Distributed Computing
Authority at Commit Time: Reject-and-Rerun Semantics for Governed Agentic Systems
Jesus Salas
Enterprise
agents can compute for seconds or hours from a snapshot of policy, facts, task state, models, tools, and verifiers that may change before their work reaches the world. Completion alone therefore cannot confer institutional authority. We treat
agents as proposal producers and a logically authoritative service as the sole authority for governed effects. At admission, the service compares a proposal's declared dependencies and component manifest with current state, applies the current verifier, and either reserves an idempotent effect for outbox dispatch or rejects affected stale work for resynchronization and rerun. Unaffected work may continue; lost responses resolve through canonical verdicts and receipts. The protocol composes optimistic concurrency control, semantic verification, and transactional outbox dispatch. Its distinctive read set includes normative authority and the model components that produced the proposal. We evaluate the design through an implemented governed-work lifecycle, a local authority, outbox, and reconciliation composition, and a multi-process fallback that delegates effect authority when complete mediation is unavailable. The local composition preserved selective fallback, unaffected work, retry and restart recovery, and package reconstruction. Complete mediation produced 18 authority commands and 18 external effects. Direct-credential arms produced eight additional effects, all detected with independent journal reconciliation and none falsely detected in the mediated control. The results support commit-time reject-and-rerun semantics and expose the clock, lease, ordering, and quorum obligations introduced by delegated admission. They do not establish prevention or reversal of bypass effects, external-journal completeness, production throughput, wide-area availability, semantic completeness, or verifier correctness.
Read the full abstract ▾
Number Theory
Further Results on Sum-Freedom of Binary and $q$-ary Functions
Xiang-dong Hou and Shujun Zhao
The notion of sum-freedom of binary functions was introduced recently by C. Carlet as a generalization of the APN functions used in cryptography; the $q$-ary version of the notion is a natural extension. For each integer $k$ with $0\le k\le n$, there is a $k$th order sum-free function on $\Bbb F_{2^n}$. It is also known that when $k/n$ is not close to 0 or 1, the multiplicative inverse function on $\Bbb F_{2^n}$ is not $k$th order sum-free. We generalize these two results to $q$-ary functions. APN functions have a coding theoretic characterization. We generalize the characterization to sum-free functions of arbitrary order over any finite field. It is well known that the Welch functions is 2nd order sum-free. We give an alternative proof for this result which leads to a more general algebraic question. We also investigate that the 3rd order sum-freedom of the Welch function and power functions of algebraic degree 3. We formulate a
conjecture about the 3rd order sum-freedom of the Welch function which is supported by strong numerical evidence.
Read the full abstract ▾
Quantum Physics
Connes Distance in Discrete-Time Quantum Walks
Rayan Trabelsi, A. F. Reyes-Lega and Pablo Arrighi
We study the
geometric spread of discrete-time
quantum walks using Connes' distance formula. We introduce its discrete analogue by adapting the continuous spectral metric to the
lattice, with the Dirac operator replaced by the
unitary step operator. Focusing first on a (2+1)-dimensional
quantum walk, we evaluate the dynamical spread of localized wavepackets to compare propagation in free space against that under a uniform
magnetic field. We find that while the free walk shows linear ballistic growth, gauge-induced phases in the
magnetic walk constrain the admissible test operators, suppressing the long-time
geometric expansion. Furthermore, we apply this discrete framework to a (1+1)-dimensional plastic
quantum walk simulating an inhomogeneous spatial metric. The results demonstrate that the discrete Connes distance naturally captures the underlying spatially varying metric, providing a robust measure of the wavepacket's spread that adapts directly to the local simulated geometry.
Read the full abstract ▾
Probability
Singular Backward SDEs for Optimal Control with State Constraints
Mathieu Lise and Nizar Touzi
We investigate a class of backward
stochastic differential equations (BSDEs) with at most quadratic growth which explode at a possibly unbounded random horizon, defined through the first hitting of zero of an adapted Ito process. In contrast with the classical theory of singular BSDEs, the explosion is generated by the nonlinear dependence of the generator on the martingale integrand, rather than by a superlinear coercivity condition in the solution component. We construct a minimal singular solution and derive two-sided estimates on its explosion, together with weighted BMO estimates for the martingale integrand. We also obtain the uniqueness and exact explosion rates under additional structural assumptions. For
Hamiltonian generators, we establish a verification
theorem for an infinite-horizon
stochastic optimal control problem with possible non-Markovian state constraints, showing that the BSDE feedback induces the unique optimal constrained law. We finally specialize the theory to exit times of uniformly
elliptic Markov
diffusions and recover the connection with large solutions of viscous Hamilton-Jacobi equations.
