Computer Vision
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Friday 28 August
Large Vision-Language Models (LVLMs) remain prone to hallucinations, producing responses that are irrelevant or inconsistent with the multimodal input. Existing mitigation methods mainly rely on external supervision, output calibration, or attention regulation, leaving the internal representation dynamics of autoregressive generation underexplored. We identify an inference-time failure mode in which cross-modal representations degrade across decoder layers and drift across generation steps, destabilizing token prediction and increasing hallucination risk. We propose Dynamic Alignment Compensation (DAC), a training-free inference-time method that detects representation divergence and selectively applies lightweight residual compensation. DAC combines Layer-wise Semantic Compensation to mitigate inter-layer degradation with Sequential Semantic Correction to constrain temporal drift. Experiments on nine hallucination-focused and general-purpose multimodal benchmarks across multiple LVLM backbones show that DAC consistently reduces hallucinations while maintaining strong overall performance.
High Energy Physics (Phenomenology)
Quantum computation of partonic Drell-Yan scattering cross sections and interference effects
We probe the possibilities of efficiently constructing simple Feynman diagrams into quantum devices. More precisely, we study Drell-Yan lepton pair creation at the partonic level of the form q qbar -> gamma/Z -> l- l+. We develop quantum gates that build up the relevant diagrams using simple Feynman rules, such as vertex and propagator gates V and P. We show how the quantum circuit may compute simultaneous amplitudes in the phase space and how to reach the full integrated cross section from the outputs. In addition to this, we also show how the circuit is able to simultaneously isolate the interference effects of the contributing diagrams by a simple basis rotation. The circuit design is made to be general, and thus this work constitutes a step towards the implementation of arbitrary scattering process computations and efficient interference analyses.
Read the full abstract
Systems and Control
Anytime Primal–Dual Certification of the Maximum Disturbance Radius in Robust MPC
Adjustable-set robust model predictive control (MPC) characterizes a state-dependent maximum disturbance radius, which can be interpreted as a certified robustness reserve. Existing methods primarily focus on optimizing and propagating this reserve under in-set disturbances. This letter investigates how much reserve remains after a finite out-of-set disturbance without immediately re-solving the full optimization problem. To this end, we propose a two-sided reserve-depletion envelope formed by independent primal and dual correction hierarchies, which supply monotone lower and upper bounds, respectively. The envelope remains valid across active-set changes, has an online-computable width, and contracts monotonically under subspace expansion. Leveraging its finite-step exactness, we present a basis-first adaptive algorithm that operates in an anytime manner: every completed reduced solve returns a valid certificate, enabling early termination once a prescribed tolerance is reached. Across the tested cases, numerical studies report no certificate violations and median speedups of up to 4.97x over warm-started full re-optimization at a 2% certificate-width tolerance.
Read the full abstract
Soft Condensed Matter
Weakly non-linear creep of amorphous polymers near their glass transition, comparisons between models and experiment
The non-linear mechanics of amorphous polymers near the glass transition reveals a stress-induced acceleration of stress relaxation of nanometric sub-units. Recent theoretical work predicts that the local acceleration within these nano-domains should scale as the exponential of the squared local stress, a behavior now supported by experiments. However, this local dynamics has some complex consequences on the macroscopic mechanical response, as dynamical heterogeneities generate complex stress and strain fields in polymers close to the glass transition. In this study we consider the non-linear creep of an amorphous polymer near its glass transition and evaluate the relation between local and global acceleration and the emerging load-carrying structure, by comparing experimental data with predictions of three models of increasing complexity: a two-states (2S) model, a self-consistent (SC) model and a finite-element (FEM) model. The experimentally observed trend of accelerated creep under increasing applied stress is reproduced by the SC and FEM models, while the 2S model overestimates stress localization. The macroscopic, homogenized acceleration is predicted to be close to the microscopic one, albeit with an apparent yield stress that depends on compliance. The FEM model evidences the development of a load carrying sub-structure that occupies a small fraction of the total material volume driven by the interaction of sub-domains. This work shows that complexity and heterogeneity emerge due to non-linear interactions and that their adequate representation is essential for predicting the macroscopic mechanical response of amorphous polymers near their glass transition.
