Papers

We investigate emission from polycyclic aromatic hydrocarbon (PAH) dust molecules in 224 MIRI-selected star-forming and quiescent galaxies at $z=0.5-0.8$ in the GOODS-South field using multi-wavelength data from the AstroSat/UVIT, JWST, HST, Herschel, and MUSE. We define an observed diagnostic parameter, f$_{\rm PAH}$, as the ratio of JWST/MIRI F1280W to the average of F770W and F2550W band fluxes, and demonstrate that it effectively traces the PAH mass fraction (q$_{\rm PAH}$) by capturing 7.7$\mu$m PAH emission at our targeted redshift. We observe a bimodal distribution of similar nature in both f$_{\rm PAH}$ and q$_{\rm PAH}$, suggesting that PAH enrichment in galaxies goes through a rapid transition phase from a PAH-poor to a PAH-rich system, at about 70% solar metallicity. For the star-forming galaxies (SFGs; i.e., $\sim$78% of the total sample), we find that f$_{\rm PAH}$ increases with stellar mass and rest-frame FUV luminosity, identifying massive, UV-bright galaxies to have higher PAH mass fraction. However, the majority of the massive quiescent galaxies (QGs) show lower PAH mass fraction relative to that of SFGs of similar mass, suggesting PAH and smaller dust grains are more susceptible to destruction during quenching processes. Among SFGs, we observe an anti-correlation between f$_{\rm PAH}$ and the O32 emission line ratio, suggesting PAH photo-dissociation in strongly ionized ISM. This anti-correlation, along with a correlation between f$_{\rm PAH}$ and gas-phase metallicity 12+log(O/H), indicates that low-mass, metal-poor SFGs, which are increasingly common at even higher redshifts, host more extreme conditions less favorable for PAH production and survival.

Despite the promise and potential of quantum computing, most of the current day developments of quantum computing have several limitations for relevant practical use. As a result, hybrid classical-quantum computing has emerged as a viable solution. This paper discusses the software aspects of hybrid classical-quantum computing solutions to understand and examine how hybrid computing solutions can be expected to work in practice. This paper examines existing software solutions (cloud applications, middleware frameworks, and API gateways) that can enable classical and quantum computing to work together. For practitioners the paper provides a curated list of algorithms, full stack libraries, simulators, and cloud service providers that are likely to be central to the development of any hybrid computing solutions. We conclude by identifying an agenda for future research on classical-quantum hybrid software engineering, and some near-term priorities for development of functional, robust hybrid software systems. The paper argues that due to the deep dependency of quantum execution on classical pre- and post-processing, hybrid computing may represent a near-permanent outcome of classical-quantum computing rather than a transitional workaround.

World models must predict how an environment evolves under sequences of actions, enabling agents to compare possible futures and reason about counterfactual actions before acting. Long-horizon prediction is commonly obtained by recursively applying a one-step transition model, but intermediate errors can compound over time. Multi-step dynamics models instead condition on a sequence of future actions and predict their consequences directly, but become harder to learn as horizon grows: the model must track interacting state changes across the trajectory, endpoint supervision provides weak credit assignment, and intermediate predictions can remain plausible while losing information needed for later states. We show that these challenges can be addressed by casting the internal evolution of a multi-step transition as structured reasoning over textual world states: reasoning over sparse state changes reduces the burden of state tracking, a predictive-gain objective rewards the learned state for improving over a matched predictor that conditions on raw history instead, and intermediate predictive rewards supervise each state along the trajectory. Because these intermediate states are explicit textual representations of the world, they provide semantically meaningful targets that can be inspected, scored, and corrected during training. Across ScienceWorld, Jericho, and CEO-Bench, our approach achieves the strongest average long-horizon performance against recursive and non-recursive baselines that condition directly on raw history, with gains increasing at longer horizons. In a controlled counterfactual study, our model is also the only one with statistically significant sensitivity to future actions.

