passage does not reproduce the drop. We release the Returned-Object Profile (ROP), an executable record of the target, returned-ID field, cutoff, membership rule, and complete expected population, with data and an offline replay.
Read the full abstract

All authors Siddharth Vohra, Runmin Jiang, Xiaomo Li, Min Xu

Optimization and Control

Centered Permutation Prefixes for SGD with Random Reshuffling: Sharp Rates, Hölder Geometry, and Composite Proximal Extensions

We study stochastic gradient descent with random reshuffling for finite sums \[ F(x)=\frac1n\sum_{i=1}^n f_i(x). \] For fresh reshuffling with a constant component stepsize, if each $f_i$ has an $L$-Lipschitz gradient and the average $F$ is $\mu$-strongly convex with a Lipschitz-continuous Hessian, we prove the last-epoch rate \[ \mathbb E[F(y_K)-F(x_\star)] =\widetilde O\!\left(T^{-2}+n^2T^{-3}\right), \qquad T=nK, \] matching the known quadratic lower bound in its $(n,K)$-dependence. The components may be nonconvex, and no componentwise Hessian continuity or separate bounded-iterate assumption is required. More generally, a $\nu$-Hölder-continuous average Hessian adds only $\widetilde O(n^{1+\nu}T^{-2-2\nu})$, so every $\nu\ge 1/2$ preserves the quadratic rate. Under convex components, a decreasing-stepsize result removes the large-epoch requirement and recovers the same two-term scale once $nK$ exceeds the condition-number scale. We also analyze epoch-wise ProxRR for $\mathcal P=F+\psi$. Writing $x^\dagger$ for the composite minimizer and $\beta_\star=\|\nabla F(x^\dagger)\|$, we prove \[ \mathbb E\|y_K-x^\dagger\|^2 =\widetilde O\!\left( \frac{\beta_\star^2}{K^2} +T^{-2}+n^2T^{-3} +n^{1+\nu}T^{-2-2\nu} \right). \] For $\nu\ge 1/2$, we show that the $\beta_\star^2/K^2$ splitting term is unavoidable and obtain a matching lower bound up to logarithms in the stated constant-stepsize regime.
Read the full abstract

Machine Learning

Optimizer Memory Schedules for Outscaling the Overtraining Axis

We investigate how optimizers scale across the overtraining axis and show that relative optimizer performance and optimal hyperparameters change substantially with training horizon. In particular, we study how matrix-preconditioned methods (Muon and SOAP) and a momentum-scheduled method (ADANA) scale relative to AdamW. We compare these four optimizers across models from 51M to 253M parameters and overtraining (OT) factors from 1x to 256x, sweeping the base learning rate at every setting. The preferred learning rate schedule can reverse across the overtraining axis, the best weight decay coefficient scales approximately as sqrt(OT), and longer horizons generally favor longer fixed memory. ADANA's scaling advantage over AdamW persists after tuning AdamW's fixed memory separately at each horizon. Log-time weight decay and momentum cooldown provide substantial gains for ADANA that compound as training increases. With this treatment, ADANA outscales AdamW with an exponent advantage close to that predicted by DANA theory on power-law random features. Muon and SOAP instead provide roughly constant token-efficiency advantages over AdamW across most of the measured range, although SOAP may gain further at the highest overtraining factors. ADANA begins behind both matrix-preconditioned optimizers but closes the gaps as training increases, surpassing Muon and becoming competitive with SOAP at our highest OT factors. These results establish training horizon as an essential axis for optimizer evaluation and design.
Read the full abstract

Classical Physics

Cartesian isotropic tensors revisited

An alternative, streamlined methodology is presented for deriving the isotropic Cartesian tensor bases under the special orthogonal group $\text{SO}(3)$. By shifting the traditional interpretation of the isotropy condition to analyze it as an algebraic system of equations for the rotation matrices themselves rather than the tensor components, lower-rank bases can be explicitly established in just a few lines without requiring finite coordinate rotations or complicated contractions of infinitesimal generators. Furthermore, a transparent combinatorial interpretation is provided for the number of tensors in the general higher-rank spanning sets. Finally, the Gram-matrix method is advocated as an efficient computational sieve to resolve the subsequent linear dependence.
Read the full abstract