Read the full abstract
Strongly Correlated Electrons
Spin-Selective Spectral Flattening and Wave-Packet Dynamics in a Flux-Engineered Lieb Lattice
We investigate reversible internal-state-selective wave-packet transport induced by spin-dependent Peierls phases in a two-dimensional nearest-neighbor Lieb lattice. The two conserved spin components experience effective fluxes $\alpha_{\sigma}=\alpha_{0}+s_{\sigma}\alpha_{s}$, where $s_{\uparrow,\downarrow}=\pm1$. At the working point $\alpha_{0}=\alpha_{s}=1/4$, the spin-up and spin-down components experience $\alpha_{\uparrow}=1/2$ and $\alpha_{\downarrow}=0$, respectively. A band-resolved calculation in the $q=2$ magnetic unit cell shows that the spin-up spectrum contains two zero-energy flat subbands associated with the sublattice-imbalance flat-band sector, whereas the remaining four subbands retain finite bandwidths. The half-flux sector therefore fails the all-bands-flat condition and does not realize exact Aharonov–Bohm caging for a generic localized initial state. Nevertheless, real-time simulations reveal a pronounced suppression of spin-up propagation relative to the dispersive spin-down component, manifested by a smaller mean-square displacement and an enhanced finite-region retention probability over the pre-reflection time window. Reversing the state-dependent flux interchanges the slow and fast spin channels, while the dynamical contrast remains robust against moderate flux detuning. These results establish spin-dependent synthetic flux as a reversible means of controlling internal-state-resolved matter-wave transport without spin-flip processes or interactions, and provide complementary spectral and real-space criteria for distinguishing exact caging from finite-time dynamical slowing in atomic and photonic flat-band simulators.
Read the full abstract
Computation and Language
CNeo-Bench: Diagnosing Large Language Models on Chinese Neologisms
Chinese neologisms exploit diverse and unique linguistic mechanisms, such as phonetic substitution (e.g., 886 for “bye-bye”) and visual character decomposition that are rare in other languages. We introduce CNeo-Bench, a benchmark of 4,759 such neologisms with reference definitions, organized into five top-level categories and nine subcategories by the linguistic mechanism behind each expression. CNeo-Bench is paired with a two-tier evaluation framework that separates whether a model can describe a neologism from whether it can operate on its underlying mechanism. Evaluating 18 LLMs, we find that Chinese neologisms remain an open challenge; most models fall below 40% on definition generation, and on several subcategories a systematic recognition-manipulation gap emerges: models describe neologisms correctly but, in source-form restoration tasks, substitute a semantic equivalent (paraphrase) for the source form rather than producing the source form itself. A few-shot analysis on 1,058 hard items shows that in-context examples can solve many difficult cases, but leave a noticeable portion of errors remaining, indicating challenges beyond prompting alone can address.
Read the full abstract
Machine Learning
Explainable Uncertainty Estimation for Reliable Medical AI
Artificial intelligence has strong potential to support clinical decision-making, yet its adoption in healthcare remains limited due to a lack of trust. Uncertainty estimation can signal unreliable predictions, and explainable AI (XAI) can clarify how predictions are made but existing methods treat them separately, providing no feature-level insight into why a prediction is uncertain or which tests to prioritize to reduce it. To address this gap, we propose explainable uncertainty estimation, which unifies uncertainty estimation and XAI to both quantify uncertainty and explain feature-level contributions. We introduce the Expected Gradients Reconstruction Uncertainty Estimate (egRUE), which incorporates prediction explanations into its uncertainty computation and decomposes uncertainty into feature-wise contributions. We prove theoretical properties of egRUE and show through experiments that it improves reliability and interpretability compared to existing methods. A user study with medical experts further demonstrates that egRUE's explanations improve calibrated trust over uncertainty scores alone, increasing confidence in correct predictions and reducing confidence in incorrect ones. By combining prediction uncertainty with feature-level explanations, egRUE strengthens decision-making support in safety-critical healthcare settings, clarifying both when predictions may be unreliable and which features drive that uncertainty.