Many machine learning applications involve sensitive data and therefore require training under differential privacy (DP). However, DP training often degrades model utility. In some cases, first pre-training the model on "public" data before finetuning with DP on the sensitive data can reduce the drop in utility. However, the success of this depends on how relevant the selected public dataset is to the sensitive data. We introduce the first pipeline that privately learns the mixture of several public datasets to pretrain on for a given sensitive downstream task. Our key insight is that we can privately find the best mixture of multiple public datasets by privately learning a low-dimensional linear model. We tested our method on the NIH dataset for X-ray classification and the ENRON email dataset for language modeling. Applying our method to find tailored mixtures of X-ray datasets to pretrain on for diseases in the NIH ChestX-ray14 dataset, we improved macro AUC by up to 0.037 across privacy budgets compared to the baselines, with gains as large as +22.8% relative AUC on Cardiomegaly at $\epsilon=1$. For DP training on the ENRON dataset, pre-training on our mixture of The Common Pile (a collection of public-domain text datasets) decreased test perplexity by 16% relative to the baseline mixtures.

Galaxy richness maps provide a useful tracer of the stellar mass distribution in galaxy clusters, but identifying cluster members from multiband photometric data can be challenging. In this work, we investigate multiband imaging data obtained with the Super-Pressure Balloon-borne Imaging Telescope (SuperBIT) during its science flight in April 2023. Using spectroscopically identified cluster members across the SuperBIT cluster sample, we benchmark the performance of the three colors available from the three SuperBIT bands: $(u-b)$, $(b-g)$, and $(u-g)$. We quantify the performance of each color using the $1\,\sigma$ scatter of the spectroscopically identified members in color and the purity of our cluster-member identification algorithm. We find that $(b-g)$ consistently exhibits the smallest color scatter and highest purity across the sample and over the redshift range probed by SuperBIT. We therefore identify $(b-g)$ as the preferred color for red-sequence-based cluster-member identification, providing a basis for future analyses of the cluster galaxy population using SuperBIT multiband photometry.

The Brightest Cluster Galaxies (BCGs) in the most massive galaxy clusters of our Universe are, at present, mostly quiescent, elliptical, and red. However, their formation histories are unclear and likely different from those of massive ellipticals in lower-mass haloes. Simulations face the challenge of simultaneously requiring large volumes to sample massive clusters and sufficient resolution to model their stellar populations. Therefore, we make use of TNG-Cluster, one of the highest-resolution cosmological simulations of galaxy clusters, which returns 352 BCGs at $z=0$ with stellar masses $M_{*, <30 \mathrm{pkpc}} = 10^{11.6-12.9} ~\mathrm{M}_\odot$ and total halo masses $M_{\mathrm{200c}} = 10^{14.3-15.4}~\mathrm{M}_\odot$. 93 per cent of the BCGs are quenched at $z=0$, whereas at redshift $z>3$ all their main progenitors were still star forming. The star-formation histories of individual BCGs are complex, including multiple (re)quenching episodes. They rejuvenate on average $2.1 \pm 1.8$ times since $z=3$. BCGs quench for the first time $9.9^{+0.9}_{-1.4} ~\mathrm{Gyr}$ ago (10th-90th percentiles; $z=1.7^{+0.5}_{-0.5}$), while the last and final quenching event occurred $7.0^{+2.8}_{-5.0} ~\mathrm{Gyr}$ ago ($z=0.8^{+0.8}_{-0.6}$). We find a large scatter in the halo, stellar, and SMBH masses at both first and last quenching, suggesting no characteristic mass at which quenching occurs. Instead, BCGs quench for the first time when the cumulative kinetic energy injected by their SMBH equals the BCG binding energy, corresponding to a delay of $2.3^{+1.2}_{-1.2} ~ \mathrm{Gyr}$ after the first SMBH kinetic energy injection. Namely, TNG-Cluster BCGs quench when they experience significant gas expulsion. Our work shows that BCGs have complex and varied star-formation histories, with multiple rejuvenation episodes, but they ultimately quench due to SMBH feedback.

