Complex Variables
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Tuesday 10 November
Let $E$ be a regular compact set with connected complement, and let $f$ be holomorphic in a neighborhood of $E$. We establish Hadamard-type asymptotic formulas for products of errors of best uniform rational approximation of $f$ on $E$, formed from the first $m+1$ rows of the Walsh table. When the corresponding meromorphy radii are finite, we also show that the extremal exponential rates of the approximation errors along these rows are attained along a common subsequence. We also prove a sharp asymptotic comparison theorem for ratios of best rational approximation errors along fixed rows of the Walsh table on $E$ and its closed Green sublevel sets, and derive corresponding sharp asymptotic estimates for ratios of their products. These results extend earlier product theorems from continua with connected complement and Jordan boundary to the more general setting of regular compact sets with connected complement.
Nuclear Theory
Coherent control of quadrupolar order in solid spin-1 systems:\ majorization bounds, orientation-indexed adiabatic inversion, and order lifetime
Tensor polarization $\Pzz=1-3p_0$ of spin-1 nuclei in dynamically polarized solid targets is limited, under a single spin temperature, to $\Pzz^{\rm eq}=2-\sqrt{4-3P^2}$ ($\Pzz\simeq0.20$ at vector polarization $P=0.5$). Radio-frequency (rf) semi-saturation of one quadrupolar Pake-doublet branch, with and without slow sample rotation, has reached $\Pzz\simeq0.36$. Using the first-order quadrupole Hamiltonian, spin temperatures of the Zeeman and alignment (quadrupolar-order) reservoirs, and single-transition operators, we analyze these manipulations. Five results follow. (i) Under unital operations (rf pulses, rotation, saturation), $1-3\lambda_{\max}\le\Pzz\le1-3\lambda_{\min}$, where $\lambda_{\min,\max}$ are the smallest and largest initial populations. Saturation achieves at most half the available transfer; selective $\pi$ inversion reaches either bound ($\Pzz=-0.85,+0.65$ at $P=0.5$). (ii) In a powder, the two single-quantum transitions overlap except in the pedestal region. Adiabatic frequency sweeps confined there invert one transition per bond. Inverting each bond exactly once requires packing antipodal caps on the orientation sphere. The optimal single-axis arrangement has four stops at the magic angle $54.74\degree$ (cube body diagonals; 73% coverage); six icosahedral axes give the global optimum (90%). (iii) Rotation-induced Landau–Zener crossings of deuteron pairs are adiabatic at current rates and scramble the inversion pattern; transits must be fast for the strongest, geminal CD$_2$, pairs. (iv) A parity theorem determines the inverted bonds solely from the endpoints of continuous adiabatic passage: rf during rotation yields at most two caps, and a full turn none. (v) The enhanced state decays by dipolar flip-flops, with $T_{1Q}\le T_{1Z}/3$ and a duty-cycle break-even of $\simeq35$~min. Magic-angle spinning cannot replace this scheme.
Read the full abstract
Category Theory
Pre-Model Structures on the Category of $C_p$-Sets
It is a folk theorem of Goodwillie that there are exactly 9 model structures on the category of sets, coming from 18 pre-model structures. We investigate model structures and pre-model structures on the category of $G$-sets for a finite group $G$. We outline a strategy for finding all model structures in this category, and partially execute this strategy for the case of $C_p$-sets, where $C_p$ denotes a cyclic group of prime order $p$. In particular, there are exactly 1777 pre-model structures on the category of $C_p$-sets.
Read the full abstract
Cryptography and Security
Stinging the Swarm: Extracting Attacker Intelligence from a Twin-Controller OT Honeypot
Autonomous language-model agents can drive industrial controllers over fieldbus protocols, probing a plant at machine speed and in coordinated swarms. We present Talos, an inline gate that judges every agent request to a controller in milliseconds and routes agents it flags as hostile to a device-faithful decoy – a mirror twin of the real plant – where their attack continues harmlessly and is read for intelligence. We evaluate the full pipeline (detect -> quarantine -> mirror -> extract) against 2,400 autonomous attacker agents, 600 in each of four tiers that escalate the attacker's effort to stay hidden, from no concealment to an explicit non-disclosure rule and a trained cover story. No write reached the real plant in any tier: the gate denies the damage path and a device-level envelope check backstops it, while unmapped registers are denied without exception and the gate still permits safe-direction stop commands. Decisions are model-driven in every tier with no inference timeouts, closing a common objection to small on-device judges. Beyond blocking, the pipeline reads the attacker. We find a behavioral structure along two axes – how persistently an agent attacks and how much it discloses – and show that the system has a capture channel for three of the four resulting regions: hardened, silent attackers escalate most and are read on the mirror; compliant, talkative attackers volunteer their identity to a diagnostic form before they are ever quarantined; only the paranoid, low-yield region resists both. Attacker identity fields are self-authored cover stories, but involuntary tells – the self-reported toolchain fingerprint and the attacker's own list of targeted registers – accumulate across the swarm into a high-confidence profile drawn from agents under explicit orders to stay silent. The gate, mirror, and harness are released under Apache 2.0 with a synthetic device family.
