Papers

Artificial intelligence offers an unprecedented opportunity to augment human capabilities, yet progress at the frontier has focused primarily on advancing model capabilities. We introduce StudentBench, a suite of AI teaching evaluations and a public platform that enables large-scale data collection with over 175,000 student-AI messages to study whether large language models (LLMs) produce learning gains equivalent to human tutoring. Using StudentBench, we measured learning gains on Quantitative and Verbal GRE questions across 2,383 human participants receiving AI tutoring, human tutoring, or no tutoring. We establish that AI tutoring is statistically equivalent to expert human tutoring for GRE learning gains (p = .015), and in five of the seven GRE domains, the best performing AI tutor surpassed the human tutor, on average. In a second study, expert human tutors compared LLM-generated lesson plans and practice problems through 2,028 pairwise rubric evaluations. Together, the two studies clearly separate AI tutors across: (1) lesson planning, (2) practice-problem creation, (3) conversational pedagogy, (4) cost, and (5) engagement. Surprisingly, one AI tutor achieved learning gains equivalent to human tutoring (p = .044) at 918 times lower cost (USD 0.0052 for AI versus USD 4.81 for human, per percentage point gained). For Quantitative GRE sessions, faster AI replies correlated with more student messages, more messages with more correct practice, and more correct practice with larger learning gains (all p < .002). To support future research, we open-source the de-identified data collected in our studies.

Social navigation typically assumes a specified goal and focuses on reaching it while respecting social conventions, whereas robot group joining requires predicting where to join based on the group's real-time activity and formation. This is a highly semantic task, yet an important capability for applications such as robotic guide dogs and autonomous mobility scooters. We formulate language-grounded robot group joining: given an observation and a natural-language description of a target group, the robot identifies the relevant group members and predicts socially compliant joining poses. For grounding, we generate structured candidate subsets through recursive spectral partitioning and rank them with a language-conditioned image–geometry model. Given the grounded group, a goal predictor leverages human-formation priors to produce a multimodal energy–orientation map over feasible robot poses. Experiments on conversations, queues, and audiences across varying group sizes, crowd densities, and visual ambiguities show that our method achieves competitive grounding accuracy with sub-second inference and outperforms all baselines in joining-pose prediction. Real-robot experiments further demonstrate group joining in both static and dynamically changing interactions.

Advances in generative models have improved video fidelity, enabling long-horizon generation, interactive world modeling, and evolving visual environments. Autoregressive (AR) video generation extends visual sequences through causal rollouts. However, a fundamental bottleneck emerges: as the generated sequence expands, practical models must operate under strictly bounded context windows, storage, and computational limits. Consequently, critical historical information, e.g., entity identities, dynamic states, and intervention-induced causal changes, often leaves the active context long before its relevance diminishes. Overcoming this limitation and maintaining temporal persistence constitutes a fundamental memory problem. We present a systematic and comprehensive review of memory mechanisms in AR video generation. We formulate memory operationally as persistent historical information maintained across outer AR steps, capable of influencing future generation even after the originating evidence is no longer locally accessible. Building upon this unified framework, we organize the literature through five complementary perspectives: (I) Forms, the representational carriers of history; (II) Functions, the specific semantic and physical information requiring preservation; (III) Operations, the lifecycle of writing, reading, updating, managing, and integrating memory; (IV) Learning, the optimization of memory behaviors under closed-loop rollouts; and (V) Evaluation, the paradigms for diagnosing genuine memory capabilities. We conclude by synthesizing open challenges, including composable and resource-aware memory architectures, trustworthy state updating, self-rollout learning, and standardized evaluation. By bridging representations, mechanisms, and learning paradigms, this paper establishes a structured foundation for developing reliable, memory-conditioned video generation systems.

We introduce Sharper Transductive Local Complexity (STLC), a localized complexity method for transductive learning under uniform sampling without replacement. The construction starts from a Bernstein-type concentration inequality for the supremum of the test–train empirical process. Its proof uses the modified log-Sobolev inequality for the swap walk and a two-parameter entropy closure. A peeling argument with a surrogate localization functional then gives excess-risk bounds with the same fixed-point and confidence terms as the classical inductive local Rademacher-complexity bounds, without the additional logarithmic confidence factor in earlier transductive results. For realizable learning over a binary class of VC dimension $\dVC$, with training size $m$, test size $u$, and $u\ge m\ge\dVC$, STLC yields $\cO\{\dVC\log(me/\dVC)/m\}$. This matches the standard inductive rate and, when $m\ge9$, is within a logarithmic factor of the transductive minimax lower bound of order $\dVC/m$. For transductive kernel learning, STLC gives a spectrum-adaptive excess-risk bound without the multiplicative imbalance factors appearing in the earlier local-complexity bound.