Interactions of very-high energy gamma-rays during their propagation through the intergalactic medium result in a delayed secondary “pair echo” signal. Detection of such pair echo can be used for measurements of intergalactic magnetic fields (IGMF) in the voids of the Large Scale Structure of the Universe. We notice that a combination of data of LHAASO and CTAO/LST-1 telescopes on the gamma-ray burst GRB 221009A is consistent with the pair echo model. The LST-1 measurement of the source signal 1.33 days after the burst can be interpreted as the pair echo produced by very-high-energy gamma-rays propagating through intergalactic medium with magnetic field of the strength log(B/G)=-16.2+/- 0.7, if its correlation length lambda_B is larger than 20 kpc. If the field correlation length is shorter than 20 kpc, the field strength estimate is modified upwards by a square root of (20 kpc / lambda_B). In the specific case of short correlation length IGMF that is relic from cosmological phase transitions, the pair echo model of the LST-1 signal implies the field strength in the 1e-14-1e-12 G range. Such magnetic field parameters are consistent with all known constraints on IGMF, including the lower bounds from gamma-ray observations of active galactic nuclei.

Generative speech enhancement models can produce cleaner and more natural-sounding speech than conventional discriminative approaches, but may hallucinate by changing speech content or speaker identity, even with transcript conditioning. We present AuraSE, a flow-matching framework that addresses hallucination through complementary modality and inference designs. First, a double-stream-to-single-stream multimodal Diffusion Transformer (MMDiT) allows transcript and acoustic representations to interact while preserving a dedicated pathway for the degraded input. Second, we find that the best decoder configuration, governed by guidance scale, sampling temperature, and step count, varies substantially across utterances. This observation motivates Inference Policy Optimization (IPO), an online, on-policy preference optimization method. IPO generates multiple candidates from the current model under different inference configurations, ranks them with a multi-objective reward, and learns from their relative preferences. AuraSE-IPO ranks first on 11 of 12 metrics across the synthetic test sets and obtains the highest DNSMOS and blind-listening scores among the evaluated systems on the real DNS blind test set. At deployment, it uses a fixed $10$-step ODE decoder without classifier-free guidance (CFG) or per-utterance configuration search.

In this work, we study transparent boundary conditions for the time-fractional heat equation on a metric star graph. This equation is the standard continuum model for subdiffusive heat and mass transport in disordered and porous branched media. Using the Laplace transform to solve the exterior branch problems, we demonstrate that the non-local vertex operator observed by the incoming bond is an Abel-type convolution kernel of order $\beta/2$, which generalizes the classical half-derivative kernel recovered at $\beta=1$. Our main result is that the diffusion-coefficient sum rule $\kappa_1=\kappa_2+\kappa_3$, known to eliminate thermal backflow for the classical heat equation, remains unchanged for every fractional order $\beta$: the transparency of the junction depends only on the network's diffusivities and not on the memory of the transport law. We rigorously prove this and uniqueness for the coupled vertex problem. We confirm our results numerically at the level of the underlying coupled partial differential equation system with an implicit L1 finite-difference scheme. We compare the star-graph solution to a reference solution on an unbranched line with matching diffusivity.

CoVariance Neural Networks and their extensions have emerged as effective tools for processing multivariate data, deriving graph shift operators directly from second-order statistics. These architectures, however, are designed for independent and identically distributed observations and do not fully capture the joint structure of temporal and cross-variable dependencies in multivariate time series. In this work, we introduce Inverse Cross-Spectral Neural Networks (iCSNNs), a class of graph neural networks for stationary multivariate time series whose shift operators are the inverse cross-spectral density (iCSD) matrices. These operators encode frequency-specific conditional relationships among variables, exploiting the decomposition provided by the spectral representation theorem. Leveraging spectral smoothness, frequencies are grouped into bands sharing a single iCSD operator, yielding a compact parametrisation that retains the frequency-dependent structure of the process. We further propose a joint learning procedure to estimate both the Fourier-domain dependence structure and the iCSNN parameters, adapting the iCSD operators to the downstream task. When tested on synthetic data, iCSNN outperforms baselines from different methodological families.

Vector field tracking for quadrotors provides a systematic framework for specifying complex behaviors at the translational level and transferring them to the full vehicle dynamics. Objectives such as safety, for example, can be encoded into a low-dimensional vector field, which is then embedded into the full dynamics through appropriate tracking conditions. The development of such tracking controllers for quadrotors can be facilitated by the fact that their translational dynamics depend only on the thrust direction and are independent of the vehicle's heading. Thus, reduced-attitude representations, leveraging the geometry of the unit sphere, provide a natural framework for vector field tracking on quadrotors. In this work, we propose two vector field tracking controllers for quadrotors using reduced-attitude representations based on dynamic extension and geometric control, respectively. Next, we illustrate how safety can be encoded into the desired vector field using control barrier functions and enforced for the full system. The effectiveness of the proposed approaches is demonstrated via simulations and experimental flight tests.