Machine Learning

Training-Free Halving of Activated Experts in Fine-Grained Mixture-of-Experts Models

Modern fine-grained Mixture-of-Experts (MoE) models route each token to a small number of experts and renormalize their router probabilities. We show that this renormalization implicitly calibrates expert output gain to the training top-$k$: reducing $k$ at inference changes not only which experts are used but also the strength of the expert branch. We separate these effects by activating the top $k_1$ experts while normalizing by the probability mass of the top $k_2$ experts, introducing one integer with no parameters, training, or measurable compute overhead. On Qwen3.6-35B-A3B, reducing from 8 to 4 experts causes a 4.65-point MMLU drop under standard renormalization but only 0.35 points with $k_2=16$, while halving routed-expert compute. The result replicates on the $11\times$ larger Qwen3.5-397B-A17B, where reducing from 10 to 5 experts loses only 0.55 points with an appropriate reference set. Removing renormalization entirely is catastrophic, showing that preserving a suitable reference mass is crucial. We further find that perplexity and downstream accuracy favor different $k_2$, cautioning against selecting MoE compression settings using unlabeled text alone. Analyses also show that expert identity matters substantially more than expert weighting, while balanced and domain-specialized routing leaves limited room for expert pruning.
Read the full abstract

Information Retrieval

MURAL: Multimodal Uncertainty-aware Recommendation via Adaptive edge Learning

Multimodal Graph Neural Networks have become standard for recommendation by augmenting sparse interaction data with content features. Yet current architectures face two bottlenecks: structural rigidity, from a reliance on static precomputed similarity graphs that cannot adapt to evolving preferences; and semantic fragility, where noisy modality signals are indiscriminately fused, distorting the collaborative signal. We propose MURAL (Multimodal Uncertainty-aware Recommendation via Adaptive edge Learning), a unified framework that shifts multimodal recommendation from fixed structural augmentation to dynamic topology discovery. To address structural rigidity, an Adaptive Edge Learner combines a differentiable retrieval-augmented strategy with an approximate nearest neighbor search to discover latent item-item correlations that are both semantically adaptive and computationally scalable (O(NlogN)). To address semantic fragility, an Uncertainty-Aware Fusion module models the aleatoric uncertainty of heterogeneous modalities, dynamically down-weighting unreliable features while prioritizing high-confidence signals as a defense against cross-modal noise. We further employ a contrastive teacher-student alignment that anchors modality-specific representations to stable behavioral signals, ensuring optimization stability without gradient leakage. Experiments on large-scale benchmarks including TikTok and Amazon show that MURAL significantly surpasses both structural and generative state-of-the-art baselines, achieving superior accuracy while offering interpretability through domain-specific modality dominance and robustness under extreme data corruption.
Read the full abstract

All authors Ahmad Mousavi, Majid Alikhani, Yeon-Chang Lee, Roberto Corizzo, Yeganeh Abdollahinejad

Distributed Computing

Improving Progressive Compression with Adaptive Interpolation and Coefficient Decomposition

Exascale simulations generate data far faster than it can be stored or analyzed, making efficient data reduction essential. Error-controlled lossy compression offers high compression ratios under user-specified error bounds, but the target tolerance must be fixed at compression time. Progressive compression relaxes this restriction, yet existing methods still rely on fixed refactoring strategies and do not fully exploit correlations among decomposed coefficients, limiting the efficiency of progressive retrieval. In this work, we present an adaptive progressive compression framework that improves retrieval efficiency for two common targets, namely error-bound and peak Signal-to-Noise ratios. Our contributions are fourfold. (1) We propose to leverage two complementary interpolation schemes for adaptive progressive compression toward different targets, and we optimize them to achieve high efficiency. (2) We propose coefficient decomposition, a novel method that exploits the commonly overlooked spatial correlations among decorrelated data, which further improves the efficiency. (3) We develop the adaptive progressive compression workflow with automatic selection of the best-fit refactoring pipeline and tailored optimizations. (4) We evaluate the proposed framework on five real-world scientific datasets against three state-of-the-art progressive compressors. Experimental results demonstrate that the proposed framework improves the compression ratio by up to $42.3\%$ under the same requested error tolerance and up to $92.5\%$ at the same PSNR, compared with the best-performing existing methods. When transferring $512$ GB of scientific data to remote sites, the framework delivers up to $1.26\times$ speedup in the end-to-end data transfer performance. Furthermore, our method achieves the highest visualization quality while retrieving the least amount of data from storage.
Read the full abstract

