Peer Revenue Play credits. Real papers.

Papers

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

All of arXiv

Tuesday 8 September

Computer Vision

A Joint 2D-3D Statistical Shape Model for Orthopedic Reconstruction

Three-dimensional femoral reconstruction from radiographs supports surgical planning, implant sizing, and post-operative follow-up, but remains ill-posed as X-ray projections discard depth information. Existing methods often incorporate a 3D statistical shape model (SSM) as a shape prior to guide reconstructions toward anatomically plausible shapes, relying on iterative 3D-to-2D projection matching. Yet, these approaches are computationally expensive and constrain their SSM to a single dimensionality, leaving the statistical relationship between 2D observations and 3D geometry largely unexploited and unexplored. We instead propose a joint 2D-3D SSM that explicitly captures the co-variation between 2D and 3D segmentations in a shared latent space. During training, 2D and 3D segmentations are registered to a common 3D template and its corresponding 2D projections, and the resulting stationary velocity fields are jointly decomposed using principal component analysis (PCA). This joint modeling allows the 2D-to-3D mapping to be learned directly from data rather than computing correspondences at inference time. For unseen subjects, the 3D shape is recovered directly by lifting the 2D latent coordinates to the 3D PCA subspace, thereby eliminating the need for iterative 3D-to-2D projection. Experiments on NMDID demonstrate that the proposed joint 2D-3D SSM outperforms a widely-used 3D-only SSM baseline while achieving inference approximately 4 times faster, at under 3 seconds per subject. The code is available at: https://github.com/florence-dellaniello-picard/joint2d3d-ssm.
Read the full abstract

All authors Florence Dell'Aniello Picard, Pranav Poudel, Nairouz Shehata, Frédéric Lavoie, Herve Lombaert

Machine Learning

Let It Go or Learn to Self-Correct: Continuous Diffusion for Constrained Discrete Tasks

Denoising Diffusion Probabilistic Models (DDPMs) generate samples by starting from noise and repeatedly denoising while keeping each update close to the current noisy state. This behavior is effective in many continuous domains, but its role is less clear for globally constrained discrete tasks, such as Sudoku, graph connectivity, Latin squares, and N-queens. In such settings, early discrete errors can be difficult to undo. As a result, standard diffusion sampling may preserve early mistakes, even when the model's clean predictions are informative. We compare standard samplers to sampling directly from the model's clean prediction. Without retraining, this single change improves Sudoku validity from 31% to 95%, with consistent gains across the other discrete tasks. We hypothesize that staying close to the current noisy state is harmful because the reverse trajectory can drift off the forward noising distribution the model was trained on. To reduce this train-test mismatch, we further introduce self-correction training, which exposes the model to its own predictions, improving robustness to errors that arise during inference. This substantially improves the performance of standard samplers. Our results suggest that continuous diffusion models can learn nontrivial global constraints, but discrete reasoning tasks require better alignment between training and inference: either through samplers that reduce commitment to early decisions, or through training that teaches the model to correct its own inference-time errors.
Read the full abstract

Earth and Planetary Astrophysics

Combined JWST and HST Deep Imaging to Characterize the Smallest Known Trans-Neptunian Objects

We present optical-near-infrared (NIR) color measurements for small ($\leq 40$ km) Trans-Neptunian Objects (TNOs) using coordinated and nearly simultaneous observations from the James Webb Space Telescope (JWST) and the Hubble Space Telescope (HST). JWST/NIRCam data provided detections and NIR photometry for faint TNOs, while HST/ACS and WFC3 imaging enabled recovery in the optical, together yielding optical-NIR colors spanning 0.35-3.2 $\mu$m. Thirteen JWST-detected TNOs were recovered in the HST observations, and trailed PSF photometry was used to derive mean magnitudes, colors, and rotational lightcurves. The color distribution of our small cold classical TNO discoveries is narrow and consistent with the occupation of a single reflectance (color) sequence previously identified for larger cold classical TNOs. We find no evidence for a change in this sequence at smaller sizes. In contrast, the dynamically excited TNOs we discovered exhibit a broader range of colors consistent with multiple compositional classes seen at larger sizes. Lightcurve amplitudes were generally low for both dynamical groups in our sample. 2015 GK56, a previously known TNO in our field, displays a large-amplitude, structured lightcurve consistent with a contact binary. Our results indicate that the characteristic color distribution of TNOs extends to smaller sizes than previously studied, suggesting a primordial origin rather than size-dependent collisional processing.
Read the full abstract

