Searcharxiv⌕ Search

arXiv subjects

Qian Yang

Publications and source records attributed to Qian Yang.

At least 19 recordsLinked to original sources

Rethinking Actions: Continuity-Aware Priors for World-Model Planning

Actions in expert demonstrations exhibit temporal structure. An ordered-versus-shuffled analysis across five environments confirms this dependence. We ask whether action proposals that better reflect expert continuity can improve visual world-model planning. Rather than imposing generic smoothness, we learn continuations from the temporal structure of demonstrated actions. Our locality-gated deep-kernel Gaussian-process prior, LG-DKGP, combines recent and longer-range executed action history, predicts a neural continuation, and learns a temporal Gaussian-process (GP) residual. A training-derived exploration scale initializes search around the corrected mean; a frozen world model evaluates progress toward the goal. Training first fits the neural mean, then jointly optimizes the prior with the GP residual using expert future actions. The resulting planner achieves 87.07\% macro success across five simulated tasks spanning native and fragment-level goal attainment, compared with 71.03\% for INTACT, 61.66\% for PRISM, and 58.66\% for LeWM. Component interventions and expert-relative action-structure comparisons support demonstration-informed continuations and reveal task-dependent effects of visual conditioning, history length, and planning horizon.

cs.RO↗

The SDSS-V Black Hole Mapper Reverberation Mapping Project: Short Lags in the COSMOS Field

We present the first reverberation mapping (RM) measurements in the COSMOS field included in the SDSS-V Black Hole Mapper Reverberation Mapping Project. Using $\sim$6 months of plate-based spectroscopy (from late 2020 to mid-2021) and accompanying ZTF+DECam \emph{gri}-band photometric light curves, we analyze the variability of over 250 active galactic nuclei (AGNs) out to $z \sim 3$. We successfully recover reliable $\lesssim$ 200d time lags for 22 AGNs ($z \lesssim 2.5$) in H$α$, H$β$, MgII, or CIV, using a combination of multiple lag detection methods and significance tests. Our measurements are consistent with the radius-luminosity relations from recent large-scale RM programs, providing measured masses of $10^{7-9} M_{\odot}$ for these AGNs. These results and upcoming long-baseline SDSS-V spectroscopy lay the groundwork for large-sample AGN studies in conjunction with other facilities covering the COSMOS field, including the Vera C. Rubin Observatory and Roman Space Telescope.

astro-ph.GA↗

Layer-by-layer water filling in molecular-scale capillaries monitored by AFM

Under ambient humidity, water spontaneously condenses in pores only a few nanometers in size, making nanoscale capillarity central to numerous natural phenomena and technological applications. At these dimensions, water may no longer be treated as a continuous fluid, yet the consequences of molecular discreteness for capillary condensation and filling remain poorly understood. Here we study two-dimensional (2D) nanocapillaries fabricated by van der Waals assembly and, using atomic force microscopy (AFM), monitor their wall deformations during humidity-driven water uptake. We observe two distinct regimes: layer-by-layer filling of flexible capillaries and abrupt filling of more rigid ones. Flexible walls deform in around 3 Å steps, consistent with the entry of individual molecular layers of water. The different filling regimes are explained by a competition between wall deformation energy and oscillatory wall-water interactions. These results, together with the theoretical framework developed here, show how molecular discreteness can profoundly affect ubiquitous capillary phenomena, with wall rigidity determining whether this discreteness is exposed or hidden within an apparently classical, Kelvin-like response.

cond-mat.mes-hall↗

Memory as Belief: Bayesian Evolution of Memory for Self-Evolving LLM Agents

Large language model (LLM) agents increasingly rely on external memory to accumulate past experiences for continual adaptation. However, effectively reusing memory requires accurately assessing its value, as retrieved experiences may become unreliable when transferred across task contexts. Despite its importance, memory valuation remains largely underexplored, with existing methods relying on historical retrieval success rates that overlook the context dependency of memory usefulness and the varying reliability of estimates under different amounts of evidence. Inspired by the Bayesian brain hypothesis, we propose BEAM (Bayesian-style Evolutionary Agentic Memory), a framework that reformulates memory value evaluation as a parameterized cognitive inference process, where memory value is represented as a context-conditioned dynamic belief rather than a passive historical statistic. Specifically, BEAM infers memory value beliefs from current task contexts and continuously calibrates them using execution feedback as empirical evidence, transforming memory from passive storage into an adaptive cognitive mechanism for continual self-evolution. Extensive experiments across reasoning and code generation benchmarks demonstrate that BEAM consistently improves memory utilization and enables continual self-evolution of LLM agents.