We present a phenomenological study of the longitudinal unpolarized structure function $F_{UU,L}$ in semi-inclusive deep inelastic scattering (SIDIS) and its ratio with the transverse structure function, $R_{\, \rm SIDIS}=F_{UU,L}/F_{UU,T}$. We study their transverse-momentum dependence, combining for the first time low- and high-transverse-momentum contributions across the kinematic range relevant to SIDIS measurements. At low transverse momentum, $F_{UU,L}$ is power suppressed, making it sensitive to next-to-next-to-leading-power dynamics in the unpolarized sector. We find that this contribution can be sizable, with $R_{\, \rm SIDIS}\simeq 0.3$ at low transverse momentum and hard scale, producing a correction of the order of $10\%$ to transverse-momentum-integrated observables. Although these estimates remain subject to theoretical uncertainties, our results indicate that accounting for $F_{UU,L}$ is likely to be important for the interpretation of SIDIS multiplicities and motivate dedicated measurements of $R_{\, \rm SIDIS}$ to better constrain the longitudinal contribution.

The information bottleneck (IB) seeks a representation $U$ of a source $X$ that retains as much information as possible about a target $Y$, subject to a constraint on $I(U;X)$. A classical argument shows that it suffices to consider representations with at most $|\mathcal{X}|+1$ symbols, and this bound is known to be tight whenever $|\mathcal{X}| \geq 3$. We show that the binary case behaves differently: if $X$ is binary and $Y$ is finite, then for every joint distribution of $(X,Y)$ and every rate constraint, the IB optimum is attained by a binary $U$. Hence the bound $|\mathcal{U}| \leq |\mathcal{X}|+1$ sharpens to $|\mathcal{U}| \leq |\mathcal{X}|$ for binary sources. The proof combines a separating hyperplane argument with the observation that, for a binary source, the ratio of the second derivatives of the two entropy functions involved is concave.

We investigate the real-time dynamics of the Riesz gas (RG), a paradigmatic system of particles with power-law interactions that encapsulates several many-body interacting classical systems. While the equilibrium properties of the system have been extensively studied the connection between microscopic dynamics, macroscopic evolution and thermalization has remained largely unexplored. Combining extensive large-scale Newtonian simulations with a hydrodynamic framework formulated in terms of three coarse grained fields – density, velocity, and temperature – we establish a direct and quantitative correspondence between the two descriptions. The hydrodynamic theory incorporates two phenomenological transport coefficients, the bulk viscosity $\xi$ and thermal conductivity $\kappa$, which encode dissipative effects at a macroscopic level. By analyzing representative classes of initial conditions, such as dome-like and Newton-cradle-type profiles, we find excellent agreement between microscopic and hydrodynamic evolution over a broad range of timescales, starting from relatively early all the way up to times at which steady state is attained. Notably, this agreement persists across a substantial window of $\xi$ and $\kappa$, demonstrating the robustness of the emergent hydrodynamic description. Our results provide a systematic validation of hydrodynamics in the RG and establish a concrete bridge between Newtonian dynamics and continuum theories in systems with power-law interactions, starting from first principles.

Large language models (LLMs) annotate and scale political text or constructs by generating text tokens. A new class of models, which TypeSafe markets as "System One" models, instead returns decisions and probability distributions across a user-supplied fixed answer set. A commercial model, JEV, is advertised as having a dramatic cost and speed advantage over traditional LLMs along with better calibrated decisions. As such, it might be useful for social scientists looking to quickly and cost-effectively annotate or scale large corpora of text and have a reliable indicator of a classifier's uncertainty. Yet, the accuracy of these claims and the broader model accuracy in social science text-based tasks are not yet established. In this paper, we do just that and hope to establish the suitability of JEV for social science tasks. We compare JEV with LLMs and human coders from published research, and with a current mid-tier commercial LLM (GPT-6 Luna) and an open-weight alternative (Qwen3.8-27B). We find that JEV matches, or comes close to, the capabilities of both LLMs in a variety of tasks. However, we find no cost advantage over GPT-6 Luna at OpenAI's batch prices. Further, we find that, when each question is asked once, JEV's probabilities are better calibrated than GPT-6 Luna's token probabilities, but not consistently better than Qwen3.8-27B's. We conclude that unless researchers have a need for speed, JEV's only obvious advantage is ease of parsing the underlying choice probabilities.

