Peer Revenue Now accepting positions

Papers

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

All of arXiv

Monday 9 November

Robotics

UniOMA: Unified Optimal-Transport Multi-Modal Structural Alignment for Robot Perception

Achieving generalizable and well-aligned multimodal representations remains a core challenge in robot learning. While recent approaches have attempted to align modalities by modeling conditional or higher-order statistical dependencies, they do not explicitly consider the structural coherence across modalities. In this work, we propose a novel multimodal alignment method that augments existing contrastive losses with a geometry-aware Gromov–Wasserstein (GW) distance-based regularization. To this end, we encode intra-modal geometry for each modality and reduce cross-modal structural discrepancies through a dynamically learned GW barycenter, thereby encouraging structural consistency in addition to instance-level contrastive alignment. We apply this optimal-transport-based alignment strategy to robot perception tasks involving underexplored modalities such as force and tactile signals, where modality data often exhibit varying sample densities. Extensive experiments on five robotic benchmarks show that our method improves structural alignment and downstream performance over state-of-the-art multimodal contrastive baselines across a wide range of downstream tasks, including robot and environment state prediction. Moreover, our GW-based regularization can be seamlessly integrated into most InfoNCE-like objectives.
Read the full abstract

All authors Xinrui Zu, Mark Hoogendoorn, Kevin Sebastian Luck, Shujian Yu

Fluid Dynamics

Why fluid relaxation modes matter: Nonlinear frequency response, dynamical integrity, and dynamic pull-in of electrostatic microbeams with squeeze-film damping

The squeeze film between an electrostatically actuated microbeam can both induce damping and add stiffness to the structure. In addition, the dynamics of the film can interact with the nonlinear response of the microbeam. This paper extends a reduced-order model built on the complex modes of the coupled fluid-structure problem to the computation of periodic responses and their stability for clamped-clamped microbeams. Periodic orbits are continued in the frequency domain, and their stability is determined from their Floquet multipliers. The robustness of the response is assessed through basins of attraction, integrity measures, and dynamic pull-in maps in the plane of the excitation parameters. The model is validated against the measured and computed frequency response of a microswitch. Closed-form expressions are derived for the fluid relaxation modes, together with a correction for strongly coupled films, and define a parameter that characterizes the regime of the film. The approach is applied to different microbeams including cases where fluid modes relax on the timescale of the oscillation. For all simulated devices, a single fluid mode is necessary and sufficient, whereas adding coupled modes cannot replace it, whether the fluid modes respond quasi-statically or carry their own dynamics. Removing the fluid mode displaces both bifurcations that bound the pull-in band, changes the size of the basins of attraction and the position of the attractor within them, misses an abrupt erosion of the safe basin under increasing excitation, and shifts the whole region of dynamic pull-in in the excitation plane. The reduced model, with three real unknowns, makes these global analyses affordable for a fully coupled fluid-structure system.
Read the full abstract

Combinatorics

Sufficient conditions for spanning trees with bounded total excess in tough graphs

Let $G$ be a connected graph and $T$ a spanning tree of $G$. The toughness of $G$ is defined by $t(G)=\min\Big\{\frac{|S|}{c(G-S)}:S\subseteq V(G), c(G-S)\geq2\Big\}$ for $G\neq K_n$. A graph $G$ is said to be $t$-tough if $t(G)\geq t$. The size, the adjacency spectral radius and the signless Laplacian spectral radius of $G$ are denoted by $e(G)$, $\rho(G)$ and $q(G)$, respectively. The $k$-excess of a vertex $v$ in $T$ is defined as $\max\{0,d_T(v)-k\}$. The total $k$-excess $\mbox{te}(T,k)$ is defined by $\mbox{te}(T,k)=\sum\limits_{v\in V(T)}{\max\{0,d_T(v)-k\}}$. A tree $T$ is called a $k$-tree if $d_T(v)\leq k$ for every $v\in V(T)$, namely, the maximum degree of a $k$-tree is at most $k$. It is obvious that $T$ is a spanning $k$-tree if and only if $\mbox{te}(T,k)=0$. This paper investigates a generalization of spanning $k$-trees using a concept called total $k$-excess and puts forward three lower bounds based on $e(G)$, $\rho(G)$ and $q(G)$ in a connected $\frac{k}{k(k-2)+b+3}$-tough graph $G$ to guarantee that $G$ has a spanning tree $T$ with $\mbox{te}(T,k)\leq b$, where $k\geq3$ and $b\geq0$ are two integers. As corollaries, we obtain three sufficient conditions for a connected $\frac{k}{k(k-2)+3}$-tough graph to has a spanning $k$-tree.
Read the full abstract

