Quantum Physics
Joseph Tindall
Strong dynamical
symmetries provide an algebraic description of persistent oscillations in open
quantum systems governed by the Lindblad equation. For finite-dimensional systems with a full-rank stationary state, such
symmetries are necessary and sufficient for non-decaying oscillatory modes to appear in the Liouvillian's spectrum. These modes can be directly constructed by acting on stationary states with the identified
symmetry operators. Here we establish an asymptotic counterpart of this
algebra for collective
spins. Taking a single
spin $S$ in a
magnetic field, subject to dissipation through linear jump operators, we identify the exact conditions on the field and dissipative couplings under which non-decaying oscillatory modes emerge as $S \rightarrow \infty$. We then write down a simple, closed form for the corresponding `
symmetry' operator which can be applied to the steady state to construct these modes, which are eigenvectors of the Liouvillian, with imaginary
eigenvalues, up to $O(1/S)$ corrections. A second operator, an emergent conserved quantity, completes the
algebra, and its slow relaxation determines how
quantum noise selects the stationary state. The construction can be directly extended to a pair of interacting
spins sharing a common reservoir, with the resulting
quantum algebra making the
entanglement during a quench analytically accessible.
Read the full abstract ▾
Adaptation and Self-Organizing Systems
How Mobile Oscillators Reshape Dynamical Robustness in Complex Networks
Sayantan Nag Chowdhury and Srilena Kundu
While the robustness of collective dynamics in static networks is well established, how
agent mobility and network
topology jointly shape dynamical robustness remains largely unexplored. In this letter, we investigate the robustness of mobile Stuart-Landau oscillators performing random walks on networks and interacting diffusively when co-located. In the rapid-mobility regime, we reduce the dynamics to a mean-field framework linking network degree statistics to an effective global coupling strength. We show analytically that increased degree heterogeneity in the network enhances the effective coupling strength, and strictly lowers the critical threshold of functional degradation for amplitude death. Numerical simulations confirm the predicted heterogeneity-dependent ordering of the transition thresholds, showing an earlier collective collapse in scale-free networks than in random or regular networks. Our results establish mobility and network heterogeneity as key determinants of robustness in mobile oscillator populations, with implications for resilient biological and engineered collectives.
Read the full abstract ▾
High Energy Physics (Phenomenology)
From Earth to Mars and Beyond: AI-Driven Cosmic-Ray Validation and Prediction with CorsikaAgent
Zhuo Chen, Zhizheng Zhao, Jinhong Shen et al. (+5 more)
We calculate the sea-level
cosmic-ray
muon flux with CORSIKA and develop CorsikaAgent, an auditable large-
language-model-based system for configuring, executing, and validating air-shower simulations, and for predicting particle fluxes in unexplored environments. The physics calculation uses
proton primaries described by the
solar-minimum spectrum of Papini et al., with UrQMD at low interaction energies and either Sibyll 2.3e or QGSJET-III at high energies. The predicted momentum spectra at approximately 0 and 75 degrees are compared with experimental measurements of the atmospheric
muon flux and with commonly used analytic parametrizations. The two interaction-model configurations give similar spectra and describe the principal momentum and zenith-angle dependences of the data. The
agent independently reconstructs the same observable through a different sampling strategy and keeps every steering card, executable, seed, log, and analysis script bound to the result it produced; its output is consistent with the direct calculation at the level resolved by the present comparison. Transferred to Mars, the workflow reproduces the CORSIKA 8
benchmark, and transferred to Venus it produces the first prediction of the surface
muon flux, an environment for which no measurement exists. The
agent thus extends the workflow from auditable reproduction to first-principles prediction: it autonomously reconfigures the simulation(atmosphere,
magnetic field, electric field)for new environments, while the
proton-only primary model, limited high-momentum statistics, and an incomplete systematic-uncertainty treatment bound the physical precision that can currently be claimed.
Read the full abstract ▾
All authors Zhuo Chen, Zhizheng Zhao, Jinhong Shen, Leyun Gao, Cheng-en Liu, Qite Li, Zijian Wang, Qiang Li
Geometric Topology
Homotopy Self-Equivalences of $\mathbb{R}P^2\times S^2$ and a Braid for non-orientable $4$-Manifolds
Mehmetcik Pamuk
We compute the group of based homotopy classes of self-homotopy equivalences of the non-orientable $4$-
manifold $\RP^2\times S^2$. It is an elementary abelian $2$-group of rank $5$. The subgroup of classes whose based lift to the universal cover $S^2\times S^2$ has degree $+1$ has rank $4$ and is isomorphic to the group of free homotopy classes. The group maps onto the group of self-equivalences of the algebraic $2$-type, which has order $8$, with kernel $\pi_4(\RP^2\times S^2)\cong(\Z/2)^2$ generated by pinch-fold maps; that these maps are independent is shown by an adjunction argument. We then construct a non-orientable analogue of the Hambleton–Kreck braid of exact sequences in dimension $4$, for non-orientable
manifolds with
spin universal cover, stated for finite fundamental group. We apply the braid to $\RP^2\times S^2$ and to $\RP^4$, and for the Hambleton–Hillman
manifold with fundamental group $(\Z/2)^2$ we determine the normal $1$-type and show that self-equivalences act on $\pi_2$ through diagonal sign
matrices. For $\RP^2\times S^2$ the pinch-fold maps are detected by twisted
spin bordism, and, assuming facts from surgery theory stated as hypotheses, exactly one of the three non-trivial pinch-fold maps is homotopic to a homeomorphism.