All authors Anastasia N. Morgan, Marielle R. Eduardo, David E. Trilling, Wesley C. Fraser, John A. Stansberry, Gary M. Bernstein, Bryan Hilbert, Matthew J. Holman, William M. Grundy, Kevin J. Napier, Stephen C. Tegler, Cesar Fuentes

High Energy Physics (Theory)

5d SCFT fixtures

In the context of the atomic classification of 5d SCFTs building blocks with non-simply laced global symmetries are lacking. In this work we fill this gap by introducing a new class of 5d UV fixed points (the 5d SCFT "fixtures"). Physically, 5d SCFTs fixtures are the result of a Higgs branch RG flow, realized geometrically by M-theory on singular threefolds which we identify explicitly. In particular, this enables the construction of 5d conformal matter SCFTs with non-simply laced flavor symmetry. Along the way, we discuss their circle reduction in relation to 4d theories of class-S type.
Read the full abstract

Optics

Stimulated Brillouin Scattering in InGaP-on-Insulator Waveguides

Stimulated Brillouin Scattering (SBS) is a nonlinear interaction between optical and acoustic waves in solids. First regarded as a parasitic process in optical fibers, it has gathered significant interest in microwave photonic applications such as optical signal processing and narrow linewidth lasers. While many Brillouin system demonstrations have been done on photonic chips, they struggle to provide simultaneous high Brillouin gain and narrow linewidth in a scalable platform, capable of integration with established photonic integrated circuits. Here, we present a novel integrated InGaP-on-SiO$_2$ platform, where a single crystalline InGaP waveguide layer offers superior material properties and strong nonlinearities to support SBS near 1550 nm wavelength. We demonstrate backward SBS with a measured high Brillouin gain coefficient of $588\, \mathrm{W}^{-1}\,\mathrm{m}^{-1}$ and a record narrow linewidth of 5.2 MHz at 9.346 GHz frequency shift. This work establishes a wafer-scale complementary metal-oxide semiconductor compatible fabrication process, paving a scalable path for high gain and narrow linewidth Brillouin photonics with applications such as precision signal processing.
Read the full abstract

All authors Yuyang Xue, Lisa-Sophie Haerteis, Ryan L. Russell, Choon Kong Lai, Benjamin J. Eggleton, Michael J. Steel, Glenn Solomon, Kevin L. Silverman, Moritz Merklein, Andreas Boes, Nima Nader

Computer Vision

DXPR: Depth-Based Vision-LiDAR Cross-Modal Place Recognition Using Vision Foundation Models

We present DXPR, a depth-based cross-modal place recognition (CMPR) framework that uses vision foundation models (VFMs) to match monocular camera queries against a LiDAR map without modality-specific encoders. This enables robots and autonomous vehicles to robustly localize using only cameras within pre-built LiDAR maps, even under severe seasonal, weather, and illumination changes. The key idea is to convert both camera images and LiDAR scans into a unified depth image representation so that a single VFM backbone with an aggregation head can learn modality-invariant global descriptors. To make pairwise metric learning faithful to scene geometry, we introduce a geometry-aware overlap miner: after cross-modal scale alignment of camera and LiDAR depth, we forward-warp measurements between views to compute a pixel-level overlap score. This score relabels ambiguous pairs and adaptively modulates the positive margin in a multi-similarity loss to avoid overfitting on weakly overlapping views. Extensive experiments on KITTI odometry and Boreas demonstrate strong performance and robustness across seasons, weather, and day/night. On KITTI, DXPR achieves near-perfect Recall@1 on most sequences and outperforms prior CMPR baselines. On Boreas, DXPR achieves intra-sequence performance on par with a strong single-modal baseline (DINOv2-SALAD), while showing clear improvements in the more challenging inter-sequence setting. Compared with RangeBEV, our method consistently performs better in both intra- and inter-sequence evaluations, demonstrating robustness under diverse seasonal and illumination changes.
Read the full abstract

All authors Yungsoo Han, Youngseok Jang, Seungwon Roh, Jeongyeon Seo, H. Jin Kim

Computation and Language

Evaluation of Contextual Understanding in Large Language Models

Large Language Models (LLMs) demonstrate impressive performance across diverse NLP tasks, yet their ability to exhibit genuine contextual understanding remains uncertain. Traditional evaluation metrics such as perplexity, BiLingual Evaluation Understudy (BLEU), or surface-level accuracy fail to reveal how well LLMs extract, integrate, and reason over contextual information–a gap particularly critical in question answering, where models must align responses with contextually grounded knowledge rather than memorized associations. We propose a novel knowledge graph-based evaluation framework introducing Semantic Structural Similarity for KGs (S3KG), a hybrid similarity measure integrating structural and semantic similarity into a continuous evaluation score, alongside a diagnostic framework for categorizing reasoning errors. To validate this pipeline, we evaluate S3KG against established metrics on a curated question-answer (QA) benchmark, demonstrating its effectiveness in measuring correctness, faithfulness, and interpretability in LLM-generated responses.
Read the full abstract