Let $n\ge2$, and let $X=(X_1,\ldots,X_n)$ be a centered Gaussian vector with $\mathrm{Var}(X_i)=1$ for every $i$. Let $Z_1,\ldots,Z_n$ be independent standard Gaussians, and put $\overline{Z}=(Z_1+\cdots+Z_n)/n$. We prove $\mathbb{P}\{\max_i X_i\le t\}\ge\mathbb{P}\{\sqrt{n/(n-1)}\,\max_i(Z_i-\overline{Z})\le t\}$ for every $t\in\mathbb{R}$, and for each fixed $t>0$ equality holds only when $\mathrm{Cov}(X_i,X_j)=-1/(n-1)$ for all $i\ne j$. The right side is the distribution function of the maximum of the regular simplex vector. Equivalently, among all simplices containing a given centered ball, the regular simplex circumscribed about the ball has the least standard Gaussian measure, as conjectured by Balitskiy, Karasev, and Tsigler. In our preceding paper we proved this comparison after both maxima are smoothed by independent Gaussian noise of variance $1/(n-1)$, which suffices for the Weak Simplex Conjecture; here we remove the smoothing, which is what probabilities at a single threshold require. As an application we consider $n$ equally likely signals of equal energy in Gaussian noise, where the transmitter may also send nothing. At every positive false-alarm level, and for every law of a common nonnegative random amplitude not concentrated at zero, the regular simplex uniquely maximizes the average probability of correct identification whenever the signal dimension is at least $n-1$. A Lean formalization is available at https://github.com/abhmul/full-simplex-conjecture-lean.

We study dynamic pricing problems under inter-temporal price constraints. We have resources with limited capacities. At each time period, we decide which products to make available and what prices to charge for the available products. The sale probability for a product depends on its price. If we make a sale for a product, then we collect a revenue reflecting the price and consume the capacities of a combination of resources. We work with two types of inter-temporal constraints. In price monotonicity, the prices charged for a product at different time periods have to be monotone. In promotion fatigue, we can discount a product at most once over each time interval of a fixed length. Computing the optimal policy is intractable. We use fluid approximations to construct policies. Traditionally, policies from fluid approximations make randomized decisions at each time period by following an optimal solution to the fluid approximation, but such randomized decisions easily violate price monotonicity or promotion fatigue constraints. We develop policies that sample price paths according to an optimal solution to the fluid approximation, while satisfying the inter-temporal constraints. Letting $c_{\min}$ be the smallest initial capacity of a resource and $L$ be the maximum number of resources used by a product, our policies have a performance guarantee of $\max\Big\{ \frac{1}{8L}, \, \frac{1}{2} - \sqrt{\frac{\log c_{\min}}{2 \,c_{\min}}} - \frac{L}{c_{\min}}\Big\}$. Thus, under large resource capacities, our policies are guaranteed to obtain at least half of the optimal total expected revenue. The latter performance guarantee is tight in the sense that no policy can, in general, obtain more than half of the optimal objective value of the fluid approximation even under large resource capacities. We unify our approach to open the path for extensions to other inter-temporal price constraints.

Large language models (LLMs) are increasingly used in coding tasks, but their ability to reason about code execution remains unclear. Existing repository-level QA benchmarks mainly evaluate static code understanding and often rely on LLM-based evaluation, while execution-reasoning benchmarks are mostly limited to snippets or functions. We introduce SWE-Flux, a repository-level benchmark for dynamic execution reasoning containing 480 execution-grounded instances across 12 real Python repositories, with gold answers automatically harvested from instrumented test executions rather than written manually or judged by LLMs. The benchmark covers singletest and multi-test questions over control flow, loops, program state, dataflow, exceptions, and program invariants. Evaluating five LLMs shows that this task remains challenging. The best model achieves only 37% accuracy. Models perform better on localized behavior such as invariants, intra-procedural control flow, exceptions, and simple loops, but struggle with dataflow, inter-procedural execution, precise state reasoning, and suite-level aggregation. Finally, we show that the oracle-harvesting pipeline can generate fresh benchmark variants using input perturbation. It successfully harvests valid variants for almost 90% of the selected instances, and the resulting variants are substantially more challenging for the evaluated models.