Read the full abstract
Solar and Stellar Astrophysics
CoBiToM project. III. Physical characterisation of the potential merger candidates AL Lep, ASAS J082243+1927.0, and ZZ PsA
Contact binaries (CBs) are key systems for investigating the late evolutionary stages that may lead to stellar coalescence. Systems with an extremely low mass ratio and high fill-out factor are of particular interest as potential merger candidates and possible progenitors of luminous red novae. A combined photometric and spectroscopic study of the CB systems AL Lep, ASAS J082243+1927.0, and ZZ PsA is presented. The aim is to derive self-consistent orbital and physical parameters, assess their evolutionary state, and examine whether they have reached the terminal and unstable phase of their evolution. High-resolution VLT/UVES spectra were analysed to derive radial velocity values, spectroscopic mass ratio, and metallicity. The results were combined with multi-band light-curve modelling within a Monte Carlo framework to determine the system geometries and absolute parameters. Orbital period variations were investigated through O-C analysis, while the evolutionary status of the systems was assessed using the Darwin instability criterion, critical mass ratio estimates, fill-out factor, and orbital period variation. Mass ratio values of q = 0.1332, 0.1041, and 0.0789 were derived for AL Lep, ASAS J082243+1927.0, and ZZ PsA, respectively. The corresponding primary mass values are M1 = 1.656 +/- 0.081 Msun, 1.275 +/- 0.078 Msun, and 1.795 +/- 0.111 Msun. The locations of their components show significant deviations from single-star ZAMS and TAMS reference curves, indicating strong binary interaction and energy redistribution through the common envelope. The spin-to-orbital angular momentum ratio for AL Lep and ASAS J082243+1927.0 remains below the standard Darwin instability threshold, while ZZ PsA lies close to the instability regime and may be consistent with an unstable configuration within the derived uncertainties.
Read the full abstract
Human-Computer Interaction
Too Much of the Same: From Algorithmic to Human Bias in Learning to Defer
Learning to Defer (LtD) extends supervised learning by allowing a Machine Learning (ML) model to defer harder or less confident decisions to a human expert. Despite being geared for human-AI collaboration, LtD strategies neglect the potential negative interference of human cognitive biases. Our contribution is twofold. First, we demonstrate that standard LtD strategies show class-dependent sampling bias in classification tasks in practice, and thus may disproportionately defer the minority classes when applied to imbalanced datasets. Second, we show that such asymmetries in task delegation may trigger human biases, ultimately leading to poorer downstream decision making. Specifically, we conduct a user study ($N=226$) where participants complete a classification task on a set of deferred items, with conditions presenting different levels of class imbalance. Our results show that participants exposed to a highly imbalanced rejection set achieved lower classification accuracy in the majority class compared to those exposed to a more balanced set, regardless of which class constituted the majority. Exploratory analyses suggest that this may be an instance of the Test-taker's effect, which stems from a mismatch between the actual distribution of classes and the participants' expectations about that distribution. Finally, we discuss the implications of these findings for the deployment of LtD algorithms.
Read the full abstract
High Energy Astrophysical Phenomena
Spherical Accretion onto Konoplya Zhidenko Black Holes: Sonic Point Analysis and Fluid Dynamics
We investigate the spherical accretion onto a static Konoplya Zhidenko black hole and examine how the deformation of the spacetime affects the dynamics of the accreting matter. By applying the conservation laws of particle number and energy-momentum, we formulate the accretion problem and derive the corresponding Hamiltonian describing the fluid motion. The critical points and sonic transitions are then analyzed for ultra stiff, ultra relativistic, radiation, and sub relativistic fluids. We study the dependence of the critical point structure and flow trajectories on the KZ deformation and compare the results with those obtained for asymptotically safe, Schwarzschild MOG, and Reissner Nordström black holes. The comparison shows that the location and behavior of the critical points vary significantly with the underlying BH geometry and the equation of state of the fluid. In particular, the KZ deformation produces noticeable changes in the radial structure of the accretion flow, with different effects observed for the considered fluid models. These results provide a comparative picture of spherical accretion in different black hole spacetimes and help clarify how modifications of the background geometry influence matter dynamics in the strong field region.
