Peer Revenue Publish or profit

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Thursday 3 September

Artificial Intelligence

From Answers to Interpretations: Rethinking Ambiguity-Induced Aleatoric Uncertainty Estimation in LLMs

A key challenge in reliable LLM deployment is recognizing when uncertainty reflects irreducible variability in the task rather than limitations in the model's knowledge. In language tasks, a central source of such aleatoric uncertainty is input ambiguity or underspecification, where multiple interpretations remain plausible. Existing decomposition methods estimate aleatoric uncertainty by generating multiple clarifications of the input, querying the model for an answer under each clarification, and comparing the resulting answers. We argue that answers are not necessary for identifying ambiguity: they are often redundant, add avoidable cost, and can mislead through epistemic leakage. We support this claim theoretically, and propose a clarification-only approach that estimates this ambiguity-induced component directly from the space of plausible interpretations, without answers to the clarified inputs. Using ambiguity detection as an operational evaluation across three benchmarks, this direct approach improves AUROC (63.34 vs. 60.85), reduces computational cost by 4-26x in output tokens and 2.2-3.5x in API calls, and yields estimates with substantially lower correlation with epistemic uncertainty. Overall, our results suggest that ambiguity-induced aleatoric uncertainty is better estimated from the interpretation space than from the response space.
Read the full abstract

All authors Omer Nahum, Niv Nayman, Jonathan Fhima, Alon Zolfi, Jeremy Levy, Shai Mazor, Paolo Favaro

Human-Computer Interaction

Matched Starts, Divergent Objects: Research-Object Formation in Human-AI Scholarly Work

Human-artificial intelligence interaction can shape what scholarly inquiry becomes about. This study examines research-object formation in two isolated scholarly trajectories developed by the same expert researcher with Claude and ChatGPT under matched starting conditions. Both used the same five-session, 843-turn project corpus, frozen research problem, initial prompt, publication objective, and conduct rules. The subsequent scholarly-development conversations were reconstructed ex post through consequential-event analysis, a Socioduality relational-process overlay, and tracing into completed manuscripts. Both trajectories shifted the continuity problem from recall toward usability, then developed different research objects: continuity labour and distributed, evolving project state. Failed analytical units, rejected explanations, counterexamples, and changes in analytical scale shaped these objects and propagated into research questions, methods, findings, evidence logic, and theoretical contributions. Consequential scholarly change, reciprocal relational continuity, and local substantive re-formation had distinct boundaries; continued interaction did not itself establish epistemic endorsement. The findings demonstrate empirically constrained research-object formation under matched scholarly starts. For systems supporting sustained scholarly knowledge work, they motivate recoverable analytical provenance, inspectable proposal status, and source-faithful contribution records.
Read the full abstract

Artificial Intelligence

Data-Driven Discovery of Composition-Dependent Constitutive Models for Hyperelasticity and Viscoelasticity of Digital Materials

Digital materials fabricated by multi-material 3D printing are designed as controlled mixtures of stiff and compliant constituents, yielding effective responses that span more than an order of magnitude in apparent stiffness and exhibit strongly nonlinear, composition-dependent, and rate-dependent dissipative behavior. Classical finite-strain viscoelastic models represent such behavior with closed-form strain energy functions for equilibrium and non-equilibrium stresses as well as evolution of internal variables, which may limit flexibility when a single constitutive model is expected to generalize across materials and loading rates. Here, we present a data-driven multi-material constitutive modeling framework that generalizes a formulation by Bergström and Boyce. The proposed framework retains the structure of the classical model, namely multiplicative kinematics, invariant-based strain-energy functions, and a scalar dissipative evolution law directed along the normalized nonequilibrium deviatoric stress. For the equilibrium branch, the data-driven discovery framework either directly predicts closed-form model parameters as functions of composition or automatically constructs a polyconvex strain-energy function using neural ordinary differential equations (NODEs). The nonequilibrium branch kinetics are learned similarly, either by directly identifying closed-form parameters across compositions or by using appropriately constrained artificial neural networks. Using multi-rate uniaxial compression data across multiple material compositions, we show that the proposed formulation captures rate-dependent stiffness and hysteresis across compositions while preserving thermodynamic consistency.
Read the full abstract

All authors Josué García-Ávila, Beijun Shen, Manuel K. Rausch, Mary C. Boyce, Adrián Buganza-Tepole