We study the nonequilibrium dynamics of a minimal recurrent transformer with $N$ normalized tokens, $Q=K=I$, and a negative value map $V=-I$. Similarity-based attention selects nearby representations, while the negative value map drives tokens away from the selected field. This feedback can continually reorganize both the representation geometry and the attention network. For $d=2$, the tokens lie on a circle, where the regular polygon is an exact fixed point. As the attention feedback strength $\gamma$ is increased, the polygon loses stability through a flip bifurcation, giving rise to period-two motion, chaos, and cluster-exchange or cluster-flip states. Despite this temporal complexity, attention remains diffuse as $N\to\infty$ at finite fixed softmax sharpness $\beta$. Attention condensation instead emerges in the scaling regime $\beta\sim N^2$. In the hard-routing limit, repulsive updates amplify local perturbations and routing-partner switches transmit them ballistically, producing an emergent butterfly cone in representation space. High-dimensional geometry provides a distinct route to localization. For $d=N\to\infty$, simulations from Gaussian initial conditions provide evidence for a condensation transition at $\beta=O(1)$, driven by dynamically generated finite overlap gaps. Depending on $\gamma$, the resulting phases include diffuse simplex-like states, consensus flips, condensed active routing with signatures of chaos, and fragmented cluster flips. These results establish temporal activity, attention condensation, and geometric clustering as distinct collective phenomena, and show that sparse attention can sustain persistent dynamics rather than freeze it.

We propose and analyze a fully discrete finite element scheme for the Beris-Edwards system of nematic liquid crystal dynamics, in which the incompressible Navier-Stokes equations are coupled to a gradient flow for the Landau-de Gennes Q-tensor. The scheme combines a linearly implicit formally second-order accurate backward differentiation formula in time with an incremental Chorin projection step for the incompressibility constraint, conforming finite elements in space, and the invariant energy quadratization approach with mass lumping for the nonlinear bulk potential. Each time step requires the solution of one linear system and one Poisson problem. We show that the scheme is uniquely solvable, that it preserves the symmetry and trace-free structure of the discrete Q-tensor and molecular field, and that it satisfies a discrete energy law without any restriction on the time step. Our main result is that, as the mesh size $h$ and the time step $\Delta t$ tend to zero subject to $h^{2} = o(\Delta t)$, the approximations converge along a subsequence to a weak solution of the Beris-Edwards system. The convergence proof addresses two difficulties: the projection method produces two velocity approximations, only one of which is uniformly bounded in $L^2(0,T;H^1_0(\Omega))$, and the coupling term $\mathcal{H}\nabla Q$ requires strong convergence of $\nabla Q$, which we obtain from the structure of the equation for $\mathcal{H}$ rather than from any discrete $H^2$-bound. Numerical experiments in two dimensions exhibit approximately second-order convergence in space and time, and reproduce the splitting of a $+1$ point defect into two $+1/2$ defects and the transport and deformation of a skyrmion induced by a constant pressure gradient.

Reordering a set of mathematical rules without changing its meaning should preserve the correct answer, but must a model's internal representations stay invariant too? We investigate this question using synthetic multi-step function-composition problems, each presented under multiple rule orderings with the same correct answer. We measure accuracy and permutation signal-to-noise ratio (SNR), which quantifies how distinctly ordering patterns are represented relative to variation across problem instances. Across 16 language models ranging from 1B to 8B parameters, we find a pattern: models that solve reordered problems more accurately represent different rule orderings more distinctly. Layer-averaged permutation SNR is positively rank-correlated with accuracy in every synthetic setting we evaluate, with Spearman correlations reaching 0.86. These findings highlight a distinction between answer invariance and representation invariance: successful mathematical rule composition can accompany distinct internal representations between equivalent rule orderings. This motivates distinguishing answer invariance from representation invariance, and offers a representational perspective on mathematical reasoning beyond answer accuracy alone.