Read the full abstract
Hardware Architecture
AI Hardware Accelerators for Large Language Models: Architectures and the Memory Wall
Large language models (LLMs) place unprecedented and still-growing demands on the hardware that trains and serves them. This review surveys the full landscape of AI hardware accelerators for LLMs, including general-purpose GPUs, custom ASICs such as TPUs, Trainium, Groq, and Cerebras, reconfigurable FPGAs, processing-in-memory and near-memory architectures, and emerging neuromorphic and photonic approaches across cloud and edge deployment. Using the transformer's computational structure and roofline analysis as a common framework, we show that the decisive constraint on LLM acceleration is not arithmetic but memory: the autoregressive decode phase is bandwidth-bound, the key-value cache can rival the model weights in size, and data movement dominates energy. Comparing platforms on compute, memory, energy, programmability, and scalability, we find that no single architecture is optimal across workloads: GPUs remain the flexible default and the workhorse of training; domain-specific ASICs win at scale for stable, high-volume workloads; processing-in-memory is the most promising near-term response to the memory wall, entering systems as a heterogeneous complement; and neuromorphic and photonic computing, while promising, are not yet production-ready at frontier scale. Future progress depends on hardware-algorithm co-design and heterogeneous, memory-centric systems: for large language models, the memory system has become the computer.
Read the full abstract
Nuclear Experiment
Doppler-shift attenuation method (DSAM) lifetimes in $^{54}$Cr - a re-evaluation
Background: Contemporary stopping powers for fp-shell nuclei slowing in tantalum can differ by a factor of two from the 1963 Lindhard, Scharff and Schiøtt (LSS) theory [Mat. Fys. Medd. Dan. Vid. Selsk. 33 no. 14, (1963)] used in Doppler-shift attenuation method (DSAM) lifetime measurements dating back to the 1970s. In recent work [Phys. Rev. C 113, 044306 (2026)], it was found that anomalously high collectivity in the $4^+_1 \rightarrow 2^+_1$ transition of 58Fe given in the literature [Nucl. Data Sheets 111, 897 (2010)] could be traced back as due to the use of these historical stopping powers in the Doppler-shift lifetime measurements. Satisfactory agreement with shell-model calculations was obtained from a re-analysis of the measurement of Bolotin et al. [Nucl. Phys. A 311, 75 (1978)] once the stopping powers were replaced by up-to-date values. Purpose: The DSAM measurement on 54Cr by the same group [Nucl. Phys. A 337, 1 (1980)], which used the same experimental methods and procedures, is re-examined. Method: The computer code used in the original DSAM analysis has been rebuilt and upgraded with the capacity to use contemporary stopping powers. E2 transition strengths derived from the revised lifetimes are compared with shell-model calculations. Results: The lifetimes increase by 16% to 27%, with the larger increase generally corresponding to shorter lifetimes where electronic stopping dominates. Conclusions: Revised lifetimes, based on current stopping powers, imply E2 transition rates between states up to the 6$^+_1$ state in 54Cr that compare well with shell-model calculations. The lifetime data together with branching ratios and the shell-model calculations strongly suggest revision of the spins assigned to the levels at 3.786 (currently (4)$^+$) and 4.043 MeV (currently 5$^+$). These levels are suggested to be the 5$^+_1$ and 6$^+_2$ states, respectively.
Read the full abstract
Statistical Mechanics
Emergent aggregation from collective foraging
Collective behaviour in living systems is usually modelled as the outcome of a direct social drive: agents are rewarded, or hard-wired, to align with or approach their neighbours. Here we show that aggregation can instead emerge from an indirect objective. We let reinforcement learning foragers, initially performing a random walk, optimize their dynamics from a purely individual reward for finding replenishable targets, while perceiving only their conspecifics and never the targets themselves. As the visual range grows, the agents undergo a sharp crossover from an environment-tuned individual search to a scale-agnostic collective one, and this crossover coincides with the onset of spatial aggregation. Thus a collective phase arises as a by-product of optimal foraging, without any direct reward for grouping. A minimal analytical first-passage model reproduces the transition as a crossover between the two search strategies. Our results identify indirect, resource-driven reward as a generic route to emergent collective phenomena.
Read the full abstract
Group Theory
Solvability of unimodular equations in groups and Lie algebras
Our results imply, in particular, that a finitely generated solvable group $G$ is nilpotent if and only if it contains a solution to any unimodular equation, i.e., an equation of the form $\prod g_ix^{n_i}=1$, where $g_i\in G$ and $\sum n_i=\pm1$. A similar fact turns out to be true for Lie algebras. We also exhibit an example of a unimodular equation $w(x)=g$ over a finitely generated group $G$, which has a solution (in $G$) for any $g\in G$, but the solution is not unique for some $g\in G$. We show that, for nilpotent groups $G$, the set of unimodular mappings $G^n\to G^n$ (which are defined naturally) forms a group under the composition.