Machine Learning

Mitra-v2 Technical Report

We introduce Mitra-v2, a tabular foundation model that delivers state-of-the-art performance on real-world classification and regression problems, from credit-risk scoring and clinical prediction to equipment-failure detection and house-price estimation. Mitra-v2 is trained only on synthetic data, with a pretraining distribution that is much larger and more diverse than Mitra-v1's. Built on a small 2D Transformer backbone, Mitra-v2 supports longer contexts and larger feature spaces. Improved optimization lets it learn from this larger task distribution. We evaluate Mitra-v2 on the TabArena and TALENT benchmarks, comprising more than 300 real-world datasets under two evaluation protocols. On the full TabArena benchmark, Mitra-v2 delivers state-of-the-art performance at the level of the industry-scale TabFM and EXAONE Tabular models, while surpassing TabPFN-3 by a wide margin in both classification and regression. Mitra-v2 matches the 1.6B-parameter TabFM with only 5% of its size (77M parameters), delivering frontier performance at a fraction of the cost. On TALENT, Mitra-v2 remains among the leading models, clearly outperforming TabPFN-3 and TabICLv2. It also ranks first on classification tasks with more than ten classes, even though it was pretrained only on tasks with at most ten classes. These results make Mitra-v2 one of the strongest and most broadly applicable open tabular foundation models released to date. We release the model weights, the inference and fine-tuning code, and our evaluation results under the Apache-2.0 license.
Read the full abstract

All authors Yefan Tao, Xiyuan Zhang, Xinyi Liu, Boran Han, Danielle Maddix, Haoyang Fang, Zhen Han, Jiading Gai, Xuanqing Liu, Michael Bohlke-Schneider, Yuyang Wang, Gerald Friedland, Kevan Mah, Chris Lee, Chris Kong

Computation and Language

A Calibrated Reflection Approach for Enhancing Confidence Estimation in LLMs

A critical challenge in deploying Large Language Models (LLMs) is developing reliable mechanisms to estimate their confidence, enabling systems to determine when to trust model outputs versus seek human intervention. We present a Calibrated Reflection approach for enhancing confidence estimation in LLMs, a framework that combines structured reasoning with distance-aware calibration technique. Our approach introduces three key innovations: (1) a Maximum Confidence Selection (MCS) method that comprehensively evaluates confidence across all possible labels, (2) a reflection-based prompting mechanism that enhances reasoning reliability, and (3) a distance-aware calibration technique that accounts for ordinal relationships between labels. We evaluate our framework on diverse datasets, including HelpSteer2, Llama T-REx, and a proprietary conversational dataset, demonstrating its effectiveness across both conversational and fact-based classification tasks. This work contributes to the broader goal of developing reliable and well-calibrated confidence estimation methods for LLMs, enabling informed decisions about model trust and human judgement.
Read the full abstract

All authors Umesh Bodhwani, Yuan Ling, Shujing Dong, Yarong Feng, Hongfei Li, Ayush Goyal

Fluid Dynamics

Lagrangian dynamics unveil polymer conformation in viscoelastic flows

The complex flow behavior of viscoelastic fluids emerges from a feedback between fluid stress and the out-of-equilibrium conformational dynamics of their constituent polymer chains. While constitutive models can predict flow and stress fields in simple geometries, a fundamental understanding of the mechanisms regulating viscoelastic flows with mixed kinematics and non-trivial polymer dynamics remains elusive. Quantifying the history-dependent conformation of polymer chains in flow is essential to resolve these complex systems. To address this knowledge gap, we employ direct Lagrangian tracking of individual DNA molecules to reveal their conformational dynamics in microfluidic viscoelastic flows for different polymer contour lengths and concentrations. Comparing the measured molecular extension and orientation reveals discrepancies with canonical constitutive models, which do not fully capture the transient dynamics. Our measurements show a distinct anisotropy in polymer relaxation and shape, which couple the polymer's rotation and extension through hydrodynamic drag, regulating their Lagrangian mechanics. These findings emphasize the importance of Lagrangian polymer history and the need to account for non-trivial intra-molecular mechanics to improve constitutive descriptions of viscoelastic flows.
Read the full abstract

All authors Louison Thorens, Gabriel Juarez, Jeffrey S. Guasto, Paulo E. Arratia

Game Theory

Sparse Disapproval Guarantees a Nonempty Hare Core

An approval committee is Hare-core stable if no coalition meeting the Hare quota can strictly improve by moving to another candidate set. Whether every approval election has such a committee remains open. We prove nonemptiness when each voter disapproves at most two candidates, with no bounds on the numbers of candidates, seats, or voter types. The result also permits arbitrary positive rational voter weights. Our deterministic rule represents a committee by its missing set. It first maximizes weighted coverage of two-candidate disapproval sets and then maximizes total disapproval incidence. An exact coverage inequality excludes targets one seat below the committee. The incidence objective excludes unanimous equal-size targets, while targets of size at most $k-2$ cannot improve any voter. Two implementation-level independent verifiers audit overlapping finite grids. The symbolic proof, not this bounded enumeration, establishes the theorem's unbounded quantifiers. The argument identifies complement-side coverage as a tractable mechanism for a broad parameter range within a sharply defined preference domain.
Read the full abstract