We prove stability of nonrelativistic fermionic matter in the presence of fixed nuclei \(\{R_j\}\) on an asymptotically Euclidean Riemannian manifold \((M,g)\). Upper and lower bounds on sectional curvatures are assumed, but no specific sign conditions on the curvature are imposed. The proof follows the route of Lieb-Thirring and electrostatic inequalities, the latter being proved by means of a Voronoi decomposition type argument which is however based on the Green functions \(\{G(\,\cdot\,,R_j)\}\) rather than the distance function.

Building on our earlier program-evolution workflow guided by large language models (LLMs), we study weight-five bivariate bicycle (BB) and perturbed bivariate bicycle (PBB) codes. The resulting catalogue contains 1,142 distinct code proposals, including 1,081 nonbaseline proposals attributable to LLM-generated programs. Across the catalogue, we certify connected Calderbank–Shor–Steane (CSS) realizations [[96,4,10]], [[140,6,10]], and [[180,4,14]]. A post-search comparison certifies seven imported Lin–Pryadko archive constructions. For leading parameter triples also represented in that archive, we provide exact distance evidence, explicit bivariate presentations, and verified component reductions. A basis-independent connectivity analysis identifies 409 of the 1,142 catalogue entries as disconnected and shows that 73.1% of the classes with exact distance certificates contain repeated connected components. Algebraic analysis organizes the connected CSS classes into order-3, order-7, and order-15 cyclotomic-kernel strata. The strongest exact connected PBB parameter point is [[216,4,10]], attained by two distinct component classes. Among the 936 distinct CSS proposals from the LLM-guided campaign with a recorded positive distance, 816 (87.18%) are certified at $d\geq5$. For comparison, three random-search controls each sample 6,444 CSS code proposals uniformly without replacement, using the same per-lattice and encoded-dimension sample counts as the LLM-guided campaign. In these controls, 4,672–4,785 proposals (72.50–74.26%) meet the same criterion. The LLM-guided campaign has the higher certified yield, while the random controls cover more connected classes. Together, these results extend LLM-guided discovery to a more constrained code family and provide a reproducible structural and exact-distance account of its strongest candidates.

GPU-based optimization faces two communication bottlenecks: coordinating frequent requests and delivering dense models that exceed device memory. We present a discrete simulated bifurcation (dSB) architecture that addresses both through NVIDIA DOCA GPUNetIO. For resident models, a persistent service receives field updates, executes each solve within one GPU thread block, and returns the result. Exact integer coupling sums, GPU work queues, and batched transmission keep the receive–solve–reply path on the device without a dedicated CPU data-path core. In comparisons with socket-based servers using the same solver, the largest latency gains occur under concurrent load. As the offered load increases from 400 to 800 thousand requests per second, median round-trip latency rises by only 6%. At the highest tested load, median and 99th-percentile latencies are 189 and 218~$\mu$s, compared with 288 and 609~$\mu$s for the tuned persistent CPU proxy across repeated runs. For models larger than device memory, a streaming solver retains dynamical state on the GPU and reuses incoming coupling tiles across replicas. It evaluates ten-million-variable dense binary matrices at approximately 307~Gb/s, consuming a 12.5-TB logical matrix through a 64-MiB packet buffer. Ground-state recovery on planted instances and agreement with reference executions verify the computation. Together, the two modes scale dSB to concurrent requests and dense models beyond GPU memory.