Data Structures and Algorithms

How Hard Is Tokenisation? Algorithms and Complexity Beyond NP-Hardness

Given a collection of strings and a vocabulary budget, how well can we minimize its total token count, and how does the answer change with the tokenisation model? For binary direct tokenisation, previous work proved APX-hardness but left constant-factor approximability unresolved \citep{kastreva2026}. We rule out every polynomial-time $M^{1-\varepsilon}$ approximation, for fixed $\varepsilon>0$, unless $\mathrm{P}=\mathrm{NP}$, even on distinct equal-length strings with $K=m-1$. Here $M$ is the total input length and $m$ is the number of input strings. We complement this lower bound with a deterministic polynomial-time $O(M/\log M)$ approximation algorithm and stronger bounds for selected vocabulary budgets. For the different objective of maximizing compression savings, GreedTok was shown to attain a constant fraction of the optimum on unrestricted alphabets \citep{baliga2026}; we establish a deterministic EPTAS for direct savings on every fixed alphabet and analyze greedy selection by true marginal gains. For exact optimization, we give a binary direct algorithm running in $2^{O(n\log\log n/\log n)}\poly(I)$ time, where $n$ is the total length of the distinct input strings and $I$ is the input encoding length. Finally, we resolve two unary merge-model hardness questions left by the bounded-alphabet study \citep{kastreva2026}: standard unary bottom-up tokenisation is APX-hard, and unary optimal pair encoding is strongly NP-complete, strengthening the previously known weak hardness.
Read the full abstract

All authors Andrei Robert Ion, Rares Mateiu, Adrian Miclaus, Alexandru Popa

General Economics

Fueling Inflation Risks: An Oil Shortage Scenario

Central banks routinely use alternative scenarios to assess risks to the economic outlook, but these are usually generated from linear models that are unable to speak to the likelihood of extreme outcomes. We develop a nonlinear modeling framework that turns a narrative oil market scenario into a fully probabilistic assessment of the dynamics of oil and gasoline prices, aggregate inflation, and household inflation expectations. It combines an expert-based path for a tail event, a nonparametric model of the global oil market, and nonlinear Phillips curves for the energy, food, and core components of PCE inflation. We apply it in real time to quantify energy price and inflation risks of a severe oil shortage caused by the closure of the Strait of Hormuz, modeled as a rapid drawdown of global petroleum inventories, using only information available in February 2026. Our analysis shows that once oil reserves hit rock bottom, oil and gasoline prices rise sharply, with risks tilted to the upside. Headline PCE inflation peaks near 4.8% in late 2026, and households expect inflation to rise by about 3 percentage points. Two years out, inflation is still highly likely to exceed the Fed's 2% target.
Read the full abstract

Computer Vision

ReVMap: Vectorized Global Mapping via Connectivity-Aware Local Map Fusion

Large-scale annotation of vectorized high-definition (HD) maps is labor-intensive and costly, motivating local-to-global auto-labeling approaches that incrementally construct global maps from local vectorized predictions. However, existing approaches often overlook prediction uncertainty and rely on heuristic map integration, potentially propagating errors and introducing topological inconsistencies. To address these limitations, we propose ReVMap, a reliable local-to-global framework that iteratively refines local map predictions and global map construction through uncertainty-guided prior utilization and connectivity-guided global map revision. Specifically, ReVMap leverages point-wise localization uncertainty to incorporate global map priors into local map prediction, improving robustness against inaccurate prior information. It then predicts local-to-global connectivity and selectively updates the global map through uncertainty-aware fusion based on the reliability of local predictions. By iteratively coupling local prediction and global revision, ReVMap progressively improves both local map accuracy and global map consistency. Extensive experiments on the nuScenes and Argoverse 2 benchmarks demonstrate that ReVMap achieves state-of-the-art performance in both local map prediction and global map construction.
Read the full abstract