Information Theory

Deep Reinforcement Learning for Optimization of STAR-RIS Phase and Energy Splitting Coefficients in OTFS-NOMA Framework

This paper considers a downlink communication framework comprising a simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS)-aided by orthogonal time frequency space (OTFS) and non-orthogonal multiple access (NOMA) technologies. Further, delay-Doppler mobility in such frameworks renders classical alternating optimization impractical for per-coherence interval reconfiguration. To mitigate such issues, the STAR-RIS phase-shift and energy-splitting design is formulated as a constrained, non-convex sum-rate maximization problem with closed-form maximum ratio transmission beamforming and fixed NOMA power allocation. To circumvent the per-interval re-optimization burden, a deep reinforcement learning (DRL) approach is adopted that maps observed channel realizations to STAR-RIS configurations through a single forward pass. Specifically, Beta-Space Soft Actor-Critic (SAC-BSE), a maximum entropy DRL agent, is proposed. Simulation results, with two NOMA-multiplexed users on each STAR-RIS branch, confirm rapid convergence, limit the sum-rate degradation to roughly 10% across a 128-fold user-speed range, and yield consistent gains over OTFS-only, NOMA-only, STAR-RIS-only, fixed-split, and mode-switching baselines as transmit power and the number of STAR-RIS elements increase.
Read the full abstract

Software Engineering

An Empirical Analysis of CodeQL False Positives and Query Refinements for Java Vulnerabilities

Static application security testing (SAST) tools help developers find vulnerabilities before deployment, but false positives create substantial triage effort. We study whether CodeQL false positives in Java security analysis form recurring, explainable patterns that can be reduced by refining the analysis. We run CodeQL's Java security query suite on 167 CVE instances from 110 projects, focusing on the ten queries with the highest false positive rates. We manually review 500 sampled false positive paths and locations and construct a source-level taxonomy. The five categories are Missed Path Constraint or Sanitization (36.6%), Benign Execution Context (29.4%), Missing Trust Boundary Modeling (27.6%), Imprecise Concurrency Modeling (5%), and Imprecise Sink Modeling (1.4%). Guided by these findings, we implement CodeQL refinements that detect and filter recurring false positive patterns at the query level. The refinements remove 81.8% of reviewed false positives. Across the full selected-query dataset, they remove 15.8% of reported paths and locations while retaining 7 of 8 true positives. This shows that many false positives can be reduced in the analysis, although fixed refinements often depend on project-specific context. To address this generalization gap, we evaluate whether agentic coding tools can adapt refinement patterns to new projects. Given our patterns as templates, the two tools succeed on 56% and 62% of tasks, with query compile-pass rates above 90%. Without this guidance, both succeed on only 28%, while compile rates fall to 30-36%. These results support a refinement-oriented SAST workflow in which recurring false positives are modeled in CodeQL queries and automatically adapted to different project contexts, reducing repeated triage.
Read the full abstract

Numerical Analysis

Adaptive Multilevel Discontinuous Galerkin Methods on GPUs

I present a matrix-free symmetric interior penalty discontinuous Galerkin method for adaptively refined Cartesian meshes on GPUs. At non-matching interfaces, auxiliary shadow cells represent the adjacent coarse polynomial on the fine level. This approach converts non-matching interfaces into matching faces, permitting uniform face evaluation throughout the mesh. I prove that this construction is equivalent to the standard non-matching formulation. Furthermore, this representation yields a local geometric multigrid method in which frozen shadows provide inter-level boundary data and carry residual contributions to coarser levels. A primal–dual pairing eliminates shadow assembly from the Krylov iteration. Numerical experiments with cubic elements ($p=3$) on an NVIDIA A100 show stable multigrid convergence under increasing refinement depth and efficient GPU execution.
Read the full abstract

Cryptography and Security

Repeat-After-Me: Black-Box Adaptive Visual Prompt Injection

Prompt injection is widely recognized as a major security threat to AI agents that interact with untrusted external data, such as websites, documents, and emails. Prior work has shown that, in the text domain, black-box prompt injection can achieve near-perfect attack success rates (ASRs). In the image domain, however, existing visual prompt injection methods are substantially less effective in attacking frontier commercial VLMs for materially harmful behavior. Achieving such outputs is hard because it requires a long and/or format-compliant target string, such as a precise, parseable native tool call with exact function names and arguments. We present Repeat-After-Me, a black-box adaptive visual prompt injection attack that can reveal personally identifiable information or make malicious tool calls. Across both open-weight and commercial frontier VLMs, including Qwen3.6-27B and GPT-5.5, our method achieves ASRs exceeding 82% and 47%, respectively, under a realistic setting in which the benign user prompt is semantically unrelated to the injected task and does not verbally authorize it. Our optimized injection has non-trivial attack transferability across commercial VLMs and benign samples. We show our attack works in cases where adaptive textual prompt injection fails. In a real-world OpenClaw agent connected to Discord, an untrusted user can use a minimally injected image from our attack to overwrite TOOLS.md, enabling future sensitive behaviors like remote code execution and secret exfiltration. We discuss potential defenses.
Read the full abstract