Read the full abstract
Performance
Characterization of Request and Token Energy Costs for LLM Inference Workloads on GPU Platforms
Large language model (LLM) inference serving is priced by tokens, but GPU energy is consumed over inference windows. This accounting mismatch makes token-normalized metrics incomplete, since average output-token energy can decrease even when total request energy increases. We characterize this behavior with a decomposed energy model: a fixed one-time prefill with a fixed generation setup cost, while each output-token generation step adds marginal step energy. We evaluate this LLM inference energy model on NVIDIA H100 and H200 GPUs across dense and mixture-of-experts (MoE) models, reporting both request energy and token energy as functions of model type (M), phase (P), batch size (B), context length (C), and output length (N). For Llama-3.2-1B on H200 at batch-16 and context-4K, increasing output length from 10 to 512 tokens reduces token energy from 7.46 to 0.72 J/token while total batched inference-window energy increases from 1.19 to 5.93 kJ. Batching also reduces token energy, but the gain is context-bounded: at 10 output tokens, the batch-16 to batch-1 gain falls from 6.31x at context-512 to 1.17x at context-4K. MoE models amplify this effect: sparse routing and fragmented expert execution increase fixed energy at low concurrency, while batching spreads that energy across more generated tokens and substantially narrows the dense-vs.-MoE token-energy gap. These results show that energy-aware serving should jointly optimize both request energy and token energy, rather than only reducing per-token energy cost.
Read the full abstract
Nuclear Theory
Nuclear surface energy in a semiclassical Extended Thomas-Fermi approach with finite-range interactions
Using the Gogny finite-range interaction, we investigate a series of semiclassical approximations to the Fock term entering the calculation of the nuclear surface energy. By comparing these approximations with the exact results obtained from the full Hartree-Fock solution in semi-infinite nuclear matter, we derive a simple pocket formula that can be incorporated into fitting protocols to estimate the surface energy coefficient with excellent accuracy.
Read the full abstract
Computation and Language
SimpCue: Cue-Based Prompting for Multilingual Text Simplification
Text simplification aims to make complex texts easier to understand while preserving their original meaning. Recent large language models can perform simplification through prompting, but it remains unclear whether adding explicit linguistic information about sentence complexity to the prompt improves their outputs. We investigate this question for multilingual sentence-level Easy-to-Read simplification in Catalan, Spanish, and Italian. Using Qwen3-8B, we compare a baseline prompt, a gold-cue prompt enriched with gold linguistic cues, and a predicted-cue prompt enriched with automatically predicted cues. We evaluate the outputs using SARI, BLEU, chrF, and BERTScore, and complement this evaluation with a manual qualitative analysis. Predicted-cue prompting obtains the best overall scores across all four metrics, although the gains over the baseline are small. Gold-cue prompting does not consistently improve over the baseline, and results vary across languages. These findings indicate that cue-based prompting can influence multilingual Easy-to-Read simplification, but its benefits are modest, metric-dependent, and language-dependent.
Read the full abstract
Soft Condensed Matter
Mechanics and statistics of a solvable model of an autophoretic colloidal chain
Equilibrium statistical mechanics owes much of its analytical tractability to symmetry: detailed balance, gradient flows, and the resulting vanishing of steady-state entropy production follow directly from the structure of the underlying dynamics, not from any smallness of the driving. Exact solutions of this kind are rare away from equilibrium. Here we identify a class of far-from-equilibrium active colloidal chains – coupled via roto-translational, autophoretic (monopolar) interactions – that admit an exact quasi equilibrium description: at fixed chain geometry, the orientational equations of motion for every monomer are derivable from a scalar potential, detailed balance holds exactly in the orientational sector, while the positional sector breaks the equilibrium structure. The associated steady-state entropy production rate (EPR) vanishes identically for this sector, even though the full system is manifestly driven and dissipative. We solve this reduced dynamics exactly for dimers and semi-analytically for general $N$-mers, obtain the orientational fluctuations and the full-system EPR in closed form, and show that all dissipation is carried by the translational (center-of-mass) sector. We further examine the effect of dipolar chemical emission – expected from asymmetric micelle deposition at the monomer scale – and find that the equilibrium structure holds exactly for dimers, whereas for longer chains no such description is possible. A purely dipolar coupling instead producesa genuinely non-equilibrium state with no static attractor, sustaining non-monotonic drift with no fixed limit, and an EPR that itself never reaches steady state. Monopolar coupling remains necessary and sufficient for the polarized state; dipolar coupling alone breaks the quasi-equilibrium structure without replacing it with a new static one.