The adoption of CMOS sensors in mobile photography is frequently compromised by the rolling shutter (RS) effect, which introduces geometric distortions and motion artifacts. Particularly, recent rolling shutter with global reset (RSGR) mode, while mitigating some RS issues, also incurs major limitations, including reduced capture speed and compressed dynamic range. To address these problems, we propose a novel dual reversed scanning setup utilizing both RSGR and inverted RSGR views. This solution not only handles the inherent flaws of RSGR by synchronizing complementary exposures to balance the dynamic range across the frames but also introduces an effective method for HDR photosequencing under highly dynamic scenes. Our proposed network first accommodates row-wise complementarity and manages visual shifts by row-adaptive feature alignment. Subsequently, the hallucination module, built upon a correlation-guided mixattention block, integrates the mutually reinforced features to recover missing details. In addition, we construct a coaxial imaging system to collect a real-world dataset, enabling robust training and evaluation beyond numerical simulation. Experimental results demonstrate the twofold benefits of our solution in mitigating RSGR limitations and advancing HDR reconstruction techniques.

We study minimal-norm interpolation and $\ell_2$-regularized logistic-loss minimization for binary classification by univariate two-layer ReLU networks. We give complete geometric characterizations of the optimal classifiers in function space, resolving how the solutions depend on whether hidden-layer biases are included in the parameter norm. When biases are unpenalized, the minimal-norm interpolators are exactly the continuous piecewise-affine functions that hug every label switch and have kinks of the appropriate convexity. When biases are penalized, the minimizer is unique in function space, has exactly one kink in each intermediate same-label segment, and is therefore a sparsest positive-margin classifier. We further show that adding a free affine skip connection leaves these function-space solutions unchanged but fundamentally improves the parameter-space landscape: every KKT point of the constrained problem becomes globally optimal, whereas suboptimal KKT points can occur without the skip connection. We establish analogous global-optimality and geometric results for sufficiently weak $\ell_2$-regularization of the logistic loss. In the unpenalized-bias case, we identify an additional sparsity-like restriction, implying that most minimal-norm interpolators cannot arise as small-regularization limits of margin-normalized logistic-loss minimizers. Numerical experiments across varying dataset complexity and network width support the predicted landscape and sparsity phenomena.

Online information is increasingly consumed in video format. Much of this comes in the form of *raw video*: continuous footage taken on a cell phone, with a hand-held camera, or via CCTV, which is then directly uploaded to social media platforms and content sharing services. Whereas professional or even amateur-edited footage tends to feature scripted speech, chyrons, graphics, and metadata that help contextualize its subject matter, raw video typically contains none of these things, making it a much more challenging medium for information retrieval and machine understanding. To facilitate progress in this domain, we release MultiVENT-Raw, a multilingual collection of nearly 120,000 primarily raw videos (over 5,300 total hours), paired with 130 events and 222 event-centric queries, along with human-annotated video relevance judgments and human-extracted key facts for relevant videos. MultiVENT-Raw supports both a retrieval task—to identify videos in the collection relevant to a query event—and a generation task—to summarize event-related videos into a coherent report for a target user. We benchmark strong baselines on MultiVENT-Raw, showing both tasks to be challenging even for some of the latest multimodal models.

Adolescent Idiopathic Scoliosis is defined as a lateral curvature of the spine that develops during adolescence, without known cause. The condition can result in significant pain and disability, and often progresses rapidly during adolescence. The objective of this paper is to predict the progression of the condition in a temporal sequence from ages 9 to 24, as measured from a sequence of Dual X-ray Absorptiometry (DXA) scans. To this end, we train a transformer model that takes in the curve of the spine to predict curve progression. The model is trained using a large-scale synthetic dataset of spine curves and their time series, covering different curve types and different progression patterns. We show that the model is able to generalise from synthetic to real data by evaluating it on a dataset of real DXA scans covering multiple time points. We find that fine-tuning the model on real data gives a significant boost to performance. The model is able to accurately predict spine curve progression in both scoliosis and normal cases.