All authors Sizhe Chen, Yu-Lin Tsai, Ivan Evtimov, Kamalika Chaudhuri, Raluca Ada Popa, David Wagner, Arman Zharmagambetov

Strongly Correlated Electrons

Layer Architecture Shapes Electronic, Magnetic, and Lattice Interactions in Ruddlesden-Popper Nickelates

The discovery of superconductivity in Ruddlesden-Popper nickelates has raised a central question: how does layer architecture shape the electronic, magnetic, and lattice interactions relevant to pairing? Here, we report a detailed comparative study of the two polymorphs of La3Ni2O7–the alternating monolayer-trilayer (LNO-1313) and bilayer (LNO-2222) structures–and the related trilayer compound La4Ni3O10, using both Ni L3- and O K-edge RIXS. We find that LNO-1313 and La4Ni3O10 share strikingly similar electronic, magnetic, and lattice excitations, whereas bilayer LNO-2222 exhibits distinct features. Compared to LNO-2222, LNO-1313 and La4Ni3O10 have weaker orbital polarization, enhanced 3d8L character, a reduced out-of-plane magnetic-exchange scale, and stronger EPC. Within an effective local-moment framework, an entangled-dimer scenario provides a natural description of the spin excitations generated by strong antiferromagnetic interlayer coupling. Its advantage over conventional spin-wave theory is clearest in bilayer LNO-2222, where the interlayer coupling dominates the intralayer interactions. These findings provide critical experimental constraints for future theoretical models for the low-energy physics relevant to superconductivity in these layered nickelates.
Read the full abstract

All authors W. He, X. Guo, X. Luo, J. Thomas, J. Sears, Sophia F. R. TenHuisen, Ziqiang Guan, Xinglong Chen, D. A. Dahlbom, B. Zager, J. Pelliciari, Yi-Feng Zhao, H. LaBollita, Hong Zheng, M. K. Lajer, J. F. Mitchell, V. Bisogni, A. S. Botana, M. Mitrano, S. Johnston, M. P. M. Dean

Machine Learning

Distilled Continuous Diffusion Language Models Can Write Code in Few Steps—or One

Language generation is almost universally treated as a sequential process: autoregressive models emit one token at a time, while diffusion language models replace token-level seriality with a long trajectory of iterative refinement. In this work, we introduce PlaidQ, a 0.7B continuous diffusion language model for code generation, and show that its trajectory can be aggressively distilled into only a few denoising steps—or even one, enabling efficient code generation. PlaidQ repurposes a pretrained autoregressive model as a bidirectional denoiser over continuous token embeddings. We distill PlaidQ with distribution matching for few-step generation and paired-trajectory supervision for one-step generation. At matched model scale, PlaidQ is competitive with discrete diffusion language models on code generation. Distillation then shifts the quality–compute frontier: a 16-step student reaches 31.78 and 40.49 pass@10 on HumanEval and MBPP+, surpassing the same PlaidQ teacher sampled for 512 steps. At the extreme, paired-trajectory distillation achieves 7.07 pass@1 on HumanEval with a single denoising step, producing functionally correct programs. Together, these results establish continuous diffusion as a viable path to few-step and one-step code generation. Broadly, continuous diffusion is not merely another representation for language: it provides an interface through which language models can inherit the acceleration and distillation machinery of continuous diffusion modeling. Training and inference code and model checkpoints are available at https://github.com/pengzhangzhi/plaidq.
Read the full abstract

Machine Learning

An Energy-Based Conservative-Dissipative Latent Neural Evolution Operator for Magnetization Dynamics

We introduce a reduced-order learning framework that couples a convolutional autoencoder to a structured latent neural ordinary differential equation. A learned scalar generates the latent dynamics through an antisymmetric operator and a positive-definite dissipative operator. The antisymmetric component permits motion along scalar level sets, while the dissipative component decreases the scalar. Adding a coercive quadratic term gives bounded continuous-time latent trajectories and approach to the scalar's critical set under the stated assumptions. The scalar provides a structural constraint rather than an approximation of physical energy. The encoder, decoder, scalar, and operators are trained jointly on short snapshot windows without time-derivative supervision, physical-energy labels, or dissipation penalties. At inference, one initial state is encoded and evolved continuously in latent space, with decoding only at requested times. We demonstrate the framework on micromagnetic magnetization trajectories from NIST $\mu$MAG Standard Problem 4 and compare it with dissipative-only dynamics and an unstructured latent neural differential equation. Repeated training runs show competitive observed-horizon accuracy, while selected longer rollouts illustrate improved retention of approximately stationary states relative to the unstructured baseline. These results motivate scalar-generated latent dynamics as an inductive bias for reduced-order prediction and other learning tasks where bounded evolution and equilibrium behavior are important.
Read the full abstract

Rings and Algebras

Factorization of central polynomials over division algebras

The factorization problem is a classical topic in the theory of polynomial rings. In this paper, we obtain an irreducibility criterion for central polynomials over division algebras. An example demonstrating the computation of factors for a central polynomial is also presented.