Read the full abstract
Computation and Language
A Shaky Voice Is Not Always a Dodge: Benchmarking Textual and Vocal Evasion Detection in Earnings Calls
Existing approaches to evasion detection in earnings calls focus on textual transcripts, treating evasion as a single-dimensional phenomenon. We argue that evasion in spoken communication is inherently multidimensional: beyond what executives say, how they say it carries independent and complementary information. To study these dimensions jointly, we introduce DualEvasion, a benchmark for evasion detection across text and audio in earnings call Q&A. The benchmark contains 505 annotated question-answer pairs from 60 earnings calls, each with two independent labels: textual evasion (direct vs. evasive) and vocal cues operationalized as speaker confidence (confident vs. unconfident). Our experiments show that state-of-the-art multimodal models struggle to detect vocal confidence, particularly on unconfident responses. Our analysis suggests these models interpret acoustic cues in isolation rather than relative to each speaker's baseline. Providing speaker-level references yields modest improvements, but a substantial gap with human performance remains.
Read the full abstract
Solar and Stellar Astrophysics
A decade of radial-velocity monitoring of rho Leo: moment analysis and periodic variability
We investigate the origin of long-term spectroscopic and photometric variability in the blue supergiant rho Leo, with particular emphasis on distinguishing between intrinsic pulsations and variability induced by a possible companion. Our analysis is based on an 11.5-year spectroscopic time series obtained at Tartu Observatory, complemented by high-cadence, high-resolution spectroscopy from the Hertzsprung SONG telescope and space-based photometry from TESS. We studied line-profile variability using normalised moments of the HeI 6678, HeI 5875, and SiIII 4552 lines. Periodic signals were identified using the generalised Lomb-Scargle periodogram with iterative pre-whitening, and their temporal stability was examined with the weighted wavelet Z-transform. We detect a persistent periodic signal at P = 16.46 d in the first and third moments, present throughout the full observing interval, with a radial-velocity amplitude of 3.9 km/s. This signal is also present in the SONG data and is visible in multiple spectral lines, indicating a global origin. Photometric observations reveal a dominant variability timescale near 33 d, approximately twice the spectroscopic period. The stable 16.46-day period, present throughout the entire observing interval, most likely results from non-radial pulsations of the supergiant. A binary origin of the signal is not excluded, but distinguishing between these scenarios is complicated by the supergiant's complex variability pattern.
Read the full abstract
Algebraic Geometry
A non-klt counterexample to Shokurov's index conjecture
We give a counterexample to Shokurov's index conjecture for minimal log discrepancies in the non-klt setting.
General Relativity and Quantum Cosmology
Devoured by a Hairy Gargantua: Probing Massive Scalar Charges with Non-minimal Curvature Coupling with Extreme-Mass-Ratio Inspirals
Ultralight scalar fields nonminimally coupled to curvature can endow rotating black holes with stationary hair and alter nearby orbits. Using recently constructed hairy black-hole spacetimes, we model the transition to plunge of an extreme-mass-ratio inspiral. For scalar masses $\mu M=0.01$ and $0.1$ and dimensionless curvature coupling $\zeta=10^{-3},10^{-4}$, we find $\mathcal{O}(10)$ rad of gravitational-wave dephasing relative to general relativity over $\mathcal{O}(10^3)$ orbits. A multimode Fisher forecast at $\mu M=0.2$, where neighbouring hair solutions permit a numerical $\mu$-derivative, suggests that $\zeta$ and $\mu$ could be measured to precisions of $\simeq0.2\%$ and $0.6$–$1.1\%$, respectively, for a source with signal-to-noise ratio of about 80 in LISA.