All authors Wenbo Li, Xuan Wu, Qian Gong, Pu Jiao, Jieyang Chen, Qing Liu, Norbert Podhorszki, Scott Klasky, Xin Liang

Thursday 3 September

Computation and Language

Many Paths to the Same Mistake: Generating Diverse Reasoning Paths Behind the Same Student Mistake

Realistic evaluation of AI systems requires user simulators that capture not only cooperative behavior, but also the errors and variability that arise in real interactions. In educational technology, simulating step-by-step flawed student reasoning is essential, yet datasets mapping these pathways are scarce, and Large Language Models (LLMs) often auto-correct rather than sustain simulated mistakes. To address this, we present a generate-and-filter pipeline that produces diverse and realistic student error traces. Our approach utilizes forward consistency, a metric verifying if a flawed reasoning trace can be used to recover the student's actual incorrect answer, paired with an LLM-as-a-judge to ensure realism. Our evaluation suggests that while many LLMs struggle to maintain incorrect reasoning mid-trace, smaller supervised fine-tuned models successfully capture these divergent paths. This work provides a framework for high-fidelity error trace generation, advancing realistic, non-cooperative user simulation.
Read the full abstract

All authors Fareya Ikram, Mengxuan Hu, Parker Glenn, Zhiqi Huang, Alfy Samuel

Computation and Language

NYUAD at ArGuard Shared Tasks: Multimodal Embedding Models for Detecting Arabic Hateful Memes and Unsafe Prompts

The rapid growth of social media and generative AI has significantly increased the volume of Arabic textual content, memes, and user prompts shared across online platforms. While these media facilitate communication, creativity, and information exchange, they can also be exploited to disseminate hate speech, harmful content, and unsafe instructions. As a result, there is an increasing need for reliable automated methods to identify hateful Arabic memes and harmful (unsafe) prompts. This paper leverages pre-trained and fine-tuned multilingual embedding models to learn rich semantic representations from textual and visual content. The evaluation is conducted using a dataset of Arabic memes and prompts proposed in the ArabicNLP ArGuard 2026 challenge. The results underscore the capacity of large language model (LLM)-based embedding models to deliver superior performance. Developing robust detection systems is essential for improving content moderation and fostering a more secure and trustworthy online environment for Arabic-speaking users.
Read the full abstract

Computation and Language

Justification of Emotional Appeal in Computational Argumentation

Emotional appeals (EAs) are often treated as fallacious in computational argumentation, yet they may be justified when warranted by the context. We introduce the task of predicting EA justification and present the first human-annotated dataset for this task, covering 1,176 argument–context pairs. Grounded in philosophical literature, our framework evaluates EA justification along logical and emotional dimensions. Human annotations show that justification criteria form a shared latent judgment space, with evidential proportionality and normative acceptability most strongly associated with overall justification. We further find that EAs without logical support are not always judged fully unjustified, and stronger EAs are not necessarily judged less justified. Finally, LLM experiments show that EA justification prediction remains challenging.
Read the full abstract

Computation and Language

Syntactic Complexity Convergence in Dialogue: Analysis of the Phenomenon and Application to LLM Detection

We revisit the under-explored phenomenon of syntactic complexity convergence in dialogue. Extending prior topic-level analyses to full conversations, we examine five human-human datasets and a set of human-LLM dialogues. We integrate and evaluate a multidimensional set of syntactic complexity metrics to quantify structural variation in dialogue. Our analyses reveal robust syntactic complexity convergence across human-human datasets, especially in task-oriented interactions, and statistical tests confirm that these effects are not due to chance. In contrast, human-LLM dialogues show no significant convergence under individual metrics; however, linear-CKA uncovers a distinct cross-metric convergence pattern. Motivated by these differences, we use syntactic complexity as a feature for detecting LLM-involved dialogue, achieving 96% accuracy. Our classification results demonstrate its potential without relying on likelihood-based signals such as surprisal, which require a language model as the estimator.
Read the full abstract