Read the full abstract ▾
Combinatorics
$2$-large sets are sets of nil-Bohr recurrence
Ryan Alweiss and Sohail Farhangi
A set $S$ of positive integers is $2$-large if every two-coloring contains arbitrarily long monochromatic arithmetic progressions with common difference in $S$. We prove that every such set meets every nil-Bohr neighborhood of $0$. This proves a
conjecture of the first author and gives a sharp answer to a question of Host, Kra, and Maass. We also show that, for every fixed nil-Bohr neighborhood \(B\) of \(0\) and every \(\eps>0\), restricting the common difference to \(B\) increases the extremal number for \(k\)-term arithmetic progressions by at most a factor of \(1+\eps\), provided \(k\) is sufficiently large.
Read the full abstract ▾
Robotics
Design and Analysis of a Nonlinear Parallel Mechanism Coupling for Dynamic Humanoid Legs
Rohit Kumar, Heiner Peters, Felix Wiebe et al. (+5 more)
Humanoids increasingly demand lightweight leg mechanisms for agile, dynamic, and energy-efficient locomotion. Reducing distal inertia through proximal actuation has thus become a common design goal. Yet, compact transmissions that preserve torque capability and efficient force transmission remain hard to achieve. This paper presents a novel series-parallel leg design combining a knee parallel mechanism, an ankle parallel mechanism, and a nonlinear knee-ankle coupling where all actuators placed closer to the hip. As the knee flexes, the coupling passively assists ankle dorsiflexion for foot clearance and, as it extends it assists plantarflexion for push-off, redistributing ankle load through the knee transmission. We derive the coupling's nonlinear transmission models and validate them in simulation and on hardware. The design shows that the shared transmission lets both joints draw on each other's actuator capacity by easing the knee's peak torque demand under load and the ankle's peak speed demand during fast motion. %Together With the leg's proximal actuator placement, the leg achieves an effective reflected mass of 0.37kg at the foot for a 6kg leg, demonstrating very low distal leg inertia for dynamic
humanoid robots.
Read the full abstract ▾
All authors Rohit Kumar, Heiner Peters, Felix Wiebe, Franek Stark, Janne Janzen, Dennis Mronga, Shubham Vyas, Frank Kirchner
Number Theory
The ring of values of $E$-functions is not a field
Stéphane Fischler and Tanguy Rivoal
Let $\mathbf{E}$ be the ring of values taken at algebraic points by $E$-functions. The exponential function and the Bessel function $J_0$ are classical examples of $E$-functions. We prove that $\mathbf{E}$ is not a field. More precisely, we prove that it is neither noetherian nor factorial, and construct non-units of $\mathbf{E}$ explicity, including $e+1$, $J_0(1)$, and ${}_1F_1(a;b;1)$ for rational parameters $a$ and $b$ such that $0
Read the full abstract ▾
Earth and Planetary Astrophysics
An updated view of Jupiter's UV main auroral emission: large-scale analysis of UV and particle data from Juno's first 60 orbits
Linus Head, Denis Grodent, Bertrand Bonfond et al. (+9 more)
The properties and origins of Jupiter's UV main auroral emission are, at present, not yet fully understood, particularly the extent to which its two distinct zones, the Zone-I and Zone-II aurora, originate from a common source. In this work, remote-sensing and in-situ data from the Juno spacecraft are used to investigate and produce maps of the average properties of the main emission over Juno's first 60 orbits. It is found that the peak in mean energy of electrons precipitating in the main emission is consistently poleward of the energy-flux (or brightness) peak, in both hemispheres and at all local times, and is co-located with the statistical location of the Zone-II aurora. Evidence is presented that the flux-energy relationship of the main emission is adequately characterised by theoretical power laws describing electron acceleration from quasi-static potentials or Alfvénic
turbulence and that it shows a modulation with ionospheric electron density consistent with theoretical predictions. Finally, the average energy spectra of precipitating electrons in the Zone-I aurora remain power-law-like between an energy flux of 1 and 100 mW m$^{-2}$, whereas the average Zone-II spectra develop a distinct “plateau” or “flattop” shape beyond an energy flux of $\sim$5 mW m$^{-2}$. Our results provide some evidence that the Zone-I and Zone-II aurorae are phenomenologically distinct, and the consequences of this interpretation for future investigation of the main emission are discussed.
Read the full abstract ▾
All authors Linus Head, Denis Grodent, Bertrand Bonfond, Nicholas Kruegler, Ali Sulaiman, George Clark, Bilal Benmahi, Guillaume Sicorello, Julie Vinesse, Thomas Greathouse, Rohini Giles, Barry Mauk
Computation and Language
MixLID: Token-Level Language Identification in Mixed-Language Text
Seongtae Hong, Youngjoon Jang, Seungyoon Lee et al. (+2 more)
Processing
multilingual text requires both identifying a broad range of
languages and distinguishing the
language of each
token in inputs containing multiple
languages. We introduce MixLID, a
token-level
language identification model supporting 111
language-script combinations. MixLID follows a two-stage training process: it first learns to distinguish individual
languages using the
language labels of monolingual examples, then continues training on mixed-
language data. The second stage uses concatenated examples from existing code-switching data alongside augmented examples created through word substitution based on correspondences between English source words and loanwords. For monolingual identification, MixLID outperforms all baselines in both accuracy and macro F1 on four
benchmarks within the
language coverage shared with each baseline. For mixed-
language identification, it outperforms the baselines in identifying the
language of individual
tokens on every evaluated
language pair. MixLID achieves high performance across a broad range of
languages in both monolingual and mixed-
language settings. The model is publicly available at https://huggingface.co/nlpai-lab/MixLID.