All authors Hongjae Shin, Jungho Kim, Donghyuk Kwak, Myeongjun Kim, Seunghoon Yu, Jiyong Oh, Jun Won Choi

Computer Vision

TMT: Adapting CLIP to Infrared Imagery via Text-Derived Modality Knowledge Transfer

Vision-language models such as CLIP achieve strong performance across computer vision tasks, yet their image-text alignment is less reliable for infrared (IR) imagery because large-scale pre-training is dominated by visible-spectrum data. Due to the high cost of the infrared-text labeling, we investigate adapting CLIP to IR through only paired infrared-visible images, which requires considering modality-specific knowledge while preserving scene content and CLIP's image-text alignment. To achieve this, we propose TMT (Text-Derived Modality Knowledge Transfer), which adapts CLIP's image encoder through a text-defined infrared-to-visible direction and supervision from paired IR-visible images. First, we found that modality discrepancy can be described by corresponding text embeddings, thus using the direction of text embedding differences to guide the knowledge transfer between visible and infrared modalities without erasing infrared-specific information. To enhance the representation of semantics and details of infrared images, we introduce Generative-Perceptual Co-Supervision that combines semantic alignment from CLIP visible image encoder with diffusion noise-prediction feedbacks. On four source-disjoint IR classification benchmarks, TMT achieves the highest mean zero-shot accuracy among the compared methods (48.92%). It also achieves the best IR-text retrieval results among the compared methods without infrared-text supervision during adaptation and the highest open-vocabulary segmentation mIoU on both CART and SemanticRT among the CLIP-like methods.
Read the full abstract

All authors Ruofan Wang, Puzuo Wang, Xin Chen, Zibo Yu, Jungang Yang, Yi Hou

Statistics Applications

The Fragility Of Synthetic Populations Over Time: An Evaluation Of Forecasting Strategies For Agent-Based Models

Agent-based simulation models are widely used to assess the future mobility projects. However, their effectiveness hinges on the accuracy of synthetic populations, which are often generated using data from a single reference year. This study explores the temporal validity of these populations by evaluating their performance over multi-year datasets in a simulation-like environment. We focus on multi-dimensional agents, ensuring that all attributes critical to the simulation are comprehensively represented. To assess the quality of synthetic populations, we compare three generation methods: direct use of original training data, Bayesian Network sampling, and diffusion model. We then evaluate four agent-forecasting approaches: static projection (both idealistic and realistic), resampling, and dynamic projection. While dynamic projection holds potential, it requires complex longitudinal data to model the evolution of agent attributes. To address this, we introduce a heuristic model based on aggregated statistics, though its limitation lies in ignoring attribute correlations, which reduces its accuracy. Our results demonstrate that synthetic populations lose relevance over time, emphasizing the necessity of effective agent-forecasting. Among the methods tested, idealistic static projection yields the best performance but requires data that is difficult to obtain. The other methods show limited improvements, highlighting the ongoing challenges in accurately forecasting agent attributes.
Read the full abstract

All authors Vianey Darsel, Etienne Côme, Jeppe Rich, Francisco Camara Pereira, Latifa Oukhellou

Physics and Society

The role of administrative costs in policies to improve cooperation

We use an agent-based model for studying policies of an abstract government to increase the level of cooperation. In situations where defection is the rational choice, agents receive a bonus if they choose to cooperate. The government taxes agents' payoffs from a repeated prisoner's dilemma game to establish a budget that covers the payment of bonuses, but also the administrative costs involved. We consider costs to identify cooperators, and costs to mitigate the consequences of defective behavior. Governmental interventions are fallible, i.e. cooperators and defectors are correctly identified only with a certain accuracy. Hence, the critical bonus needed to let agents cooperate depends on the accuracy, but also on the current level of cooperation. We analyze the cost constraints for the government to pay the critical bonus and different policies to adjust the accuracy dependent on the level of cooperation. We find that perfect cooperation becomes impossible. Instead, high levels of cooperation can be achieved with optimal adjustment policies.
Read the full abstract

Signal Processing

Robust Beamforming and Antenna Position Optimization for Movable Antenna-Assisted Covert Communications