Read the full abstract
High Energy Physics (Phenomenology)
Thermal $CP$ dynamics in a four-fermion model: mean-field and functional renormalization group analyses
We investigate a one-flavor four-fermion model with scalar, pseudoscalar, and mixed scalar-pseudoscalar interactions at finite temperature and vanishing chemical potential. Our analysis combines the mean-field approximation, which determines the equilibrium condensates, with the functional renormalization group (FRG), which describes the scale dependence of the interactions and the fermion mass phase. Within the mean-field approach, we find that, in the absence of explicit symmetry-breaking sources, the dominant interaction determines the direction of condensation. In particular, a pseudoscalar condensate spontaneously breaks the ordinary $P/CP$ symmetry and disappears through a continuous thermal phase transition. For regular high-temperature FRG trajectories, the dimensionless mixed coupling tends to zero in the infrared, while its dimensionful counterpart remains finite. Increasing temperature also suppresses the scale evolution of the fermion mass phase. Together, these results clarify the distinct roles of thermal fluctuations in suppressing spontaneous symmetry breaking and fermionic instabilities, without necessarily eliminating explicit symmetry breaking effects.
Read the full abstract
Physics Education
AI is ready, but are the students. From answer-making to sense-making
Students in introductory physics often struggle with interpreting kinematics graphs, especially when connecting graphical features to physical meaning. We investigated AI tool use in kinematic graphing analysis as an intervention to support active sensemaking. This study examines GPT-5 performance on the Test of Understanding Graphs in Kinematics (TUG-K) and the way students interacted with GPT-5 during introductory physics activities involving graphical analysis. For the first part of the study, GPT-5 responses were evaluated across repeated TUG-K trials and compared to previously reported GPT-4 performance. Following that analysis, student-GPT-5 dialogue episodes were collected and analyzed using an inductive coding scheme. GPT-5 performance results show markedly improved accuracy. Student-GPT-5 dialogue episodes evolved from mostly procedural answer-making requests early in the term, to more reflective sensemaking interactions at the end of the term, suggesting that instructional framing is needed for AI tools to function as meaningful peer-like learning supports.
Read the full abstract
High Energy Physics (Phenomenology)
Two-loop differential cross section for $l^{+}l^{-}\rightarrow\gamma\gamma$ with power corrections in the lepton mass
We calculate the two-loop QED amplitude for $l^{+}l^{-}\to\gamma\gamma$ and the corresponding unpolarized differential cross section in the high-energy region, retaining power corrections in the lepton mass. For both this process and the inverse process $\gamma\gamma\to l^{+}l^{-}$, these corrections can be important for accurate predictions when the lepton mass cannot be neglected, particularly in the near-threshold region. The amplitude is presented in a tensor basis and can therefore be evaluated for arbitrary polarizations of the external particles. All results are expressed in terms of classical polylogarithms $\mathrm{Li}_n$ for efficient numerical evaluation. Numerical tests indicate that the truncated expansions remain accurate for $s,|t|,|u|\gtrsim40m_l^2$. At leading power in the lepton mass, the two-loop amplitude agrees with the prediction of QED mass factorization.
Read the full abstract
Artificial Intelligence
Living constitutions: Participatory alignment for sovereign foundation models
In Constitutional AI, a plain-text "constitution" encodes values and gets incorporated into a model via post-training. However, the procedure for creating such a constitution poses significant challenges: who contributes, who decides which norms are included, and to what extent must the process be participatory or democratic? These challenges can be paralyzing. Yet, we argue that taking action is essential: without a deliberate constitution, sovereign models would risk misalignment or default to norms set by frontier model providers. We propose treating the development of a constitution as an evolving, participatory process akin to citizen science, iteratively drafted, tested, and refined. We therefore formulate five desirable principles: iterative, modular, justifiable, positional, and legitimate. We further argue that such a document must remain responsive not only over time but across the people it governs, and propose that provisions be marked as settled, contested, or underdetermined, so that durable disagreement is recorded in the constitution rather than resolved away. Empirically, we show via agent-based modelling that proposed process for maintaining living constitutions is robust against 50% bad actors among citizens - assuming a share of 75% participants with good intent in the deliberative body. This indicates that the proposed process tolerates a substantial share of adversarial participants.