Read the full abstract ▾
All authors Seongtae Hong, Youngjoon Jang, Seungyoon Lee, Jungseob Lee, Heuiseok Lim
Geometric Topology
Werner Stanggassinger
For p, q >= 3 with gcd(p,q) = 1 we show that the
symmetry group of the Turk's head
knot Th(p,q), the closure of the braid (sigma_1 sigma_2^{-1} sigma_3 sigma_4^{-1} ... sigma_{p-1}^{+-1})^q, is the dihedral group of order 4q. This proves
Conjecture R of Di Prisa and Şavk for q >= 3; the case p = 3 is due to Sakuma and Weeks. The argument is diagrammatic: by the flyping
theorem of Menasco and Thistlethwaite, the
symmetries of a
knot with a
prime, reduced alternating diagram without non-trivial flypes are the
symmetries of the diagram (we write out a proof for the diagrams at hand), and we show that the standard diagram of Th(p,q) has no non-trivial flypes and exactly 4q
symmetries. The chirality of Th(p,q) for even p, which also follows from a
theorem of Murasugi, is visible in the group. We compare the argument with the
geometric one of Sakuma and Weeks for p = 3.
Read the full abstract ▾
Soft Condensed Matter
Solubility of poorly soluble species in bulk and nanopores from grand-canonical Wang-Landau sampling
Russell Kajouri, Adyant Agrawal, Philipp Stärk et al. (+2 more)
Poorly soluble species can have mean populations below one in small systems such as nanopores. In this regime, the grand-potential minimum, which identifies the most probable particle number, differs from the ensemble-averaged solute loading. We use grand-canonical Wang-Landau sampling (GCWLS) to calculate the density of states $\Omega(N_\mathrm{A}, N_\mathrm{B})$ for a binary Lennard-Jones mixture of solute B in solvent A, both in bulk and in slit pores. One set of GCWLS simulations gives ensemble averages over a range of chemical potentials, whereas conventional grand-canonical Monte Carlo (GCMC) simulation requires a separate simulation at each state point. Independent GCMC simulations confirm the GCWLS ensemble averages. The solvent number distribution is nearly Gaussian, whereas the dilute solute distribution is skewed and Poisson-like. The most probable solute number therefore lies below the ensemble-averaged loading $\langle N_\mathrm{B}\rangle$ and is zero whenever $\langle N_\mathrm{B}\rangle<1$. Reading the solute loading from the grand-potential minimum underestimates it and can falsely indicate that no solute is present. Defining the oversolubility as $\mathcal{O}_\mathrm{B}=x_\mathrm{B}^{\mathrm{conf}}/x_\mathrm{B}^{\mathrm{bulk}}$, where $x_\mathrm{B}$ is the solute mole fraction, we find $\mathcal{O}_\mathrm{B}\approx12$ in a pore of 3.5 particle diameters width when both wall interactions are repulsive. Adding wall-solute attraction moderately raises it to $\approx13$, whereas wall-solvent attraction lowers it to $\approx3.4$. This comparison shows that the wall-solvent interaction controls most of the enhancement. As the pore widens, $\mathcal{O}_\mathrm{B}$ decreases toward unity. In bulk and in nanopores alike, the solubility of poorly soluble species should be obtained from ensemble averages rather than from grand-potential minima.
Read the full abstract ▾
All authors Russell Kajouri, Adyant Agrawal, Philipp Stärk, Alexander Schlaich, Christian Holm
Computer Vision
Detecting Chain-of-Thought Hijacking in Backdoored Multimodal Large Reasoning Models
Zhifang Zhang, Qi Wei, Weitong Chen et al. (+5 more)
Chain-of-thought (CoT)
reasoning enables
multimodal large
reasoning models (MLRMs) to solve complex visual problems, but also exposes them to CoT hijacking: an adversary implants a
backdoor through poisoned fine-tuning so that a visual trigger induces a hijacking turn, redirecting an otherwise faithful
reasoning chain to an attacker-chosen answer. However, since the hijack acts within the
reasoning chain, existing
multimodal backdoor defenses designed for MLLMs that answer directly struggle to detect it. Moreover, inspecting the chain itself does not reliably expose the hijack, since the hijacking turn can resemble honest self-correction in form. In this paper, we find that honest
reasoning tends to revise its answer at high-
entropy "aha moments", whereas a hijack can overturn a confident answer through a near-deterministic turn. To quantify this contrast, we formalize this phenomenon as belief inertia, the cumulative next-
token entropy of a
reasoning span minus the answer-belief revision it induces, with markedly lower values for hijacked spans than for honest
reasoning spans. Since hijacking
tokens are also image-dependent, we further improve detection by restricting the measure to image-dependent spans, yielding visual belief inertia (VBI). Building on this, we propose InertiaGuard, a training-free, test-time detector that flags a trace when its VBI falls below a threshold calibrated on a small set of clean traces. Across various attacks and
benchmarks, InertiaGuard achieves higher detection performance than existing baselines, while generalizing to different model architectures and trigger types.