All authors Yu Wang, Yanran Chen, Yifan Wang, Steffen Eger, Hendrik Buschmeier

Computation and Language

Enhanced English-Sanskrit Translation using Multilingual NMT Approach

Neural Machine Translation (NMT) is a widely accepted technique of automatic translation tasks of Natural Language Processing. However, NMT requires a sufficient amount of parallel corpora to attain a reasonable translation quality. In this work, we have considered a low-resource English-Sanskrit pair for English-to-Sanskrit and Sanskrit-to-English translation. We have performed a comparative analysis by exploring bilingual and multilingual approaches using LSTM-based Sequence-to-Sequence, and fine-tuning a pretrained model IndicTrans in a multilingual NMT approach. To investigate multilingual transfer learning, we additionally construct multilingual translation settings by incorporating Hindi along with English and Sanskrit using language-tagged data. Experimental results demonstrate that IndicTrans-based multilingual models outperform baseline results in terms of automatic evaluation metrics such as BLEU and ChrF for both directions of translation.
Read the full abstract

Plasma Physics

Shock wave transmission across solid-water interfaces in underwater electrical foil explosions

The transmission of shock waves generated by underwater electrical foil explosions across solid materials with varying acoustic impedances and thicknesses is investigated. Copper foils with thicknesses of 10 and 18 $\mu$m are exploded on planar substrates of aluminum, copper, and 3D-printed PLA with thicknesses up to 5 mm using a multi-gigawatt pulsed-power electrical discharge. The resulting shocks are visualized in water using high-speed shadowgraphy at 5 million frames per second, while high-resolution fiber optic hydrophone measurements provide direct quantitative measurements of shock strength after transmission through the impedance-mismatched materials. One-dimensional magnetohydrodynamic simulations are used to reproduce the foil explosion dynamics and provide insight into the thermodynamic, acoustic, and electrical properties of the foil during the explosion. The results show that shock transmission is governed not only by the impedance mismatch at the solid–water interface but also by the finite thickness and dynamic response of the supporting material. These findings provide a quantitative framework for understanding and controlling shock transmission in underwater electrical foil explosions and establish the feasibility and limitations of using solid substrates as structural supports in complex shock-generating geometries.
Read the full abstract

All authors Francesc Hernández Garcia, Brian Bertacco, Tobia De Benedictis, Sebastián Rojas Mata, Jergus Strucka, Michael Liverts

Game Theory

Exact Sizewise Least Cores for Candidate-Disjoint Approval Chains

For approval-based committee elections, the sizewise least-core value $\Gamma(P)$ minimizes the maximum excess support of a deviation over size-$k$ committees. Deviations are nonempty and contain at most $k$ candidates, with support measured relative to their Hare quota. A negative value certifies core nonemptiness, while $-\Gamma(P)$ is the largest worst-case quota shortfall secured by any committee. We quantify this margin on candidate-disjoint approval chains, a domain whose qualitative core nonemptiness is already known. We first factor each committee's deviation score over disjoint components. For chains, a closed local table yields the exact value and an attaining committee in polynomial time for rational weights. For nonnegative normalized real weights, let $q_{\mathrm{act}}\ge1$ count the connected components of the positive-support incidence graph. Then \[ \Gamma(P)\le-\frac{1}{k\min\{q_{\mathrm{act}},k+1\}}. \] Explicit extremal profiles show that this dependence on the active-component count is sharp. The margin is $1$-Lipschitz in total variation on the full ballot space. Each negative margin therefore certifies a neighborhood with nonempty core, even when perturbations leave the chain domain.
Read the full abstract

Group Theory

Soficity of One-Relator Groups and Reducible Presentations

We prove that every one-relator group is sofic, answering a question of Nate Brown. More generally, every group admitting a reducible presentation without proper powers is sofic, with one proper-power relation allowed at the final reducible step. The proof starts from the endpoint-preserving edge replacements of Poulin–Wróbel. We retain the free-group word carried by each terminal replacement and use it to define an \(F(S)\)-valued cocycle repair; in its relative form, the repair preserves the previously chosen generator coordinates while controlling the new relator defect. Coinduction transports the lower cocycle through successive one-relator-product extensions, allowing the construction to be iterated along reducible presentations. For a final relator \(w^m\), cyclic translates of the repaired set amplify a zero-defect set of measure close to \(1/m\) to one of measure close to one. A cocycle criterion then converts arbitrarily small relator defect into finite permutation approximations via a treeable skew-product orbit relation.
Read the full abstract