All authors Subavarshana Arumugam, Mamta Nallaretnam, Kithuni Wickramasinghe, Chamath Gunapala, Pragatheeswaran Vipulanandan, Uthayasanker Thayasivam, Kamal Premaratne

Atomic Physics

Simultaneous sub-Doppler laser cooling and optical trapping of bosonic $^{39}$K-$^{133}$Cs and $^{41}$K-$^{133}$Cs mixtures

We report simultaneous sub-Doppler cooling and optical dipole trapping of $^{39}$K-Cs and $^{41}$K-Cs mixtures. Both mixtures are cooled to temperatures $\sim10$ $\mathrm{\mu}$K, achieving performance comparable to that obtained with each species individually. To the best of our knowledge, this constitutes the first realization of a laser-cooled and optically trapped $^{41}$K-Cs mixture. For the $^{39}$K-Cs mixture, we additionally implement parallel spin-resolved Feshbach spectroscopy enabled by Stern-Gerlach separation. For potassium isotopes, we implement sub-Doppler cooling with $D_1$-line gray molasses, while for cesium the entire cooling sequence is implemented using $D_2$ transitions. We spin-polarize the atoms and confine Bose-Bose mixtures in a 1064 nm optical dipole trap. Using the resulting $^{39}$K-Cs samples, we observe 14 heteronuclear Feshbach loss features. Five agree with resonances reported previously, while nine have, to the best of our knowledge, not been observed experimentally before, including $p$-wave features and resonances in additional spin channels. The shorter lifetime of the $^{41}$K-Cs mixture in the optical dipole trap currently hinders systematic Feshbach spectroscopy, which we therefore do not pursue in this work. To characterize this limitation quantitatively, we study the decay dynamics of $^{39}$K-Cs and $^{41}$K-Cs mixtures under comparable temperature and density conditions, revealing a substantially stronger nonexponential loss in the $^{41}$K-Cs mixture. The demonstrated preparation of ultracold $^{41}$K-Cs mixtures provides a starting point for Feshbach-resonance and photoassociation spectroscopy of this previously unexplored isotopologue. Such measurements are essential for identifying suitable pathways for magnetoassociation and coherent optical transfer, and ultimately for the production of ultracold ground-state $^{41}$KCs molecules.
Read the full abstract

All authors Mateusz Bocheński, Jakub Dobosz, Koray Dinçer, Paweł Arciszewski, Mariusz Semczuk

Databases

Factorized and Vectorized Execution: Optimizing Analytical and Semantic Queries over Relations

Many-to-many joins are central to analytical and semantic workloads such as fraud detection, network analysis, and recommendation, where insights arise from relationships between entities. These workloads often suffer from an explosion of intermediate results, sometimes orders of magnitude larger than the inputs. Factorized representations address this problem by exploiting conditional independence among attributes to encode intermediates more compactly. In some cases, they can reduce the output size asymptotically below the worst-case output size. However, adopting factorization in modern vectorized query processors remains challenging: factorized representations are hierarchical, whereas vectorized execution is built around flat, block-oriented processing. Prior approaches either rely on full materialization or support only restricted factorization layouts, sacrificing much of the benefits of both factorization and vectorization. We present FFX, a novel engine for Fast Factorized eXecution. FFX is the first pipelined engine to support arbitrary factorization schemes while preserving full vectorization. The engine introduces packed factorized vectors and operators that maintain cache-friendly, contiguous layouts. Beyond analytics, FFX also co-optimizes semantic operators by serializing factorized intermediates into compact prompts for large language models (LLMs), substantially reducing token usage and inference cost while maintaining output quality and, in some cases, improving it. Together, these contributions enable efficient execution of join-heavy analytical queries, including queries augmented with semantic operators.
Read the full abstract

Artificial Intelligence

Deposon: An Auditable, Conservation-Guaranteed, Game-Theoretically Tested Scattering Layer over LLM Reasoning Paths