In this paper, we study a movable antenna (MA)-enhanced covert communication system, where a legitimate transmitter (Alice) serves multiple legitimate users in the presence of multiple non-colluding Willies that attempt to detect the covert transmission from Alice. To ensure data covertness with imperfect knowledge of the Willies' channel state information (CSI), we develop a robust covert transmission framework with a norm-bounded CSI error model. In particular, we aim to jointly optimize the transmit beamforming vectors, artificial noise (AN) covariance matrix, and MA positions to maximize the weighted minimum signal-to-interference-plus-noise ratio (SINR) among all legitimate users subject to worst-case covertness constraints. To tackle this challenging problem, we develop a block coordinate descent (BCD)-based optimization framework, incorporating the S-procedure, semidefinite relaxation (SDR), bisection search, and successive convex approximation (SCA). Numerical results demonstrate the effectiveness and robustness of the proposed algorithm in improving the performance of MA-enhanced covert communication systems.
Read the full abstract

All authors Xiangyu Dong, Ran Yang, Weidong Mei, Zheng Dong, Yue Xiu, Zhongpei Zhang, Tony Q.S. Quek

Algebraic Geometry

Cylindricity of log del Pezzo surfaces in characteristic five

Over an algebraically closed field of characteristic five, we prove that a log del Pezzo surface of Picard rank one contains no cylinder if it is not log liftable. We also characterize Du Val del Pezzo surfaces that contain no cylinder in positive characteristic.

Instrumentation and Detectors

Usage of GPUs for ALICE Reconstruction in the ALICE Online Compute farm and in the Grid

ALICE, the dedicated heavy-ion experiment at the LHC, records Pb-Pb collisions at an interaction rate of up to 50 kHz. It was the first LHC experiment to leverage GPUs for online data processing during LHC Runs 1 and 2, and its Run 3 online event reconstruction is now almost fully based on GPUs. In order to use its online processing server farm also for offline processing in an efficient way while the LHC is not operating, ALICE has been running the offline TPC tracking on GPUs since 2023. Since then, ALICE has been working on offloading more offline reconstruction steps to GPUs. In particular, porting the track model decoding and ITS tracking to GPUs has improved the throughput by 29%. Starting in 2026, ALICE can use the GPUs at external grid sites beyond the ALICE online computing farm for offline reconstruction. A first production campaign ran on the NVIDIA GPUs of the NERSC Perlmutter cluster, marking the first use of GPUs for LHC offline reconstruction at an external grid site.
Read the full abstract

Machine Learning

Evidence or Marginal? What Moment-Vector Propagation Computes in Polynomial Density Neurons

When a neuron's input is lost, or known through a belief $q$, what should it output? For a neuron storing a joint polynomial density $N$ with input density $D$, two formulas compete. $p_\mathrm{CD}=E_q[N]/E_q[D]$ costs one matrix-vector product; Hierarchical Correlation Reconstruction introduces it as a constant-denominator approximation to $p_\mathrm{EU}=E_q[N/D]$. For a density model both are exact answers to different questions: $p_\mathrm{CD}$ is Bayesian updating with the belief as virtual evidence; $p_\mathrm{EU}$ installs the belief as a new marginal (Jeffrey's rule). With a missing input $p_\mathrm{CD}$ returns the model's conditional; $p_\mathrm{EU}$ generally does not. This post hoc reading is proved, and illustrated outside the pair family up to $k=8$ inputs; it has no pre-registered test, only one exploratory real-data pilot. Post hoc, closed forms fall within a factor 2 of all 112 measured gaps, and negative density begins near input correlation $1/k$. In a pre-registered synthetic study (up to 32 inputs), 5 of 15 hypotheses are supported, 2 inconclusive and 8 not supported, including all three on transfer beyond equicorrelated Gaussians. Post hoc, where $p_\mathrm{EU}$ is defined, truncation usually outweighs the gap for Beta beliefs in pair neurons; triple neurons under flat and data beliefs are exceptions.
Read the full abstract