Read the full abstract ▾
Rings and Algebras
The Hilbert Basis Theorem for Non-Associative Skew Power Series Rings and Non-Associative Skew Laurent Series Rings over a Non-Associative Coefficient Ring
Darwin P. Mangubat, Jocelyn P. Vilela and Johan Richter
In this paper, we give conditions on a unital non-associative ring $R$ and additive map $\sigma: R\longrightarrow R$ so that the non-associative skew power series ring, denoted by $R[[X;\sigma]]$ and the non-associative skew Laurent series ring denoted by $R((X;\sigma))$ are left (resp. right) Noetherian, respectively. Consequently, we establish versions of
Hilbert Basis
Theorem of $R[[X;\sigma]]$ and $R((X;\sigma))$, respectively, for left (resp. right) Noetherianity . We also give a counterexample showing that the
Hilbert Basis
Theorem does not hold in general in the non-associative case. In establishing the
Hilbert Basis
Theorems for $R[[X;\sigma]]$ and $R((X;\sigma))$, we introduce and utilize the right (resp. left) quasi-associative property of $R$, which is equivalent to $rR$ (resp. $Rc$) being one-sided ideals for all $r\in R$ (resp. $c\in R$).
Read the full abstract ▾
Cryptography and Security
Verifiable Randomness for Blockchain-Based Lottery Systems
Gonçalo Ferreira, André Zúquete and Paulo Bartolomeu
As lotteries and other high-stakes decentralized applications increasingly depend on unpredictable randomness for their operations, the lack of a secure and transparent on-chain random number generator that is verifiable by all participants remains a critical open problem. Various approaches to blockchain-based random number generation have emerged over the years, each with their own strengths and limitations, and have consistently been superseded as blockchain technology evolved. This paper surveys existing approaches to on-chain randomness and proposes a new platform that builds upon the well-known commit-and-reveal scheme while directly addressing its principal vulnerability, the last revealer attack, in which the final participant can withhold their reveal in order to bias or abort the output upon seeing an unfavorable result. We further compare this solution with prior approaches and evaluate its
entropy properties. The proposed architecture combines a web-based front-end with a Solidity smart contract deployed on the Polygon 2.0 blockchain. Implemented and tested on the Amoy testnet, the prototype is low-cost and simple to deploy, providing a practical, accessible proof-of-concept for verifiable on-chain randomness.
Read the full abstract ▾
Information Theory
COFI-DQI: Curve-based Optimal Function Intersection via Decoded Quantum Interferometry
Gretchen L. Matthews and Julia Shapiro
In 2025, Jordan et al. introduced Decoded
Quantum Interferometry (DQI), a
quantum algorithm for combinatorial optimization based on decoding. They considered Reed-Solomon decoding and its associated optimization problem, called Optimal
Polynomial Intersection (OPI), which may be viewed as a
polynomial regression problem over a finite field. DQI exhibits provable speedups on certain problem instances and establishes a connection between decoding problems and optimization tasks. Leveraging the well-understood dual structure and decoding theory of algebraic geometry codes from other curve families, we introduce COFI: Curve-based Optimal Function Intersection. By considering two-point Hermitian codes, Suzuki codes, and extended norm-trace codes, we broaden the range of algebraic geometry codes used in DQI and identify families that offer further improvements over one-point Hermitian codes in the Hermitian Optimal
Polynomial Intersection framework considered by Jordan and Gu. Depending on the family and parameter regime, these curves can reduce
quantum resource requirements or increase the number of constraints that can be considered.
Read the full abstract ▾
Chemical Physics
Scaling Density Functional Theory with Gaussian Splatting
Andrés Guzmán-Cordero, Cindy Zhang, Majdi Hassan et al. (+3 more)
Density functional theory (DFT) strikes a practical balance between accuracy and computational cost in many problems of computational chemistry and materials science. However, many DFT calculations are limited by fixed atom-centered basis sets, which dictate how accuracy and cost scale with system size. We propose Gaussian Splatting for Density Functional Theory (GS-DFT), which represents molecular orbitals as a cloud of Gaussians whose positions, shapes, and mixing coefficients are optimized jointly by gradient descent to minimize the energy without training data. Conceptually, GS-DFT is 3D Gaussian splatting with the renderer replaced by
quantum mechanics. We introduce two key solver components: adaptive density fitting with screening for efficient evaluation of two-electron integrals, and a regularized differentiable orthogonalization of the molecular orbitals. Empirically, the optimized basis reaches the accuracy of the largest conventional basis sets with a fraction of the parameters, converging systematically in energy, density, and
nuclear forces. At equal parameter count, it captures the stretched-bond and anion physics that fixed bases only recover with specialized basis augmentation. The resulting solver exhibits quadratic peak
memory scaling in the cloud size, allowing us to simulate systems of up to 2,742 atoms (10,406 electrons) without any modifications at triple-
zeta scale using a single four-GPU node.