Dexterous manipulation demands long-term foresight and rapid reactive control. Vision-Language-Action (VLA) models, while proficient in high-level reasoning, often lack a fine-grained understanding of physical dynamics and spatial perception. Conversely, World-Action Models (WAMs) typically suffer from high inference latency due to iterative generation. These deficiencies result in a critical temporal misalignment where the model's intent fails to adapt to rapid physical contact changes. To overcome this fundamental bottleneck, we propose LiMA, an asynchronous dual-system generative framework that systematically decouples intent planning from reactive execution. LiMA organizes computation into a multi-scale hierarchy: a slow system handles sparse long-horizon spatiotemporal intent generation, while a fast system focuses on dense high-frequency motion refinement. To align sparse intent predictions with dense action trajectories, we introduce a Latent Schrödinger Bridge Coupling mechanism that formulates refinement as an entropy-regularized probabilistic transport process. LiMA reduces inference latency by 45.8% compared with Cosmos-Policy via asynchronous decoupling. Evaluated across six bimanual dexterous manipulation tasks spanning multiple horizons, LiMA achieves an overall success rate of 70.8% and an average subtask success rate of 78.9%, while maintaining performance in unseen scenarios. The project website is available at https://ccdcs.github.io/LiMA_repo/

This work addresses continuous depression-severity score prediction from clinical interview transcripts under data scarcity. We propose a sequential low-rank adaptation (LoRA) protocol for cross-scale transfer: a Qwen3 backbone with a bounded regression head is first fine-tuned on the English DAIC-WOZ dataset (189 avatar-mediated sessions, PHQ-8), and the adapter then initializes fine-tuning on the Chinese PDCH dataset (100 real clinical consultations, HAMD-17), where a reinitialised, scale-specific head predicts the clinician-assigned score. All configurations use patient-level stratified 5-fold, 2-repeat cross-validation. On the data-scarce HAMD-17 target, the sequential protocol attains the best point-estimate MAE , RMSE, and macro-$F_1$ on both 0.6B and 1.7B backbones, outperforming target-only training and non-LLM baselines—4.96/6.59/0.36 with Qwen3-0.6B and 4.38/5.62/0.46 with Qwen3-1.7B. Ablations suggest that correctly aligned source supervision gives the best point estimates (unsupervised exposure and shuffled-label controls also show partial gains), that native-Chinese target input outperforms machine-translated English input, and that the reversed order yields no clear gain within run-to-run variance. The study is an exploratory, single-site internal evaluation: it does not establish screening or diagnostic utility, nor separately identify the contribution of the scale, language, or paradigm shifts. To our knowledge, no prior study evaluates this specific DAIC-WOZ-to-PDCH sequential transfer setting.

An always-on robot faces an endless stream that never resets: instructions arrive and lapse, the scene changes, and its own past actions reshape what it must reason about. Today's action models are built for the opposite: a fixed instruction, no mid-task intervention, single-step reasoning. In an open-ended world a robot must watch a live stream for far-future cues, recall its own far-past actions, and act on them under dual-arm concurrency. We present ARMS (Always-on Robot in Multi-modal Streams), a deliberately simple streaming policy: a single pretrained $\pi$0.5 backbone augmented by three lightweight modules that turn live perception, embodied states, and the robot's own past actions into context the backbone reads before it acts. The modules update this context asynchronously, so watching and recalling never block acting and the two arms act at once. Rather than inventing new mechanisms, ARMS integrates these learned context providers with an agent-causal self-history that logs which arm did what, and when. To supervise them without extra annotation, we build ARMS Dataset, whose staged construction script itself labels every module from real dual-arm teleoperation. Trained on it, ARMS reaches 45% on the combined task against 28% for the strongest of our four main baselines, and ablations confirm the memory module, the embodied-state head, and asynchronous concurrency are each necessary.

Personalizing exoskeleton assistance across operating conditions is constrained by the time and physical effort required to collect user feedback. We examined whether a user's preference landscape varies smoothly across operating conditions and when this continuity supports learning from limited feedback. We propose Context-Continuous Preference Learning (CCPL), a Gaussian-process preference model that shares observations across nearby contexts while retaining context-specific utility estimates. We evaluated CCPL through simulations and retrospective analyses of ankle and elbow exoskeleton preference data from nine healthy adults. In simulations, CCPL improved reconstruction and preference-based Bayesian optimization relative to independent learning when preferences varied smoothly, but showed negative transfer when continuity was weak. In both human studies, full-data reference landscapes estimated separately for each participant and context tended to be more similar between nearby operating conditions. With five exposures per context, CCPL increased mean reconstruction correlation with these references from 0.644 to 0.720 for ankle assistance and from 0.476 to 0.526 for elbow assistance relative to independent learning. The five-exposure budget was approximately 37% lower for ankle and 17% lower for elbow than the estimated independent-learning budgets needed to match these correlations. CCPL also improved held-out response prediction relative to independent learning, while benefits over pooled learning varied. These findings support context continuity as a basis for sharing preference observations under limited feedback, although benefits for online personalization in humans remain to be established.