Multi-step LLM reasoning lacks a machine-recheckable ledger: discarded reasoning paths leave no auditable record. We propose the Deposon scattering layer, which binds each node of an LLM-generated concept-decomposition graph to a two-parameter Deposon state; paths undergo three-channel scattering – transmission, reflection, irreversible dissipation – obeying T+R+A=1 for arbitrary parameters, with a maximum per-path energy-audit deviation of 2.2E-16 (machine epsilon). We report all three evidence tiers honestly. On synthetic trap benchmarks the path-filtering gain is closed (pre-registered): unified reaches 100% versus a decoy-capture baseline at 7%/10%. On real benchmarks the layer is indistinguishable from a trivial six-keyword rule filter (GSM8K 0.87 >= 0.85, McNemar p=0.5; StrategyQA 0.899 = 0.899); no difference is detected here, so we sharpen the claim to "the differential value lies solely in machine verifiability." Fusion yields a second negative result: convex combinations with a semantic prior never improve (physics 0.484 -> 0.452), and the apparent lambda=2 gain is an anti-field artifact; any fusion gain must be nonlinear. Modeling the reverse dynamics as a potential game on the graph, we evidence an auditable scalar's monotonicity and near-gradientness and quantify the empirical coordination ratio (ECR). The three formalized dynamical-equivalence propositions (P1a/P1b/T-P1c) are falsified under the pre-registered kill protocol, and the potential-game claim is downgraded to approximate (cyclic-graph median residual 0.669): only consistency-level evidence survives at the dynamical level. Code: github.com/zeroandcat/Deposon.
Read the full abstract

Systems and Control

Intelligent Fault and Lightning Detection Algorithm for VSC-MTDC grids based on ResNet with Hybrid Attention Mechanism

To address the existing challenges in fault detection for voltage source converter-based multi-terminal DC (VSC-MTDC) grids, this paper proposes an intelligent fault and lightning detection algorithm based on S transform and Residual Network with hybrid attention mechanism (RWHAM).The DC line double-ended initial current traveling waves (ICTWs) are first converted into time-frequency matrices by performing S transform and then visualized as a two-dimensional image. The image effectively characterizes the time-frequency features of ICTWs under different fault and lightning conditions, making it easier to extract critical features for the models. Next, the RWHAM model is constructed and the images are fed into the model to detect internal and external faults and lightning interference. The weight of the key information in the image is increased by hybrid attention mechanism, which in turn improves the fault detection ability of the RWHAM model. Extensive simulations involving 21,620 distinct cases on PSCAD/EMTDC validate the algorithm's high accuracy and rapid response across different types of faults and lightning interference. Furthermore, it exhibits excellent generalization ability when the parameters of VSC-MTDC grids change.
Read the full abstract

Computer Vision

Concentrate After Imagination: Text-Conditioned Evidence Grounding for Partially Relevant Video Retrieval

Partially Relevant Video Retrieval (PRVR) retrieves untrimmed videos when queries describe only short moments. Although recent methods improve local representations, uncertainty modeling, and global context, final ranking often still trusts the strongest local response; a coincidentally similar fragment can therefore produce an unsupported peak. We identify this failure as the query-agnostic concentration bottleneck and propose TRACE, a score-level evidence verification operator for PRVR. Given a query and global video registers, TRACE activates query-relevant registers, routes their support to frame-level evidence, and smoothly marginalizes alternative query-to-register-to-frame paths before localized temporal selection. Unlike representation-level feature fusion, TRACE uses this evidence only as a query-conditioned residual calibration of the original local score. On ActivityNet Captions, Charades-STA, and TVR, TRACE achieves the best SumR on all three benchmarks and improves the DreamPRVR backbone by 1.2, 1.1, and 1.5 points, respectively. Ablation, routing-corruption, hard-negative, and cross-backbone transfer analyses support the interpretation that the gains arise from query-conditioned evidence verification rather than a generic score offset.
Read the full abstract

All authors Shuaiqi Cheng, Siyu You, Yanbi Wu, Yuxi Chen, Jiahao Zhang, Xuming Hu

Quantum Physics

No information transmission through quantum channels above capacity

We show that the capacity of a quantum channel demarcates a phase transition: while reliable transmission below capacity is always possible, any attempt to transmit information above it fails catastrophically. Specifically, we prove exponential strong converse theorems for unassisted quantum and classical communication over arbitrary finite-dimensional memoryless quantum channels. At rates beyond the respective capacity, the entanglement-generation fidelity and the success probability for classical communication decay exponentially with the number of channel uses. This rules out transmission above capacity even when one tolerates arbitrarily large errors. Our proof follows the classical Arimoto strategy, augmented by a crucial new ingredient: integral representations of Rényi information measures that lead to asymptotic continuity bounds for Rényi capacities.
Read the full abstract

High Energy Physics (Phenomenology)