Read the full abstract ▾
All authors Andrés Guzmán-Cordero, Cindy Zhang, Majdi Hassan, Marta Skreta, Kirill Neklyudov, Matija Medvidović
Artificial Intelligence
"AI is (not) the new...": A Diagnostic Analogy Framework for Generative AI's Cultural Impacts
Rida Qadri, Vinodkumar Prabhakaran and Remi Denton
Generative AI is reshaping the cultural infrastructures through which knowledge is found, synthesized, and held accountable. To make sense of this shift, scholars and policymakers reach for historical analogies of technologies such as the printing press, steam power or electricity. But these comparisons are typically imprecise about which property of the technology carries the comparison, and imprecise analogies produce imprecise governance by designing interventions against the wrong property of the system. This paper offers a diagnostic framework for analyzing how generative AI can transform epistemic and cultural practice. This paper offers a diagnostic framework for analyzing how generative AI can transform epistemic and cultural practice. We decompose each intervention into three coordinates: the epistemic site at which a technology acts, the governing logic by which it organizes its object, and the technical mechanism through which the logic is instantiated. This framework allows us to distinguish between structural cultural consequences, which follow from the mechanism itself, from contingent ones, which remain open to design and institutional choice. Applying the framework to information discovery and knowledge synthesis, we show how the shift from indexicality to inference and from editorial authority to statistical consensus produces specific, traceable cultural effects and reveals governance levers that gestalt analogy obscures.
Read the full abstract ▾
Mathematical Physics
Rotationally Symmetric Zoll Metrics with a Cubic Integral
Holger R. Dullin, Vladimir S. Matveev, Gleb P. Palshin et al. (+1 more)
We introduce a new method for constructing Zoll metrics that admit first integrals
polynomial in the momenta. The method uses Funk's form of the metric together with an algebraic relation between the first integrals. This reduces the problem to an algebraic system from which all unknown functions can be determined explicitly. In the case of a non-trivial cubic integral, we present an explicit two-parameter family of metrics that covers all such rotationally symmetric Zoll metrics on the $2$-sphere: every such metric is isometric to a member of this family. We compare these metrics with the earlier results of Valent, Duval, and Shevchishin.
Read the full abstract ▾
All authors Holger R. Dullin, Vladimir S. Matveev, Gleb P. Palshin, Serena Scapucci
Materials Science
Ferroelectric Switching in ZnO/Zn1-xMgxO Heterostructures: Atomistic Insights into Interfacial Coupling and Layer Architecture
Alireza Sepehrinezhad, Ali Mohammadi Dinani, Ece Gunay et al. (+4 more)
Ferroelectric switching in
heterostructures couples composition, layer
topology, temperature, and interfacial boundary conditions. ReaxFF molecular dynamics isolates these variables in ZnO/ Zn1- xMgxO/ZnO and Zn1-xMgxO/ZnO/ Zn1-xMgxO stacks. Within pristine, initially single-domain models, coupling to switchable Zn1-xMgxO reduces the applied field required to reverse ZnO by up to fivefold. Temperature generally lowers the coercive field, whereas Mg concentration produces a nonmonotonic response. At equal ZnO and Zn1-xMgxO (ZMO) proportions, structures with ZnO at the center switch at lower fields than those with ZMO at the center at all four temperatures examined, demonstrating a
topology-dependent response. Layer-resolved trajectories reveal
topology-dependent switching sequences with direction-dependent redistribution of normal stress near the heterointerfaces, consistent with a stress-assisted cooperative pathway. Limiting MgO-containing structures exhibit sequential multilevel switching or low-polarity trapping, depending on thickness and temperature. Fixed-charge atomistic simulations complement previous continuum descriptions by resolving structural, energetic, and local stress
evolution under a common applied field. TEM and STEM-EDS observations provide experimental structural context for the modeled architectures. Together, the results establish layer
topology and interfacial mechanical confinement as design variables for wurtzite ferroelectric
heterostructures.
Read the full abstract ▾
All authors Alireza Sepehrinezhad, Ali Mohammadi Dinani, Ece Gunay, Elizabeth C. Dickey, Jon-Paul Maria, Susan Trolier-McKinstry, Adri C.T. van Duin
Number Theory
Ternary Perfect Forms Over Imaginary Quadratic Field
Zachary Parker and Dan Yasaki
In this work, we enumerate the ternary perfect forms over all imaginary quadratic fields of absolute discriminant up to $91$ and study the number and combinatorial types of their associated polytopes. We prove a bound on the number of combinatorial types that can arise regardless of the discriminant. We analyze the data and make several observations concerning the number and complexity of the perfect forms that arise in the scope of our calculation.
Read the full abstract ▾
Number Theory
Irreducible rootless unimodular lattices generated by vectors of norm $3$
Hong-Jun Ge, Jack H. Koolen, Jing-Yuan Liu et al. (+1 more)
The classical ADE classification implies that $E_8$ is the unique irreducible unimodular Euclidean
lattice generated by vectors of norm $2$. In sharp contrast, for every positive integer $m$, we construct an irreducible odd unimodular
lattice of rank $24m$ and minimum norm $3$ that is generated by its minimal vectors.