Read the full abstract
Software Engineering
SPARK: Semantic-preserving Prompt Mutation for Code Generation using Consensus-based Verification Framework
Automated code generation with large language models (LLMs) has improved in recent years. With the lack of test cases in real-world usages, generated code is not automatically verifiable and often requires manual validation. Previous studies have leveraged consensus verification, where code solutions and corresponding test cases are generated and evaluated based on functional consistency. Building on this, SPARK incorporates semantic-preserving mutations to problem descriptions such as paraphrasing, grammatical adjustments, function renaming, and other alterations to enhance the robustness and correctness of LLM-generated code. For each mutated description, SPARK generates a set of code solutions and test cases. These candidate solutions are evaluated using consensus-based verification, where the solutions' quality is measured by how consistently they pass generated tests. We assess the effectiveness of SPARK on three benchmarks (LiveCodeBench, BigCodeBench, and DLBench) across three LLMs (o4-mini, DeepSeek-R1-Basic, and Gemini-2.5-Pro). Our evaluation demonstrates that SPARK outperforms state-of-the-art consensus-based verification approaches by up to 4.1%. Our generalization experiment highlights the practical potential of SPARK as an effective strategy for enhancing the reliability and correctness of LLM-based code generation systems in scenarios where traditional validation resources are scarce, while our qualitative analysis shows insight into how SPARK's mutation helps LLMs generate better code solutions.
Read the full abstract
Statistics Applications
Orthographic depth at two scales, with English also matched for lexicon size
A reader can struggle to read in only one of their languages, and one proposed reason is how predictably spelling maps to sound. We measure that predictability as conditional entropy, the pronunciation uncertainty left once the spelling is seen, at two grain sizes (unit sizes): the alignment unit, an aligned spelling-sound chunk, and the syllable, bounded in the pronunciation. We analyse word lists in twelve alphabetic orthographies and report five in full, with English pronunciations from a dictionary and the rest from the synthesizer eSpeak-ng, length marks removed. Among the five, English has the most uncertainty and Finnish the least at both grains. Across all twelve, Portuguese and Danish exceed English per alignment unit, and German and Italian order differently by grain and transcription. These orders hold when English is transcribed by eSpeak-ng with u allowed to spell /j/ plus a vowel. Uncertainty per unit has no shared direction as units grow: it falls in Spanish and, less securely, German, is level in Finnish, depends on transcription in Italian and on the word list in English, and rises in Norwegian and Swedish. Summed uncertainty per word falls in all twelve, but that score is handed the pronunciation's boundaries and mixes context with in-sample memorisation, so it is no evidence of what a reader computes. An exact identity states when discarded spelling boundaries could reverse that fall. A uniform 20,000-word English sample holds 35% of the dictionary's syllable spellings and has syllable entropy 0.098 bits lower, so lexicons must be matched in size. In an exploratory check over eleven languages, syllable-scale entropy tracked first-year reading accuracy (rho = -0.74, falling to -0.25 if English is given the lowest entropy instead), without separating the syllable scale from the alignment unit. These estimates describe sampled writing systems, not readers.
Read the full abstract
Fluid Dynamics
Lift-element-based dynamic network of vortex gust-airfoil interactions
We construct a lift-element-based (LE) network for analyzing extreme vortex gust-airfoil interactions. The edge weights quantify the magnitude of changes in local lift elements induced by vorticity perturbations. Network broadcast mode analysis then extracts forcing and response modes that characterize input-output pathways over the evolving network. We consider three-dimensional flows around a NACA 0012 airfoil with a Taylor vortex gust at a Reynolds number of 1000 and compare the LE modes with those of a Biot–Savart (BS) network. The leading BS modes highlight the vortical structures formed through the gust-airfoil interactions, whereas the leading LE modes connect the gust impingement on the airfoil to the shear layer thickened by the gust, which lies on the pressure or suction side depending on the gust condition. Perturbations along the leading BS forcing modes amplify or attenuate the gust-induced lift variation, whereas those along the leading LE forcing modes yield a sustained lift increase or decrease. These input-output mode pairs, representing cause-and-effect relationships within the dynamic network, can guide the selection of actuation locations and timing for manipulating transient aerodynamic loads.
