Science & Mathematics

Video-Language Models (VLMs) excel at visual recognition but struggle with causal reasoning. Current evaluations are flawed: multiple-choice QA allows shortcuts, and open-ended QA lacks robust automatic evaluation. We introduce Causal Chain Generation (CCG), a novel task where models have to generate the sequence of intermediate events connecting a cause and an effect. This structured setup demands genuine reasoning and enables reliable automatic scoring. We accompany this task with two complementary metrics: SODA-Embedding, which measures the semantic matching of a generated chain against a reference, and Causal-QA, which probes the chain's causal coherence by asking whether it contains the specific causal facts needed to reconstruct the narrative. We provide two test sets: SF20K-CC, a manually annotated diagnostic on 20 short films, and CMD-CC, a movie clip benchmark derived from audio descriptions. Our analysis reveals that state-of-the-art VLMs struggle with narrative causality understanding, frequently hallucinating or omitting key events. To address this, we fine-tune VLMs with GRPO using a composite reward that combines a format reward, SODA-Embedding, and Causal-QA accuracy, demonstrating substantial improvements over strong baselines.

Nowadays, it is easy to set up any element of a VR experience so users can interact with it physically, except for interactive VR characters. Can we extend this to animated VR characters? In this work, we studied whether physics-based character animation (PBCA) controllers can produce high-quality movement in VR-compatible environments (Unity3D, WebXR). We evaluated the quality of the results, as well as the losses at each stage of the processing pipeline: data ingest, training and inference. For data ingest, we computed accuracy metrics related to the retargeting method used. For training, we computed accuracy metrics related to different training policies and different training environments. We developed new JAX implementations of state-of-the-art PBCA controllers (DeepMimic, AMP and ADD) which can be trained in MuJoCo-MJX in minutes, 50 times faster than in a traditional CPU-based training environment. The benefit of these particular implementations is that the resulting control policies can be readily transferred to Unity3D and support cross-platform compilation, including Android, something critical for use in VR production. For inference, we tested these controller policies in Unity3D and WebXR. Our results suggest that DeepMimic and ADD trained in MuJoCo-MJX generate controller policies with better animation fidelity than other methods. To simplify the comparison between controller policies, and different reference animations, we implemented an animation fidelity index (afi). Its results are analogous to other more common metrics, but afi is more intuitive to interpret and thus simplifies the comparison of training methods across the successive stages of the processing pipeline.

Compact quantum neural networks have been explored for cloud-cover parameterization, but their predictive value depends on how physical knowledge is incorporated and which classical models provide the reference. We investigate an offline residual formulation that retains a fitted \XR\ cloud-cover estimate and learns a correction to its logit in transformed target space. Using released ICON/DYAMOND-derived data, we compare ten model variants in six- and eight-input settings, with three matched replicates and 10,000 training samples per replicate. All quantum circuits use local Z readout and noiseless exact expectation values. Under a common 30-epoch protocol, residual prediction reduced physical-space test mean squared error relative to direct QNN prediction by 54.3\% and 38.8\% for six-input XYZ and ZZXY circuits, and by 49.2\% and 38.9\% for their eight-input versions. Residual QNNs also outperformed the selected compact residual multilayer perceptron in every paired replicate. A larger, longer-trained Grundner-Q1 adaptation nevertheless achieved lower error: residual Q1 reached mean MSEs of 0.00778 and 0.00791, compared with 0.01124 and 0.01082 for residual XYZ. The results support residual prediction for compact simulated QNNs within the tested budget and establish a clear boundary set by stronger classical references. They do not establish computational quantum advantage or online climate-model skill.