Artificial Intelligence

Towards a universal language of concepts: A survey

Humans can learn and generalize novel concepts from sparse data because they express knowledge in rich structural formats. In this paper, we propose that programs are a strong candidate for universal representation of concepts. We review computational models of concept learning that use programs as their concept representation and evaluate their contribution toward a universal representational language.
Read the full abstract

Chemical Physics

Learning Memory-Dependent Neural Network Correlation Potentials for Accurate Electron Dynamics

Time-dependent density functional theory, though exact in theory, is in practice applied in an adiabatic approximation using exchange-correlation functionals with only local temporal dependence. Simultaneously, the exact correlation potential formally depends on, among other quantities, the time-history of electron densities. Here we develop a framework to learn neural network models of correlation functionals that feature explicit memory-dependence. Our framework includes two different approaches, both linked through their use of adjoint-based optimization. One approach decouples the learning of the functional from inversion to find ground truth values of the correlation potential. The other approach learns the functional directly without requiring inversion. We apply these methods to modeling the electron dynamics of two-electron systems in two spatial dimensions. In both cases, our methods yield correlation functionals with low test set propagation error, outperforming standard local density and generalized gradient approximation functionals by one to two orders of magnitude. Overall, our framework points at one strategy to move beyond the adiabatic approximation and develop memory-dependent correlation functionals that yield accurate propagation for excited state and/or non-equilibrium dynamics.
Read the full abstract

Computation and Language

Scale-QLoRA: Code-Invariant Adapter Merging for Native 4-bit Microscaling LLMs

Merging a LoRA adapter into its base model is standard deployment practice: it removes the runtime adapter's per-forward overhead and leaves a single standalone checkpoint any serving stack can load. On a native 4-bit microscaling checkpoint (NVFP4, MXFP4) that step stops being free. The merged weights must be written back through a quantizer, which re-derives the checkpoint's discrete E2M1 code plane (roughly 90% of the artifact's bytes), so the deployed artifact becomes coupled to one quantization convention, and every later code-touching event in its lifecycle can move it. Done naively the step is worse than fragile: it deletes the adaptation, by up to 39 pp, because against an already-on-grid base the reconstruction optimum is that base. Scale-QLoRA instead adapts only the native per-block scale field, trains those scales on the deployment grid, and freezes every E2M1 code. Within a fixed native format, scale grid, block layout and code plane, merging is then a bit-exact identity and the merged artifact is code-invariant. Across four models and four tasks, Scale-QLoRA and merge-aware QAT-LoRA are both accuracy-lossless, so we claim no accuracy ordering between them; they differ structurally, in that QAT-LoRA re-derives the code plane through a quantizer while Scale-QLoRA preserves it exactly. That difference is what the lifecycle prices: nearest-rounding implementations disagree by about a point on the measured task, and more extreme rule mismatches can drive the weight-space artifact to ~0%, which we report as a sensitivity bound rather than a deployment frequency. Preserving the code plane also drops the weight-space straight-through estimator from training (3.9x per step on the dense 8B model) and enables exact rollback, code-plane deduplication, and a ~125x faster scale-only task swap.
Read the full abstract

All authors Tung-Ling Li, Jiale Huang, Lee-Chi Wang, Janaki Ram Gotei

Audio and Speech Processing

Discriminative Flow Matching: Beyond Time-Conditioning in Generative Restoration via Flow-State Representations

Existing Conditional Flow Matching (CFM) formulations describe transport progress using an explicit interpolation coordinate, commonly interpreted as time, assuming that a single global variable adequately represents a sample's position along the generative trajectory. In restoration tasks, however, transport progress is sample-dependent because the initial distribution may exhibit varying statistical dependencies with the target distribution. Thus, samples at the same interpolation coordinate can differ substantially in degradation level, distance to the target distribution, and restoration difficulty. We investigate whether signal representations learned by discriminatively trained models provide a meaningful description of generative transport state in CFM-based restoration. Through systematic latent-space analysis, we show that discriminative representations organize according to degradation severity and follow a consistent trajectory toward the clean-data manifold during generation. Motivated by these observations, we introduce the Discriminative Flow-State Hypothesis, which posits that discriminative representations encode a transport state governing generative restoration. Based on this hypothesis, we propose Discriminative Flow Matching, which conditions the Flow-Matching velocity field on Discriminative Flow-State Representations rather than explicit time coordinates. Experiments on speech enhancement and image denoising show that these representations characterize restoration progress, enable adaptive inference, and consistently outperform CFM and diffusion-related baselines. Our findings suggest that discriminative representations provide an effective state-aware alternative to explicit time conditioning and offer a novel perspective on the relationship between discriminative and CFM-based generative modeling.
Read the full abstract