Recursive state estimation often executes approximate numerical solutions inside a feedback loop, where highly accurate local steps do not guarantee better overall results. For a fixed linear Kalman model, we characterize when a correction within a prescribed subspace and norm budget can meet a local admissibility tolerance, and how the defects actually executed affect the finite-horizon covariance response. Centering each defect on the exact gain for the implemented covariance separates current solve error from inherited gain drift. Expanding the exact residual-drift identity reveals opposing quartic contributions beyond the quadratic response: innovation-covariance inflation enters positively, while local-gain reoptimization enters subtractively. Under matched initialization, an absolute sixth-order remainder bound, uniform over bounded defect sequences at fixed horizon, gives sufficient conditions for quadratic under- or overprediction. Machine learning proposes bounded corrections, while a learner-independent residual certificate and verified fallback govern execution of classical and quantum candidates without changing the reference estimator. In a power-grid tolerance study, learned correction lowers the minimum conjugate-gradient iteration count for deployment without fallback relative to uncorrected solves under the same residual certificate. Gains reconstructed from a variational quantum linear solver and from an annealing-based binary encoding, with small-scale terminal measurements on superconducting hardware and sampling on a quantum annealer, are executed through the same interface. By linking local repairability to nonlinear error propagation, the framework evaluates approximate solvers and learned corrections through independent certification and finite-horizon response, providing a practical basis for studying hybrid quantum–classical computation.

Catalog skin recolouring has to change pigment and leave shading alone. The classical Reinhard map does not make that split: it rescales lightness by the ratio of standard deviations, and a flat reference swatch therefore flattens the limb. This paper formalises the correction used in our pipeline, the skin-restricted Reinhard transform. It is the diagonal affine map in CIE Lab that translates lightness, matches the chromatic mean, and clamps the chromatic gain to [0.72, 1.18], with moments taken on the central 84% of each channel. A diagonal affine map has six real parameters. The shading constraint forces the lightness gain to +1 and the lightness shift to the difference of means; one-dimensional quadratic optimal transport on each chromatic axis, followed by Euclidean projection onto the gain interval, fixes the other four. Inside that family the four conditions determine every parameter. The content of the result is the forced lightness gain; it is not a uniqueness claim outside the diagonal affine class. For Gaussian marginals the chromatic step is not merely the best affine map: it is the unrestricted Wasserstein-2 map. The same formulae with trimmed moments remain optimal because a positive affine image commutes with quantile trimming. On hands, arms, legs, and feet of nine photographs and three reference tones, the map keeps the lightness contrast ratio at 0.974 +/- 0.029 with chromatic error 0.77 CIE Lab units. Reinhard matching, the linear Monge map, and histogram matching reach a smaller chromatic error only by cutting lightness contrast to about half.

Recent advances in large language models (LLMs) have enabled agents to tackle long-horizon tasks across diverse environments. To further improve agent performance, existing language world models typically predict environment observations, yet reconstructing high-entropy, execution-dependent tool responses offers limited value when real feedback is available. Meanwhile, agents suffer from task-state contamination, where unsupported assumptions and outdated plans persist in history and distort subsequent decisions. We propose the Agent-Editing World Model (AEWM), which models how reasoning and actions shape future task progress rather than simulating tool responses. AEWM combines Action Judge to distinguish Critical, Exploratory, and Noisy decisions with State Revision to edit noisy reasoning–action continuations from the same observed history. EditAct integrates these capabilities with real execution, directly changing the state underlying subsequent decisions rather than merely providing critiques. We train AEWM across Search, Terminal, and Software Engineering through mid-training and supervised fine-tuning. AEWM achieves 70.5% macro-F1 on our Action Judge benchmark, exceeding the strongest frontier baseline by 10.6 points. Across six benchmarks and three agent backbones, EditAct improves average scores by 3.2–6.7 points over the strongest baseline. Furthermore, rejection sampling fine-tuning on verified EditAct trajectories, termed AEWM-RFT, improves over Self-RFT by 2.2–2.6 points across three domains without online AEWM guidance.