Target propagation (TP) is a backpropagation alternative in which local targets carry the global learning signal. Prior works claim that TP performs Gauss-Newton (GN) optimisation on these targets, but the correspondence has only been established under unrealistic assumptions where hitting them exactly is trivial. Practical variants instead descend on a least-squares problem via a single gradient step. In this paper, we rederive TP from a trust region under an implicit, previously unexamined norm, and introduce CGTP (Conjugate Gradient Target Propagation), which uses conjugate gradients to solve the layer-local subproblems in parallel. A single adaptive parameter jointly controls this trust region and a matching bound on the weight update; KFAC emerges exactly as the special case where the per-sample Jacobians are replaced by their batch mean, giving its factored damping a variational grounding. On MNIST and Fashion-MNIST MLPs, CGTP matches the accuracy of ADAM and KFAC while needing orders of magnitude fewer outer iterations than ADAM (about 10x fewer backward-pass-equivalents overall), while remaining stable across depth.

Second-order cone programs (SOCPs) are a fundamental class of convex optimization problems widely applied across engineering domains. Classical interior-point methods are robust and accurate, but their iterative nature can be costly in real-time or repeated-solve settings. This has motivated learning-based solvers that approximate optimizer solutions with neural networks. Recent work has increasingly studied graph neural networks (GNNs) for linear, quadratic, and second-order cone optimization, leveraging their ability to model sparse relational structure among variables, constraints, and objective functions. Most existing GNN-based solvers compress the optimization process into a single forward pass, predicting the optimizer directly without using the intermediate trajectory produced by classical solvers. We introduce SOCPNet, a recursive GNN framework that uses solver trajectories to learn an iterative SOCP solution procedure. A shared GNN model is applied repeatedly, refining the current iterate at each step and enabling inference for a variable number of iterations. Rather than predicting the solution in a single pass, SOCPNet learns from intermediate iterates to progressively refine its estimates using a trajectory-regularized training procedure. We evaluate SOCPNet on multiple datasets spanning standard SOCP problem classes and show that it consistently reduces the final solution error relative to a SOCP-GNN baseline. Finally, we demonstrate how trained SOCPNet models can be used in a hybrid optimization pipeline by warm-starting a classical interior-point solver, reducing solver iterations without loss of accuracy.

Online robust zero-sum Markov games pose a coupled operator-learning and game-solving problem: from nominal interaction data, the learner must recover a worst-case minimax Bellman operator while maintaining strategic coverage and controlling stage-game approximation error. Existing robust reinforcement-learning theory does not yet provide such an operator-learning perspective beyond specialized multi-agent settings. We develop an optimistic dual robust Bellman framework that combines dual robust Bellman fitting, optimistic minimax planning, backward-validated confidence sets, and exploiter-mix data collection. Our analysis is organized around the robust Bellman residuals induced by the candidate value class and yields a regret reduction that separates exploration complexity, robust operator-estimation error, stage-game solution error, and misspecification. Under explicit interpretable assumptions, we obtain conditional theorem interfaces for general and projected function approximation, together with closed results in several structured regimes, including tabular, epochwise cross-fit linear, self-normalized pure-online linear / finite-rank kernel, and projected RKHS truncation. Small-scale numerical experiments illustrate robustness calibration, strategic coverage, and the return-gap trade-off predicted by the framework.

Modern robotic photography requires understanding photographic intent, planning camera motions, and producing aesthetically pleasing images in complex environments. We present RoboPhoto-VLA, a unified Vision-Language-Action framework that jointly models perception, reasoning, and camera control for autonomous robotic photography. RoboPhoto-VLA incorporates PhotoCoT for structured action planning and Aesthetic Direct Preference Optimization (ADPO) to align camera behaviors with human aesthetic preferences. To facilitate training and evaluation, we construct OpenPhotography, a robotic photography dataset with language-guided demonstrations. Extensive real-world experiments demonstrate that RoboPhoto-VLA produces photographs better aligned with human aesthetic preferences while generalizing effectively to unseen subjects, viewpoints, poses, and complex photographic instructions.