Quantum Physics

DPRQ: A Dynamic Programming-based Qubit Routing Algorithm for Collective Communication in Distributed Quantum Computing

Distributed quantum computing (DQC) offers a promising approach to scale quantum computing by overcoming the resource limitations of a single quantum processor. However, inter-node communication remains a major bottleneck of DQC due to inefficient and error-prone entanglement distribution. Optimizing inter-node communication can not only reduce the amount of entanglement resource needed to execute a quantum circuit but also improve execution speed and accuracy of the results. This paper proposes DPRQ, a qubit routing algorithm for minimizing inter-node communication in distributed quantum circuits divided into collective communication blocks. Unlike current approaches that utilize greedy block-level qubit routing strategies, DPRQ employs a dynamic programming-based technique focused on global circuit-level optimization, while capturing inter-block dependencies. We evaluated DPRQ on four sets of quantum circuits and a variety of DQC configurations. The results demonstrate that DPRQ's innovative routing strategy achieves an average of 24.40% reduction with a maximum of 85.06% reduction in inter-node communication, when compared to the state-of-the-art collective communication-based DQC compiler QuComm.
Read the full abstract

Artificial Intelligence

MaxKernel: Agentic Kernel Generation for TPUs

Designing and authoring high-performance custom kernels for accelerators is a complex task that requires deep hardware-level expertise. Large Language Models (LLM) can be leveraged together with real-time compiler feedback to build agentic systems for kernel generation. In this work, we present MaxKernel, a multi-agent system that implements three distinct paradigms for TPU kernel development: (1) a Human-in-the-Loop (HITL) agent for collaborative, step-by-step design; (2) an Autonomous (Auto) agent that executes a fully automated, metric/trace-driven optimization loop; and (3) a Graph-Based Autonomous Search that scales the Auto agent for global exploration of the design space. All three paradigms leverage a shared pool of specialized sub-agents to handle planning, implementation, self-debugging, testing, and hardware profiling. We evaluate MaxKernel on JaxBench, a comprehensive suite of 50 diverse kernel tasks for TPUs, alongside complex, real-world workloads from state-of-the-art open-source models. We demonstrate that MaxKernel consistently generates highly optimized implementations, matching expert hand-tuned baselines and delivering significant performance across the benchmark. Our agent is open-sourced and available https://github.com/AI-Hypercomputer/accelerator-agents/tree/main/MaxKernel.
Read the full abstract

All authors Shangkun Wang, Nina Cai, Charles Hoong, Julian Walker, Gerson Kroiz, George Vanica, Deepak Patil, Andi Gavrilescu, Hassan Sipra, Sethu Sankaran

Cryptography and Security

Hoss: Fast Oblivious Semantic Search with Heterogeneous GPU-CPU-TEE Architecture

Semantic search is widely deployed in modern AI systems, but protecting both data contents and access patterns remains challenging. The current state-of-the-art system, Compass, achieves oblivious semantic search by building an optimized ORAM over HNSW graphs. However, even with aggressive optimizations, it still incurs large overheads. Closing this performance gap is fundamentally difficult: Compass has already removed most cryptographic overheads, leaving ORAM accesses as the dominant cost, which are constrained by well-known Omega(log N) bandwidth lower bounds. Our key insight is that traditional ORAM overhead stems from the assumption of limited private memory, whereas modern GPU TEEs provide large private memory (Pmem) that blinds internal access patterns (Hunt et al., NSDI '23). This shift opens a new design space. We therefore propose Hoss, a first-of-its-kind oblivious semantic search system with a heterogeneous CPU-GPU TEE architecture that supports fast, scalable search with low cost of ownership. In Hoss, the GPU TEE's large Pmem hosts the hot-path HNSW traversal, while the lower layers of the graph, if they exceed GPU capacity, are offloaded to CPU TEEs. The system invokes oblivious primitives only when accessing these lower layers. The availability of large Pmem also enables new optimization opportunities. For example, Hoss features a host-access ORAM mechanism that goes beyond traditional performance constraints and incorporates several data-dependent optimizations that are not possible in prior designs. We implement a prototype of Hoss and benchmark it against Compass. Our results show that Hoss achieves up to 67x speedup while maintaining high recall, with larger gains at scale.
Read the full abstract

All authors Jianzhang Du, Weijie Huang, Chenghong Wang, Nicolas Tsagareli, Yukui Luo, XiaoFeng Wang, Zhongshu Gu

