arXiv:2605.17899v1 Announce Type: new Abstract: AlphaFold3 introduces a diffusion-based architecture that elevates protein structure prediction to all-atom resolution with improved accuracy. This state-of-the-art performance has established AlphaFold3 as a foundation model for diverse generation and design tasks. However, its iterative design substantially increases inference time, limiting practical deployment in downstream settings such as virtual screening and protein design. We propose DCFold, a single-step generative model that attains AlphaFold3-level accuracy. Our Dual Consistency training framework, which incorporates a novel Temporal Geodesic Matching (TGM) scheduler, enables DCFold to achieve a 15x acceleration in inference while maintaining predictive fidelity. We validate its effectiveness across both structure prediction and binder design benchmarks.
Science Journals
arXiv:2510.27446v2 Announce Type: replace Abstract: We develop a wave packet molecular dynamics framework for modeling the structural properties of partially-ionized dense plasmas, based on a chemical model that explicitly includes bound state wavefunctions. Using hydrogen as a representative system, we compute self-consistent charge state distributions through free energy minimization, following the approach of Plummer et al. [Phys. Rev. E 111, 015204 (2025)]. This enables a direct comparison of static equilibrium properties with path integral Monte Carlo data, facilitating an evaluation of the model's underlying approximations and its ability to capture the complex interplay between ionization and structure in dense plasma environments.
arXiv:2505.17152v2 Announce Type: replace Abstract: Vector search underpins modern AI applications by supporting approximate nearest neighbor (ANN) queries over high-dimensional embeddings in tasks like retrieval-augmented generation (RAG), recommendation systems, and multimodal search. Traditional ANN search indices (e.g., HNSW) are limited by memory constraints at large data scale. Disk-based indices such as DiskANN reduce memory overhead but rely on offline graph construction, resulting in costly and inefficient vector updates. The state-of-the-art clustering-based approach SPFresh offers better scalability but suffers from reduced recall due to coarse partitioning. Moreover, SPFresh employs in-place updates to maintain its index structure, limiting its efficiency in handling high-throughput insertions and deletions under dynamic workloads. This paper presents LSM-VEC, a disk-based dynamic vector index that integrates hierarchical graph indexing with LSM-tree storage. By distributing the proximity graph across multiple LSM-tree levels, LSM-VEC supports out-of-place vector updates. It enhances search efficiency via a sampling-based probabilistic search strategy with adaptive neighbor selection, and connectivity-aware graph reordering further reduces I/O without requiring global reconstruction. Experiments on billion-scale datasets demonstrate that LSM-VEC consistently outperforms existing disk-based ANN systems. It achieves higher recall, lower query and update latency, and reduces memory footprint by over 66.2%, making it well-suited for real-world large-scale vector search with dynamic updates.
arXiv:2605.08475v2 Announce Type: replace Abstract: Mechanistic accounts of in-context learning (ICL) have identified iterative algorithms for linear regression and related linear prediction tasks, often using linear or ReLU attention variants. For nonlinear ICL, prior work has related softmax and kernelized attention to functional-gradient-type dynamics, but it remains unclear whether a standard transformer with softmax attention can implement a convergent solver with an end-to-end prediction-error guarantee. In this paper, we study in-context kernel ridge regression (KRR) with Gaussian kernels and show that a standard softmax-attention transformer can approximate the KRR predictor during its forward pass by implementing preconditioned Richardson iteration on the associated kernel linear system. Under bounded-data assumptions, we construct a single-head transformer with $O(\log(1/\epsilon))$ blocks and MLP width $O(\sqrt{N/\epsilon})$ that achieves $\epsilon$-accurate prediction for prompts of length $N$. Our construction reveals a functional decomposition within the transformer architecture: softmax attention produces a row-normalized Gaussian-kernel operator needed for cross-token interactions, while ReLU MLP layers act locally to approximate the intra-token scalar arithmetic required by the update. Empirically, we train GPT-2-style transformers on Gaussian-process regression tasks to further test the preconditioned Richardson interpretation. Through linear probing, we compare the transformer's layer-wise predictions with the step-wise outputs of classical KRR solvers and find that its error profiles align most consistently with preconditioned Richardson iteration. Ablation studies further support this interpretation. Together, our theory and experiments identify preconditioned Richardson iteration as a concrete mechanism that softmax-attention transformers can realize for nonlinear in-context Gaussian-kernel regression.