Software Engineering

From Formal Methods to Fuzzing and Back

For the first twelve years of my career in computer science I mainly worked on formal methods. I had been drawn to the field as an undergraduate by its promise of a way to build software with strong, mathematically based correctness guarantees. But after many years of struggling to make formal methods scale, I was seduced by randomised testing—also known as fuzzing. Fuzzing is much less principled than formal methods, but it just works: it can be readily applied to very large systems, and it is extremely effective at finding real bugs that developers care about and fix. In this personal account I trace my path from formal methods to fuzzing, by way of the GPUVerify project on verifying GPU kernels and the GraphicsFuzz project on randomised testing of GPU compilers. I argue that the oracle problem provides an intellectual link between fuzzing and formal methods, describe my continuing, targeted uses of formal methods in the domain of GPU programming language design, and discuss work at the boundary between formal methods and fuzzing. Finally, I reflect on whether generative AI might be the missing ingredient that allows formal methods to scale, and set out my plans for large-scale formal methods via the Keystone project, through which my collaborators and I aim to find out how much of that promise can be realised in practice.
Read the full abstract

Human-Computer Interaction

From LLM Hallucinations to Inspiration in Creative Work

Large language models (LLMs) can introduce unexpected ideas through hallucinations, yet little is known about when creative practitioners find these errors inspiring or how these judgments shape interactions. We conducted a retrospective qualitative study with 20 experienced practitioners in research, design, and social media content creation, thematically analyzing 41 hallucination cases and 20 semi-structured interviews. Our framework links inspiration judgments of hallucinations based on goal relevance, task stage, content adaptability, and type-specific factors to behaviors across not-inspired, inspired-by, and inspired-to states. We characterize downstream interaction consequences, showing how hallucination-inspired experiences vary across creative stages and reshape role-differentiated trust, and propose these consequences as a dimension of hallucination evaluation. Finally, we derive design recommendations for stage-sensitive control of divergence and adherence, user-defined constraint priorities, differentiated hallucination labels, and salvage support for preserving useful ideas while correcting errors. These contributions deepen HCI understanding of hallucinations and responsible LLM design in creative work.
Read the full abstract

All authors Xinrui Lin, Heyan Huang, Shuning Zhang, Qi Wang, Youchao Zhou, John Vines

Analysis of PDEs

A Silverman-Meadows-type criterion for the null controllability of time-dependent coupled heat equations with non-diagonalizable diffusion matrices

This paper is devoted to a uniform spectral Silverman-Meadows rank condition for the nullcontrollability of coupled systems of heat equations with time-dependent diffusion, coupling and control matrices. The proof relies on an adaptation of the Lebeau-Robbiano strategy combined with uniform finite-dimensional observability estimates associated with each Laplace eigenmode.
Read the full abstract

Nuclear Theory

Reabsorbed Urca neutrinos suppress bulk-viscous damping of neutron star merger oscillations

Reabsorption of Urca neutrinos suppresses the bulk-viscous damping predicted for neutron star merger oscillations. Below the direct-Urca threshold the capture neutrinos have mean energies of 19-46 MeV and are absorbed within metres. An energy-resolved kinetic treatment gives 2.7 s for a 1-kHz, 10-km oscillation at nuclear density and 4 MeV, against 19 ms for freely escaping neutrinos. It holds across the sampled states, three equations of state and the rate variations, with nearly frozen composition. This absorption is present in full-kinematics rates and absent from elastic opacity tables.
Read the full abstract

Artificial Intelligence

Rethinking Radar 3D Perception: Representation Learning from Raw Spectra

Raw radar spectra are widely viewed as too cluttered and noisy for direct representation learning, motivating handcrafted reductions, sparse point clouds, or compact Doppler descriptors in radar perception. We challenge this assumption and argue that the main limitation lies not in the raw representation itself, but in the absence of a learning objective that separates structured signal from noise. Handcrafted reductions partially address this problem but impose rigid assumptions about what constitutes signal, discarding weak yet informative structure while retaining artifacts that fit the reduction model. Once removed, such information cannot be recovered by supervised learning. We propose an alternative: structure is the component of a radar spectrum that remains reconstructable under localized corruption, while noise is the component that does not. This perspective motivates self-supervised masked modeling for dense radar spectra without predefined signal-noise separation. We instantiate it with Leakage-aware Spectral MAE, a masked autoencoder for raw radar cubes using a corruption strategy inspired by cross-bin energy spreading caused by spectral leakage. We also demonstrate the importance of Doppler organization: treating Doppler as a primary spatial dimension in an x-y-D representation, with height encoded as features, substantially outperforms naive channel stacking and enables effective learning from dense 4D spectra. Without handcrafted reductions, our approach consistently outperforms state-of-the-art radar-based detectors and point-cloud-based self-supervised learning (SSL) counterparts. These results establish raw radar spectra as an effective representation for 3D perception when paired with an appropriate pretraining objective.
Read the full abstract