Data Structures and Algorithms

A Polymatroidal Perspective on Random Contraction

Karger's elegant random contraction algorithm for finding a global mincut in a graph has been highly influential. More recent work has obtained several different (nonuniform) random contraction algorithms for mincut in hypergraphs and hedgegraphs. Motivated by the conceptual goal of understanding these algorithms in a unified fashion, we study random contraction algorithms for finding a minimum quotient of a polymatroid. We introduce the notion of quotient-bounded polymatroids and show that several existing results can be derived and understood under a common algorithmic framework for quotient-bounded polymatroids.
Read the full abstract

Physics and Society

Crowding controls the scaling of bus frequency with demand

Cities must allocate limited resources to maintain mobility, with uncertainties about the resulting state of the system. Analyzing roughly 3,000 bus routes with more than 4 billion yearly riders across 19 metropolitan areas worldwide, we uncover a robust scaling law of the form $f \sim (d/t)^\alpha$ with exponent $\alpha \in [1/2,\,2/3]$, linking the service frequency $f$ to passenger demand $d$ and route duration $t$. We show that this scaling emerges from a simple optimization principle: cities implicitly minimize total passenger waiting time under a fixed operational budget when both schedule frequency and crowding are taken into account. This mechanism produces two universal regimes: a frequency-dominated regime with $\alpha = 1/2$ when crowding is negligible, and a capacity-dominated regime with $\alpha = 2/3$ when most routes are overloaded. Intermediate exponents arise when only part of the network operates near capacity. Furthermore, we find that the benefits of additional investment are highly uneven across systems. For instance, our model suggests that a $20\%$ budget increase yields nearly a 5-minute reduction in daily waiting time per passenger in Boston, compared to only about 1 minute in Paris. These findings place urban transit within a broader class of constrained capacity-allocation problems, while highlighting a distinct regime in which prescribed route demands shape the allocation of limited service resources. The resulting scaling laws show how simple optimization principles can generate systematic exponents in complex transport systems, beyond the dissipation-based frameworks usually considered in physical and biological flow networks.
Read the full abstract

All authors Siddharth Patwardhan, Şirag Erkol, Filippo Radicchi, Marc Barthelemy

Fluid Dynamics

Quasi-Two-Dimensional Simulation of a Rotating Detonation Engine Combustor and Injector

A numerical simulation of an annular rotating detonation engine with stoichiometric hydrogen-oxygen is performed. A generic, well posed, and easily implemented approach using a quasi-two-dimensional method to model the area variations through the rotating detonation engine's injector and combustor is presented. The detonation–injector interaction is studied for the case with a ratio of four between the combustor and injector's throat areas. A shock wave is formed in the divergent portion of the injector due the high back pressure created by the detonation in the combustor. A Favre-averaged steady-state analysis of stream lines and particle paths reveals that the shock causes an irrecoverable loss of stagnation pressure. Stagnation pressure gain in the combustor is insufficient to make up for the loss and the flow leaves the engine with lower stagnation pressure than in the plenum.
Read the full abstract

2609.04519 · 3 Sep · This is the authors' version of an article submitted to the Journal of Propulsion and Power on 2021-06-10, resubmitted on 2023-04-04, revised 2023-04-13, 2023-07-10, accepted 2023-07-16, and published online 2023-09-25

Artificial Intelligence

Multi-Harness Reinforcement Learning for Coding Agents: Training Gains and Harness Dependence

Recent systems train coding agents with reinforcement learning (RL) on trajectories collected through multiple execution harnesses. However, gains under the training harnesses leave open whether the improvements transfer to unseen interfaces and how they depend on reward normalization. We present a controlled study that measures RL gains under fixed harnesses and compares the same models across harnesses. We train two Qwen3-8B RL variants from the same supervised fine-tuning (SFT) checkpoint using identical replay data. They differ only in whether rewards for each task are normalized within each harness (Within) or across harnesses (Cross). On SWE-bench Verified, we identify three main findings: (i) Multi-harness RL yields modest gains over the SFT baseline under both normalization rules. Across the four training harnesses, the estimated mean improvements are +0.48 and +0.77 percentage points for Within and Cross, respectively. (ii) Cross increases the fraction of trajectories with nonzero advantage but provides no consistent advantage over Within on an unseen harness across training seeds and evaluation budgets; neither rule significantly improves over SFT on this harness. (iii) Performance varies substantially more across harnesses than with RL training. After averaging over model variants, the highest and lowest source-harness solve rates differ by 7.13 percentage points, far exceeding the estimated training gains. Even after multi-harness RL, the observed performance differences across execution interfaces remain substantially larger than either the gains from training or the differences between normalization rules.
Read the full abstract

All authors Chenqian Le, Jiayi Cheng, Qijia He, Runhao Li, Yinghao Li, Xupeng Chen