arXiv:2505.07813v2 Announce Type: replace Abstract: Large-scale, diverse robot datasets have emerged as a promising path toward enabling dexterous manipulation policies to generalize to novel environments, but acquiring such datasets presents many challenges. While teleoperation provides high-fidelity datasets, its high cost limits its scalability. Instead, what if people could use their own hands, just as they do in everyday life, to collect data? In DexWild, a diverse team of data collectors uses their hands to collect hours of interactions across a multitude of environments and objects. To record this data, we create DexWild-System, a low-cost, mobile, and easy-to-use device. The DexWild learning framework co-trains on both human and robot demonstrations, leading to improved performance compared to training on each dataset individually. This combination results in robust robot policies capable of generalizing to novel environments, tasks, and embodiments with minimal additional robot-specific data. Experimental results demonstrate that DexWild significantly improves performance, achieving a 68.5% success rate in unseen environments-nearly four times higher than policies trained with robot data only-and offering 5.8x better cross-embodiment generalization. Video results, codebases, and instructions at https://dexwild.github.io
arXiv:2504.14820v2 Announce Type: replace Abstract: For peg-in-hole tasks, humans rely on binocular visual perception to locate the peg above the hole surface and then proceed with insertion. This paper draws insights from this behavior to enable agents to learn efficient assembly strategies through visual reinforcement learning. Hence, we propose a Separate Primitive Policy (S2P) to learn how to derive location and insertion actions simultaneously. S2P is compatible with model-free reinforcement learning algorithms. Ten insertion tasks featuring different polygons are developed as benchmarks for evaluations. Simulation experiments show that S2P can boost the sample efficiency and success rate even with force constraints. Real-world experiments are also performed to verify the feasibility of S2P. Ablations are finally given to discuss the generalizability of S2P and some factors that affect its performance.
arXiv:2605.18327v1 Announce Type: new Abstract: AI agents deployed into SRE workflows currently derive their understanding of environment state from raw observability telemetry at query time, paying a semantic-interpretation tax in tokens, latency, and inferential reliability. We propose Causely, a causal intelligence layer that maintains a structured representation of environment topology, attribute dependencies, and causal relationships that are anchroed to a ontological representation of the managed environment. Causely transforms raw telemetry into a live, queryable model providing the semantic and causal foundation AI agents require to diagnose, evaluate impact, and act safely in production. We evaluate this value proposition through a benchmark study conducted in a controlled setting with injected faults in a 24-microservice OpenTelemetry demo application. Our experiments compare four agent configurations (Claude Code, OpenAI Codex, HolmesGPT with Sonnet and Gemini backends). Experiments are run with and without access to Causely under two scenarios: an active incident and a healthy baseline. On the active-fault scenario, causal grounding reduces mean time-to-diagnosis by 63\%, mean token consumption by 60\%, and mean tool-call count by 78\%, compressing the investigation footprint by 4.8$\times$ and lowering direct API cost per run by 57\%; root-cause-diagnosis accuracy rises from 75\% to 100\%.
arXiv:2605.16556v1 Announce Type: cross Abstract: The determination of crystallographic texture through elastic wave propagation offers a cost-effective, nondestructive means of obtaining through-thickness information with minimal sample preparation. Existing ultrasonic approaches rely on either bulk-wave or guided-wave velocity measurements for texture inversion. These strategies impose geometric constraints: bulk-wave methods become impractical for thin specimens, whereas guided-wave techniques are limited to relatively small thicknesses. Furthermore, many formulations assume orthotropic symmetry of the aggregate, thereby restricting their applicability to materials with higher anisotropy. In this work, a full-field wave fitting strategy is developed in which the transmitted ultrasonic field is simulated and directly compared to experimental measurements. Because the approach does not rely on bulk-wave or plate-wave approximations, it remains applicable across a broad range of specimen thicknesses. Furthermore, no macroscopic symmetry assumptions are imposed on the aggregate, enabling the characterization of generally anisotropic materials. The effective elastic response is computed using a Hashin-Shtrikman homogenization framework, which provides tighter bounds than classical Voigt-Reuss-Hill averages and constrains the admissible search space during optimization, thereby improving convergence. The nonlinear inverse problem is solved using a GPU-accelerated optimization scheme. The methodology is validated on materials with hexagonal and cubic crystal symmetry over a range of specimen thicknesses. The inferred texture coefficients show consistent agreement with independent diffraction measurements. Additionally, textures with weak elastic anisotropy are successfully recovered, demonstrating the robustness and versatility of the proposed method. Complete measurement and inversion are achieved within approximately 10 minutes.