Read the full abstract
Networking
Network Topologies for QKD Networks
Quantum key distribution (QKD) is a method for distributing cryptographic keys between remote endpoints, enjoying security based on quantum physics. This paper makes a first step towards studying topologies for QKD networks. QKD networks imply several required characteristics for their reliability and efficiency. We express such properties in terms of graph structure. For the comparison of potential graphs, we describe cost functions that allow us to identify families of graphs that are reliable and efficient. To make the discussion realistic, we summarize representative field-reported QKD performance numbers and illustrate the sensitivity of secret-key rate (SKR) to a few dB of additional loss. Last, we present methods to construct large efficient graphs through connecting smaller graphs.
Read the full abstract
Quantum Physics
Physics-Constrained Conditional Generative Learning for Quantum State and Process Tomography
Quantum state and process tomography constitute essential diagnostic tools in quantum information science, yet their standard formulations suffer from prohibitive computational scaling as the number of qubits grows. In this work, we introduce a physics-constrained conditional generative adversarial network that bypasses iterative constrained inversion by directly learning a forward generative mapping conditioned on Pauli expectation values. The generator embeds a differentiable Cholesky layer at its output, which enforces Hermiticity, positive semidefiniteness, and unit trace by construction. Our experiments reveal that the strength of the $L^1$ penalty critically governs the emergence of GHZ coherence during training: an excessively large penalty postpones the coherence onset and yields a prolonged low-fidelity plateau, whereas an intermediate value enables the fastest stable convergence. Moreover, for high-temperature thermal states, an over-complete measurement basis proves necessary to prevent sustained late-stage fluctuations. By extending the same Cholesky constraint to the Choi-matrix representation, the framework naturally accommodates quantum process tomography. For systems with $n \ge 6$ qubits, the exponential growth of the underlying $2^n \times 2^n$ density matrix remains the fundamental bottleneck; we discuss how integrating tensor-network structures can contain the per-iteration cost while preserving reconstruction fidelity. Altogether, these results suggest that physically constrained generative learning offers a scalable and amortizable pathway toward data-driven tomography for noisy intermediate-scale quantum devices.
Read the full abstract
High Energy Astrophysical Phenomena
The Geometry of Ultra-Fast Outflows Probed by Soft X-ray Variability in PDS 456
Constraining the location and geometry of ultra-fast outflows (UFOs) is essential for identifying where they are launched and how they are accelerated. We investigate the soft X-ray variability of the luminous quasar PDS 456 using simultaneous March 2024 observations with XRISM/Xtend and NuSTAR. A model-independent comparison between the flare and quiescent phases reveals spectral variability around 1 keV in the rest frame, while the hard X-ray spectral shape remains nearly unchanged. Broadband spectral fitting shows that the soft X-ray structure is well described by a partial-covering low-ionization UFO with $\log (\xi/(\mathrm{erg~cm~s^{-1}})) \simeq 3.1$ and $v_{\mathrm{out}} \simeq 0.30c$. Time-sliced spectral analysis further reveals significant variability in the covering fraction of this absorber. Interpreting this variability as transverse motion across the X-ray source, we constrain the crossing velocity to be $v_{\mathrm{cross}} \lesssim 5 \times 10^{-3}c$ and derive a lower limit on the absorber distance of $r \gtrsim 4 \times 10^3 R_{\mathrm{g}}$. This location is substantially farther out than the high-ionization UFO previously inferred at ~200-600 $R_{\mathrm{g}}$, while the two phases have comparable outflow velocities. The resulting velocity-distance structure disfavors a self-similar magnetocentrifugal wind and instead suggests either radiation-pressure acceleration following a Castor-Abbott-Klein-like velocity law or compact magnetic acceleration through magnetic reconnection. These results demonstrate that soft X-ray partial-covering variability can provide a geometrical probe of UFOs and directly connect spectral variability to wind acceleration.