Read the full abstract
Quantitative Methods
The Dose Makes the Poison: Exposure-Aware Evaluation of Generative Drug Design
Generative models are increasingly used to propose molecules for drug discovery. They can produce far more candidates than can be experimentally tested, making computational selection essential. Most selection methods rely on predicted molecular properties, such as potency and toxicity, but overlook how these properties translate into drug exposure at a given dose. As a result, molecules that appear promising on paper may require impractical doses or raise safety concerns. We investigate this gap by using physiologically based pharmacokinetic (PBPK) simulations to re-evaluate molecules selected from three PARP1-targeted generative pipelines. Of 60 shortlisted molecules, seven fail to achieve sufficient target exposure, while 28% of the remaining candidates show predicted safety risks at their required dose. Moreover, 85% of candidates with an achievable dose are outperformed by molecules overlooked during static selection. These findings suggest that accounting for dose and exposure can substantially change which generated molecules are considered promising for drug development.
Read the full abstract
Information Theory
The Optimal Constant for One-Shot Channel Simulation and Strong Functional Representation Lemma
We prove a tight bound for one-shot channel simulation via a stronger version of the strong functional representation lemma: for any random variables $X,Y$, there exists a random variable $Z$ with $I(Z;X)=H(Y|X,Z)=0$ and $H(Y|Z)\le I+\log I+(1-\gamma)\log e-\log\log e+O(I^{-1}\log I)$, where $I=I(X;Y)$ and $\gamma$ is the Euler-Mascheroni constant. A more convenient bound is $H(Y|Z)<I+\log(I+2)+1$. The first bound is notable in the sense that the constant term $(1-\gamma)\log e-\log\log e$ is optimal, and cannot be improved for any bound stated only in terms of $I$. Hence, this is the tightest form of the strong functional representation lemma up to the third order. We also prove that the Poisson functional representation is almost optimal up to a constant gap, and provide a simple finite-blocklength achievability bound that implies the $(1/2)\log n$ redundancy of the simulation of memoryless non-singular channel. These are proved via a generalization of the functional information, in particular, the exponential functional information, which closely approximates the cost of one-shot channel simulation.
Read the full abstract
Software Engineering
The AI Era's Lines of Code? A Case Study on What an Engineer's Token Spend Measures
Organizations use engineers' token spend to monitor AI adoption, set budgets, and measure productivity, but the metric's relationship with engineering output is unclear. We link five weeks of token usage in coding agents to the commits of 404 engineers at JetBrains, and analyze a company survey of 412 employees. We find that token spend varied mostly between engineers, and which engineers spent the most stayed largely the same over the five weeks. Among weeks with similar commit counts, the 90th percentile of spend was 39 to 240 times the 10th. Within an engineer, doubling spend was associated with about 10% more commits. Available work metrics explained 38% of the variation in spend between engineers and 13% of the week-to-week variation within an engineer. In the survey, respondents in technical roles reported in nearly nine of ten agent evaluations using agents for tasks beyond coding, such as analysis, research, and planning. Token spend therefore measures how much engineers use agents across many kinds of work and not the outcome and value of that work. We recommend using token spend at most for resource accounting, and evaluating agent-assisted work with evidence about tasks and outcomes.
Read the full abstract
Software Engineering
A Taxonomy of Failing Bug Reproduction Tests on SWT-Bench
Automatic Program Repair (APR) tools require high-quality test generation models to validate candidate patches. Despite the advancements of automated bug reproduction testing tools, they continue to struggle with issues of higher complexity. These struggles may be caused by a framework's lack of problem-specific context. We address the unclassified nature of these framework failures by providing a taxonomy of bug reproduction tests. This study uses the AssertFlip and OpenHands frameworks, the two highest-scoring open-source tools on SWT-Bench Verified. The significance of model selection on the causes of failure is also explored by utilizing GPT-5-mini and GPT-4o-mini on both tools. The error traces of these frameworks are categorized to determine granular modes of failure. Through this process, mechanical and implementation failures were the most common failure classes. Overall, a tool's generation approach has a significant impact on precision and coverage, while the selection of the model alters the quality and quantity of generated tests. Specifically, more powerful models with higher reasoning capability tended to present fewer failures due to a lack of technical or problem-specific context. Our experiment also shows evidence that our taxonomy can be used as concrete guidance to augment the issue description to improve the generation quality.