Mathematical Finance

Climate aware lending allocation under NGFS scenarios - A Monte Carlo approach

This paper examines climate-sensitive credit risk and lending allocation over a five-year horizon by mapping NGFS short-term transition and physical-risk scenarios into obligor default probabilities. NGFS shocks enter mean-reverting log-production dynamics, yielding analytical terminal default probabilities. Monte Carlo simulation with covariance-based PCA estimates joint losses, while constrained nonlinear optimization accounts for exposure-dependent default risk and compares alternative lending policies. Under a synthetic 1,000-obligor portfolio, mean five-year terminal default probabilities range from 13.88% (HWTP) to 13.95% (DAPS). Relative to conventional allocation, climate-aware lending reduces expected credit losses by 0.55–0.71% under HWTP, SWUC, and DiRe, with negligible improvement under DAPS. Tail risk is sensitive to dependence assumptions, and fixed-PD allocation generates markedly greater concentration. Results are conditional on NGFS scenario inputs and synthetic borrower calibration. Sectoral shocks cannot fully represent within-sector differences in emissions, geography, technology, or adaptation, motivating future institution-specific validation. The framework helps banks link NGFS scenarios to sector-level credit risk and portfolio loss outcomes. It can support ICAAP and climate stress testing by highlighting exposures most sensitive to climate risk, adding a forward-looking overlay to existing models. The study advances climate-aware credit allocation by linking NGFS sectoral projections to analytical obligor-level default probabilities and endogenous lending decisions. It provides a unified framework for examining how climate-related credit risk influences lending allocations, expected losses, and portfolio concentration, supported by quantitative policy comparisons and robustness analysis.
Read the full abstract

Machine Learning

SW-KAN: Kolmogorov-Arnold Networks with Stieltjes-Wigert q-Orthogonal Polynomials

Kolmogorov-Arnold Networks (KANs) represent a paradigmatic shift in deep learning by replacing fixed node activations with learnable univariate functions on edges, offering enhanced interpretability and parameter efficiency. While recent polynomial-based KAN variants have addressed the computational overhead of original B-spline implementations, they introduce a fundamental yet underexplored challenge: the domain mismatch between unbounded real-valued inputs and the bounded or semi-infinite support of orthogonal polynomial bases. To address this limitation, we propose the Stieltjes-Wigert Kolmogorov-Arnold Network (SW-KAN), a novel architecture that employs Stieltjes-Wigert q-orthogonal polynomials defined on the semi-infinite domain (0, infinity). We introduce a smooth exponential-of-tanh mapping that stably bridges the domain gap while preserving well-conditioned gradients, and leverage a numerically stable three-term recurrence that evaluates polynomial expansions in O(N) operations without special-function calls. Through comprehensive experiments spanning image classification and continuous function approximation, we demonstrate that SW-KAN achieves superior accuracy-efficiency trade-offs across diverse tasks. The log-normal weight structure and learnable q-parameter of Stieltjes-Wigert polynomials provide a distinct inductive bias that enables robust performance under resource-constrained conditions, including reduced feature dimensionality and limited training data. The proposed architecture not only outperforms established polynomial KAN baselines on standard benchmarks but also exhibits strong representational capacity for approximating complex multivariate functions with remarkably few parameters, making it a compelling alternative for efficient function approximation and classification in resource-constrained settings.
Read the full abstract