Read the full abstract
Computer Vision
ZipMVS: Multi-View Stereo with Compressed Cost Volumes
Multi-view stereo (MVS) methods typically deliver highly accurate 3D reconstructions from multiple registered RGB images, thanks to the highly informative, geometric constraints between them. However, their substantial memory requirements remain a major obstacle for deployment in domains such as aerospace and autonomous systems, where resource efficiency is critical. In this work, we introduce ZipMVS, an MVS method specifically designed for efficient high-quality reconstruction. We propose a novel depth-hypothesis strategy that enables substantial compression of the cost volume, hence greatly reducing GPU memory consumption while preserving reconstruction accuracy. Experiments on the DTU and Tanks and Temples datasets show that ZipMVS achieves competitive reconstruction quality compared with other efficiency-oriented MVS methods, while achieving a competitive balance between reconstruction quality and GPU memory usage. The code is available at https://github.com/JihnGlyn/ZipMVS
Read the full abstract
Optics
Density-gradient effect in high-harmonic generation in gases
High-harmonic generation (HHG) in gaseous targets is the most widespread method to produce coherent extreme-ultraviolet (XUV) pulses with sub-femtosecond duration. However, this process has intrinsically low efficiency, and substantial research and development is devoted worldwide to increase the achievable photon flux through this highly nonlinear light–matter interaction process. In this work, we show the strong interplay of phase matching and absorption in gas-pressure gradients, substantially affecting macroscopic HHG efficiency. Through detailed experimental analysis and supporting numerical studies, we highlight their significance, particularly at the boundaries of the interaction volume. The concluded results have implications in the massively expanding application possibilities of HHG sources requiring high photon flux, for example in the semiconductor industry, in nanoscale imaging of biological and industrial samples, or in nonlinear optics in the XUV regime.
Read the full abstract
Data Structures and Algorithms
Beyond the Bethe Approximation of the Permanent
The canonical Bethe approximation gives a deterministic approximation to the permanent of every nonnegative matrix within a factor of $(\sqrt{2})^n$. We improve the base of this exponential factor: for some absolute constant $c<\sqrt{2}$, there is a deterministic polynomial-time $c^n$-approximation for the permanent of every nonnegative matrix. This shows that the canonical Bethe guarantee is not a barrier for deterministic approximation of the permanent. The proof augments the Bethe lower bound with a new certificate tailored to matrices on which that lower bound loses nearly the full factor. The author supplied the high-level plan of attack, and the proof was developed in an interaction with ChatGPT 5.6 Sol Pro. The author subsequently verified the results. Codex assisted with proof checking, manuscript assembly, and typesetting.
Read the full abstract
Other Quantitative Biology
Expanding the Human Ancestry Ontology to include under-represented populations and ethnicities for broader utility in annotations
Successful discovery, integration and reuse of data relies on the availability of rich, well-structured and machine-readable metadata to describe every aspect of the data, from sample sources to collection processes to experimental protocols. The use of standardised terminologies to express concepts in a harmonised fashion lies at the core of high-quality data annotation, increasing the FAIRness of the data, facilitating data integration and promoting reproducibility. Here, we describe the Human Ancestry Ontology (HANCESTRO), originally developed to improve standardised reporting of genetic ancestry genomic resources such as the NHGRI-EBI GWAS Catalog and the Human Cell Atlas through high-level population descriptors, and more recently expanded to include diverse and previously under-represented populations in genomics and genetics research. HANCESTRO provides a framework for population descriptors that includes both ancestry based on the analysis of genetic information and self-reported ethnicity, which is based on social and cultural factors that don't necessarily align with genetic populations. By enabling the accurate and interoperable representation of population-related data, it promotes inclusive, representative and reproducible science.
Read the full abstract
General Relativity and Quantum Cosmology
Dynamical friction on black holes in scalar field environment
Astrophysical black holes exist within non-vacuum environments, and their motion through these environments generically result in a force on these black holes through processes such as dynamical friction and Bondi accretion. We explore these forces numerically for black holes moving through a complex scalar field of mass $m$ with constant acceleration $a$, while also revisiting the constant velocity case considered in the existing literature. The former is modeled by the C-metric, while we use Painlevé-Gullstrand metric with an additional divergence and vorticity free velocity field to mimic the constant velocity motion. Our simulations reveal several novel and interesting aspects of the drag force in both cases. For the constant velocity case, the force saturates at late times, with a value that depends on velocity in a manner distinct from known dependence. For the constant acceleration case, the force increases to a maximum value $F_\star$ and then reduces drastically, with $F_\star \approx 0.3~ m a$. Moreover, the density wake in this case shows revivals separated by decreasing time intervals.