cs.LG↗

AI-Mediated Self: How HCI Defines and Relates to the Self

How might AI alter how we understand and experience the self? This scoping review analyzes 102 papers to examine how the self is defined in the field of human-computer interaction (HCI), how AI-self relationships are conceptualized, and what risks emerge when AI becomes entangled with selfhood. Our synthesis makes three contributions. First, we define AI-mediated self as a conceptual umbrella that connects dispersed work across education, workplace, health, and creative practices. Second, we consolidate six framings of the self with four domains of ethical risk-agency/autonomy, identity/authorship, relational capacity, and meaning-making-into a conceptual map that provides a reusable vocabulary across contexts. Third, we introduce the Inclusion of AI-Self framework, which situates AI-self relationships along a spectrum of proximity. Together, these contributions position selfhood as a central design space in HCI.

cs.HC↗

MPC-Based Trajectory Tracking for a Quadrotor UAV with Uniform Semi-Global Asymptotic Stability Guarantees

This paper proposes a model predictive trajectory tracking approach for quadrotor UAVs subject to input constraints. Our proposed approach relies on a hierarchical control strategy with an outer-loop feedback generating the required thrust and desired attitude, and an inner-loop feedback regulating the actual attitude to the desired one. For the outer-loop translational dynamics, the generation of the virtual control input is formulated as a constrained model predictive control problem with time-varying input constraints and a control strategy, endowed with uniform global asymptotic stability guarantees, is proposed. For the inner-loop rotational dynamics, a hybrid geometric controller is adopted, achieving semi-global exponential tracking of the desired attitude. Finally, we prove that the overall cascaded system is semi-globally asymptotically stable. Simulation results illustrate the effectiveness of the proposed approach.

eess.SY↗

How and What to Imagine? Visual Thinking in Unified Multimodal Models for Cross-View Spatial Reasoning

Cross-view spatial reasoning remains a weak spot for vision-language models (VLMs): they reason in language and discard the fine-grained geometry the task requires. Thinking with images aims to fix this by generating an intermediate thinking-image, but recent work shows the visual evidence in these traces is largely ignored. We therefore ask how to make visual thinking matter, and what kind of visual thinking works best. We ask these questions for unified multimodal models (UMMs) that natively support interleaved image-text generation. For the how, we propose View Dropout (VDrop), a training-time intervention that hides parts of one input view from the answer span while leaving it visible to the thinking-image tokens. This incentivizes the model to make use of the thinking-image when answering, rather than answering based on the input views only. With the thinking-image now being used in answer prediction, we ask which kind of visual thinking works best. We frame this as a Learnability-Informativeness (L-I) tradeoff and compare three thinking-image variants: top-down, panoramic, and point-matching renderings. Trained on synthetic scenes and evaluated on five real-world out-of-domain benchmarks, panoramic visual thinking with VDrop is the only configuration that is simultaneously informative and learnable, and achieves the best out-of-domain generalization.

cs.CV↗

PRISM-Bench: An Audio-Centric Diagnostic Benchmark for Text-to-Audio-Video Generation

Text-to-audio-video (T2AV) generation has advanced rapidly, but its evaluation still underestimates the audio modality. Existing benchmarks either treat audio as an auxiliary component of video quality or assess it in isolation from audiovisual grounding, making it difficult to diagnose where current systems truly succeed or fail in audio generation. We present PRISM-Bench, the first audio-centric diagnostic benchmark for T2AV generation. Built from a rigorously curated dataset of 900 human-verified samples, PRISM-Bench factorizes audio evaluation along two orthogonal axes: audio type (Speech, Music, and Sound) and sound-source visibility (On-screen vs. Off-screen). It evaluates generated content across four perceptual dimensions (Audio-Visual Coherence, Audio Quality, Audio Expressiveness, and Prompt Following) with 35 fine-grained criteria. To ensure reliable assessment, we adopt an enhanced MLLM-as-a-Judge protocol based on blind, side-by-side comparison against ground-truth references, demonstrating strong alignment (over 70% mean agreement) with human raters. Our evaluation of recent T2AV systems highlights a significant performance gap between frontier and open-source models. Furthermore, we demonstrate that current generation paradigms overfit to perceptual fidelity while struggling with complex grounding and control tasks, particularly in generating music and synchronized On-screen audio.