Spectral variants of low-rank adaptation (LoRA) choose both a subspace and which factor to freeze. We separate these choices by freezing the input factor A or output factor B on the top or bottom singular directions of pretrained weights, with learning rates selected separately. At rank 2, the same-band advantage of freezing A is larger than either within-factor band difference on all four task-model pairs with complete comparisons. Freezing B also trails comparable-budget free LoRA by 8-18 percentage points on five pairs spanning a formatting task and OpenBookQA. The A-frozen advantage persists in a single-GPU-model replication and within individual MLP module groups, including controls with equal or greater trainable counts for B frozen, and when A is frozen on a random orthonormal basis. The factor contrast weakens with rank. On OpenBookQA / Qwen2.5-1.5B at rank 16, PEFT's MiCA implementation trails comparable-budget LoRA by 3.08 points under a shared training recipe transferred from the MiCA paper. A trained oracle output subspace largely removes the low-rank deficit; partial warm-up gains recur across three direction seeds. The factor-versus-band ordering is descriptive; an approximate multiplicity audit weakens several earlier significance claims. These results extend known factor asymmetry by showing how its magnitude depends on spectral placement, rank and training conditions.

Global condition-number dependence can substantially overestimate the difficulty of preparing normalized quantum linear-system solutions. Many existing beyond-conditioning approaches are favorable when globally ill-conditioned directions have limited relevance to the target solution. Here we study a harder regime for time-evolution quantum Krylov linear solving (QKS), where a vanishing eigenvalue remains solution-relevant and must be retained. We identify a QKS regime in which the solution-relevant spectrum is compressed into a compact reduced space where the ill-conditioning is concentrated along a soft direction aligned with the solution, so that normalization removes the divergent amplification along that direction and leaves only a bounded transverse response. Consequently, the normalized-state complexity need not inherit the divergence of the global condition number. Our contributions are threefold: (1) We show that the solution-relevant information admits a compact and stable QKS representation, with bounded subspace dimension, evolution time, and reconstruction overhead as the relevant eigenvalue vanishes. (2) We show that normalized-state difficulty is governed by whether inverse amplification changes the solution direction, rather than by the small eigenvalue alone; even an arbitrarily ill-conditioned reduced system can remain directionally stable when the amplification is solution-aligned. (3) We prove that this stability persists under finite projected-system errors and transfers to the actual QKS output, yielding an end-to-end complexity bound without inverse dependence on the retained small eigenvalue. An explicit separation family further shows that a full inverse-amplification criterion can diverge while the corresponding QKS quantities remain bounded.

To investigate the design space where children might design AI chatbots for their own toys, we developed ToyTalk, a technology probe that positions children as designers of LLM-enabled toys. Children begin with a familiar toy, configure its AI-enabled version through a no-code interface, and then interact with and test the character. We deployed ToyTalk with 76 children aged 7-9 across five elementary schools in the southeastern U.S. We examine how children define their toy, probe what it becomes, and respond when behavior diverges from expectations. Children predominantly designed toys with socially positive personalities, supportive roles, and interpersonal rules. In conversation, they most often probed identity and knowledge, while also testing capabilities, memory, and relationships. When mismatches arose, children typically responded through correction, persistence, and retesting, while few returned to reconfigure the system. We discuss implications for children's design agency, testing practices, and expectations of coherence in child-facing generative AI.

Characterizing the frequencies of resonant modes is critical to a wide variety of experimental applications. In this work we demonstrate a heterodyne-enhanced sensing scheme capable of precisely measuring the frequency spacing between longitudinal modes of a 19 m optical cavity. The technique utilizes two resonant lasers to sense the differential phase accumulated between them as they propagate through the cavity by measuring their interference beatnote before their injection to the cavity and in transmission of it. With these measurements we were able to mitigate the effects of noise in our control loops in post-processing by up to a factor of 30 in the frequency range from 1 mHz to 100 Hz while suppressing injected calibration signals by over two orders of magnitude.