Read the full abstract ▾
All authors Hong-Jun Ge, Jack H. Koolen, Jing-Yuan Liu, Kiyoto Yoshino
Fluid Dynamics
A parallel solver for vortex-induced vibrations at zero mass ratio
Arthur Bawin, Joan Le Pouhaër, Stéphane Étienne et al. (+1 more)
We present a parallel strategy for the study of the
vortex-induced vibrations (VIV) of a rigid, massless cylinder mounted on translational springs, in two and three dimensions. Without mass or damping, its motion is governed solely by the fluid forces and the spring restoring force, and is an immediate response to the flow. We focus on low spring stiffness, yielding high reduced velocity and maximal oscillation amplitude. A monolithic finite element method in an arbitrary Lagrangian-Eulerian framework with implicit time stepping is proposed, solving the fluid flow, cylinder dynamics, and mesh motion simultaneously. Mesh movement is modelled with a linear elastic pseudo-solid with variable Lamé coefficients, and the no-slip condition on the cylinder is enforced with a Lagrange multiplier coupled to the position degrees of freedom; three strategies are explored to enforce this force-position coupling in a distributed framework. To the best of our knowledge, this work is the first to propose a distributed solver for three-dimensional fluid-structure interaction at zero mass ratio. The solver is verified with manufactured solutions, and exhibits second-order accuracy in time. A numerical study of two- and three-dimensional VIV for Reynolds numbers from 100 to 300 reproduces the VIV responses obtained in previous two-dimensional studies, and predicts nonperiodic 3D oscillations starting at $Re = 300$. Using the direct solver MUMPS, the proposed method scales to up to 768 compute cores.
Read the full abstract ▾
All authors Arthur Bawin, Joan Le Pouhaër, Stéphane Étienne, Cédric Béguin
Dynamical Systems
Non-autonomous dynamics of inner functions: mixing and entropy
Vasiliki Evdoridou, Núria Fagella, Philip Rippon et al. (+1 more)
We generalise a mixing result due to Pommerenke for non-autonomous dynamical systems of forward compositions of the boundary values of centred inner functions; the proof depends on a precise bound on the orbits of such forward compositions in the unit disc. This enables us to give new criteria for such systems to have positive metric
entropy, lower
entropy and topological
entropy. In the opposite direction, we construct a forward composition of centred degree two Blaschke products with topological
entropy zero.
Read the full abstract ▾
All authors Vasiliki Evdoridou, Núria Fagella, Philip Rippon, Gwyneth Stallard
Optics
Direct Observation of Switching-Wave Dynamics in 20 GHz Normal Dispersion Spiral Microresonators
Van Doan Le, Julien Fatome and Erwan Lucas
Integrated Kerr frequency combs are powerful tools for microwave photonics, spectroscopy, and optical communications. While traditional architectures rely on the anomalous-dispersion regime, this typically requires thick, highly strained silicon nitride layers that complicate standard CMOS-foundry fabrication. Thinner layers circumvent these fabrication constraints but yield normal-dispersion microcombs that generally require specific trigger mechanisms, such as deterministic seeding, due to the absence of spontaneous modulational instability. Here, we demonstrate the generation of a 20 GHz microcomb in the normal-dispersion regime, driven by a synchronized dual-pump scheme via electro-optic sidebands modulation. The platform leverages an Archimedean spiral geometry on a foundry-compatible, 350 nm-thin SiN platform. By employing adiabatic
curvature engineering, the resonator balances a compact footprint with an intrinsic quality factor exceeding $7 \times 10^6$, while effectively eliminating avoided mode crossings with higher order modes. This strong-dispersion architecture yields a low-repetition-rate microcomb featuring high power-per-line and picosecond-scale temporal profile, which enables the direct optical sampling and characterization of the out-coupled switching wave waveforms. The measured dynamics across a range of pump desynchronizations demonstrate excellent agreement with simulations. Our work establishes a scalable, strategy for footprint-efficient normal-dispersion microcomb generation at microwave frequencies, with potential for scaling to other wavelength ranges.
Read the full abstract ▾
Quantum Gases
Spatiotemporal Non-Hermitian Skin Effect with Floquet-Engineered Ultracold Atoms
Weijie Liang, Weiping Zhang and Keye Zhang
Non-Hermitian skin effect, the accumulation of bulk eigenstates at edges under open boundary conditions, has been observed across classical,
quantum, and synthetic platforms. Whether it can be generalized to the genuine time dimension remains open. We propose a spatiotemporal NHSE in the Floquet phase space via Floquet engineering of a ring-trapped ultracold atomic gas. Placing spatial and temporal degrees of freedom on an equal footing, we show that a single parent skin effect projects onto either domain. The spatial projection yields an NHSE with arbitrarily tunable skin locations, eliminating the need for physical open boundaries. The temporal projection gives time-domain nonreciprocity and periodic temporal funneling, consistent with causality. This establishes a unified spatiotemporal paradigm for NHSE and opens a route to non-Hermitian topological physics in time.