Final-state effects on the transverse-momentum spectra of charged hadrons in $p+Pb$ and $Pb+Pb$ collisions at $\sqrt {s_{\rm{NN}}}~=~5.02$~TeV using the modified Tsallis distribution

The ATLAS Collaboration has reported measurements of the transverse momentum ($p_{\rm{T}}$) spectra of charged hadrons in proton-proton ($p+p$), proton-lead ($p+Pb$), and lead-lead ($Pb+Pb$) collisions at a nucleon-nucleon center-of-mass energy of $\sqrt{s_{\rm{NN}}}~=$~5.02~TeV within the rapidity interval $-2.5<y<2.0$. In the present work, we investigate medium effects on charged-hadron production in $p+Pb$ and $Pb+Pb$ collisions using a phenomenological modified Tsallis parametrization that accounts for transverse collective flow in the low-to-intermediate $p_{\rm{T}}$ region and medium-induced parton energy loss at high $p_{\rm{T}}$. The modified parametrization provides a consistent description of the measured spectra over the full investigated $p_{\rm{T}}$ range across different centrality classes. The extracted energy-loss exponent $\alpha$, which characterizes the energy dependence of parton energy loss, is found to lie in the range 0.59$-$0.73 for $p+Pb$ collisions and 0.32$-$0.59 for $Pb+Pb$ collisions. In addition, the extracted parameters exhibit a clear centrality dependence, reflecting enhanced collective effects in central collisions. These results provide quantitative insight into the interplay between transverse collective flow and medium-induced parton energy loss in charged-hadron production at LHC energies.
Read the full abstract

All authors Kapil Saraswat, Deependra Singh Rawat, Manoj Kumar Singh, Prashanta Kumar Khandai, Damini Singh, Venktesh Singh

Mesoscale and Nanoscale Physics

Probing Antialtermagnetism via Orbital-Field-Induced Spin Splitting

Layer compensation can conceal spin polarization behind a spin-degenerate bulk spectrum in antialtermagnets, preventing spectroscopic identification of the underlying magnetic order. Here we show that the orbital effect of an in-plane magnetic field converts the hidden spin texture into an observable spin splitting with momentum parity opposite to that of the underlying exchange order, thereby restoring spectroscopic access. Once the parity is determined, complementary response functions in the weak-field regime further resolve its wave character. Within a minimal model, we illustrate that a sum rule over the difference of spectral functions between opposite spins isolates the momentum gradient of the form factor and differentiates hidden even-parity orders, whereas for odd-parity sectors the induced net spin polarization exhibits characteristic field-amplitude and angular dependences. These results establish orbital-field-induced spin splitting as a generic route to identifying the hidden altermagnetic order parameter without layer resolution.
Read the full abstract

General Relativity and Quantum Cosmology

Second-order Teukolsky calculations for nonspinning, quasicircular binaries

Currently, the only second-order gravitational self-force calculations have been based on directly solving the perturbative Einstein equations in the Lorenz gauge. That method relied on the complete separability of the Einstein equations in a Schwarzschild background. In this paper, we present a new scheme based on the second-order Teukolsky equation. Crucially, this method promises to extend (reasonably straightforwardly) to the more realistic case of a Kerr background. Here we implement the scheme in the simplest setting of quasicircular orbits around a Schwarzschild black hole. In addition to working with the Teukolsky equation, our scheme incorporates several other advances over previous second-order self-force calculations: compactified hyperboloidal slicing, transformation to a Bondi-Sachs gauge, and a combination of spectral and variation-of-parameters methods. We also use these tools to re-examine the infrared divergences that arise in second-order Lorenz-gauge calculations, showing they are less pronounced in the Teukolsky case and completely eliminated in the Bondi-Sachs gauge. We conclude by calculating the asymptotic energy fluxes and benchmarking them against previous Lorenz-gauge calculations.
Read the full abstract

All authors Benjamin Leather, Andrew Spiers, Adam Pound, Samuel D. Upton, Barry Wardell, Leanne Durkan, Niels Warburton

Combinatorics

Strongly separating graph edges with $10n$ paths

A family of paths strongly separates the edges of a graph if every two distinct edges are separated in both directions by paths in the family. Bonamy, Botler, Dross, Naia, and Skokan proved that every $n$-vertex graph admits such a family of at most $19n$ paths. We improve this bound to $10n-o(n)$.
Read the full abstract

High Energy Physics (Phenomenology)

A Vector-Like Lepton Interpretation of the High-Energy Nuclear Recoil Candidate in LUX-ZEPLIN