All authors Miao Zhang, Mohamed Abdelsamad, Bin Yang, Sherif Abdulatif, Benedikt Loesch, Marco Altmann, Abhinav Valada, Bin Yang

Artificial Intelligence

AMSchem: Automated Schematic Visualization of AMS Circuits via Verifiable Agentic Generation

For decades, the netlist-to-schematic gap has remained a persistent challenge in design automation, especially for analog and mixed-signal (AMS) circuits. Netlists encode connectivity, not visual presentation, making their manual conversion into readable schematics time-consuming and dependent on expert conventions. The same gap leaves LLM-based design tools without structured visual feedback, forcing them to iterate on circuits they cannot see. We present AMSchem, a training-free, verifier-guided agentic framework that automatically generates human-readable schematics from AMS netlists. AMSchem separates circuit-aware visual planning (DrawingIntent) from executable schematic geometry (SchematicIR), and applies localized visual feedback with bounded, revalidated repair patches. Deterministic validators enforce device preservation, pin-net consistency, and geometric correctness throughout. For evaluation, we construct AMSchem-Bench comprising 64 circuits of 12 types. On the 40-circuit schematic subset, AMSchem with gpt-5.5 achieves 85.0% validity (vs. 32.5% direct generation). When integrated as a visual feedback module in LLM-based analog design, AMSchem resolves all 13 previously failed cases from AnalogCoder benchmark across five models, substantially improving the success rate. Our dataset and benchmark are available at https://www.amschem.com/benchmark
Read the full abstract

All authors Yichen Shi, Zhiyuan Wang, Jianing Cui, Yao Lai, Jinting Zhang, Zhuofu Tao, Hongyang Wang, Zhiping Yu, Ting Jung Lin, Lei He

Computers and Society

Do Machine Learning Models Predict?

Machine learning (ML) models are said to predict. Philosophers draw a line between accommodation and prediction, depending on whether or not the evidence a theory succeeds on was used in constructing it. Does this distinction survive when a theory is replaced by an ML model? We argue so, yet the dividing line becomes fuzzy. Whether an output of an ML model counts as an accommodation or as a prediction hinges upon the relationship between the patterns the input instantiate and those on which the model was selected through its training. Once this relation is taken into account, most successful outputs of ML models, we suggest, bear a strong accommodation character. We further trace the consequences of our account for confirmation, novelty, and scientific realism.
Read the full abstract

Analysis of PDEs

Persistence in weighted Sobolev spaces for the Calogero-Moser derivative nonlinear Schrödinger equation

This paper is devoted to the study of invariant weighted Sobolev spaces for the Calogero–Moser derivative nonlinear Schrödinger (CMDNLS) equation on the real line. For \(s\geq 1\) and \(0\leq r\leq s\), we first prove that the \(H^s(\mathbb R)\) flow preserves the condition \((1+x^2)^{r/2}u\in L^2(\mathbb R)\). The proof uses truncated weighted energy estimates and weighted interpolation. We then establish the persistence of weighted regularity for spatial derivatives. For integers \(p\geq 3\) and \(1\leq m\leq p-2\), and for \(0\leq r\leq p-m \), the \(H^p(\mathbb R)\) flow preserves the conditions \((1+x^2)^{r/2}\partial_x^j u\in L^2(\mathbb R)\) for \(0\leq j\leq m\). This result combines weighted energy estimates, the unitary equivalence and the integrable structure of the CMDNLS equation.
Read the full abstract