arXiv:2605.16376v1 Announce Type: cross Abstract: Kelvin is a lightweight learned pre-encoder that sits in front of an unmodified libx264 encoder. It applies content-adaptive pixel adjustments, bounded at +/-1/255 per channel, so that the encoder allocates bits where they matter most perceptually, while emitting a standard H.264 bitstream compatible with every existing decoder, player, and CDN. On the seven-sequence 1080p UVG benchmark, Kelvin v1.0 achieves a mean BD-VMAF of -27.62% (7 of 7 wins) and BD-VMAF-NEG of -5.18% (6 of 7 wins) relative to baseline libx264 at preset medium. On the 30-sequence MCL-JCV public set (28 unseen by training), the same checkpoint wins on 28 of 30 clips by BD-VMAF; with the two diagnosable failures removed the mean is -27.70% BD-VMAF and -5.37% BD-VMAF-NEG, consistent with UVG to within one percentage point. A central engineering challenge is the non-differentiability of H.264: we describe a hybrid codec proxy that combines a calibrated differentiable rate estimator (Spearman rho = 0.986 vs. real libx264 bits-per-pixel) with a U-Net distortion proxy trained on real encoder outputs. We publish full per-sequence rate-distortion data, a named failure-mode taxonomy on MCL-JCV (rate-floor violation, distribution shift, metric saturation), a five-baseline sanity panel (hqdn3d, unsharp, -tune psnr, -tune ssim, x265 medium), and honest positioning: x265 medium beats Kelvin on every metric on the same corpus. Kelvin is therefore designed for workloads where remaining on H.264 is a constraint rather than a choice.
arXiv:2605.18025v1 Announce Type: new Abstract: While Large Language Models have achieved remarkable integration in various vertical scenarios, their deployment in the telecommunications domain remains exploratory due to the lack of a standardized evaluation framework. Current telecom benchmarks primarily focus on static, foundational knowledge and isolated atomic skills, neglecting the equipment-specific documentation and end-to-end industrial workflows essential for real-world production systems. To bridge this gap, we present TeleCom-Bench, a comprehensive benchmark comprising 12 evaluation sets with 22,678 curated samples, which evaluates LLMs across a synergistic hierarchy: (1) Multi-dimensional Knowledge Comprehension, which integrates telecommunication fundamentals, 3GPP protocols, and 5G network architecture with proprietary product knowledge across wired, core, and wireless networks via knowledge graph-driven synthesis; and (2)End-to-End Knowledge Application, which formalizes six core tasks on authentic trajectories from live network agent workflows, including intent recognition, entity extraction, event verification, tool invocation, root cause analysis, and solution generation-across network optimization and fault maintenance scenarios. Evaluations of eight state-of-the-art LLMs reveal a universal Execution Wall: while models achieve 90% accuracy in linguistic interface tasks such as intent recognition and entity extraction, performance collapses to approximately 30% in procedural execution tasks like solution generation. This capability gap demonstrates that current LLMs function competently as diagnosticians but fail as field engineers. TeleCom-Bench provides standardized diagnostics to precisely pinpoint this deficit, offering actionable guidance for domain-specific alignment toward production-ready telecom agents. The dataset and evaluation code have been released at https://github.com/ZTE-AICloud/TeleCom-Bench.