Read the full abstract
Combinatorics
Edge-coloring spectra of odd-fold graph covers and counterexamples to a conjecture
For a loopless multigraph $G$ and an odd integer $\ell\ge3$, let $\Delta(G)$ be the maximum degree of $G$, let $s_\ell(G)$ be the minimum chromatic index of a cyclic $\ell$-fold cover of $G$, and let \begin{align*} L_\ell(G):=\max\left\{ \Delta(G),\ \max_{\substack{S\subseteq V(G)\\ |S|\ge3\text{ odd}}} \left\lceil\frac{2\ell|E(G[S])|}{\ell|S|-1}\right\rceil \right\}. \end{align*} In this paper, we prove the sharp upper bound \begin{align*} s_\ell(G)\le\max\{\Delta(G)+1,L_\ell(G)\}. \end{align*} For every nonempty bridgeless cubic graph $H$, we construct two $3\ell$-regular graphs, one a multigraph and the other a simple graph. In each case, the minimum chromatic indices over cyclic and over all $\ell$-fold covers both equal the chromatic index of the multigraph obtained by replacing each edge of $H$ with $\ell$ parallel edges. Taking $H$ to be the Petersen graph gives both multigraph and simple graph counterexamples to a conjecture of Chen, van der Holst, Luo, and Ma, which asserts that $s_\ell(G)=L_\ell(G)$ always holds. When $G$ is connected and $L_\ell(G)>\Delta(G)$, we determine the common spectrum of its connected cyclic and arbitrary $\ell$-fold covers. We give polynomial-time algorithms to compute $L_\ell(G)$ and the spectrum endpoints. For fixed $\ell$, we also construct in polynomial time a connected cyclic cover with any prescribed chromatic index in the spectrum.
Read the full abstract
Distributed Computing
Plumbline: When the Loss Lies, Validating LLM Training Correctly
Correctness is a prerequisite for training large language models (LLMs) at scale, yet it is hard to ensure: a fault can sit in any part of a run, and how it shows differs across frameworks and the libraries they depend on. Some failures are loud and some are silent, and both take time to detect and trace. A single fault can waste a long run: the run crashes late, or the model scores lower only at the final evaluation. We present Plumbline, a correctness validation system for LLM training, used in production at Amazon. Plumbline checks each part of the training lifecycle, from the data a run trains on to the model it serves, down to the new layers each model brings and up to whole runs. Its diagnosis reuses these checks to trace failures to their root causes. Both rest on a systematic correctness methodology designed to carry across models and frameworks. Plumbline supports two agentic use cases: Day-0 onboarding, which validates the training of a newly released model before teams rely on it, and root-cause tracing of training failures. Across 3 frameworks and 5 models, it found 61 problems, 32 of them silent. Of these, 42 are fixed or reported upstream, 8 of them first raised by us. We further group the 61 into seven failure patterns that show where training breaks, and report how problems recur across frameworks and models.
Read the full abstract
Data Structures and Algorithms
Minimum Membership Dominating Set and Its Generalization: Leafage vs. Asteroidal Number
The Dominating Set problem is known to be fixed-parameter tractable when parameterized by leafage, a structural graph parameter defined on chordal graphs. We investigate whether this tractability extends to its variants with natural constraints. Specifically, we study the Minimum Membership Dominating Set problem and its generalization, the Min. $(\sigma, \rho)$-Dominating Set problem. Given a graph $G$ and a positive integer $\ell$, Minimum Membership Dominating Set asks whether there is a dominating set $D$ such that every vertex in $G$ is dominated by at most $\ell$ vertices in $D$. We first show a negative result: Minimum Membership Dominating Set parameterized by leafage plus $\ell$ is W[1]-hard, demonstrating that even this natural constraint serves as a severe barrier to designing FPT algorithms. To complement this, we present an XP algorithm for the Min. $(\sigma, \rho)$-Dominating Set problem parameterized by asteroidal number plus $\mu$, where $\sigma$ and $\rho$ are finite sets of non-negative integers and $\mu$ is the maximum value in $\sigma \cup \rho$. Note that asteroidal number generalizes leafage to arbitrary graphs. Combining this algorithm with our structural bound that every graph $G$ admits a dominating set such that no vertex is dominated by more than $7\mathsf{an}(G)-6$ vertices, where $\mathsf{an}(G)$ denotes the asteroidal number of $G$, we show that Minimum Membership Dominating Set admits an XP algorithm parameterized solely by asteroidal number.