Read the full abstract ▾
Artificial Intelligence
Game Arena: Strategic LLM Evaluation in Competitive Environments
Bovard Doerschuk-Tiberi, Yao Yan, Justin Chiu et al. (+59 more)
We introduce Kaggle Game Arena, an open and ever-expanding platform to evaluate large
language models (LLMs) through competitive games. Different from static
benchmarks, game arena enables models to play head-to-head matchups in structured environments where the gameplay strength naturally increases as models evolve, preventing performance saturation. This technical report details the infrastructure behind Game Arena and describes the three pilot game environments: Chess, Poker, and Werewolf. These environments span perfect information, imperfect information, and multiplayer game settings, enabling a systematic study of models' strategic planning, adaptation, and robustness under uncertainty. For each game, we provide a detailed description of the environment, evaluation metrics, and results from running full competitions across models. Through robust infrastructure and large-scale ground-truth based evaluation, Game Arena ensures reproducibility, transparency and generalizability to new games and variants over time.
Read the full abstract ▾
All authors Bovard Doerschuk-Tiberi, Yao Yan, Justin Chiu, Hann Wang, Timothy Chung, Martyna Plomecka, John Schultz, Jon Lipovetz, Clayton Drazner, Yuchen Zhuang, Jaimie Hwang, Nate Keating, Riley Jones, Andrew Lee, Oran Kelly, Ian Gemp, Michael Aaron, Laurel Prince, Kate Larson, Jeff Moser, Harrison Jobe, Chad Woodford, Siqi Liu, Andrew Wang, Bo Chang, Christopher D'Mello, Diane Chaleff, Addison Howard, Johnny Yip, Chuck Sugnet, Antonio Gulli, Meghan O'Connell, Will Cukierski, Nenad Tomasev, Dima Yeroshenko, Kinjal Parekh, Roxanne Daniel, Marc Lanctot, Domino Weir, Elsa Dong, Daniel Hennes, Melissa Nalubwama, Robert Fraser, Ryan Trostle, Jun Peng, Tom Mason, Lloyd Hightower, Chiamaka Chukwuka, Yuexiang Zhai, Phoebe Kirk, Yi Su, Yuting Han, Jie Ren, Chris Prichard, Sahand Sharifzadeh, Karim Hakimzadeh, DJ Sterling, Meg Risdal, Kate Olszewska, Ya Xu, Orhan Firat, Minmin Chen
Quantum Physics
A Three-Party W-State Quantum Secret Sharing Protocol with X-Gate Encoding and Forbidden-Outcome Detection
Timur R. Teregulov and Elena R. Loubenets
We develop a new three-party
quantum secret-sharing (QSS) protocol based on a three-
qubit W state. This protocol encodes the secret-sequence bits using X gates and employs randomly selected Hadamard operations and measurement bases to generate information and security-test rounds. We evaluate the efficiency of the proposed protocol and analyze its security against an internal adversary performing intercept-and-resend and entangle-and-measure attacks, deriving the corresponding detection probabilities. The analysis of the entangle-and-measure attack shows that extracting information through an ancillary system leads to a nonzero probability of forbidden outcomes in the test rounds, thereby enabling the detection of the attack. The proposed protocol is implemented in Python using the Qsikit framework and executed on
quantum processors. Its performance is evaluated in terms of the
quantum bit error rate (QBER) for information rounds under different secret-bit encoding configurations. The experimental results demonstrate that the QBER depends noticeably on the
quantum processor used.
Read the full abstract ▾
Chemical Physics
Communication: Becke Weights as a Partitioning Scheme for Phase Space Electronic Structure Theory
Ben Curlee, Zheng Pei, Xinchun Wu et al. (+4 more)
We implement the Becke weight for atom A ($w_A$) as the partitioning scheme needed for phase space electronic structure theory. We show that this ansatz can lead to very efficient electronic structure calculations at nonzero
nuclear momentum ($P \ne 0$). We further show that using Becke weights leads to efficient VCD calculations that are very stable with respect to basis set size. We hypothesize that, as far as
quantum chemistry simulations in an atomic orbitals basis, the choice of Becke weights as a partitioning scheme will be adopted for all future phase space electronic structure calculations in an atomic orbital basis, as theorists learn to efficiently simulate coupled
nuclear electronic dynamics through a new framework.
Read the full abstract ▾
All authors Ben Curlee, Zheng Pei, Xinchun Wu, Titouan Duston, Todd J. Martinez, Yihan Shao, Joseph E. Subotnik
Machine Learning (Statistics)
Beyond Empirical Support: Structured Outlier Generation via Sinkhorn Optimal Transport
Haixiang Sun and Andrew L. Liu
Outliers are essential for evaluating and improving the robustness of machine learning systems, especially when future distributions may differ significantly from historical training data. In high-stakes applications, robustness often depends on rare cases that finite
datasets fail to capture, making simple resampling or perturbation insufficient for stress scenario generation. Existing outlier synthesis methods typically rely on sparse neighborhoods, low support latent regions, or classifier boundary crossings, which can be heuristic, unstable, and tied to specific modalities or architectures. We therefore propose Sinkhorn Boundary Outlier Generation (SBOG), a structured framework for latent-space outlier generation that couples Sinkhorn optimal transport geometry with distributionally robust boundary modeling. The resulting Sinkhorn-induced support cost guides the sampler toward weakly supported boundary regions, while semantic constraints prevent uncontrolled drift from the intended context, yielding controlled deviations from the in-distribution reference measure rather than arbitrary sparse-region samples. Experiments on time series anomaly generation and image outlier synthesis show that our framework produces informative, semantically controlled outliers and improves downstream robustness evaluation across modalities, providing a foundation for stress scenario generation beyond empirical support.