World Action Models (WAMs) incorporate visual representations from video generation backbones to guide action prediction. Recent efficient WAMs adopt Mixture-of-Transformers (MoT) architectures and compute video representations once for reuse by the action expert. However, intra-expert iteration (\ie, multi-step action denoising) and inter-expert waiting (\ie, sequential execution of the video and action experts) still limit inference efficiency. To this end, we present RealtimeWAM, an extremely efficient WAM variant with one-step action generation and asynchronous inference, addressing these two bottlenecks. To reduce intra-expert iteration, we propose Teacher-Anchored Consistency Distillation (TACD) to address a local-global error gap: low local consistency error alone does not guarantee accurate final actions. TACD supplements local consistency with explicit supervision from the frozen teacher's multi-step rollout endpoint, enabling accurate one-step action generation. Additionally, we propose Cross-Expert Wavefront Pipelining (CEWP) to eliminate unnecessary expert-level waiting. It overlaps the two experts through block-wise sharing of the video KV cache, synchronizing only immediately before the corresponding action attention consumes it. Extensive experiments across diverse benchmarks (\eg, LIBERO, LIBERO-Plus and RoboTwin) and model variants (\eg, Fast-WAM and Faster-WAM) demonstrate the superiority of RealtimeWAM. Notably, RealtimeWAM maintains near-lossless performance (\ie, $<1\%$ drop) across these benchmarks while delivering significant end-to-end speedup (\eg, $\sim25\times$ on H100). Our code and checkpoints are available via this \href{https://github.com/ModelTC/LightX2V/tree/main/scripts/realtimewam}{link}.

The design of a video encoder determines when frames begin to interact and which frame-specific visual evidence remains accessible to the language model. Native video pathways couple neighboring frames during visual encoding, whereas image pathways preserve independently computed frame representations but incur a much larger visual-token cost when all image tokens are forwarded. We ask a basic question: whether a compact video encoder can instead be built on image representations. To answer this question, we separate three operations that are often coupled: per-frame representation, cross-frame token allocation, and temporal interaction. A frozen image encoder first produces frame-specific candidates. A question-aware selector then allocates a fixed token budget across frames using relevance, diversity, and cross-frame correspondence, after which a lightweight learned refiner reads neighboring-frame context and writes residual updates only to the retained anchors. This preserves source positions and keeps the visual output at the fixed budget. Across 13 benchmarks and three vision-language backbones, the resulting pathway matches full-image aggregate performance while using only about 28%-35% of its visual tokens. Specifically, on Qwen3-VL-8B, it achieves a 13-benchmark macro-average of 62.75 with 1,535 visual tokens, compared with 62.58 for the full Image pathway at 4,424 tokens and 59.49 for native Conv3D at 2,212 tokens. On Qwen3-VL-32B, it reaches a 13-benchmark macro-average of 66.28, compared with 66.09 for Image, while providing a 2.16x end-to-end speedup. These results show that compact video encoding does not require early temporal mixing: frame-specific evidence can be preserved first, allocated jointly, and temporally contextualized after selection.

Language models can score psychiatric questionnaires from speech, but agreement with self-report may reflect the questionnaire rather than the person. We introduce psychometric representational similarity analysis, a framework for comparing the structure of speech-derived scores, self-report, item wording and theory, and implement it alongside person-level construct scoring in COMPASS 2.0. We show how similarly worded items induce covariance without psychological signal. In pre-registered discovery and confirmation analyses of clinical interviews from 275 participants, language-derived symptom geometry resembled wording more than self-report, with no structure beyond wording detected by the registered tests. Geometric agreement with self-report survived assigning participants someone else's answers, whereas person-paired scores captured distress more than specific symptoms. Complementary analyses examined counselling quality and wording structure across 34 instruments and the Research Domain Criteria (RDoC) framework. These findings distinguish agreement about psychological structure from evidence that language-derived assessments track individual people.

As generative models and AI agents propose chemical reactions at a scale beyond expert review, feasibility verifiers decide which proposals enter synthesis planning. But do their decisions agree with chemists across different kinds of candidates? We introduce FREA, a benchmark of 751 reactions labeled by expert chemists under an explicit feasibility criterion, drawn from retrosynthesis model proposals, zero-yield experimental records, edits by large language models (LLMs), and five negative candidate generation methods. Our evaluation finds that no verifier leads across all sources: LLMs given only the criterion are competitive with dedicated verifiers, while forward models perform best on retrosynthesis proposals but reject most feasible edits of recorded reactions at the evaluated operating points. Looking beyond aggregate scores, both forward models perform below chance when separating infeasible alternative disconnections from feasible generated candidates. To study whether negative supervision addresses these weaknesses, we also release a corpus of over 14 million recorded reactions and generated negative candidates. In matched training comparisons, adding a mixture of generated negatives to forward training raises mean AUROC across sources, but these gains do not extend to retrosynthesis proposals. Varying the generation method further shows that the largest gain on generated candidates coincides with worse proposal screening. These findings motivate evaluating verifiers against experts across sources and designing negatives for transfer to model proposals.