Latent world models that integrate a flow in a frozen self supervised latent space train stably and cheaply, yet silently lose the property manipulation depends on most: motion. The pretrained flow never moves the manipulated object; retraining it with latent-only losses only trades stillness for teleport-like motion. We trace the failure to the training signal, not the representation: anchor-sparse, latent-only supervision never says where along the horizon change belongs. Decode-augmented rollout training (DART) repairs this while keeping the representation frozen, retraining only the flow with decode-path supervision. DART outperforms its latent only parent on the full protocol, restores the temporal structure of motion, and re-couples predicted motion to the scene; at larger scale it further improves prediction quality, closing nearly half the remaining gap to an oracle-informed interpolation reference. Finally, we report an unexpected finding about evaluation: pixel error alone rewards frozen predictions.

The classical Hill–Clohessy–Wiltshire equations assume an unforced Keplerian reference trajectory, an assumption that is violated by missions requiring continuous thrust. We formulate spacecraft relative motion on the $\mathrm{SE}_2(3)$ Lie group, which encodes position, velocity, and attitude in a single geometric state, linearize the gravity and body-frame control mismatches between the two vehicles, and derive analytic bounds on every neglected term. Under circular coasting Keplerian assumptions the framework recovers the Hill–Clohessy–Wiltshire equations exactly. For thrusting references the state transition matrix acquires thrust-induced attitude–translation coupling blocks that, when the two vehicles share body-frame inputs, are exact in log coordinates and independent of the attitude error, and a mismatch between the vehicles' inputs re-centers the linearization at their mean. A two-impulse rendezvous planner follows directly. A scalar coupling ratio, computable before planning, separates two regimes. Where it exceeds unity, the planner reduces the terminal error of coupling-blind planners by more than an order of magnitude, and somewhat above unity it is also more accurate than a Cartesian planner of the same computational structure; well below unity, simpler planners are adequate or better. Simulations confirm the recovery to machine precision and validate every bound along the transfer.

In this paper, we introduce a novel second-order, positivity-preserving finite element method for the Vlasov equations using a convex limiting algorithm. The method employs strong-stability-preserving (SSP) Runge-Kutta time integration and a tensor-product construction of the phase-space mesh for efficient high-dimensional implementations. The convex limiting algorithm combines the robust first-order positivity-preserving graph viscosity approach with high-order residual-based viscosity stabilization to obtain a high-order positivity-preserving scheme. Both novel first-order and high-order methods applicable to high-dimensional problems such as the Vlasov system are presented. In addition, we propose a divergence-cleaning technique for Maxwell's equations to ensure that the divergence constraints of the electromagnetic fields are satisfied. Numerical experiments are provided to demonstrate the accuracy and robustness of the proposed methods.

Vector Symbolic Algebras project data structures into a hyperdimensional vector space through the application of their vector algebras to randomly generated atomic vector symbols and fractional power encodings of real-valued data. Embedding unstructured data remains an open question. We present Clifford-VAE, a variational autoencoder that learns to project data onto a Clifford torus in arbitrary dimensions. Experiments using the MNIST, FashionMNIST, and CIFAR-10 datasets demonstrate that Clifford-VAE produces representations that are competitive with those produced by Gaussian and Hyperspherical VAEs for semi-supervised classification tasks while outperforming Gaussian and Hyperspherical counterparts in the VSA benchmark tests of self-binding and unbinding, role-filler recovery, and bundle capacity. Clifford-VAE provides a principled technique for grounding perceptual data into a symbolic reasoning framework, providing a new approach to a long-standing problem in the VSA literature.