Read the full abstract ▾
Algebraic Topology
LandscapeSHAP: Which Persistent Homology Class Gets the Credit?
Nikola Milićević
Shapley values, a solution concept from cooperative game theory, have recently become a standard tool for feature credit allocation in machine learning. They provide an axiomatically justified method to fairly distribute a model's prediction among the data features. Shapley values have not yet been applied to explain machine learning models trained on features from topological data analysis. We develop what we believe is the first such approach, focusing on the persistence landscape featurization of persistence diagrams. Because each landscape coordinate is a rank statistic, crediting a model's prediction back to individual persistent homology classes (persistence diagram points) is nontrivial. We introduce LandscapeSHAP, a method for fair credit allocation to persistence diagram points based on a model's prediction. For linear models on persistence landscapes, LandscapeSHAP has a closed form expression that gives the exact Shapley value of every persistence diagram point. In particular, there is no coalition sampling required. We further prove that the four Shapley "
fairness" axioms uniquely characterize this credit allocation for any model, not only linear ones. For a general nonlinear model, this unique value can only be calculated exactly from its defining coalition averaging formula, which requires considering all $2^N$ many coalitions, where $N$ is the number of points in the persistence diagram. This is computationally intractable for persistence diagrams of realistic size. We complement the exact linear model result with an efficient Monte Carlo sampling of persistence diagram coalitions. We give convergence rates in terms of number of samples needed to approximate to a desired degree of accuracy. We also prove stability results for the LandscapeSHAP credit allocation, for any model.
Read the full abstract ▾
General Economics
PriceBench: A Diagnostic Benchmark for Price, Quality, and Brand Preferences in LLM Booking Agents
Pavel Kireyev
LLMs increasingly act as purchasing
agents, which makes the LLM, not the user, the one choosing among the options that satisfy a request; its preferences quietly fix what gets bought and what it costs. Hotel booking is a clean instance: a high-volume choice settled on a few comparable attributes, where the pick reveals those preferences. We introduce PriceBench, a diagnostic
benchmark that recovers an LLM's price, quality, and brand preferences from its booking choices with a logit choice model, applied to 28 LLMs from 8 providers on 3,600 hotel tasks from 179 real New York City properties. We find that capability is associated with how consistently an LLM chooses, not with what it chooses: more capable LLMs hold stronger, more consistent preferences, while weaker ones either lock onto one position, exploitable by whoever controls listing order, or choose almost indifferently. What those preferences favor varies sharply across providers and even within one family: price sensitivity spans more than an order of magnitude, and the price/quality trade-off moves mean booked nightly price from \$247 to \$393 on identical tasks. What an
agent buys must therefore be measured per LLM, not inferred, and we release the tasks, code, and all 28 response sets.
Read the full abstract ▾
Statistical Methodology
A likelihood-based coefficient for biomedical independence testing: the binomial-cut composite likelihood ratio
Jing Qin
The standard dependence summaries used in biomarker studies – Pearson's r, Spearman's rho, Kendall's tau – take values in [-1, 1] with 0 indicating no linear or monotone association. Zero does not distinguish independence from non-monotone dependence, so the scale cannot represent threshold effects, heteroscedasticity, and tail shifts common in biomarker practice. We formulate independence testing as a composite Bernoulli likelihood ratio: at each threshold t, comparing the Bernoulli laws of 1(Y <= t) conditionally on X versus marginally, aggregated over cut points. The resulting coefficient xi_cut lies on [0, 1] with 0 iff X and Y are independent (under continuity of Y) and 1 iff Y is a measurable function of X. Fisher weighting arises at second order from the Bernoulli likelihood, and xi_cut equals twice the threshold-averaged mutual information between X and 1(Y <= t), giving a distribution-free lower bound on I(X; Y). A second-order expansion recovers the Fisher-weighted Dette-Siburg-Stoimenov measure, which coincides under continuity with Chatterjee's rank correlation. Estimation uses a Nadaraya-Watson plug-in with a max-over-grid bandwidth; inference is by exact permutation. In biomarker-motivated simulations T_cut substantially outperforms rank-based coefficients on W-shaped non-monotone and heteroscedastic alternatives. We illustrate on the Seattle cohort (n=70, ages 21-88) of the aging
plasma proteome
dataset, screening all 1,305
proteins for age dependence: under Benjamini-Hochberg control at q<0.05, T_cut rejects on 70
proteins, six of which are missed by Pearson, Spearman, and Chatterjee at the same FDR level.