arXiv:2605.16645v1 Announce Type: cross Abstract: Machine learning systems increasingly face requirements to forget not only individual data points, but entire domains of information, such as toxic language, copyrighted corpora, or demographic biases. This raises a fundamental dilemma of statistical-computational tradeoffs: removing all samples from an unwanted domain may be computationally prohibitive, while randomly removing a subset may not provide distribution-level statistical guarantees. We propose a statistical framework for distributional unlearning, in which domains are modeled as probability distributions, and the goal is to remove a carefully chosen subset of samples that reduces the effect of an unwanted distribution while preserving performance on a desired one. We formalize this using a hypothesis test of the edited data with the desired and unwanted domains, leading to an interpretable and robust criterion for selecting samples to remove. Within this statistical framework, we characterize the fundamental region of the allowable edited data distributions and the removal-preservation Pareto frontier for a broad class of distribution families. This includes parametric families such as shifted Gaussians of arbitrary dimension, a one-dimensional location family with log-concave noise, and the one-dimensional Poisson family. It also includes nonparametric families such as the Gaussian white noise model, a canonical model for nonparametric regression. We prove composition rules that describe how distributional unlearning behaves across multimodal unwanted domains, and introduce a central-limit behavior for the removal-preservation baselines when composing a large number of such families. Finally, we provide finite sample guarantees by providing Pareto frontiers for some selection algorithms, and observe an information-computation gap.
Toward an Origin of Human Randomness: Interaction-Driven Enhancement in the Rock-Paper-Scissors Game
arXiv:2605.18616v1 Announce Type: new Abstract: Human-generated randomness is constrained by cognitive, motor, and strategic biases. This study examines how these constraints appear in individual behavior and how they may be modified through interaction with another human. We analyzed repeated rock-paper-scissors data from 9 participants, yielding 108 human-human matches and 216 individual player sequences. Using Lempel-Ziv complexity (LZC), we compared human-human sequences with the RNG-opponent condition. In the RNG-opponent condition, the maximum human LZC value was 84, which we used as an empirical reference. In the human-human condition, most sequences remained below this value, but a small number exceeded it, producing a small high-complexity tail that was not present in the RNG-opponent condition. We introduced a sensitivity measure that captures whether a player responds to the opponent's recent frequency bias by choosing the move that beats the opponent's most frequent recent move. Partial regression showed that focal-player sensitivity positively predicted future entropy in the opponent's move sequence after controlling for the opponent's current entropy. Circular-shift surrogate analyses indicated that this relation was most clearly interaction-specific when the opponent was in a low-entropy state, where the recent move distribution contained a clear frequency bias. These results suggest that human randomness is not only an isolated individual capacity, but can be shaped by interaction in a state-dependent manner. The findings identify a local mechanism by which interaction may destabilize biased behavior and increase entropy, providing a concrete basis for future causal experiments and generative models of high-complexity human behavior.
arXiv:2602.08169v2 Announce Type: replace Abstract: Inference-time steering offers a promising way to control language models (LMs) without retraining. However, standard approaches typically rely on activation addition, which inevitably alters the hidden-state magnitudes raising concerns about representation collapse and degraded open-ended generation. In this work, we explore Spherical Steering, a training-free primitive that resolves this trade-off through activation rotation. Rather than shifting activations with a fixed vector, our method rotates them along a geodesic toward a target direction, preserving signal integrity while steering toward the target concept. To further enhance adaptivity, we incorporate a confidence gate that dynamically modulates steering strength based on input uncertainty. Extensive experiments across multiple-choice benchmarks demonstrate that Spherical Steering significantly outperforms addition-based baselines (notably by +10% on TruthfulQA, COPA, and Storycloze), while simultaneously maintaining the model's general open-ended generation quality. This work highlights the value of geometric consistency, suggesting that norm-preserving rotation is a robust and effective primitive for precise inference-time control. The code is available at: https://github.com/chili-lab/Spherical-Steering.