The LUX-ZEPLIN (LZ) experiment has recently reported a nuclear-recoil candidate at $E_R=248\pm23_{\rm stat}\pm23_{\rm sys}\,{\rm keV}$ in an exposure of $2.84\,{\rm tonne\,yr}$. At this unusually high recoil energy, the spin-independent (SI) xenon response is strongly suppressed, motivating dark matter interactions with a harder recoil spectrum. We show that elastic spin-dependent (SD) scattering through the nonrelativistic operator $\mathcal O_4=\mathbf S_{\rm DM}\cdot\mathbf S_N$ provides a simple realization of such a spectrum. We consider singlet-doublet Majorana dark matter, for which a diagonal $Z$ coupling generates the required SD interaction. The accompanying Higgs-mediated SI interaction would generically produce too many low-energy recoils, but can be suppressed along a Higgs blind spot while retaining a nonzero $Z$ coupling. We find a region of the Higgs blind spot that simultaneously predicts an $\mathcal O(1)$ event rate in the LZ high-recoil window and reproduces the observed thermal relic abundance. The same SD interaction leads to solar capture, allowing this region to be tested independently with solar-neutrino searches. Current IceCube limits already probe part of the LZ-compatible thermal parameter space, while a viable region remains. Additional xenon exposure and improved solar-neutrino searches can therefore provide complementary tests of this interpretation.
Read the full abstract

Machine Learning

Physics-Informed Deep Learning for False Ventricular Tachycardia Alarm Reduction in the ICU

False ventricular tachycardia (VT) alarms are a leading contributor to alarm fatigue in intensive care units. We propose a deep learning framework combining a 1D SE-ResNet with ICU-realistic data augmentations and a physics-informed auxiliary reconstruction task based on the three-element Windkessel hemodynamic model, implemented as a differentiable forward simulation. By requiring the network's latent representation to produce physiologically plausible arterial pressure waveforms, artifact-driven ECG patterns are penalized while true VT remains coherent across modalities. Evaluated on the VTaC benchmark under a strict real-time protocol (10-second pre-alarm window), our method achieves a 5-point Challenge Score improvement over prior state-of-the-art. Ablation studies confirm that the physics-informed objective is the primary performance driver, providing gains in accuracy, 2x label efficiency, and more localized and clinically meaningful ECG segments.
Read the full abstract

Probability

Endpoint and Vanishing-Density Asymptotics for Hardy–Szegő Zero Counts

We study the number \(N_I(L)\) of zeros of the Hardy–Szegő zero process in the horizontal window \([0,L]\times I\) as \(L\to\infty\), where \(I=[\alpha,\beta]\Subset(0,\infty)\) is fixed. We obtain sharp point-probability asymptotics at the lower endpoint of the density scale. In the fixed-count regime, for every fixed integer \(k\geq0\), we determine a full asymptotic formula for \(\mathbb P\{N_I(L)=k\}\), identifying its exponential rate, order-one correction, and \(k\)-dependent polynomial prefactor; the case \(k=0\) gives the hole probability. In the vanishing-density regime, we prove a uniform local asymptotic formula for \(b_L\leq n\leq\varepsilon_L L\), where \(b_L\to\infty\), \(\varepsilon_L\downarrow0\), and \(b_L\leq\varepsilon_L L\), identifying the large-deviation exponent, endpoint correction, and Gaussian prefactor.
Read the full abstract

Computational Geometry

A Sublinear Approximation Algorithm for Minimum Dilation Trees in the Plane

The dilation of a geometric graph measures how much longer the path between pairs of points becomes when restricted to graph edges, rather than following the direct path through the ambient space. The minimum dilation tree of a point set is the spanning tree with minimum dilation, where edge lengths in the tree are given by distances in the ambient space. In the Euclidean plane, computing the minimum dilation tree is NP-hard, but no hardness of approximation result is known. On the other hand, the minimum spanning tree is an $(n-1)$-approximation to the minimum dilation tree, but no asymptotically-better approximation algorithm is known for general point sets in the Euclidean plane. We give the first sublinear approximation algorithm for the minimum dilation tree in the Euclidean plane. Our approximation ratio is $\tilde{O}(n^{14/15})$ and our algorithm runs in polynomial time. This resolves an open problem proposed by Eppstein in 1996.
Read the full abstract

All authors Sarita de Berg, Jacobus Conradi, Peter Kramer, André Nusser, Sampson Wong