Machine Learning

Fast Polynomial Transcendentals for LLMs

Graphics processing unit (GPU) generations scale matrix, special-function, and memory pipelines at different rates, so kernel bottlenecks move as hardware evolves. FlashAttention-4 exposed this imbalance inside attention on NVIDIA Blackwell. We test whether short polynomial programs can accelerate other special-function-unit (SFU) operations in large language models (LLMs). We first compare native PyTorch evaluation with packed fused multiply–add (FMA) programs in an isolated IEEE binary16 (FP16) sweep spanning L2-resident and high-bandwidth-memory (HBM)-resident working sets. We then replace native sigmoid, tanh, and sigmoid linear unit (SiLU) with degree-3 or degree-4 bfloat16 (BF16) programs in four GB200 integration tasks: dense SiLU, tanh-softcapped attention, sigmoid attention, and routed-expert Swish-gated linear unit (SwiGLU). The programs combine analytical symmetry, target-format rounding, and packed arithmetic inside consuming kernels. The isolated paths improve by 1.19–2.19x in L2 and 1.00–1.70x in HBM. The dense-SiLU, tanh-softcapped-attention, and routed-expert substitutions improve complete training-step throughput by 2.7%, 2.9%, and 8.0%, respectively. The sigmoid-attention substitution improves complete-attention forward by 7.4% and the complete GPU step by 0.3%. Same-checkpoint open-weight ablations and one paired pre-training comparison per task extend the evaluation to model behavior. At common horizons near 100 billion tokens, the final smoothed training-loss differences (polynomial minus native) range from $-0.107$ to $+0.079$ across the four tasks.
Read the full abstract

Optimization and Control

Predictor-Based Exponential Tracking of Turbulent Solutions for the Kuramoto–Sivashinsky Equation with Input Delay

Uniform exponential tracking is addressed for nonstationary trajectories of the nonlinear Kuramoto–Sivashinsky equation subject to a constant input delay. The reference belongs to a family of complete trajectories contained in the global attractor and need not be stationary, periodic, slowly varying, or generated by a finite-dimensional exosystem. The delayed input is represented by a first-order transport equation coupled with the fourth-order tracking-error dynamics. A predictor–backstepping transformation compensates for the temporal mismatch between command generation and actuation by mapping the augmented closed-loop system into the nominal delay-free error dynamics driven by the outgoing trace of a homogeneous transport subsystem. This subsystem vanishes after one delay interval. Uniform attractor bounds permit the feedback parameters and stability constants to be selected independently of the initial time and the reference trajectory. Global well-posedness and uniform exponential stability are established in the augmented state space. Numerical results show that, for the considered configuration, uncompensated delayed feedback amplifies the tracking error, whereas predictor compensation restores sustained decay. Predictor-consistency, finite-time-extinction, and discretization-refinement diagnostics support the numerical implementation.
Read the full abstract

Artificial Intelligence

Characterizing a Configuration Where Inference-Time PRM-Pruned Fragment Grafting Is Inert: Evidence from Three Reasoning LMs

Diversity collapse in parallel chain-of-thought has motivated inference-time interventions built on a natural design: when a process reward model (PRM) prunes a chain, its high-PRM prefix is extracted and grafted verbatim as an in-context demonstration into a still-decoding sibling. We isolate this mechanism, PRM-Pruned Fragment Grafting (PPFG), as the most cost-minimal operationalization of cross-trajectory step-level transfer, and test it at the operating point where prior fragment-grafting work reports gains only under additional compensating ingredients. On Qwen2.5-7B-Instruct with Math-Shepherd on full MATH500 (n=500, three seeds), PPFG in both stagnation- and random-targeting variants is statistically indistinguishable from an independent parallel-CoT baseline on every measured axis. We characterize why: a four-bucket classification of 322 stagnation-rule injection events shows only 14% targeted a genuinely struggling chain; the rest landed on chains that had already succeeded, were near completion, or sat on a flat PRM plateau, states a rescue graft cannot change. No compound-gate refinement jointly achieves well-targeted firing and adequate density, and a random control matches the same parity at 2.4x the firing rate, so the inertness is not heuristic-specific. The finding replicates across three base LMs, six benchmarks, a second PRM, and a compatibility-gate sweep; two-one-sided-tests analysis promotes the parity to positive equivalence on all twelve Qwen/LLaMA cells. A per-event spot-check finds injected chains prune at 2.75x the matched-step rate, but a surviving-sibling counterfactual finds no population-level compensation. A hindsight oracle bounds any per-problem gain from choosing PPFG over independent at +0.13 pp. We contribute an equivalence-testing template for establishing inference-time mechanism nulls, with every claim scoped to its tested operating point.
Read the full abstract