Read the full abstract ▾
Machine Learning
Scaffold: Support Graph Theory Based Sparsification for Graph Neural Networks
Siddhartha Shankar Das, Sai Karthik Navuluru, S M Ferdous et al. (+7 more)
Graph neural networks (GNNs) rely on message passing over graph edges, making their computational and
memory costs strongly dependent on graph density. Graph sparsification offers a natural way to reduce these costs, but removing edges indiscriminately can distort important communication structure and degrade predictive performance. We introduce Scaffold, a
topology-based, unsupervised graph sparsification framework derived from support graph theory preconditioners. Scaffold explicitly controls two complementary structural quantities: dilation, which measures the length of rerouting paths induced by removed edges, and congestion, which measures how strongly these rerouted paths concentrate on the retained support. By jointly controlling dilation and congestion, Scaffold preserves short communication paths while avoiding structural bottlenecks. To our knowledge, Scaffold is the first scalable GNN sparsification framework to use a joint supporting-path dilation-congestion criterion. Across 19 homophilic and heterophilic
benchmarks spanning small to large graphs, Scaffold achieves the best aggregate rank among the evaluated sparsification and related methods. Using only 10%-50% of the original edges per sparse support, Scaffold recovers or closely approaches full-graph GNN performance while using less than half the
memory of full-graph training and reducing end-to-end training time, including sparsification overhead. We provide an open-source software package at https://github.com/siddhartha047/Scaffold.
Read the full abstract ▾
All authors Siddhartha Shankar Das, Sai Karthik Navuluru, S M Ferdous, Ryan A. Rossi, Baris Coskunuzer, Lakshman Tamil, Edoardo Serra, Alex Pothen, Robert Rallo, Mahantesh M Halappanavar
High Energy Astrophysical Phenomena
Neutron Economy and Freeze-out Dynamics in the Late-Time Cold r-Process
Mengke Li and Bradley Meyer
Standard r-process models often treat the freeze-out phase as a passive, uniform smoothing of the global abundance pattern. We demonstrate, however, that the final pattern is instead actively sculpted by a dynamic “
neutron economy” that operates well after the free
neutron source is exhausted. By analyzing cold
neutron star merger
outflows, we identify a distinct
neutron exporter-importer relationship: the second-peak region ($A \sim 130$) transitions early into a $\beta$-decay dominated regime, acting as a net exporter, while the third-peak ($A \sim 195$) and Rare Earth ($A \sim 164$) regions remain capture-dominated
neutron importers. Targeted simulations reveal that the final peak structures emerge from a competition between two opposing forces: a global “push” driven by
neutron capture shifting mass toward heavier
nuclei, and a local “pull” from beta delayed
neutron emission returning material back toward lighter masses. We identify $^{136}\text{Pd}$, $^{135}\text{Ag}$, and $^{130}\text{Ru}$ as the primary exporters driving this economy, while third-peak fine structure is governed by $N=126$ isotones (Sm to Tb). Physically, this transport reflects a fundamental drive toward shell closure, where
nuclei that overshoot the $N=82$ closed shell shed mass to migrate back toward this stable configuration, while
nuclei in the third peak capture these redistributed
neutrons to anchor themselves at the $N=126$ shell. Crucially, our models overproduce the odd-even staggering in the third peak compared to
solar data. This discrepancy suggests current models likely overestimate the $\beta$-delayed
neutron emission probabilities ($P_n$) for these $N=126$
nuclei, highlighting them as high-priority targets for theoretical refinement and future FRIB experiments.
Read the full abstract ▾
Geophysics
Explaining Ground-Motion Residuals at Two Strong-Motion Stations in Southeastern New York: Sediment Resonance and Topographic Amplification
Aser Abbas, Patrick Daniele, James Kaklamanos et al. (+3 more)
Sites hosting strong-motion stations in the Central and Eastern United States are commonly characterized using proxy-based parameters, limiting the ability to identify the physical causes of large ground-motion residuals. This study investigates the amplification mechanisms at two New York strong-motion stations that recorded some of the largest positive 1-Hz pseudospectral-acceleration residuals during the 2024 Mw 4.8 Tewksbury, New Jersey, earthquake. Station N62A is located at Caumsett State Historic Park in the Atlantic Coastal Plain, whereas station PAL is located at the Lamont-Doherty Earth Observatory atop the Palisades cliffs. Detailed geophysical site characterization at both sites combined active-source and ambient-noise surface-wave testing with HVSR measurements. At Lamont-Doherty, additional ambient-noise arrays were deployed across the Palisades ridge to evaluate topographic amplification. At Caumsett, surface-wave testing resolved a thick sedimentary column overlying a major impedance contrast, with a spatially representative fundamental site frequency of 0.93 Hz. Amplification in the same frequency range was also independently evident in observed amplifications computed from site-to-site (S2S) residuals. The fundamental-resonance band also encompasses the 1-Hz frequency at which the large Tewksbury residual was observed, supporting deep sediment resonance as the dominant mechanism. At Lamont-Doherty, surface-wave testing indicated hard-rock conditions with a thin, laterally variable sediment cover. The ambient-noise arrays showed repeatable, directional amplification near 2 Hz at PAL, and the earthquake-based S2S amplifications show a peak between approximately 0.6 and 2 Hz, a pattern reproduced by neither the transfer functions nor the ergodic linear amplification models. These observations support topographic amplification as the primary mechanism at PAL.