Read the full abstract ▾
All authors Zhifang Zhang, Qi Wei, Weitong Chen, Olaf Maennel, Wei Emma Zhang, zuomin ren, Lei Feng, Miao Xu
Machine Learning
DEFlow: Deep Equilibrium Flow Maximum-Entropy Reinforcement Learning
Ruslan Ageev, Huaze Tang, Runqing Miao et al. (+2 more)
Expressive flow policies can represent multiple high-value actions, but sampling them with few integration steps while retaining tractable action densities remains challenging. We introduce DEFlow, a deep
equilibrium flow policy that replaces explicit Euler sampling with implicit Euler integration and solves all intermediate actions jointly as an
equilibrium system. We combine implicit differentiation with a discrete change-of-variables formula for the final-action density, enabling off-policy training within soft actor-critic without backpropagating through solver iterations. DEFlow achieves an average relative improvement of approximately 10% in mean successful coverage over the strongest baseline across four Multi-Goal docking variants. It also achieves the highest mean time-averaged return on four of five manipulation tasks, while remaining broadly competitive across five MuJoCo tasks. These results support implicit flow policies for tasks requiring both diversity and precision, at the cost of additional nonlinear and linear solves. Primary
Read the full abstract ▾
All authors Ruslan Ageev, Huaze Tang, Runqing Miao, Jiarong Li, Wenbo Ding
Machine Learning
Towards Modality-Agnostic Seizure Classification: A Unified Deep Learning Framework for Heterogeneous EEG and Intracranial EEG Signals
Ramya Sundaram, Muhammad Aslam Jarwar, M Uma et al. (+2 more)
Seizure prediction and detection, essential for clinical treatment of epilepsy, remain the crucial task of this neurological disease.
EEG and iEEG are widely used to analyze epilepsy due to its capability of capturing discriminative
brain activities at high time resolution. However, existing works generally analyze either
EEG or iEEG seizure activity separately, where
EEG analysis has been well-researched, while more and more works consider iEEG for its higher specificity at the spatial level and signal quality. Currently, most of the existing seizure prediction systems are single-modal in nature, which limits its application to a variety of clinical conditions. Therefore, a common framework to analyze seizures with both
EEG and iEEG signals needs to be developed. This paper presents a new framework of
multimodal seizure classification integrating the heterogeneous
EEG and iEEG recordings from different subjects and sources. A specific framework, called Unified
Multimodal Signal Harmonization (UMSH), is developed for harmonizing signals with varied sampling rates and recording conditions. Moreover, to handle the extremely class-imbalanced problem in the data sets, we use a Conditional Generative
Adversarial Network (cGAN) based data augmentation technique to generate synthetic samples to balance the class distribution. The proposed framework uses the EEGNet-
Vision Transformer (EEGNet-ViT) as the feature extraction and classification network to learn the local discriminative patterns as well as global contextual information in the
multimodal seizure data. Experimental results demonstrate the effectiveness of the proposed framework for robust seizure classification, emphasizing the potential of integrated
EEG-iEEG
multimodal learning in advancing next-generation epilepsy monitoring and clinical decision-support systems.
Read the full abstract ▾
All authors Ramya Sundaram, Muhammad Aslam Jarwar, M Uma, Sunder Ali Khowaja, Qurat-Ul-Ain Mastoi
Image and Video Processing
Uncertainty-Aware Cortical Surface Reconstruction and Spherical Registration for Heterogeneous Brain MRI
Jesper Duemose Nielsen, Karthik Gopinath, Andrew Hoopes et al. (+4 more)
Surface-based cortical analysis is useful in many applications as it is a natural way to model the highly folded structure of the cerebral cortex, allowing quantities such as
curvature and thickness to be easily calculated. Existing deep learning (DL) methods have focused on 1x1x1 mm3 T1-weighted (T1w) scans, thus replicating classical pipelines. Here we present a neural network that is trained exclusively on synthetic scans with random contrast and resolution. Consequently, it can be applied to almost any kind of
magnetic resonance imaging (MRI) scan. In addition, our model provides vertex-wise uncertainty estimates of the predicted surface as well as a spherical registration to fsaverage space. Our model performs on par with other state-of-the-art DL methods on 1x1x1 mm3 T1w scans and outperforms the only other method capable of estimating cortical surfaces from low resolution, non-T1w scans. The predicted uncertainty, moreover, correlates with overall scan quality on low resolution scans and highlights areas of poor surface reconstruction on 1x1x1 mm3 T1w scans, suggesting that it can be used for quality control. Finally, the predicted spherical registrations prove comparable to other DL and classical registration methods. These qualities, when combined with the fast inference time, will provide the ability to analyze large, existing databases of clinical MRI scans, with the potential for finding disease-specific patterns that would otherwise be infeasible in a research setting.