Training-time explanation regularization is often evaluated by whether saliency maps become more concentrated within annotated regions. We argue that this criterion alone is insufficient: transformation consistency, lesion localization, and faithfulness to the classifier are distinct properties and should be evaluated separately, particularly in medical imaging. We examine this distinction through a case study of Contrastive Grad-CAM Consistency (CGC) in three-class brain tumor MRI classification. Using a single-channel ResNet-50 with five-fold patient-grouped cross-validation, we compare a cross-entropy baseline with CGC. CGC changes overall Pointing Game (PG) from $27.24\%$ to $36.02\%$ and Relevance Mass Accuracy (RMA) from $11.17\%$ to $18.01\%$, while classification accuracy changes from $93.63\%$ to $94.87\%$. Perturbation-based faithfulness metrics show inconsistent changes: Insertion Area Under the Curve (IAUC) changes from $0.73$ to $0.78$, whereas Deletion Area Under the Curve (DAUC) favors the baseline ($0.36$ versus $0.41$; lower is better), and overall Iterative Removal Of Features (IROF) is $0.57$ for both models. These results support our position that improved spatial agreement with pathology should not be interpreted as a general improvement in explanation faithfulness. We therefore recommend property-specific evaluation in which the property targeted by the explanation objective, spatial agreement with reference regions, and model reliance are measured and reported separately.

Extensive studies of hidden spin polarization have established a broad class of hidden phenomena in quantum materials. Here, we introduce the hidden piezo-Hall effect, in which strain induces finite anomalous Hall conductivities (AHCs) in individual sectors while the global AHC remains zero. At zero strain, neither the full system nor its sectors exhibit an anomalous Hall effect (AHE). Under uniaxial strain, however, opposite sector AHCs emerge while their sum remains canceled. We establish this effect for a $d$-wave hidden altermagnet using a tight-binding model and first-principles calculations for the experimentally synthesized material $\mathrm{RbV_2Te_2O}$. Rotating the strain direction from the $x$ to the $y$ axis reverses the signs of the sector AHCs without generating a net Hall response. The same symmetry mechanism may also give rise to hidden piezomagneto-optical, piezo-thermal Hall, piezo-Nernst, and piezo-Ettingshausen effects. Our work extends hidden physics to strain-controlled transport and identifies $\mathrm{RbV_2Te_2O}$ as a promising candidate for experimental verification.

Currently many Byzantine fault-tolerant (BFT) protocols use directed acyclic graphs (DAGs) to let replicas propose blocks of requests in parallel for improved throughput. Yet, the existing DAG-based designs mainly consider the atomic-broadcast layer: They typically do not describe an efficient client interaction layer, their garbage collection is either unspecified or relies on trusted external storage, and the lowest-latency designs rely on timing assumptions that may be violated in geo-distributed deployments. In this paper, we propose the Pangolin protocol: It introduces a novel low-latency DAG-based atomic broadcast layer and completes it into an efficient state-machine replication (SMR) framework. To preserve system throughput, clients issue requests to a single replica and verify replies efficiently via compact execution proofs. Furthermore, Pangolin introduces a new garbage-collection scheme for DAGs which bounds memory usage even under full asynchrony. Pangolin leverages trusted execution environments to tolerate f Byzantine faults with only n >= 2f + 1 replicas and accelerates vertex commitments significantly: Broadcasting certified vertices requires only a single communication step, and commit rules are refined to commit both frequently and early without revealing the coin-elected anchor. This way, Pangolin overtakes the fastest partially synchronous DAG protocols while being live under asynchrony: In our evaluation, Pangolin achieves an average 1.27x (up to 1.52x) latency speedup over Mysticeti.

Work on efficient large-language-model inference has concentrated almost entirely on reducing the cost of decoding once a request is complete. The quantity a user experiences, however, is the interval between submitting a request and receiving a usable response, and that interval can be shortened without making decoding itself faster. While a request is being composed it is already partially specified, yet conventional deployments begin no computation until submission. We propose and study streaming generation, a decoding regime in which the model begins producing a response from an incomplete request and revises it as the remaining input arrives, so that only a small residue of the answer remains to be computed once the request is submitted. We evaluate the regime on autoregressive and diffusion language models, characterise the conditions under which it reduces user-visible latency, and analyse the mechanism responsible.