Read the full abstract ▾
All authors Aser Abbas, Patrick Daniele, James Kaklamanos, Laurie Baise, Kyle Cannon, Ellie Meyer
Machine Learning
Uncertainty-Aware Federated Learning for Infant Movement Analysis
Edmond S. L. Ho
Infant movement analysis provides valuable biomarkers for the early identification of neurodevelopmental disorders. Recent advances in deep learning have enabled automated analysis of infant movements from
video-derived skeletal representations, achieving performance comparable to expert assessment for tasks such as General Movement Assessment (GMA). However, most existing approaches rely on centralized training, requiring data from multiple institutions to be collected and stored at a single site. Such assumptions are often impractical in clinical settings due to
privacy, governance, and data-sharing constraints. To address these challenges, we present, to the best of our knowledge, the first
federated learning framework for automated infant movement analysis and General Movement Assessment using skeletal motion data. As a clinically relevant use case, the proposed framework is evaluated on fidgety movement classification. To quantify model confidence, Monte Carlo (MC) Dropout is employed to estimate predictive uncertainty during inference. Building upon this, we propose an Uncertainty-Aware
Federated Averaging (UA-FedAvg) strategy that incorporates predictive
entropy derived from MC-Dropout into the
federated aggregation process, enabling client contributions to be adjusted according to their predictive uncertainty. Experiments were conducted using a cross-subject evaluation protocol under a three-client
federated learning setting. Results demonstrate that
federated learning substantially improves classification performance compared with independently trained local models while achieving performance approaching that of centralized training. Furthermore, UA-FedAvg and its variant incorporating validation loss generally outperform conventional FedAvg across the evaluated data-split configurations.
Read the full abstract ▾
Quantum Physics
Gottesman-Kitaev-Preskill error-correction with decohered resources
Rajendra S. Bhati, T. Bazylewicz and Jarosław K. Korbicz
Teleportation-based Gottesman-Kitaev-Preskill (GKP) error correction typically assumes a decoherence-free GKP ancilla Bell pair, an idealization that is difficult to reconcile with the ubiquitous environmental decoherence in bosonic systems. Here, we analyze the performance of the GKP protocol subject to pure-dephasing processes, ubiquitous in realistic situations. We show that, under this class of interactions, teleportation, instead of projecting the input onto the GKP subspace, transforms GKP error correction into a correlated Pauli-Gaussian noise process that can accumulate over successive correction rounds. Thus, unlike in previously studied noisy GKP scenarios, the code no longer self-corrects. Moreover, repeated noisy GKP teleportation drives the encoded modes out of the logical subspace, progressively degrading the encoded codewords. We show that the resulting leakage rapidly saturates with the number of correction rounds. Finally, we investigate the implications of our results for motional-oscillator-based GKP error-correction platforms.
Read the full abstract ▾
Computer Vision
KneePreM: Towards 3D Knee MRI Foundation Models via Large-Scale Unlabeled Pretraining and Label-Efficient Fine-Tuning
Xinxin Wang, Liam Hazan, Jing Li et al. (+3 more)
Background: Large volumes of unlabeled knee MRI scans are available across repositories but remain insufficiently leveraged. We developed KneePreM, a knee-specific 3D self-supervised model, and evaluated transfer and label efficiency for classification and
segmentation. Methods: A 3D U-Net masked autoencoder was pretrained on 19,011 unlabeled Osteoarthritis Initiative (OAI) MRI series from 4,791 participants. Downstream fine-tuning used full and reduced training sets for fastMRI+ two-label classification (1,172 examinations), Arthroscopic Partial Meniscectomy (APM) eight-target classification (1,716 examinations), SKM-TEA
segmentation (155 examinations), and APM
segmentation (25 examinations). Baselines were random initialization and SuPreM. Deployment workflow was implemented with a Model Context Protocol interface. Evaluation metrics included balanced accuracy, F1 score, ROC AUC, PR AUC, and Dice score. Statistical analysis used bootstrap confidence intervals and paired bootstrap tests for classification and Wilcoxon signed-rank tests for
segmentation. Results: KneePreM achieved higher full-data macro ROC AUC than both baselines for fastMRI+ and APM (all p < .001). For fastMRI+ classification, KneePreM achieved a ROC AUC of 0.722 using 50% of the training data, exceeding both full-data baselines. In APM classification, KneePreM reached a ROC AUC of 0.740 with 70% of the data, matching the full-data random baseline and outperforming SuPreM. For SKM-TEA
segmentation, its 70%-data Dice of 0.838 exceeded the full-data random baseline (0.835) and both same-budget comparators. In APM
segmentation, its 75%-data Dice of 0.746 exceeded the full-data random baseline (0.731) and both same-budget comparators. Conclusion: KneePreM improves transfer performance and label efficiency across knee MRI classification and
segmentation tasks, particularly when labeled training data are limited.
Read the full abstract ▾
All authors Xinxin Wang, Liam Hazan, Jing Li, Simona Rabinovici-Cohen, Xiaojuan Li, Mingrui Yang