Read the full abstract ▾
All authors Jesper Duemose Nielsen, Karthik Gopinath, Andrew Hoopes, Adrian Dalca, Axel Thielscher, Juan Eugenio Iglesias, Oula Puonti
Software Engineering
Software Testing Meets Bayesian Workflow: A Literature-Driven Artifact Study of Probabilistic Programs
Oleg Jarma, Maja Dall, Raúl Pardo et al. (+2 more)
Automated testing is an effective software quality assurance method, but it is rarely used for probabilistic models. This is despite the heavy use of probabilistic programming for data analytics in research and engineering, where the consequences of errors are dire. A key obstacle is the specification of testable correctness properties for probabilistic models, known as the oracle problem. Like machine learning, probabilistic programming is used precisely where no analytical specification exists: had we been able to write one, we would have written an
algorithm instead of fitting a model. An alternative is partial, proxy correctness specifications. The statistics literature already offers many, but as quality guidelines that require human judgment: comparing plots, visually analyzing shapes of distributions, assessing whether parameter values are ”far” or ”close enough”. These guidelines are not systematically automated. In this work, we elicit, implement, and study testable correctness properties of probabilistic models. We extract 17 broadly applicable properties for probabilistic models from one of the most widely used textbooks in
Bayesian data analysis, Statistical Rethinking by McElreath. For each property, we select and implement a test oracle, and study its utility for automatic testing, using a population of 59 probabilistic models from academic and industrial sources. We find realistic use cases (models under test) for all the properties. We observe that the vast majority of tests are not flaky in practice, despite the theoretical potential for flakiness. To the best of our knowledge, we are the first to investigate the effectiveness of mutation testing for probabilistic models, showing that standard mutation operators do not capture faults tracked by the probabilistic properties.
Read the full abstract ▾
All authors Oleg Jarma, Maja Dall, Raúl Pardo, Mahsa Varshosaz, Andrzej Wąsowski
Artificial Intelligence
Double-Rollout Reinforcement Learning with Sampling-based Reward Estimation
Haifeng Qian, David Mosallanezhad, Christian Munley et al. (+2 more)
Reinforcement learning with verifiable rewards (RLVR) implemented with Group Relative Policy Optimization (GRPO) has driven substantial progress in training
reasoning language models. However, verifiable outcome rewards provide an imperfect measure of
reasoning quality: a correct final answer may not be supported by the
reasoning trace, whereas an incorrect rollout may contain a nearly correct
reasoning trajectory. We present empirical evidence for this mismatch and hypothesize that the resulting reward noise contributes to instability in reinforcement learning for both
reasoning accuracy and
token efficiency. To address this credit assignment challenge, this paper proposes double-rollout reinforcement learning. For each sampled rollout, we truncate its
reasoning trace, sample multiple continuations from the truncated prefix, and replace the original reward with the mean continuation reward, yielding a more robust estimate of
reasoning quality. This double-rollout
algorithm incurs only a moderate increase in compute cost and yet reaches substantially higher accuracies across math and coding tasks and across Qwen and Nemotron models than baseline
algorithms with fewer RL steps. For example, using only DAPO data it trains Nemotron-3-Nano-30B-A3B to reach 95.4% on AIME 2026, versus 91.5% by GRPO and substantially above state of the art for this model size, within 30 RL steps. When combined with length-based rewards, the same
algorithm further improves the accuracy-efficiency Pareto frontier over existing efficiency optimizations. These findings suggest that the double-rollout reward estimation is an effective and cost-efficient way to improve both the accuracy and
token efficiency of
reasoning RL.
Read the full abstract ▾
All authors Haifeng Qian, David Mosallanezhad, Christian Munley, Adithya Renduchintala, Oleksii Kuchaiev
Distributed Computing
Prorsa: Unleashing the Potential of Prefill-Decode Disaggregation in Agentic Reinforcement Learning
Chiheng Lou, Xiangwei Deng, Bangjun Xiao et al. (+4 more)
Agentic reinforcement learning (RL) is commonly used to enhance the capabilities of large
language models (LLMs) in solving complex multi-turn tasks. During rollout, agentic RL exhibits a workload pattern distinct from that of traditional RL: multi-turn interactions repeatedly introduce prefill requests, causing prefill and decode workloads to coexist throughout rollout. This workload makes prefill-decode (PD) disaggregation, a design widely used for online LLM serving, promising for speeding up agentic RL rollout. This paper shows that PD disaggregation can improve rollout efficiency by (a) mitigating interference among data-parallel
attention modules, (b) enabling stage-specific parallelism, and (c) reducing end-to-end trajectory completion time. However, a fixed prefill-decode allocation may fail to sustain these benefits as workload composition varies across tasks and over time. We therefore present Prorsa, an agentic RL framework for prediction-guided prefill-decode resource allocation. Prorsa observes the lifecycle of every trajectory to forecast short-horizon prefill and decode demand. These forecasts then guide worker allocation decisions and trigger role switching of rollout engines, aligning prefill and decode capacity with anticipated workload demand. Across synchronous and asynchronous agentic RL settings, Prorsa achieves end-to-end rollout speedups of 1.11x-4.05x over prefill-decode colocation.