cs.MM↗

IP Protection in the Era of Visual Generative AI: A Survey

The rapid evolution of visual generative AI has introduced a wide range of intellectual property risks, spanning the unauthorized learning, reproduction, extraction, misuse, and redistribution of protected data and model assets. To address these risks, a growing body of technical defenses has been proposed. However, existing surveys typically organize this literature by lifecycle stage or technical mechanism, which can obscure the protective intent of different methods. This survey presents a two-dimensional taxonomy for IP protection in visual generative models. The primary axis is a Control Logic View, which classifies methods into Information Exposure Control, Generative Behavior Constraint, and Attribution & Accountability according to the risk variable they regulate. The secondary axis distinguishes Data IP from Model IP as cross-cutting asset dimensions. Under this framework, we systematically review protection methods, align evaluation protocols with protection objectives, and discuss open challenges including proactive model-level safeguards, standardized evaluation, robustness against adaptive attacks, and explainable evidence. This survey aims to offer a principled, systematic, and easy-to-follow overview for both new and experienced researchers in visual generative AI IP protection.

cs.CV↗

PACHA: Probing AGN Coronae with High-redshift AGN

The X-ray emission of active galactic nuclei (AGN) is generally attributed to inverse Compton scattering of accretion-disk photons by hot electrons in a compact corona. In local AGN, directly constraining coronal properties is challenging because the high-energy cutoff often lies beyond the NuSTAR bandpass. High-redshift, luminous quasars enable systematic constraints on the high-energy cutoff, as cosmological redshift shifts the spectal cutoff into the observable hard X-ray band. We present first results from the ``Probing the AGN Coronae with High-redshift AGN'' (PACHA) project, based on quasi-simultaneous NuSTAR and XMM-Newton observations of 13 radio-quiet AGN at $z>1$. We constrain the high-energy cutoff and coronal temperature at 90\% confidence level for 10 and 9 sources, respectively. The sample exhibits a mean cutoff energy of $E_{\rm cut}=80.8\pm8.1$ keV and a mean coronal temperature of $kT_{\rm e}=18.4\pm1.6$ keV, both significantly lower than those measured in local {\it Swift}-BAT AGN, while the mean optical depth ($τ=4.8\pm0.3$) is significantly higher. The uncertainties are at 1~$σ$. Combining our high-redshift sample with local AGN, we find a potential anti-correlation between cutoff energy and both X-ray luminosity and black hole mass, with no significant dependence on Eddington ratio. Within a hybrid coronal framework, the inferred temperatures lie well below the pair-production limits for purely thermal coronae, indicating a substantial efficient Compton cooling and/or non-thermal electron component. The detection of low coronal temperatures in high-luminosity AGN is broadly consistent with predictions from recent radiation MHD simulations that consider purely thermal electron populations, implying that non-thermal electrons may not be the primary drivers of the observed coronal properties in these systems.

astro-ph.HE↗

An Exploratory Analysis of New Large Gaia-informed Quasar Samples in SDSS-V

Quasars are luminous objects that provide insights into the physics and evolution of supermassive black holes (SMBHs) and their accretion flows, galaxy evolution, and even cosmology. In this study, we present an exploratory study based on the ongoing fifth generation of the Sloan Digital Sky Survey (SDSS-V) and its unique dual-hemisphere, wide-field, and multi-object spectroscopic capabilities, with the aim of creating a comprehensive, all-sky quasar sample. The targets were selected through two novel methods, GUA and Skewt-QSO, that rely primarily on data from WISE and Gaia, aiming to address gaps in previous large quasar samples. Our sample includes over 250,000 spectroscopically confirmed quasars reaching z~5, with tens of thousands of newly identified quasars in the southern hemisphere. The selection methods are highly pure, with well over 80% of the spectra collected being genuine quasars; the main contaminants are M-type stars. The detailed spectral decomposition procedure we employed shows that the quasars in the sample span a wide range of luminosities (Lbol~$10^{44}-10^{48} erg s^{-1}$), SMBH masses (MBH~$10^6-10^{10}$ Msun), and accretion rates (L/LEdd~0.01-1). The distributions of these properties are consistent with those of previous quasar catalogs, which are based on past generations of SDSS, once we account for potential selection biases related to the various survey depths. Our findings confirm that novel selection methods based on optical+IR colors and/or astrometry can yield a large, high-purity quasar sample over wide sky areas, including in cases where more nuanced multi-band photometry and/or multi-wavelength data in the X-ray or radio is not available. This SDSS-V sample, which will continue to grow, establishes a robust reference for future southern (time-domain) surveys, while enhancing and complementing our understanding of quasar demographics and SMBH evolution.