Read the full abstract
High Energy Astrophysical Phenomena
Flux Variation of the Crab Pulsar Observed Using Fermi LAT and NuSTAR
The flux of pulsed emission from the Crab pulsar, just like most of other pulsars, is considered to be steady over time scales much larger than the pulsation period, except for a smooth, secular decrease related to the aging of pulsars. Such a steady assumption prevails in all theoretical investigations of pulsar emission models. However, here we report the discovery of stochastic variation in the flux and photon index of the Crab pulsar gamma-ray and hard X-ray emissions utilizing archival data of Fermi LAT (0.1 – 100 GeV, spanning 17 years) and of NuSTAR (3.0 – 79.0 keV, spanning 13 years) observations. Furthermore, the flux variation is found to be anti-correlated with variation of the spectral photon index, that is, harder when brighter, in both gamma ray and hard X-ray bands. The origin of such variation is intriguing. Some possibilities are discussed.
Read the full abstract
Logic
The unavoidable minor in graphs of uncountable edge-connectivity
For any given uncountable cardinal $\kappa$, we show that every $\kappa$-edge-connected graph admits a minor embedding of the graph $T_{\kappa}\star1$ (the $\kappa$-regular tree with a new apex vertex adjacent to all its nodes), which is itself $\kappa$-edge-connected. This extends a previous result of Kurkofka regarding unavoidable minors of $\aleph_0$-edge-connected graphs to all uncountable cardinalities $\kappa$.
Read the full abstract
Machine Learning
AdjointFlow: Accelerating Scientific Inverse Problems with Physics-Based Transport
Inverse problems are ubiquitous across scientific applications ranging from seismic imaging to astrophysics to nondestructive testing, yet classical solvers often struggle with nonconvexity and repeated evaluations of expensive physical models. Diffusion models can amortize this computation offline, but largely data-driven conditional models underutilize known physics and suffer from ill-conditioned transport maps when the target distribution is concentrated. Plug-and-play (PnP) methods explicitly leverage the physical model to guide a learned prior, improving data consistency but reintroducing costly iterative computations. We instead propose AdjointFlow, a framework built on stochastic interpolants that injects the known physics into the generative transport itself. We design a data-dependent, adjoint-based initialization to reduce transport distance to the target, modify the transport velocity to remain regular as the endpoint distributions concentrate, and parameterize the network to build the initialization directly into the learned velocity. Together, these enable accurate reconstruction at substantially lower cost than generative models starting from Gaussian noise. Experiments on inverse scattering, seismic imaging, and black hole imaging show that AdjointFlow consistently produces more accurate reconstructions than conditional diffusion models while being up to five orders of magnitude faster at inference than PnP methods. Optional inference-time refinement that queries the physical model can further improve data consistency, matching or exceeding the accuracy of PnP methods without retraining.
Read the full abstract
Artificial Intelligence
Online-Adaptive Fuzzy Cognitive Maps for Neuro-symbolic Financial Forecasting
We present a neuro-symbolic framework that pairs online-adaptive Fuzzy Cognitive Maps (FCMs) with long short-term memory (LSTM) networks for financial forecasting. Financial markets exhibit regime shifts and nonlinear dynamics. Existing models often trade off one of these key properties: these models capture temporal patterns but hide causal drivers, or they recover causal graphs that cannot adapt post-training. We close this gap with two frameworks. Framework I constructs an FCM from live dual-source price series and macroeconomic indicators, and adapts its edges online through an error-driven Hebbian rule. We introduce the Asymmetric Additive Residual Fusion method to seamlessly combine the FCM and LSTM predictions under Framework I. Framework II constructs a literature-grounded, LLM-augmented FCM where every edge carries a verifiable citation that gives an auditable map of factor interactions. On NASDAQ Composite data, Framework I matched the performance of the LSTM baseline on state-directional accuracy and outperformed LSTM by 7.6 percentage points on the return-sign accuracy. Framework I outperformed standalone FCM models across both evaluation metrics. The adaptive maps converged to new regimes within 10-15 trading days, and Framework II center-node FCMs scored 68.9% to 85.7% on the state-directional accuracy.