Collective responses depend on individual differences, contact opportunities, and accumulated experience. Learning their dynamics from aggregate counts requires connecting a population's response distribution to both current observations and future behavior. We introduce Differentiable Gaussian Dynamics (DGD), which learns this connection through three components: a Gaussian mixture representing heterogeneous response propensities, differentiable aggregation of contact intensity and behavioral probabilities, and feedback recurrence that updates subsequent responses. Reparameterized integration and temporal recurrence let aggregate prediction errors jointly train the distribution, observation functions, and feedback parameters. On four windows from KuaiRand-Pure and Online Retail II, DGD achieves lower joint behavioral negative log-likelihood than a DeepAR adaptation with a joint-behavior head. In Retail 2010, its one-day behavioral-count MAE is 4.71 versus 6.88 for this adaptation. Learning the distribution reduces behavioral negative log-likelihood by 10.82% relative to a fixed Gaussian in KuaiRand's standard-recommendation window; removing feedback dynamics raises joint KL from 0.0340 to 0.2577 in a controlled experiment. These results establish the value of learning population representations and their feedback process from aggregate observations. Code is available at https://github.com/OranAi-Ltd/oransim.

Language models typically construct attention values from contextual hidden states, even when some of their content may be reusable across contexts. We investigate whether token-indexed memory can replace the dedicated value projection when complemented by contextual information. We propose Memory Attention (MA), which forms values by combining layer-specific token memory with contextual keys. The memory supplies token-specific representations, while the keys preserve context dependence. At inference, normalization can be folded into the memory tables, reducing value construction to lookup and addition. Token-indexed retrieval also enables CPU offloading with prefetching, reducing GPU parameter storage. Under matched training token budgets and with additional memory parameters, experiments across attention configurations show improved language modeling and average downstream performance.

In the classical list update problem, a set of items must be stored in a list-type structure, where accessing the $i$-th element costs $i$. Items will be queried in an IID manner according to some probability distribution $p$ on the items. We want to minimize the expected cost of each query. The optimal order is to place the items in decreasing order of probability $p_1 \geq p_2 \geq \cdots$ with expected cost $\mathsf{OPT} = \sum_j j p_j$, but the probability vector $p$ is generally unknown. Thus we use a self-organizing list following the transposition rule: an item is transposed 1 position forward whenever it is queried. Coester (2026) proved that, at stationarity measure for the transposition rule, the expected cost of a query is at most $\mathsf{OPT} + 1$. However, this Markov chain may have arbitrarily slow mixing time. We prove that, for arbitrary $p$ and arbitrary initial orderings $\sigma$, after polynomially many queries in the number of items, the expected cost of a query is at most $\mathsf{OPT} + O(1)$.

This paper addresses the control problem for robotic systems, including non-holonomic and underactuated platforms operating under unknown dynamics and strict actuator limits to satisfy complex high-level specifications. We denote these high-level specifications using Signal Temporal Logic (STL) and propose a novel time-aware Reinforcement Learning (RL) framework that leverages the geometric properties of Spatiotemporal Tubes (STTs). While traditional analytical STT controllers often struggle to enforce input constraints, and existing RL approaches rely on memory-intensive state history, our method natively overcomes both limitations. By mapping the logical and temporal complexities of the full class of STL into time-varying geometric boundaries, we directly constrain the multidimensional system state without relying on scalar robustness metrics. Augmenting the state space with time, we train a time-aware Soft Actor-Critic (SAC) agent using a continuous, geometry-aware reward function that eliminates the need to explicitly evaluate complex logical semantics during execution. The proposed framework offers a history-free, computationally efficient approach to learn continuous control policies that ensure robust satisfaction of specifications while strictly adhering to system input constraints.

Fine-tuning large language models on parallel data improves translation quality but can cause catastrophic forgetting. Mitigation methods are generally evaluated by retention on general benchmarks. We ask whether these findings transfer to machine translation (MT) fine-tuning and to MT-specific instruction following (MT-IF): instructions that modify a translation, such as formality, grammatical gender, and length control. We compare methods anchored to auxiliary data, to model outputs, and to the base model parameters, first in a screening study with Llama 3.2 1B Instruct, then on Llama 3.1 8B Instruct fine-tuned on bidirectional Arabic-English or Spanish-English data. Elastic Weight Consolidation preserves general capabilities best in both stages; on the 8B Spanish model the average score on general benchmarks drops 1.7 points versus 11.0 for standard fine-tuning, yet its scores for formality and grammatical gender control remain close to standard fine-tuning. Only data mixing with control-task examples preserves these controls, but its gains do not transfer to unseen prompts for the same task.