General Relativity and Quantum Cosmology

TARTARUS: A High-Performance Python Code for Ray Tracing in Curved Spacetimes

The recent observations of the supermassive black holes M87* and Sagittarius A* by the Event Horizon Telescope collaboration (EHT) have shown the need for robust, efficient, and accessible numerical tools to model light propagation in strongly curved spacetimes. Comparing theoretical accretion models with observational data relies heavily on ray-tracing algorithms that calculate the trajectories of photons traveling from a source plasma to a distant observer. In this paper, we present TARTARUS (Tracer for Astrophysical Ray Trajectories Around Relativistic Ultra-compact Sources), a new, highly modular, and extensible Python-based framework for computing null geodesics around compact objects. By leveraging Just-In-Time (JIT) compilation via Numba, TARTARUS bridges the gap between Python's high-level accessibility and the execution speed of compiled languages like C++ or FORTRAN. The code supports multiple native solvers (including adaptive embedded Runge-Kutta pairs, a Bulirsch-Stoer extrapolator, and a Verlet scheme), event-detection mechanics for handling structural intersections (e.g. event horizons and accretion disks), and support for both analytical and numerically generated background metrics. We demonstrate the code's physical accuracy and computational efficiency through some tests, including the evaluation of the Hamiltonian constraint conservation and ray-tracing of the shadow and the Novikov-Thorne thin accretion disks around a Kerr black hole.
Read the full abstract

Astrophysics of Galaxies

Triaxiality of the Milky Way dark matter halo and the microlensing optical depth towards the Large Magellanic Cloud

The microlensing optical depth towards the Large Magellanic Cloud (LMC), $\tau_{\rm LMC}$, bounds the compact-object fraction of the Galactic dark matter halo, but is almost always evaluated for a spherical halo. We compute it for triaxial haloes and show that the shape dependence is carried entirely by the projection of the line of sight onto the halo principal axes: for a scale-free halo $\rho\propto R^{-\alpha}$ at fixed local density it factorises exactly as $\tau/\tau_{\rm sph}=S_\odot^{\alpha}\langle S^{-\alpha}\rangle_w$, with $S(\Omega)$ the ellipsoidal shape function. The LMC lies only $0\fdg78$ from perpendicular to the Sun-Galactic Centre line, making it a near-optimal probe of in-plane elongation and a poor probe of vertical flattening. For the halo shapes favoured by Gaia we find $\tau_{\rm LMC}/\tau_{\rm sph}=0.95$–$1.15$ at fixed local density, $1.57$ for the Law-Majewski logarithmic potential, and $0.67$–$1.54$ when the major-axis azimuth is left free. For the Einasto halo required by the declining Gaia DR3 rotation curve, truncation near $25\kpc$ is offset by the larger local density it implies and $\tau$ falls by only $7$–$15$ per cent; but the median lens distance moves inwards from $13.4$ to $8.5\kpc$, where the logarithmic slope is steep and rising, amplifying the shape sensitivity: the Gaia RR~Lyrae enhancement grows from $15$ to $23$ per cent and the orientation-free range widens to $0.62$–$1.62$. The systematic on the inferred MACHO or primordial-black-hole fraction is two-sided, of order $\pm15$ per cent for the measured shapes and a factor $2.6$ under full orientation freedom, exceeding the statistical errors of the EROS-2 and OGLE bounds.
Read the full abstract

Other Condensed Matter

Exact Phase-Space Rotation in the Trapped Quantum Calogero Model

We develop a microscopic phase-space description of the quantum Calogero model in the presence of an external harmonic confining potential. Building on the quantum Lax-pair structure, we construct a Hermitian Wigner operator whose expectation value obeys the exact phase-space evolution equation d_t rho + lambda d_x rho - Omega^2 x d_lambda rho = 0 for arbitrary initial states and to all orders in the interaction strength. The resulting dynamics is a rigid rotation in phase space with period 2 pi/Omega, providing a microscopic realization of the isochronous dynamics of the trapped Calogero model. We further show that the moments of the phase-space density form rotating multiplets rather than independent conserved quantities. In particular, within the quadratic sector, the unique conserved combination is proportional to the trapped Hamiltonian, providing a nontrivial consistency check of the construction. In the limit Omega -> 0, the equation reduces to the exact free-streaming equation of the untrapped model.
Read the full abstract

Analysis of PDEs

Well-posedness and exponential stability for abstract evolution equations with delay in the nonlinear source: frictional and viscoelastic cases