astro-ph.GA↗

Symmetry-aware generative design of flat-band materials beyond known crystal-net prototypes

Flat electronic bands underlie a range of strongly correlated and topological phenomena, whose design in real materials has so far relied on a small catalogue of named geometric motifs such as kagome, Lieb, and pyrochlore nets. This discrete catalogue is by no means to exhaust the geometries that support flat bands in real compounds, as band flatness is a property of network connectivity. Here we combine a continuous geometric representation of crystal sublattices, with a symmetry-constrained generative model, to access a broader design space for materials hosting flat bands. The key step is to choose sublattice motifs that are outside the known geometric clusters, ensuring the novelty of the generated structures. We then introduce SkeleGen, which pins these unconventional skeletons to symmetry-compatible Wyckoff positions while denoising the surrounding chemistry, resulting in 9,352 crystal candidates that survive stability and flatnessscreening. Band flatness is confirmed using high throughput full DFT calculations, which agree well also with the tight-binding spectra of the isolated skeletons, supporting a geometric origin of the band flatness. We demonstrate "out-of-distribution" motifs as a new design principle to dramatically expand geometric repertoire for materials discovery, potentially beyond flat bands.

cond-mat.mtrl-sci↗

Exploring Reinforcement Learning for Fluid Transitions Between Clinical Mental Healthcare and Everyday Wellness Support

Mental health struggles wax and wane, yet clinical and wellness interventions typically operate separately, causing frequent breakdowns at care transitions. We explore reinforcement learning (RL) as a means to build digital health systems that deliver clinical and wellness interventions proactively, as part of a coherent care journey. We ask: what complexities does designing such a system involve? We built a contextual bandit that dynamically selects journaling prompts from clinical and wellness repertoires to optimize for an overarching health goal (sustained journaling) and deployed it in a four-week exploratory study (N=38). We found that, first, many benefits of RL-optimized intervention sequences appeared only after interventions ended, raising the question: Should systems that offer coherent clinical-wellness care journeys include stepping-back periods? If so, when and how? Second, participants most engaged with RL-generated interventions deepened their engagement over time, while those most engaged with a constant intervention tended to burn out and drop out later. It raises the question: When should a system blending clinical and wellness interventions reduce intensity to prevent burnout in versus sustain it to maximize treatment gains?

cs.HC↗

Attention: What Prevents Young Adults from Speaking Up Against Cyberbullying in an LLM-Powered Social Media Simulation

Interactive, multi-agent social simulation systems have shown promise for helping users practice navigating various complex social situations across domains. This paper asks: To what extent can such systems help young adult (YA) bystanders speak up publicly against cyberbullying, a task often thwarted by complex, multi-party social dynamics? We created Upstanders' Practicum, a multi-AI-agent social media simulation powered by Large Language Models (LLMs), as a probe and observed 34 YAs freely practicing public bystander intervention across three iteratively refined versions. We found that practicing public bystander intervention in the simulation was helpful, but after participants made three attention shifts: (1) from inattention to paying true attention, (2) from self-focus ("I don't usually do this'') to attending to those directly involved, and (3) from resolving the private conflict between bully and victim ("maybe I could set up the meeting between them'') to addressing the broader audience online ("public comment is about norm-setting"). Only after these shifts did practice in the simulation start to help: participants then saw a reason to speak up publicly and, through continued practice, crafted tactful public messages without explicit instruction. These findings illuminate new design and research opportunities for bystander education beyond social skill instruction, namely, designing for true attention, for fostering a vocal upstander identity, and for seeing bystander intervention as public norm setting. In addition, we open-source Truman Agents (cornell-design-aigroup.github.io/TrumanAgents/), the first-of-its-kind multi-LLM-agent social media simulation platform that Upstanders' Practicum builds upon, for future cyberbullying and social media research.