Combinatorics

Cut Conditions, Integrality, and Obstructions for Directed Multiflows

In contrast to the max-flow/min-cut theorem for single-commodity flows, the cut condition is necessary but not sufficient for the existence of directed multiflows, even with only two commodities. Poremba and Shepherd (IPCO 2023) conjecture that the cut condition is sufficient for all capacities and demands on a directed two-commodity topology if and only if it contains neither of two specific topologies as a relevant minor. We prove this conjecture. As a consequence, the fractional and integral flow-cut gaps of every directed two-commodity topology coincide and are either 1 or 2. We then investigate stronger necessary conditions for multiflow feasibility and their implications for integrality. We introduce a hierarchy of five conditions, ranging from the classical cut condition through conditions based on blocking edge sets and binary edge lengths to the linear programming dual characterization of fractional feasibility. The corresponding classes of sufficient topologies are strictly nested. We extend the Nagamochi-Ibaraki theorem: whenever one of the first four conditions is sufficient on a topology, integral capacities and demands satisfying that condition admit an integral multiflow, computable in polynomial time. Finally, we prove that recognizing topologies on which any of these four conditions is sufficient is coNP-complete, even for two commodities.
Read the full abstract

Combinatorics

Piercing a Chessboard with Restricted Slopes

Is it true that for $n\geq 3$, $n-1$ lines are needed to pierce every cell of an $n\times n$ board? This problem was independently rediscovered several times: Alkauskas (1998), Harel-Ackerman-Pinchasi (2012, without claiming the $n-1$ bound), Bárány-Frankl (2021). However, apparently the first to pose it was T. Keller (Honolulu, Hawaii), in 1980, in the problem section of The American Mathematical Monthly. In this note, we give an affirmative answer to Keller's problem when all lines have slopes in $(-\infty,-1]\cup(0,1]$.
Read the full abstract

Statistical Mechanics

The role of preparation protocols in the Mpemba effect under time-delayed Newton cooling

The Mpemba effect depends on both the relaxation dynamics and the protocol used to prepare the initial states. Here, this dependence is investigated using the time-delayed Newton cooling law, comparing protocols in which two samples are exposed to piecewise-constant bath temperatures before being placed in a common final bath. For protocols with at most two quenches and a common waiting time, exact conditions for the occurrence of the effect are derived. They rule out the Pontus protocol within this model and show that the synchronous two-reservoir protocol maximizes the effect magnitude within the class considered. The robustness of this optimum is then examined by allowing asynchronous quenches or an additional quench for one sample. Numerical optimization of these extensions identifies the synchronous two-reservoir protocol as the maximizing limit in the cases studied. This evidence motivates a conjecture that the same protocol remains optimal for more general bath histories, although that broader statement is not proved.
Read the full abstract

Dynamical Systems

Singular bifurcations in implicit discrete dynamical systems

A periodic orbit of an implicit discrete system can persist when the implicit function theorem no longer guarantees a smooth successor. We develop a joint Lyapunov-Schmidt reduction of the periodic equations and the determinants marking singular transitions. In unrestricted smooth one-parameter families, the generic event is a single evolution fold on a continued orbit. Simultaneous normal forms describe its interactions with saddle-node and period-doubling bifurcations, locating the singular transitions and, under backward regularity, determining the diverging multipliers. The results cover arbitrary periods and periodically non-autonomous systems. Exact realisations and applications to labour-market equilibrium, machining and implicit integration illustrate the theory.
Read the full abstract

Systems and Control

Tensor Network Kernel Machines for Nonlinear System Identification

Over the past ten years Tensor Network Kernel Machines have enabled efficient nonlinear system identification through the use of low-rank tensor networks. The goal of this article is to provide the reader with a basic understanding of this model class in terms of its two defining properties. Furthermore, both the optimization and regularization of such models are discussed, together with recent developments and starting points for further exploration.
Read the full abstract

Cryptography and Security

An Explicit Below Half Threshold Bound for the Local Leakage Resilience of Shamir's Secret Sharing Scheme

As a fundamental cryptographic primitive, secret sharing distributes a secret among multiple parties and allows certain authorized parties to recover the secret while preventing unauthorized parties from knowing the secret. However, local leakage from each party member allows an adversary to learn partial information from each share, which is not guaranteed by traditional security. In this paper, we study bounds on the reconstruction threshold of Shamir's secret sharing scheme over prime fields against arbitrary 1-bit local leakage. We show that, for all sufficiently large $n$ and every prime $p>n$, the $(n,t)$-Shamir's scheme is $\exp(-\Omega(n))$-resilient whenever $t/n\geq 0.489$, for any choice of distinct nonzero evaluation points. Our proof refines the dual-code list recovery analysis and combines complementary Fourier estimates according to the support sizes and biases of the leakage functions, yielding an improved explicit reconstruction threshold bound.
Read the full abstract