Deep-learning-based radar target detection has substantially improved detection performance, but its effectiveness depends critically on the correctness of the training labels assigned to radar echo samples. In practical applications, training data may contain partially mislabeled samples, which can cause detection models to learn biased supervisory information and thereby degrade detection performance. To address this issue, this paper proposes SL-RLF, a robust learning framework for deep-learning-based radar target detection with partially mislabeled training data. Specifically, we establish a probabilistic model to characterize the generation mechanisms of label errors in radar target detection and analyze the robustness of detection models under partial label errors within a risk minimization formulation. Theoretical analysis shows that, under appropriate assumptions, a loss function satisfying the symmetry condition enables the detection model in the learning framework to be robust against partial label errors. Guided by this result, we design a loss function satisfying the symmetry condition and construct the proposed robust learning framework, SL-RLF. Theoretically, under the corresponding label-error conditions, the optimal detection model learned by SL-RLF from partially mislabeled training data can achieve the same detection performance under the true data distribution as the optimal model learned from accurately labeled data. Experimental results demonstrate that, under different types and levels of partial label errors, the proposed SL-RLF consistently outperforms representative methods, including Co-teaching and TCE, and its robustness is generally consistent with the theoretical analysis.

The hopping expansion is an important approximation of the quark propagator in lattice quantum chromodynamics, but it diverges for light quarks, which limits the scope of strong-coupling and finite-density simulations. We present an artificial scientific discovery framework using gauge-equivariant neural networks in the Clifford basis to extract interpretable approximations. Our approach yields improved low-order approximations that outperform the hopping expansion at identical computational cost. In the parameter regime where the hopping expansion fails, our approach identifies convergent behavior, thus suggesting a novel series expansion. The discovered approximations generalize to unseen gauge fields and volumes, enabling immediate implementation as plug-in replacements and a new theoretical understanding of quark-propagator expansions.

Existing 3D large language models often overlook fine-grained attributes and less visually salient objects and parts, even when relevant evidence is present in scene videos. We introduce Lens3D to improve fine-grained object understanding through external visual assistance and knowledge transfer. Its LensUnd pipeline adopts 3D localization to select informative, complementary views for an external 2D vision-language model, supporting fine-grained object captioning, small-object grounding, and fine-grained object question answering. LensDistill transfers the resulting fine-grained knowledge to 3D LLMs through detailed caption supervision, enabling captioning from native inputs without external VLM calls. We also construct LensBench, a held-out evaluation set of 2,068 objects with three silver-standard reference descriptions per object. Experiments with Video-3D LLM and 3DRS demonstrate that LensDistill substantially improves fine-grained object captioning while preserving existing grounding and scene-level QA performance. These results establish the feasibility of transferring externally acquired fine-grained knowledge into native 3D LLMs.

We study contextual bandits in which a surrogate is observed after the action but before the learner decides whether to acquire the primary outcome that defines action value and regret. The value of acquiring the primary outcome depends on both decision relevance (how much the current outcome matters for comparing policies) and the residual uncertainty after observing the surrogate. The Audited Surrogate Bandit (ASB) learns a contextual policy while allocating a budget of $B$ primary-outcome acquisitions over $T$ rounds. ASB sets a pre-surrogate acquisition level from current decision relevance and, after observing the surrogate, redistributes that level using an estimate of that residual uncertainty. For a finite class of $N$ policies over $K$ actions, ASB incurs $\widetilde O[\sqrt{KT\log N}\{1+\sqrt{T/B}\}]$ regret relative to the best policy in the class. In a two-action family where the surrogate does not reveal the better action, a learner that observes the surrogate before deciding whether to acquire can achieve bounded regret, whereas any learner that must decide before seeing the surrogate incurs $\Omega(T/B)$ worst-case regret under the same budget. Synthetic experiments show that both acquisition factors matter: ASB has lower regret than variants using only decision relevance or only residual uncertainty. On a KuaiRec benchmark of user-video interactions, the regret gap relative to relevance-only acquisition widens and then narrows as the budget grows.