arXiv:2605.18572v1 Announce Type: new Abstract: Persuasive dialogue generation plays a vital role in decision-making, negotiation, counseling, and behavior change, yet it remains a challenging problem. In complex persuasion where the persuadee's internal states are not expressed clearly, the persuader must interpret responses, infer the persuadee's latent mental states (e.g., beliefs and desires), and translate them into targeted, strategy-consistent actions; however, current approaches often produce generic or weakly grounded responses even when such cues are identified. Moreover, although large language models (LLMs) can generate persuasive content, their performance varies substantially across domains due to uneven knowledge coverage and limited reasoning generalization. To address these challenges, we propose MA$^{2}$P, a meta-cognitive autonomous intelligent agent framework for complex persuasion. Specifically, we develop an autonomous multi-agent architecture that coordinates perception management, mental-state inference, strategy execution, memory maintenance, and performance evaluation. To mitigate cross-domain performance variation, we further design a meta-cognitive configurator that selects an appropriate meta-strategy from a structured knowledge base at the outset, thereby guiding subsequent reasoning and planning. Experimental results show that our approach achieves a higher persuasion success rate than baselines.
arXiv:2602.03535v2 Announce Type: replace Abstract: We introduce a dynamic sparse training algorithm based on linearized Bregman iterations / mirror descent that exploits the naturally incurred sparsity by alternating between periods of static and dynamic sparsity pattern updates. The key idea is to combine sparsity-inducing Bregman iterations with adaptive freezing of the network structure to enable efficient exploration of the sparse parameter space while maintaining sparsity. We provide convergence guaranties by embedding our method in a multilevel optimization framework. Furthermore, we empirically show that our algorithm can produce highly sparse and accurate models on standard benchmarks. We also show that the theoretical number of FLOPs compared to SGD training can be reduced from 38% for standard Bregman iterations to 6% for our method while maintaining test accuracy.We additionally show a training time reduction by about 50%, when using a sparsity-aware CPU implementation of our method.
arXiv:2605.17810v1 Announce Type: new Abstract: Weak force amplification describes the process of amplifying a faint low-frequency signal by means of an additional high-frequency modulation, which plays a vital role in quantum sensing and high-precision measurement. However, the potential enhancement of weak-force amplification in non-Markovian environments has received little attention. In this paper, we firstly study the amplification of weak forces within cavity-optomechanical systems incorporating a degenerate optical parametric amplifier (DOPA) under the Markovian assumption, which can be amplified via using two high-frequency signals via vibrational resonance through adjusting the strength and phase of the DOPA with different pumping frequencies. Moreover, we extend the study of the amplification of the weak force to the non-Markovian environment composed of an ensemble of infinite oscillators. We reveal that the amplification exhibits a conversion from the non-Markovian regime to Markovian regime by controlling environmental spectral width. Such a transition facilitates a remarkable improvement in amplification, and this enhancement originates from the excitation backflow generated via the interplay between the cavity and the non-Markovian environment. By controlling DOPA to amplify weak forces, the study achieves amplification in the non-Markovian regime, offering new directions for quantum optics research.
arXiv:2605.17808v1 Announce Type: new Abstract: We develop a unified theoretical framework for data-free one-step sampling from unnormalized target distributions based on Wasserstein gradient flows. For a broad class of standard f-divergence objectives, we show that the induced velocity field admits the universal form $\mathbf{V}(x)=w(r(x))\,\beta(x)$, where $\beta(x)=\nabla \log (p(x)/q(x))$ is shared across objectives and $w$ is determined solely by the choice of divergence. This decomposition shows that standard f-divergence drifts share the same asymptotic target distribution $p$ and differ primarily in how they redistribute transient repair effort across under-covered regions. To formalize this distinction, we derive a one-step regional-response theory for a soft under-coverage functional and obtain a compression--elasticity identity that links divergence choice to the geometry of mass transport into under-covered regions. We further extend the framework beyond the f-divergence family to the Log-Variance (LV) divergence, analyze how the reference distribution alters the resulting drift structure, and motivate a practical LV-inspired surrogate for data-free training. Based on this theory, we instantiate the framework with a KDE-based implementation and describe a complementary normalizing-flow route, enabling one-step inference after training. Experiments on multimodal Gaussian-mixture benchmarks are consistent with the theoretical predictions and demonstrate effective one-step sampling on these targets.