Quantum Physics

Lie-Wedge Stratification and Pure-State Stabilizability in Single-Channel Qubit Control

We determine when qubit controls admit an effective generator that drives every initial state to the same pure state. Lie wedges enforce nonnegative weights for dissipative generators and thereby preserve the distinction between reversible control and irreversible dissipation lost under Lie-algebra closure. For drift-free qubits with Hamiltonian controls available with either sign and at most one independently actuated dissipator generated by a traceless jump operator, we classify all nontrivial systems into thirteen structural strata. Explicit tests using the control algebra and channel data assign each system to a stratum. Necessary and sufficient conditions determine whether its Lie wedge contains a stabilizing generator and, when it does, the minimum number of control-rotated dissipative orientations required. These conditions distinguish stabilizable and non-stabilizable systems even when they share the same generated Lie algebra and stratum. In an exceptional stratum, one dissipative orientation is insufficient even with the available Hamiltonian controls, whereas averaging two orientations of the same channel produces a stabilizing generator.
Read the full abstract

Computation and Language

SCM-based Fairness and Faithful Explainability for Legal Document Classification

Transformer models such as LegalBERT are increasingly used in legal decision support, raising concerns about both fairness and the transparency of model explanations. These properties are usually evaluated separately, leaving open whether a debiasing intervention that changes fairness also changes how faithfully explanations reflect model reasoning. This study investigates that relationship on the ECtHR alleged-violations corpus from LexGLUE. It compares a LegalBERT baseline with a fairness-regularized variant that penalizes stereotypical warmth and competence representations during fine-tuning. The evaluation covers predictive performance, demographic fairness, and SHAP explanation faithfulness across five random seeds. At the performance-optimal regularization strength, the intervention does not reduce demographic disparity. This null result holds across two fairness definitions and a conventional word-pair control on the gender axis. Classification performance is largely unchanged. However, the intervention consistently degrades explanation sufficiency across all five seeds and three thresholds. A shuffled-pair control reproduces this degradation while leaving performance and fairness unchanged, indicating that the effect arises from contrastive representational regularization rather than specifically from the warmth and competence structure. The results demonstrate a dissociation between fairness and explanation faithfulness: changes in explanation behavior do not necessarily indicate changes in fairness, and fairness must therefore be evaluated directly.
Read the full abstract

Optimization and Control

Improved upper bounds for five IHTC-2024 benchmark instances via exact reoptimization of the nurse-to-room layer

The Integrated Healthcare Timetabling Competition 2024 (IHTC-2024) defined a benchmark that integrates surgical case planning, patient admission scheduling and nurse-to-room assignment. Its instance set remains live: the organizers keep a public table of best-known upper bounds that is open to new submissions. We report five improved upper bounds on that table, for instances i02 (1263, previously 1264), i05 (12744, previously 12760), m01 (3380, previously 3384), m03 (6692, previously 6697) and m04 (3315, previously 3318). All five solutions were checked with the official IHTP validator and carry no hard-constraint violations. The solutions are not built from scratch: each is derived from the corresponding solution published on the benchmark page, holding its admission days, room assignments and operating-theater assignments fixed, and reoptimizing only the nurse-to-room-and-shift layer to proven optimality. The improvements are small in magnitude, between 0.075% and 0.125%, but they carry a structural observation: on the instances where the nurse layer can be closed exactly, it closes within seconds on all but one of them, and several of the published state-of-the-art solutions fall short of optimality in that layer.
Read the full abstract

Systems and Control

New VSC-HVDC interconnection between the Iberian Peninsula and Balearic Archipelago to enable energy transition

One of the challenges of the Spanish Transmission System Operator (TSO) is the decarbonisation of the Balearic Archipelago, by means of the integration of Renewable Energy Sources (RES) in the islands, as well as increasing the transmission capacity between the Iberian Peninsula and the Balearic Islands. Since the Balearic Archipelago is an island power system, the decarbonisation brings challenges related to power system stability and operation. A new High Voltage Direct Current (HVDC) interconnection between the Iberian Peninsula power system and the Balearic Islands power system is planned to facilitate the decarbonisation of the Balearic Archipelago (PEN-BAL2 Project). The HVDC link will be based on Voltage Source Converter (VSC) technology and will consist of a bipole of 2x200 MW, a DC voltage of +-250 kVdc and +100/-150 Mvar of reactive power capacity for each converter station. One converter station will be connected to a future El Fadrell 400 kV substation (Castellón, Valencia, Iberian Peninsula), while the other converter station will be connected to the existing San Martín 220 kV substation (Mallorca Island, Balearic Islands). The link will have HVDC submarine cables of 363 km (approx.).This paper will describe the challenges for energy transition in the Balearic Archipelago, technology enablers in general and PENBAL2 VSC-HVDC interconnection.
Read the full abstract