Read the full abstract
Human-Computer Interaction
Towards Accessible, Autonomous Photosynthesis Monitoring on Living Leaves
Photosynthesis is a direct biomarker of plant health and functioning. However, measuring net photosynthesis on individual leaves requires expensive (USD 42,000 - USD 89,000) instruments and manual effort. This limits access to photosynthesis sensing and makes long-term monitoring across multiple leaves difficult. We present a 1000x low-cost, autonomous sensing system that enables longitudinal monitoring of photosynthesis on living leaves over multiple days. Guided by a formative user study with 13 plant biology researchers, the system integrates an automated, leaf-attached chamber with a low-cost CO2 sensor. Built using a 3D-printed clamp and thermoformed dome, the chamber periodically closes to measure CO2 drawdown and reopens to refresh the air. Each chamber costs about USD 62 and operates independently, allowing multiple leaves to be monitored simultaneously without pumps or gas tubing. We characterize each chamber and validate its measurements against a research-grade gas exchange instrument (CIRAS-4) across 29 sessions (R^2 = 0.90). We also demonstrate monitoring of C3 and CAM photosynthesis plants, adaptability to different leaf geometries, and a light-response experiment. In a nine-day deployment, six chambers captured simultaneous CO2 exchange from individual leaves of the same plant during drought and rewatering. Each leaf showed different patterns of nighttime CO2 uptake and recovery, which a whole-plant CO2 measurement instrument would obscure. To our knowledge, this is the first demonstration of such leaf-level photosynthesis dynamics using independent, low-cost autonomous chambers. Our work opens new opportunities for plant science, education, and data-driven agriculture.
Read the full abstract
Human-Computer Interaction
Designing Accessible and Rigorous ML Experiments: No-Code Visual Workflows and Structured Guidance for Beginners
Designing machine learning (ML) experiments requires beginners to plan a methodologically sound workflow and implement it in a concrete environment. We evaluated two forms of support in two within-participant studies with 18 predominantly technically trained participants: mlguide, which provides structured, evidence- backed guidance for workflow design, and AI Sandbox, a visual no-code environment for workflow implementation. After using mlguide, participants produced better high-level workflow designs, but their scores on a quiz about specific methodological choices did not improve significantly. Compared with Google Colab, AI Sandbox received higher hedonic user experience ratings, while perceived workload and acceptance did not differ significantly. These findings suggest that different forms of interactive support help with differ- ent aspects of beginner ML experimentation. The paper contributes a reusable evaluation task with instructions, a modified dataset, and a quiz, as well as empirical evidence on structured guidance for workflow design and on the experience of implementing workflows in visual and notebook-based environments.
Read the full abstract
Combinatorics
Cluster-theoretic Newton–Okounkov polytopes of flag varieties of types $A_4$ and $B_2$
Newton–Okounkov bodies have been used for studying toric degenerations of flag varieties. We classify the Newton–Okounkov polytopes associated with cluster valuations and the anticanonical line bundle on full flag varieties of types $A_4$ and $B_2$. In type $A_4$, the $672$ seeds of cluster type $D_6$ yield $162$ affine unimodular equivalence classes, which coincide with the combinatorial equivalence classes. We determine the subgroup of exchange-graph automorphisms preserving every polytope class. It is a Klein four group of index six in the automorphism group of the exchange graph of cluster type $D_6$. In type $B_2$, we give defining inequalities for all six seeds. At the anticanonical weight, the two combinatorial equivalence classes split into three affine unimodular equivalence classes. Their respective class-preserving automorphism groups have orders four and two.
Read the full abstract
Computer Vision
How Hyperbolic Are Hyperbolic Vision-Language Models?
Hyperbolic vision-language models (VLMs) have emerged as a rapidly growing area of research, reporting strong zero-shot performance alongside an interpretable embedding space in which radius orders representations by generality. But how hyperbolic are hyperbolic VLMs in practice? We investigate this at the exponential map, the single operation through which curvature enters an otherwise Euclidean network. We show that across MERU, HyCoCLIP, ARGENT, and PHyCLIP, the exponential map is well approximated by the identity function, i.e., pairwise distances and angles in the pre-map Euclidean space barely change under projection, and removing the map at inference leaves evaluation scores unchanged. Motivated by this observation, we train Euclidean VLMs on CC3M from scratch under the same hierarchy losses. Over roughly the final 90% of the training steps, the training statistics are nearly indistinguishable from those of their hyperbolic counterparts, and the resulting models match the evaluation scores and recover the same radial distributions. Our findings establish Euclidean space as a sufficient substrate for hierarchical vision-language learning as it is currently practiced, and motivate new paradigms in which hyperbolic geometry can deliver on its theoretical promise. Code is available at https://github.com/fbyrman/hyperbolic-vlms.