arXiv:2605.17887v1 Announce Type: new Abstract: We propose OASIS, an outlier- and sink-aware technique built on inter-layer null signaling. As AttnResidual architectures introduce an additional depth-wise normalization channel, they improve inter-layer routing flexibility but also exacerbate attention sinks, activation outliers, and the resulting degradation in inference stability and quantization robustness. OASIS addresses this issue by introducing a Softmax1-based null space and coupling token-level null evidence to depth routing through an inter-layer null signal, thereby reducing sink-dominated routing and improving structural robustness. Theoretically, we show that the dual-normalization design of AttnResidual intensifies sink formation and quantization brittleness. Experimentally, we compare OASIS against five baselines on three real-world datasets and observe consistent improvements in both attention sink and post-quantization performance. Notably, OASIS achieves an average reduction of 9.26% in maximum infinity norm and 2.60% in average kurtosis across the evaluated settings, while lowering perplexity by 75.85% under W8A8 and improving GSM8K Pass@1 by 12.42% under W4A4.
arXiv:2605.17801v1 Announce Type: new Abstract: A measure for the complexity of a differentiable function f(x) on an interval is introduced. It is based on approximations of the function by piecewise constant functions. The measure takes into account the quality of the approximation and the number of intervals in the approximating function. This measure, called the V-complexity of f(x), is shown to formalize some intuitions about the simplicity or complexity of f(x). The V-complexity is then compared to another measure of complexity, namely how compressible an approximation of f(x) is. It is hypothesized that V-complexity is equivalent to the compression measure, in the case of the Run Length Encoding and the Lempel Ziv 77 algorithms. V-complexity can be used as an ingredient in the definition of the Effective Complexity (EC) of a Complex System. When the perceived regularities of such a system are described by a differentiable function on an interval, the EC can be defined as the V-complexity of that function. EC is applied to the model of diffusion of cream in a cup of coffee. The perceived regularity of this model is given by the diffusion equation. The V-complexity of the solution of the equation starts at zero, quickly increases to a maximum and then decreases back to zero as the liquid reaches its equilibrium state. It is shown that this is also the result when a cellular automaton approach and the concept of Apparent Complexity is used.
arXiv:2605.17799v1 Announce Type: new Abstract: Long-tailed out-of-distribution (LT-OOD) detection is often addressed with specialized training, including auxiliary out-of-distribution (OOD) data, abstention heads, contrastive objectives, energy losses, or gradient-conflict control. We show that these training mechanisms can obscure a simpler issue: frozen long-tailed representations may already contain useful OOD evidence, but raw Mahalanobis distance is distorted by frequency-coupled feature radius and poorly supported tail covariance. We propose Hyperspherical Pooled Mahalanobis (HPM), a post-hoc detector that normalizes features onto the unit sphere and replaces class-specific covariance with a pooled, ridge-regularized metric while keeping class means as semantic anchors. In CIFAR-LT experiments and an ImageNet-100-LT near-OOD boundary analysis, HPM improves raw Mahalanobis scoring; for Prior-Calibrated ERM (PC-ERM), it raises AUROC from 46.49 to 85.67 on CIFAR-10-LT and from 50.40 to 78.35 on CIFAR-100-LT. This simple PC-ERM+HPM pipeline also achieves the best Log Efficiency Score (LES; 3.08) on CIFAR-100-LT, retaining roughly 95% of the best CIFAR-100-LT AUROC observed among the compared post-hoc scores at substantially lower training-time cost. These results argue for evaluating representation quality, detector geometry, and training complexity as separate factors in LT-OOD detection.
arXiv:2605.17967v1 Announce Type: new Abstract: This paper explores a scientific question in supervised fine-tuning (SFT): why SFT is broadly effective for small-scale deep neural networks, yet can produce inconsistent or even detrimental effects when applied to large language models (LLMs). Recent advances in interaction-based explanations suggest that interactions between words/tokens provide a faithful metric for quantifying the inference patterns encoded by LLMs. We find that the evolution of interactions during SFT can effectively explain the inconsistent effectiveness of SFT for LLMs. Specifically, we find that (1) SFT primarily removes noise-like interactions, while rarely acquiring reliable new interactions. (2) This denoising stage is extremely brief, after which continued fine-tuning tends to introduce overfitted interactions. We validate these findings across multiple LLMs and datasets. Our findings provide new insights into early stopping and offer practical guidance for LLM training.