Read the full abstract ▾
All authors Chiheng Lou, Xiangwei Deng, Bangjun Xiao, Xiaodong Ji, Yikai Zhao, Xin Jin, Hailin Zhang
Computer Vision
JtT: Rethinking Point Trajectory Diffusion for Diverse and Rigid Motion from an Image
Gabrijel Boduljak, Yushi Lan, Christian Rupprecht et al. (+1 more)
We study
forecasting rigid-body motion from a single image: generating diverse and physically plausible future trajectories for arbitrary points in a scene, without building
rigidity into the architecture or relying on
geometric inputs such as depth. Like
video generators, recent trajectory generators fail at this task, even in simple synthetic Kubric scenes, despite scaling up model size and training on real data. We show how to solve this with the right architecture and parameterization. Our model, JtT, is a simple
diffusion transformer that operates directly in coordinate space and decodes arbitrary and dense query patterns, without the trajectory VAEs or fixed grids assumed in prior work. Through a systematic study of the design space of point trajectory
diffusion, we find that diffusing per-frame velocities is the only parameterization that captures
rigidity, that it must be paired with the right timestep schedule, and that adding full-resolution RGB at the query points resolves the boundary failures of prior methods. However, evaluating such 2D models rigorously is itself non-trivial. To our knowledge, we are the first to observe that physical plausibility and diversity do not improve together during training: larger models, as used in prior work, memorize the single ground-truth future per training image, becoming more rigid while covering fewer possible futures. Yet prior work has paid little
attention to this, so we show how model rankings depend on model size, metric, and stopping time, and adopt a stricter, checkpoint-averaged evaluation protocol. Under this protocol, JtT improves over prior methods on both distributional and
rigidity metrics and approaches ground-truth
rigidity, with only about 50M parameters.
Read the full abstract ▾
All authors Gabrijel Boduljak, Yushi Lan, Christian Rupprecht, Andrea Vedaldi
Mesoscale and Nanoscale Physics
Admittance extraction of quantum dot arrays for charge sensing
Conor Power, Mathieu Moras, Andrii Sokolov et al. (+2 more)
This work demonstrates admittance extraction of
quantum dot arrays using a rate equation frame- work matching experimental reflectometry measurements. Using a semiclassical rate equation, we consider a perturbation of the source/drain chemical potential to model reflectometry readout in a
quantum dot array. This is a time-dependent extension of the rate equation which has been employed in literature for pumped arrays of
quantum dots and capacitance extractions. Here we extend this to a double
quantum dot for admittance extraction. The simulations are calibrated us- ing
Fermi-Hubbard parameters extracted from experimental data, which it is then shown to closely approximate to.
Read the full abstract ▾
All authors Conor Power, Mathieu Moras, Andrii Sokolov, Robert Bogdan Staszewski, Elena Blokhina
Machine Learning
HyBRA-EEG: Hyperbolic Global-Local Brain-Region Aggregation for Self-Supervised EEG Representation Learning
Xin Liu, Feng Lu and Wei Li
Electroencephalography (
EEG) representation learning is challenged by noisy signals, inter-subject variability, and heterogeneous electrode montages in different
datasets and recording settings. Recent self-supervised
EEG pretraining methods improve transferability by converting heterogeneous recordings into unified
token or sequence representations, but channel-aware designs may still encode montage- or
dataset-specific regularities. We propose HyBRA, a self-supervised framework based on
hyperbolic global-local
brain-region aggregation. HyBRA maps heterogeneous channel observations into a fixed regional
token space by combining electrode coordinates, region priors, and feature-dependent
hyperbolic distances. A channel-evidence injection branch preserves fine-grained discriminative cues during aggregation. The resulting
tokens are processed by a Region-Time
Transformer and pretrained with masked reconstruction of latent features and region-level signals. HyBRA outperforms the compared methods in average balanced accuracy on six
benchmark datasets. Under a controlled six-fold evaluation on OpenBMI-MI, HyBRA matches CSBrain in balanced accuracy, improves AUROC by 1.84 percentage points and AUPRC by 5.23 percentage points, and uses 96.8% fewer parameters. Matched comparisons on TUAB and TUEV show higher balanced accuracy for
hyperbolic matching than for Euclidean distance or cosine similarity.
Read the full abstract ▾
Trading and Market Microstructure
Demand Under a Mechanical Price: Elasticity and Welfare in Ethereum's Data-Availability Market
Farbod Ghasemlu
Ethereum sells data-availability capacity to rollups through "blobs" priced by a controller: the blob fee rises and falls exponentially with cumulative usage above or below a target, and no auction sets it. We measure what this mechanical price costs. The Pectra upgrade of 7 May 2025 doubled blob capacity on a date fixed weeks in advance, while the
market was congested, and the blob fee fell from 0.82 mETH per blob (about USD 1.40) to its 1-wei floor. Rollups' blob purchases rose 24%. The 95% interval, built from the same
estimator at 516 placebo dates, runs from 8% to 40%, so removing the entire price raised purchases by at most 40%. The implied elasticity at the pre-shock price is -0.19, and no value below -0.31 is consistent with the interval. The estimate includes any rise in L2 activity that coincided with the fork, so it bounds the price response from above. With demand this inelastic, the controller works mainly as a transfer. Rollups paid USD 14.8 million in blob fees under the first blob regime, 67% of it in two quarters when capacity was full and any
market-clearing mechanism would have collected comparable scarcity rents. Where capacity was slack, the reserve price introduced in December 2025 cost at most 2.6 ETH (USD 8,039) in deadweight loss under the most elastic demand the data allow. Mempool records for 5.4 million blob transactions show that when the fee sits at its floor, capacity is slack: transactions miss about one block and are almost never priced out.