All authors Javier Renedo, Silvia Sanz Verdugo, Antonio Cordón, Belén Segura, David Castañeda, Rosalía Rivas, Patricia Labra

Databases

Zengram-Lite: An In-Browser Agentic-Memory Framework - Semantic Knowledge, Session Tracking, and Token-Budgeted Context

AI agents increasingly run in the browser, and they need somewhere to keep what they learn. The client-side state of the art, however, is a vector index - nearest-neighbor search over embeddings - with the rest of an agent's memory left to application code: the conversation history is an array in localStorage, context management is hand-rolled truncation, and there is no shared notion of a fact's confidence, its provenance, or the session that produced it. We present zengram-lite, an agentic-memory framework compiled to a single ~2.95 MB gzipped WebAssembly artifact that runs entirely in a browser tab. It provides three tiers over one transactional store. The knowledge tier offers hybrid vector-and-full-text recall with a fact lifecycle - confidence that rises and falls as facts are confirmed or contradicted, importance that decays over time, supersession that treats a restated fact as an update, and scope namespacing. The session-tracking tier models the conversation itself as first-class data: sessions, turns, typed content parts, and tool calls with state and timing, all as queryable tables. The context-assembly tier turns that structure into a token-budgeted prompt through a six-phase assembly that packs system instructions, knowledge, and recent history under a caller's budget, and returns a stable fingerprint that signals when a prompt prefix can be reused. The bundle is a superset of the zeta-lite SQL engine - the same .wasm re-exports the full Postgres-compatible surface, MVCC snapshot isolation, and copy-on-write database branching - so memory inherits transactional consistency across all three tiers. Because the wasm surface is synchronous while the framework's canonical operations depend on an asynchronous LLM and embedder, zengram-lite exposes a bring-your-own-result seam that runs the framework's real code paths over results the application computes in JavaScript.
Read the full abstract

Multiagent Systems

The Delegation Danger Band: Why Mid-Capability Sub-Agents Over-Trust Inherited Stale State

Agent frameworks increasingly delegate work by forking sub-agents; a common default makes the child inherit the parent's full working context. We measure how the effect of inherited state changes with capability, where $C_m$ denotes clean fork-fresh accuracy. We compare 3 inheritance policies: Reset (fork fresh: base evidence only), Selective (curated handoff: + the useful prior conclusion), and Full (implicit fork: + the useful conclusion and $d$ copies of a superseded conclusion) over a same-family ladder (Qwen3 0.6/1.7/4/8B) on a frozen, closed-set, action-scored benchmark. Every task is solvable from the base evidence, so performance loss can be attributed to reliance on stale state. (1) Deference to superseded state falls sharply with measured capability $C_m$ (the slope's confidence interval, CI, excludes zero on every family) across 2 synthetic primitives plus MuSiQue and HotpotQA. (2) On the Qwen3 synthetic ladder, net inheritance harm follows a nonmonotone pattern: a mid-capability model (Qwen3-1.7B) is a statistically significant local minimum of net harm, falling below its fork-fresh baseline ($\Delta(32)=-0.19$ [-0.25, -0.12]) and both neighbors, while the weakest model stays near-neutral and the strongest models stay robust. We call this harmful capability range a danger band. A within-model counting-difficulty sweep shows that the effect depends on model class even at matched $C_m$, and a live parent-to-child fork reproduces the mid-model harm. (3) Curated Selective handoff improves average accuracy over Full on all 3 datasets, largest at the in-band model, while the fixed-threshold capability router fails on the other datasets; a transferable router would need to predict the balance between reuse benefit and stale-context penalty. The benchmark is frozen and version-hashed.
Read the full abstract