arXiv:2605.17976v1 Announce Type: new Abstract: Scientific discovery is increasingly constrained by costly experiments and limited resources, underscoring the need for efficient optimization in AI for science. Bayesian Optimization (BO), though widely adopted for balancing exploration and exploitation, often exhibits slow cold-start performance and poor scalability in high-dimensional settings, limiting its applicability in real-world scientific problems. To overcome these challenges, we propose LLM-Guided Bayesian Optimization (LGBO), the first LLM preference-guided BO framework that continuously integrates the semantic reasoning of large language models (LLMs) into the optimization loop. Unlike prior works that use LLMs only for warm-start initialization or candidate generation, LGBO introduces a region-lifted preference mechanism that embeds LLM-driven preferences into every iteration, shifting the surrogate mean in a stable and controllable way. Theoretically, we prove that LGBO does not perform significantly worse than standard BO in the worst case, while achieving significantly faster convergence when preferences align with the objective. Empirically, LGBO consistently outperforms existing methods across diverse dry benchmarks in physics, chemistry, biology, and materials science. Most notably, in a new wet-lab optimization of Fe-Cr battery electrolytes, LGBO attains \textbf{90\% of the best observed value within 6 iterations}, whereas standard BO and existing LLM-augmented baselines require more than 10. Together, these results suggest that LGBO offers a promising direction for integrating LLMs into scientific optimization workflows.
arXiv:2605.18480v1 Announce Type: new Abstract: The computation of chance constraints in stochastic model predictive control is often numerically challenging due to the non-Gaussian nature of the disturbances. To overcome this problem, we propose an optimization computational framework applicable to non-Gaussian disturbances. This framework employs a numerical inversion method, utilizing the characteristic function of the disturbance distribution to compute the probability in the chance constraint as well as its gradient. To improve efficiency, it vectorizes integral points and reuses intermediate computations in Gauss-Kronrod quadrature. The framework is implemented within the YALMIP toolbox to perform chance constraint calculations for arbitrary non-Gaussian disturbances, applicable to both single-component distributions and mixture models. It allows the user to simply specify a distribution type and its parameters for the disturbance and directly compute the probability and its gradient to solve the optimization problem. The method is validated through a numerical example of a stochastic model predictive control application.
arXiv:2605.18281v1 Announce Type: new Abstract: Modern deep learning science often assumes that neural networks learn from a fixed data distribution. However, many practically important learning problems involve data distributions that change throughout training. How does such non-stationarity impact the inductive biases of deep learning towards models with different structural, generalisation, and safety properties? A fruitful testbed for studying inductive bias is in-context linear regression sequence modelling, where small transformers display strikingly different generalisation patterns depending on the diversity of the (fixed) training task distribution. In this paper, we explore the effect of diversifying the task distribution across training time, finding that such temporal diversity leads to an increased bias towards generalisation over memorisation.
arXiv:2604.07630v2 Announce Type: replace Abstract: The final size of an earthquake typically cannot be predicted from its ongoing seismic radiation. Expanding observations reveal distinct exceptions, such as slow earthquakes, injection-induced seismicity, and earthquake swarms, in which fault slip has an upper bound. A common thread among these anomalies is the diffusive migration of their active areas. Here, we report a unified scaling relation for these diffusional earthquakes. By tracking prolonged earthquake swarms in Northeast Japan, we constrained the time evolution of their active seismicity areas and cumulative seismic moments. Their moment-duration trajectories coincide with the final states documented for global swarms and induced seismicity across various scales. When plotted as seismic moment versus seismicity area, their trajectories collapse onto those of slow earthquakes, uniformly explained by a diffusional constant-slip model. This constant-slip scaling carves out a unique class of diffusional earthquakes, where the final available seismic energy is predetermined by slip distance.