In this paper, we establish the well-posedness and exponential stability of a class of semilinear neutral abstract evolution equations with constant time delay. Two damping mechanisms are considered: frictional damping of the form $CC^*u_t$ and viscoelastic damping with memory. Under suitable assumptions, we prove global well-posedness and exponential decay of solutions for sufficiently small initial data. This result is of considerable importance, as it provides a unified framework for the analysis of a wide class of neutral systems arising in structural mechanics, seismic isolation, and control theory. The abstract results are illustrated by some concrete examples. The analysis relies on a combination of semigroup theory, energy estimates, Duhamel's formula, and Gronwall-type arguments.
Read the full abstract

Analysis of PDEs

Non-uniqueness for the Complex Ginzburg-Landau Equation

We show that singularities of the three-dimensional cubic complex Ginzburg-Landau equation developing from smooth initial data can induce non-uniqueness of solutions after the time of blowup. The solutions we study start as backward self-similar solutions, and the source of the non-uniqueness is an unstable eigenvalue of the linearization around the forward self-similar profile generated by the singularity. We establish the existence of this eigenvalue, and hence the non-uniqueness, by a computer-assisted proof for specific values of the parameters in the equation. The work is in part motivated by similarities to conjectured behavior for the Navier-Stokes equation, with which the complex Ginzburg-Landau equation shares several important properties, such as an energy inequality and the scaling symmetry.
Read the full abstract

Mathematical Physics

The inverse problem of time-dependent density functional theory on the torus

We prove that, for any given sufficiently regular one-particle density, there exists a unique external potential for which the solution to the many-body Schrödinger equation has this prescribed density. This implies in particular that one can exactly reproduce the time-dependent density of an interacting system using non-interacting electrons, which is at the heart of time-dependent density functional theory. Our result covers the Coulomb interaction and densities arising from extended nuclei.
Read the full abstract

Materials Science

Time-Dependent Density Functional Theory with Coulomb Interactions

We provide the first fully rigorous justification of time-dependent density functional theory in the continuum: Given a time-dependent density, we prove that there exists an external potential, unique up to a time-dependent constant, such that the corresponding Schrödinger equation reproduces the given density. This potential can be obtained with an iteration scheme. Our argument covers Coulomb interactions and does not need any Taylor expansion in time. It instead requires analyticity in space for the density and potential, which is compatible with extended nuclei.
Read the full abstract

Human-Computer Interaction

Embedded Human-Centered Data Science in a Graduate Programming Course: A Framework and Case Study

As AI and data-driven systems pervade practice, there is an imperative for instructors to embed societal impact and ethics content into computing courses. In response, we present the Human-Centered Education for Learning in Information and eXplainable Computing (HELIX) framework for information science programs, organized around three iterative pillars - knowledge building, decision-making, and empowerment - with concrete actions for instructors and students. We applied the framework in a graduate, introductory programming course using readings, algorithmic design activities, and scenario-based reflections. We present a pilot implementation of this framework to examine changes in students' (n=22) knowledge acquisition, decision-making processes, and self-reflection regarding human-centered perspectives in data science. We release an anonymized materials kit (survey, assignments, analysis code) to support adoption. We discuss design tensions (workload, assessment, relevance to diverse information science learners) and provide guidelines for integrating human-centered content without overwhelming technical outcomes. Findings suggest that the HELIX Framework is feasible in information science contexts and future work should use comparative survey assessment to strengthen causal inferences.
Read the full abstract

All authors Victoria Chui, Kelly McConvey, Daniel Chui, Malayna Bernstein, Shion Guha

Machine Learning

Transformers as In-Context Samplers: From Closed-Form Diffusion to Estimation-Free Sampling

A growing body of work establishes that large language models are not mere statistical memorizers, but are capable of in-context learning: performing inference at test time using only examples provided in the prompt, without any parameter updates. Prior theoretical work has shown that this capability extends to supervised learning tasks such as linear regression. We prove that in-context learning extends further to data generation: frozen transformers can simulate iterative generative samplers from in-context samples. We first show that transformers can realize closed-form and smoothed closed-form diffusion samplers. The construction identifies a concrete generative role for softmax attention: it computes responsibility weights and weighted empirical averages, while feedforward layers implement Euler updates. To empirically relate these constructions to pretrained language models, we study semantic-topic sampling: prompts consisting of words drawn from a common semantic category, such as animals, foods, or cities. Across transformer layers, the normalized hidden states exhibit a two-stage geometry: they move toward a uniform spherical reference in intermediate layers and then return to structured, topic-dependent representations near the output. We further measure an interacting-particle energy on these hidden-state clouds and observe the same U-shape pattern. We then prove that transformers can approximate an energy-based sampler, constructing the same U-shape energy across the layers.
Read the full abstract

All authors Arman Adibi, Alireza Jafari, Mohammad Ghavamzadeh, Hadi Daneshmand