Mobile networks are evolving from human-centered, data-oriented communication infrastructures toward machine- and/or agent-dominant, task-oriented platforms that are to serve the rapidly increasing networked machines and embodied agents. In order to fulfill the newly arose complex task objectives, the framework of future mobile network is to be innovated to reflect and facilitate the intrinsic interplay and integration of the underlying processes of perception, communication, computation, decision, action, and feedback, etc. This survey establishes the mobile embodied agent network (MEAN) as a task-oriented networking paradigm in which networking and distributed intelligence are jointly organized around closed-loop embodied task execution. We first establish a unified MEAN framework that explains how network resources and intelligent functions jointly support embodied task execution, and extend its performance indicators from conventional communication metrics to embodied task outcomes and cross-layer system effects. With this framework, we analyze three complementary enabling mechanisms that form a cognition-interaction-adaption loop. Large artificial intelligence (AI) models provide high-level cognition, token communication connects this cognition with distributed agents through shared semantic representations, and reinforcement learning closes the loop by adapting system operation to network, task, and physical feedback. We further investigate the bidirectional interplay between MEAN and emerging technologies, showing how they mutually extend system capabilities and reshape task-oriented network design. Finally, we discuss open challenges and future research directions toward the practical realization of MEAN over real joint communication and computation platform.

The random $k$-XORSAT matrix is an $m\times n$ random matrix $\mathbf A$ over $\mathbb F_2$ that represents a random instance of the $k$-XORSAT problem. Each row of $\mathbf A$ contains precisely $k\geq3$ non-zero entries in random positions. The rows are drawn independently. The matrix $\mathbf A$ is known to have full row rank for $m/n$ below the random $k$-XORSAT satisfiability threshold, which is known explicitly [Dubois, Mandler~2002; Pittel, Sorkin~2016]. Motivated by a problem from quantum game theory and refining a result of Cooper, Frieze and Pegden (2019), we determine the limiting distribution of the rank of $\mathbf A$ beyond the satisfiability threshold.

Diamond is an important host material for functional defects such as solid-state spin qubits. During the defect creation by irradiation and subsequent thermal annealing, carbon-related defects, including vacancies and interstitials, are generated and can migrate to form new defect complexes. To investigate this kinetically driven process and reveal the formation mechanisms of the di-interstitials and di-vacancies in the neutral charge state, we fine-tuned a universal machine learning interatomic potential using a small dataset generated from density functional theory calculations. We obtained a barrier of 1.84 eV for mono-interstitial migration and found that two initially separated mono-interstitials only need to overcome a lower barrier to form di-interstitial configurations. Beyond the training data, the MLIP also identified the formation pathway for an additional di-interstitial configuration and a stepwise divacancy dissociation mechanism. These results reveal migration-assisted defect complex formation mechanisms and demonstrate how targeted MLIP training can identify reaction pathways.

Metal nitridocarbonates containing the (NCN)2- ion are established for a multitude of metal atoms throughout the periodic table. The preparation of nitridocarbonate compounds containing a (CN3)5-, (C2N4)6-, or (C3N6)6- ion remains challenging, although corresponding species have been established. The new compound La14.33(CN3)8N was synthesized in a conventional solid-state metathesis reaction from a mixture of LaF3, Na2(CN2) and Li3N in an open system at 550 °C. The reaction conditions, including the formation and de-composition of La14.33(CN3)8N are analyzed on basis of thermoanalytic studies. The crystal structure of La14.33(CN3)8N was refined by single-crystal X-ray diffraction, and the mid-IR frequencies resulting from two crystallographically distinct (CN3)5- ions in the structure are reported. 23Na, 7Li and 13C NMR studies provide additional support for the identity of the compound. The crystal structure of La14.33(CN3)8N is isotypic to that of Ce15(BN3)8N.