Read the full abstract ▾
Computer Vision
Code-Interactive Visual Reasoning for Financial Candlestick Understanding
Chenxi Zhang, Youwei Pang, Xiaoqi Zhao et al. (+4 more)
Financial candlestick understanding requires locating visual elements, aligning them across trading dates and panels, and combining observations through numerical and financial
reasoning. Answer accuracy alone does not reveal whether predictions are supported by chart evidence. We introduce Code-Interactive Financial Candlestick
Reasoning and present Fin-Thinker, a model based on Qwen3.5-9B that generates and executes code to acquire evidence and perform computations. The model uses the returned observations to guide subsequent
reasoning. To train these capabilities, we construct FinKLine-32K with 32,184 replay-verified code-observation trajectories collected through teacher-guided student exploration. These include successful solutions with and without error recovery. Our three-stage training framework combines foundational visual-code fine-tuning, financial process supervision of successful
reasoning and corrective continuations, and execution-feedback reinforcement learning. For evaluation, we introduce FinKLineBench with 1,200 questions across 138 financial task types to assess answer correctness, code execution, and evidence grounding separately. On FinKLineBench, Fin-Thinker improves strict accuracy by 13.1 and 33.2 percentage points over the best-performing direct-answer baseline among those evaluated and its base model with sandbox access, respectively. Training-stage ablations further show that financial process supervision and execution-feedback reinforcement learning yield successive improvements in both answer accuracy and GPT-5.6-judged grounded accuracy.
Read the full abstract ▾
All authors Chenxi Zhang, Youwei Pang, Xiaoqi Zhao, Ziliang Gan, Rongjunchen Zhang, Shanshan Feng, Hao Wang
Machine Learning
OctMole: Multiscale 3D Molecular Generation with Hierarchical Spatial Flows
Yu-Tong Jin, Yi-Yang Tian, Qiwei Ye et al. (+1 more)
Generating accurate 3D molecules requires jointly determining a variable number of atoms, their identities, and geometry across spatial scales: global shape emerges coarsely, whereas chemical bonds demand fine local precision. Existing atom-based
diffusion and flow models operate largely at a single scale, while dense spatial fields become costly at high resolution. We introduce OctMole, a multiscale octree flow framework for 3D molecular generation. A hierarchical occupancy flow progressively resolves sparse molecular support from coarse to fine, and a joint atom flow assigns atom types and continuous in-
cell offsets to occupied leaves. This decomposition combines efficient global structure with subcell
geometric precision, while
cell corruption improves robustness to generated support. On QM9 and GEOM-DRUG, OctMole achieves strong chemical, distributional, and
geometric quality, with particularly clear gains over Uni-3DAR on GEOM-DRUG and substantial efficiency gains over dense 3D generation. As an additional capability, intermediate occupancy states and selected regions provide spatial conditioning without task-specific retraining, supporting multiscale shape conditioning, regional editing, linker generation, and shape interpolation. Code and pretrained checkpoints are available at https://github.com/octree-nn/octmole and https://huggingface.co/Disembo/octmole, respectively.
Read the full abstract ▾
All authors Yu-Tong Jin, Yi-Yang Tian, Qiwei Ye, Peng-Shuai Wang
Fluid Dynamics
Onset condition of the ramified-jet regime: an experimental and theoretical study
Bor Zupan, Jernej Kušar and Božidar Šarler
This study experimentally and theoretically investigates the electrically induced transition of a gas-focused
liquid jet into a rim-emitting sheet. Ultra-high-speed imaging was used to examine the effects of applied voltage, gas and
liquid flow rates, and
liquid properties on sheet formation and breakup. Fully developed, quasi-stable sheets formed with ethanol, whereas isopropanol exhibited only initial sheet formation. Increasing voltage promoted sheet stretching, finer and more numerous rim jets, and a narrower spray. Higher gas flow rates shifted the transition towards the nozzle and produced finer structures, while higher
liquid flow rates generated larger sheets with thicker rims. A first-order theoretical model assumes an equipotential interface and neglects finite charge relaxation, hydrodynamic perturbations, and nonlinear
evolution. It balances destabilizing Maxwell-pressure perturbations against restoring capillary-pressure perturbations on a conducting cylindrical jet. The model identifies the
elliptic azimuthal mode, m = 2, as the first unstable mode leading to sheet formation and predicts a critical electric Bond number of 3. Predicted onset voltages agree with measurements within 20%, and the threshold was consistent with published ramified-electrospray observations. The agreement with experiments supports electrocapillary
symmetry breaking as the mechanism of onset and provides a compact, physically grounded criterion for predicting the transition from electrified jets to rim-emitting sheets.
Read the full abstract ▾
Atmospheric and Oceanic Physics
Identifying Arctic states vulnerable to rapid ice loss with a convolutional neural network
Matthew Patterson, Ioana Colfescu, Nathan Kirk et al. (+1 more)
This work investigates the incidence of rapid ice-loss events in the Arctic, in which a substantial fraction of September sea-ice is lost in only a few years. Such events are seen to occur in
climate model simulations forced with increasing greenhouse gas concentrations, but not as yet in the relatively short observational record. We train a convolutional neural network (CNN) using six large
climate model ensembles to diagnose vulnerability to rapid ice-loss events. The trained model is shown to separate higher and lower risk states in out-of-sample
climate models via a leave-one-out cross-validation. The model emphasises the role of persistently thin sea-ice, remaining sea-ice extent and sustained high upper ocean heat content as predictors of rapid ice-loss. The CNN is applied to observational data over the satelite-era showing a signal for rapid ice-loss prior to a drop around 2007 and an increasing risk from around 2015 to present. Such rapid-ice loss events would likely have dramatic consequences for Arctic ecosystems, coastal communities and global shipping.