cs.HC↗

Architecting mechanosensitive nanofluidic transport in graphite nanoslits

Mechanosensitive ion transport plays a central role in enabling living systems to perceive and adapt to their environment through the deformation of soft, embedded ion channels. In this work, we demonstrate that ion transport within a two-dimensional graphite nanoslit can be rationally engineered to achieve a bipolar, pressure-sensitive response without any structural deformation. The mechanosensitivity arises from the selective charging of one channel inlet, which acts as a reversible source of mobile charge carriers. These excess-ions can then be advected in or out of the channel by the pressure-driven water flow, thereby modulating the ionic conductance. This mechanism is captured through a comprehensive electrohydrodynamic model that analytically accounts for coupled diffusion, convection, surface transport, diffusio-osmosis, and interfacial slippage, both inside and outside the nanoslit. The theoretical framework quantitatively reproduces the experimental data, showing that a simple surface charge pattern can give rise to complex, pressure-dependent conductance. These findings reveal how rich nonlinear couplings at the nanoscale can be harnessed to design adaptive, bioinspired nanofluidic systems, exemplified here by ionic pressure sensors.

cond-mat.soft↗

Discovering Failure Modes in Vision-Language Models using RL

Vision-language Models (VLMs), despite achieving strong performance on multimodal benchmarks, often misinterpret straightforward visual concepts that humans identify effortlessly, such as counting, spatial reasoning, and viewpoint understanding. Previous studies manually identified these weaknesses and found that they often stem from deficits in specific skills. However, such manual efforts are costly, unscalable, and subject to human bias, which often overlooks subtle details in favour of salient objects, resulting in an incomplete understanding of a model's vulnerabilities. To address these limitations, we propose a Reinforcement Learning (RL)-based framework to automatically discover the failure modes or blind spots of any ``candidate VLM'' on a given data distribution without human intervention. Our framework trains a questioner agent that adaptively generates queries based on the candidate VLM's responses to elicit incorrect answers. Our approach increases question complexity by focusing on fine-grained visual details and distinct skill compositions as training progresses, consequently identifying novel failure modes in which VLMs struggle. We demonstrate the broad applicability of our framework by showcasing its generalizability across various model combinations.

cs.CV↗

The Hadronization Impact on $J/ψ$ Energy Correlators: A Pythia8 Study from Partonic to Hadronic Observables

A comprehensive study of the $J/ψ$ energy correlator as a probe of non-perturbative hadronization in color-octet $c\bar{c}$ production is performed. The energy correlator measures the energy flow as a function of the angular distance ($χ$) from the identified $J/ψ$ meson. Using the PYTHIA 8 Monte Carlo event generator, the correlator is computed at both parton and hadron levels. At high $J/ψ$ transverse momentum ($p_T > 7\ \text{GeV}/c$), the parton-level correlator in the $\cosχ> 0$ region is dominated by soft gluon emission during the hadronization of the color-octet state, a contribution clearly distinguishable from other partonic sources, such as underlying multi-parton interactions. The transition to the hadron level, however, introduces substantial modifications, suppressing the correlator in this region by approximately an order of magnitude and underscoring the complexity of the hadronization mapping. Further analysis reveals that the hadron-level observable exhibits notable sensitivity to model parameters: increasing the mass splitting between colored $c\bar{c}$ pre-resonances and the $J/ψ$ meson from 0.2 to 0.8 GeV/$c^{2}$ enhances the correlator by up to $60\%$, while extending the color reconnection range yields a milder enhancement of about $10\%$. These findings demonstrate that precise measurements of the hadron level $J/ψ$ energy correlator, when interpreted within robust event-generator frameworks, can provide novel constraints on hadronization dynamics and help clarify the production mechanisms of $J/ψ$ state.

hep-ph↗

Learning What Matters: Prioritized Concept Learning via Relative Error-driven Sample Selection

Instruction tuning has been central to the success of recent vision-language models (VLMs), but it remains expensive-requiring large-scale datasets, high-quality annotations, and large compute budgets. We propose PRioritized cOncept learninG via Relative Error-driven Sample Selection (PROGRESS), a data- and compute-efficient framework that enables VLMs to dynamically select what to learn next based on their evolving needs during training. At each stage, the model tracks its learning progress across skills and selects the most informative samples-those it has not already mastered and that are not too difficult to learn at the current stage of training. This strategy effectively controls skill acquisition and the order in which skills are learned. Specifically, we sample from skills showing the highest learning progress, prioritizing those with the most rapid improvement. Unlike prior methods, PROGRESS requires no upfront answer annotations, queries answers only on a need basis, avoids reliance on additional supervision from auxiliary VLMs, and does not require compute-heavy gradient computations for data selection. Experiments across multiple instruction-tuning datasets of varying scales demonstrate that PROGRESS consistently outperforms state-of-the-art baselines with much less data and supervision. Additionally, we show strong cross-architecture generalization and transferability to larger models, validating PROGRESS as a scalable solution for efficient learning.

cs.CV↗