Materials Science

Calculation of DFT Spin-Orbit Spillage with Quantum ESPRESSO

This work describes the calculation of spin-orbit spillage from a crystal structure. Spin-orbit spillage provides a measure of the likelihood that a material has topological character. The spillage also provides the reference quantity for the machine-learning classifier of Choudhary et al., which predicts whether the spillage exceeds a specified threshold rather than the calculation of its numerical value directly. The complete computation workflow was applied to the insulating compound BaMg2Bi2, yielding a spillage of 2.094 compared with the published VASP spillage of 2.075, corresponding to a difference of 0.9%. The calculation is described in terms of two Quantum ESPRESSO (QE) calculations of spillage performed with and without spin-orbit coupling, the role of relativistic pseudopotentials, and the subsequent wavefunction-overlap analysis. The limitations of the same calculation procedure for semimetals are also examined.
Read the full abstract

Sound

Pitch-class Steering for Diffusion-based Music Generation via Latent-space Probes

Recent work on controllable music generation has focused on autoregressive models, leaving diffusion-based systems comparatively underexplored. We present a lightweight method for steering the pitch content of audio produced by Stable Audio Open, a latent diffusion model for music synthesis. A small convolutional probe containing approximately 125k parameters is trained to decode frame-level pitch-class activations from the model's variational autoencoder latent space, using paired audio and MIDI data. At inference time, the frozen probe serves as a differentiable loss function: its gradient with respect to the denoising latent is used to nudge generation toward a user-specified pitch-class sequence, requiring no retraining or architectural modification of the base model. Across 27 evaluation trials spanning 9 text prompts and 3 target melodies, probe-guided generation increases melodic coherence by 2.4x over the unguided baseline (p < 1e-5, Wilcoxon signed-rank test), demonstrating that musically meaningful structure is both recoverable and steerable in diffusion-based music latent spaces.
Read the full abstract

Fluid Dynamics

Coherent-structure dynamics in wall turbulence from state-space trajectories

We identify dynamical processes in the near-wall region of turbulent wall-bounded flow by representing coherent-structure energy as trajectories in a low-dimensional state space and analyzing recurring trajectory patterns using network motifs. Direct numerical simulations (DNS) are performed for three configurations: a minimal flow unit (MFU) at $Re_\tau \approx 180$ to isolate the self-sustaining process (SSP), a full-scale channel at the same Reynolds number to study interactions between near-wall structures, and an MFU at $Re_\tau \approx 2200$ to investigate near-wall/outer-layer coupling. Proper orthogonal decomposition (POD) is used to identify modes corresponding to streaks, rolls, and meandering structures of the SSP, and the flow is projected onto these modes to track their energy over time. Motif analysis then identifies statistically significant dynamical pathways in the resulting state-space trajectories. Several motifs are common to all three configurations, indicating robust near-wall dynamics across different background flows. While motifs associated with the classical SSP are recovered, equally prominent motifs with no direct SSP analogue are also identified, demonstrating that preferred near-wall dynamics extend beyond the canonical regeneration cycle. Conditioning the $Re_\tau \approx 180$ MFU on outer-layer energy further shows that quiescent outer-layer states produce dynamics resembling those of the low-Reynolds-number full channel, whereas energetic outer-layer states promote high-energy bursting events. These results demonstrate that near-wall turbulence evolves along preferred dynamical pathways whose prevalence is modified, but not eliminated, by outer-layer activity.
Read the full abstract

Materials Science

Triboelectrification of a dense metal-organic framework for resilient mechanical energy harvesters

Triboelectric nanogenerators (TENGs) incorporating metal-organic frameworks (MOFs) have largely been designed around porous architectures, based on the assumption that high internal surface area is the primary driver for triboelectric enhancement. Herein, we demonstrate that a dense, nominally nonporous MOF called ZIF-zni can instead function as an effective high-loading filler within a tribopositive polyurethane (PU) matrix, offering a design strategy for harnessing interfacial electromechanical effects. A 20 wt% ZIF-zni@PU composite delivers an output voltage of 470+/-15 V and a peak power density of 1.31+/-0.03 W m-2 against polydimethylsiloxane (PDMS). The device exhibits stable performance over ~94,000 cycles under a contact force of ~100 N, and it remains operational under higher impact beyond 500 N. This concept enables the demonstration of a proof-of-concept triboelectric floor tile. Combined experimental and theoretical studies indicate that the performance enhancement arises from favourable interfacial polarization, reduced dielectric screening, and surface accessible ZIF-zni rich domains, rather than porosity alone. These features are accompanied by reduced work of adhesion and modified surface roughness, hence improving contact electrification. These findings establish dense MOFs as an effective triboelectric filler and identify interfacial electronic structure and polarization as key design parameters for engineering mechanical energy harvesters and self-powered sensors.
Read the full abstract

All authors Chuzhan Jin, Debayan Mondal, Jiahao Ye, Jin-Chong Tan