Read the full abstract
Machine Learning
Settling the Burn-in Gap: Sharp Risk Bounds for Tail-Averaged SGD in Linear Regression
We study the excess risk of constant-stepsize online stochastic gradient descent (SGD) with tail-averaging for high-dimensional linear regression. Building on the instance-dependent spectral framework of Zou et al. (2021), we establish matching initialization-explicit upper and lower bounds on the total excess risk of tail-averaged SGD in terms of the population covariance spectrum $\mathbf{H}$ and initialization $\mathbf{w}_0-\mathbf{w}^*$ whenever a trajectory spectral feedback condition holds. Existing analyses exhibit a fundamental gap in the effective variance: the upper bound of Zou et al. (2021) retains an undamped contribution from the initial error in the head eigenspace $\mathbf{I}_{0:k^\dagger}$, whereas their lower bound omits the variance induced by the head initialization altogether. We show that both the existing upper and lower bounds are loose: the single-hop fast-mode variance contribution from the head initialization is exponentially damped during the burn-in phase, whereas fourth-moment energy scattered from the head initialization into slow spectral modes during the early burn-in steps survives undamped into the averaging window. We prove that our spectral feedback condition is satisfied in two concrete constant-stepsize regimes: (i) for arbitrary covariance spectra $\mathbf{H}$ and any finite horizon, under an explicit horizon-dependent stepsize chosen prior to execution; and (ii) for polynomially decaying spectra $\lambda_i\asymp i^{-a}$ ($a>1$), under a single horizon-independent positive stepsize. Finally, we complement these positive results with explicit one- and two-dimensional separation constructions proving an exponential-versus-polynomial gap in the lower bound of Zou et al. (2021) and showing why fourth-moment assumptions alone cannot determine the asymptotic exponential decay rate without additional structural conditions.
Read the full abstract
Monday 9 November
Software Engineering
LLMs for Software Vulnerability Prediction and Prioritization: A Review
Large language models have moved beyond vulnerability detection into prediction, assessment, exploitability estimation, risk assessment and prioritization, yet the evidence for this shift has not been synthesised separately from the detection literature. In this systematic literature review, we examine how LLMs are used across those five task classes, how well they perform against established baselines, and what prevents their operational adoption. Following a prespecified protocol, we identified 197 records, removed 53 duplicates, and excluded 103 records at title and abstract stage. We assessed 41 full texts and finally retained 23 primary studies published between 2024 and 2026 after a six-question quality assessment. The review contributes a five-class taxonomy ordered by decision outcome, a method taxonomy covering prompting, tuning and retrieval or agentic architectures, an evaluation-practice audit, a bounded performance synthesis, and a deployment-barrier analysis. The main findings are that (i) assessment is the largest class (9 of 23 studies), ahead of risk assessment and prioritization (7 each); (ii) reported accuracy is highest where the label space is tightest and is not stable across evaluation periods; (iii) explanation usability is an audited evidence gap rather than a demonstrated capability; and (iv) reproducibility is the weakest reported dimension. We conclude that present systems should support analyst judgement rather than replace it, and we specify three validation studies that would settle the open questions.
Read the full abstract
Software Engineering
Optimal Development Path: An Empirical Evaluation of Tiered Dependency-Aware Sequencing
The Optimal Development Path (ODP) in D-POAF orders Feature Blocks in a Wave Candidate by choosing the eligible block with the lowest effort-and-risk-to-value ratio (ERS/BVS), while respecting declared prerequisites. We evaluate native ODP (L0) against four dependency-aware strategies: unlock count (L1), unlocked value (L2), multi-hop potential (L3), and one-step rollout (L4), alongside local search, Sidney decomposition, and an exact reference. On held-out synthetic graphs with about one prerequisite per block, L0 attains 0.978 of the surrogate optimum at 12 blocks, versus 0.993 for L2 and 0.994 for L4. At 100 blocks, L0 attains 0.942 and L2 0.983. In contrast, 373 historical Jira sprints from 21 open-source projects in TAWOS, used as scheduling proxies rather than D-POAF Waves, contain sparse recorded dependencies. On these instances, L0 attains 0.998 of the surrogate optimum and is exactly optimal in 69% of sprints. Unlock bonuses calibrated on synthetic graphs do not deliver significant average gains on these sprint-derived instances and can worsen the surrogate. L4 never worsens the declared surrogate relative to L0, but this guarantee does not extend to operational outcomes. Retrospective calendar replays using recorded Jira durations show that surrogate improvements do not reliably improve value-weighted completion. The findings support keeping L0 as the default and presenting rollout and exact references as auditable alternatives. Dependency-valid sequences expose prerequisite readiness, but their effects on interruptions, rework, costs, and delivery times remain unmeasured.
Read the full abstract
Probability
A Proof of the Achlioptas-Friedgut Convergence Conjecture
For every fixed integer $q\ge3$, we show that the $q$-colourability transition in a sparse random graph takes place at a limiting edge density, thus proving the Achlioptas-Friedgut convergence conjecture. The proof adapts the argument we recently introduced to settle the satisfiability conjecture.