Read the full abstract ▾
All authors Matthew Patterson, Ioana Colfescu, Nathan Kirk, Graeme MacGhilchrist
Dynamical Systems
Fatou Components of Attracting Skew Products: From Polynomial Classification to Transcendental Complexity
Tom Potthink
We study the Fatou components of holomorphic skew products $F \colon \mathbb{C}^2 \to \mathbb{C}^2, F(z,w) = (f(z) + r(z,w), g(w))$ near an attracting invariant fiber, introducing a refined classification of components on the fiber: bulging, partial bulging, and nowhere bulging. For
polynomial skew products, we give a complete characterization of when the basin at infinity is non-bulging, and show that in this case non-bulging implies nowhere bulging, ruling out partial bulging entirely. In the transcendental setting, the behavior is richer: we construct partially bulging wandering domains and Baker domains, show that any orbitally unbounded Fatou component can be made nowhere bulging by choosing an appropriate function $r$, and prove that for any nonconstant $r$ there exists an entire function $f$ with an escaping wandering domain that fails to be bulging. Bulging thus depends on the interplay between fiber dynamics and perturbation.
Read the full abstract ▾
Machine Learning
Characterizing and Mitigating Distributional Bias in Physics-Informed Generative Models
Tanja Bien, Samir Darouich, Lluís Pastor-Pérez et al. (+3 more)
Physics-informed generative models augment a data-fitting objective with residual penalties that encourage physical constraints to hold in generated samples. However, improving constraint satisfaction need not improve fidelity to the data distribution. We study this conflict in residual-regularized
diffusion models and identify two possible consequences. First, the same prediction error can incur different residual costs near different feasible states, preferring states with lower residual sensitivity (avoidance). Second, this preference can extend to feasible states outside the data support, where the residual also vanishes (leakage). Motivated by this analysis, we propose a data-anchored penalty that weights prediction errors by the residual Jacobian at the clean training sample and normalizes the resulting magnitude. This removes the incentive to relocate to less sensitive states without changing the model architecture or sampling procedure. Experiments on a controlled synthetic system, three PDE systems, and molecular transition-state generation with flow matching demonstrate the benefits of anchoring. Code is available at https://github.com/TanjaBien/anchored_
diffusion.
Read the full abstract ▾
All authors Tanja Bien, Samir Darouich, Lluís Pastor-Pérez, Vinh Tong, Loay Mualem, Mathias Niepert
Superconductivity
Tailored VxOy Metal–Insulator-Transition Oxides for Temperature-Triggered Smart Insulation in Superconducting Coils
Mattia Ortino, Davide Uglietti, Andrea Testino et al. (+5 more)
Smart insulation is a turn-to-turn layer that insulates in normal operation and conducts only near a hot spot. It could combine the fast ramping of insulated coils with the passive energy dump of non-insulated coils. Metal-
insulator transition (MIT)
oxides are natural candidates, but the only one tested in a coil so far, V2O3, switches at 150-170 K, far above the 30-70 K window useful for REBCO magnets. We explore the vanadium-oxygen system to lower the transition. TGA redox processing produced multi-phase powders with an effective transition near 127 K, about 25 K below V2O3. Hydrothermal V3O7, hydrated or annealed, was far too resistive and showed no usable transition. V2O3-epoxy specimens provided a robust reference and an acceptance criterion. Cryogenic measurements revealed a reproducible thermal hysteresis, and the measured specific heat of V2O3 corrects earlier model inputs by close to two orders of magnitude. A first electro-thermal model reproduces current sharing through the smart-insulation layer. Doped V2O3 and V8O15 are identified as the next targets.
Read the full abstract ▾
All authors Mattia Ortino, Davide Uglietti, Andrea Testino, Nicola Casati, Alexander Steppke, Wiktoria Wnetkowska, Andrea Masi, Kamil Sedlak
Operator Algebras
Pointwise Ergodic Theorems Along Reiter Sequences on Locally Compact Groups
Panchugopal Bikram, Sudipta Kundu, Hariharan G et al. (+1 more)
Let $G$ be a locally compact, second countable group admitting a Reiter sequence, and suppose that $G$ acts on a von Neumann
algebra $\mathcal{M}$ equipped with a semifinite trace $\tau$ via trace-preserving
automorphisms. In this paper, we prove a maximal inequality for ergodic averages along such Reiter sequences. Our approach relies on approximating the Reiter sequence by a linear combination of characteristic functions of Følner sequences using the layer-cake representation. Furthermore, we establish the bilateral almost uniform convergence of these ergodic averages. As an application, we deduce a non-commutative Wiener-Wintner type
theorem along Reiter sequences.
Read the full abstract ▾
All authors Panchugopal Bikram, Sudipta Kundu, Hariharan G, Diptesh Saha