Machine Learning

How Many Exposures Does a World Model Need? Learning Unseen Game Mechanics at Test Time

World models enable agents to plan and learn from imagined experience, but in an unfamiliar environment they must learn its mechanics from limited interaction. A mechanic is a rule for how an action or an interaction changes part of the environment. Each occurrence of a mechanic that the model observes is one exposure. We ask how many exposures a world model needs to predict the effects of a new environment's mechanics. We address this question on ARC-AGI-3, a benchmark of interactive games defined by their programs. We introduce MechEval, which labels 69 recurring mechanics across 20 games from these programs. It scores each prediction on the cells the mechanic changes at each exposure. As a test bed, we build WARC (World model for ARC-AGI-3), a 35M-parameter vision transformer that is pretrained on other games and updates all of its parameters after each transition. On held-out games, WARC learns to predict four in five mechanics with at least 80% accuracy on the cells they change and keeps that accuracy to the end of the recordings. Half of them reach that accuracy within about 80 exposures, in a single play of the game, and without updates only one in seven does at all. In closed-loop play on the public games, WARC-Agent, which pairs WARC with a fixed graph explorer, triples the explorer's official efficiency score from 0.15 to 0.46, mostly through test-time training. On the hidden games of the Kaggle leaderboard, our entry scores 0.20.
Read the full abstract

All authors Ku Onoda, Haruto Kawano, Shota Takashiro, Kohei Hayashi

Dynamical Systems

Lower intermediate dimensions of classical Denjoy minimal sets

We study the $\theta$-intermediate dimension of the minimal set of the classical $C^{1+\delta}$ Denjoy map, an orientation-preserving diffeomorphism of the circle not conjugate to a rigid rotation. This follows on from work of Kra and Schmeling, who showed how the box and Hausdorff dimensions may differ, depending on the Diophantine properties of the rotation number. The $\theta$-intermediate dimension interpolates between these two dimensions. We use the well-understood gap structure of the irrational rotation map to develop a covering strategy to get upper bounds on the lower $\theta$-intermediate dimension, and use results of Banaji to show that these upper bounds are sharp for $\theta$ in an interval.
Read the full abstract

Materials Science

Ultrafast Inter-Aggregate Energy Transfer Drives Efficient Charge Generation in Organic Homojunctions

Y6-based homojunction organic photovoltaic devices have recently achieved exceptional power conversion efficiencies, surpassing those of conventional organic photovoltaic materials. Although unusually long exciton diffusion lengths and efficient exciton dissociation have been reported, the fundamental mechanism underlying this outstanding performance remains under debate. Here, we report that structurally distinct aggregates in Y6 thin films exhibit complementary charge-transfer (CT) characters that enable directional energy transfer and enhanced exciton delocalization. Using ultrafast spectroscopy, we identify highly efficient Förster resonance energy transfer (FRET) between Y6 aggregates, with an ensemble-averaged efficiency of 65% occurring on a sub-picosecond timescale (<1 ps). This aggregate-mediated energy-transfer pathway drives long-range energy migration in Y6 thin films and accounts for the remarkable external quantum efficiencies observed in homojunction devices, even in the absence of conventional donor-acceptor bulk heterojunction morphologies and despite substantial fractions of weak-CT components. Here, using Y6 as a model non-fullerene acceptor (NFA) system, we show that structurally distinct aggregates coexisting in NFA thin films possess complementary CT characters, which are expected to enable directional energy transfer and enhanced exciton delocalization. We establish a general multi-component FRET framework that quantitatively links molecular aggregation, CT character, and excited-state dynamics in systems with multiple packing motifs. This work provides insight into the development of next-generation high-performance organic electronic materials and devices.
Read the full abstract

All authors Shanchao Ouyang, Hang Yuen Ho, Sudhi Mahadevan, Ziyao Yue, Haochen Liu, Zhiqiang Guan, Chun-Sing Lee, Shu Kong So, Sai-Wing Tsang