arXiv subjects

Yuan He

Publications and source records attributed to Yuan He.

At least 19 recordsLinked to original sources

arXiv2026

Equilibrium-energy-scaled quantum lattice Boltzmann method for variable-velocity linear advection--diffusion equations

We develop a quantum lattice Boltzmann method for linear advection--diffusion equations with prescribed space- and time-dependent velocities. Direct block-encoding can accumulate normalization factors that suppress success probability exponentially, even when the classical propagator remains bounded. We control this loss by scaling each distribution component by the inverse square root of its local equilibrium weight. For uniformly positive weights and dimensionless relaxation times $τ^*\geq 1/2$, this equilibrium energy scaling gives the Bhatnagar--Gross--Krook collision operator unit spectral norm while preserving the discrete dynamics. Streaming normalization depends on weight ratios at the departure and arrival positions and times of each step; these ratios equal one for uniform steady advection. For periodic streaming, cumulative normalization stays bounded under mesh refinement at fixed physical time if logarithmic derivatives of the weights along streaming trajectories are bounded uniformly in space, time, and mesh resolution. A single-step block-encoding with two flag qubits supports coherent propagation of the full kinetic distribution without intermediate measurements. We compare oracle query bounds for time marching with those for a history-state linear system built from the same scaled operator, under specified assumptions on initial-state preparation and coefficient access. Tests in one and two dimensions and noiseless circuit simulations verify the construction. In a controlled comparison after 50 steps, scaling increases the probability of preparing the quantum state encoding the terminal distribution by approximately $450\times$. Classical refinement diagnostics up to $N_t=16{,}384$ steps confirm bounded cumulative normalization. The scaling controls additional normalization losses while retaining physical diffusive attenuation.

math-ph↗

arXiv2026

Reciprocity formulas for generalized Dedekind character sums

In this paper, we apply the Fourier series technique to establish a new formula for the product of the generalized Bernoulli function and the Bernoulli function. As applications, we derive some reciprocity formulas for generalized Dedekind character sums. In particular, we show that the reciprocity formulas of Berndt, Can, and others follow as special cases.

math.NT↗

arXiv2026

Forecast-Dojo: Replayable Environments for Benchmarking and Training LLM Forecasting Agents

Forecasting requires large language model (LLM) agents to gather evolving evidence, revise their beliefs, and make calibrated predictions over time. Yet evaluating and improving these capabilities is challenging because real-world events unfold continuously and outcomes may take months to resolve. We introduce Forecast-Dojo, a replayable environment that reconstructs resolved prediction-market events as sequences of dated forecasting steps, enabling reproducible evaluation, longitudinal analysis, and training from agent interactions. It comprises 1,568 Polymarket events split temporally into training and evaluation sets, alongside 18.8 million dated news articles. Across 12 models, research tools consistently improve forecasting performance, and greater capture of new evidence is associated with larger improvements over time. Nevertheless, all models trail historical market forecasts, while carrying a belief notebook across dates reduces research costs without consistently improving forecast quality. We supervise fine-tuned Qwen3-30B-A3B-Thinking-2507 on collected trajectories from Qwen3-235B-A22B-Thinking-2507 to demonstrate the environment's training utility. The student's accuracy is improved from 34.8% to 42.8% on the evaluation set.

cs.AI↗

arXiv2026

Experimental determination of the double $K$-shell vacancy creation probability in the electron-capture decay of $^{55}$Fe with a microcalorimeter

The probability of double $K$-shell vacancy production following electron-capture decay is determined for $^{55}$Fe using a high-resolution microcalorimeter. The $K$-hypersatellite lines arising from double $K$-shell-vacancy filling are directly resolved within the measured spectra of X-ray emissions which are dominated by Mn $K$-X-ray diagram lines. The energies of the Mn \textit{K}-hypersatellite lines, $Kα_2^h$, $Kα_1^h$, and $Kβ_{1,3}^h$, are measured as $(6143.5\pm0.5)$ eV, $(6162.1\pm1.5)$ eV and $(6811.9\pm2.6)$ eV, respectively. The relative intensities of \textit{K}-hypersatellite to diagram X-rays are measured to be $(1.30 \pm 0.07) \times 10^{-4}$. From the relative intensities together with established fluorescence yield values, the probability of double-vacancy creation per $K$-shell capture ($P_{KK}$) is determined as $(1.22 \pm 0.09) \times 10^{-4}$. Comparisons with previous experimental results and theoretical calculations are discussed.

nucl-ex↗

arXiv2026

AlphaBet: Reinforcement Learning for LLM Event Forecasting Agents

Forecasting an unresolved event requires finding relevant evidence and deciding how strongly it supports each outcome. Yet outcome-based training of Large Language Model (LLM) forecasters often begins with preassembled evidence at a single prediction date, leaving evidence acquisition outside the learning loop. We introduce AlphaBet, an agentic reinforcement learning (RL) framework that trains a single LLM policy to search, reason, and report probabilities across multiple pre-resolution dates. Each dated forecast is scored by a proper scoring rule against the eventual outcome, allowing one resolved event to supervise decisions under different information states. We decompose forecasting regret into an evidence-acquisition gap and a probability-reporting gap, and show that Group Relative Policy Optimization (GRPO)'s standard-deviation normalization can distort the reporting target under multi-class proper scores. On GLM-4.5-Air, AlphaBet reduces the Brier score from 0.731 to 0.626 on Forecast-Dojo and from 0.612 to 0.528 on FutureX. Controlled ablations attribute these gains to multi-date training and show that they come from improvements in both information acquisition and probability reporting. Our empirical results validate that GRPO's std normalization degrades proper-scoring-rule performance while leaving top-outcome accuracy largely unchanged.

cs.LG↗

arXiv2026

Efficient Provably Private Classification with a Tabular Foundation Model

Tabular data underpin prediction and decision-making in medicine, finance, government and science, but often contain sensitive individual-level information, creating a need for accurate prediction while preserving privacy. Traditional private learning provides formal privacy guarantees, but requires slow dataset-specific optimisation, suffers substantial utility loss under strong privacy, and is often difficult to apply correctly. Tabular foundation models adapt rapidly to new datasets, but existing models lack formal privacy guarantees, and are highly vulnerable to membership-inference attacks, limiting their use on sensitive data. Here we introduce PrivTab, an easy to use tabular foundation model for differentially private classification that embeds a privacy mechanism within its architecture. Pretrained on simulated datasets, PrivTab uses in-context learning to transform sensitive rows into compact, provably private summaries---effectively learning how to learn under privacy. PrivTab outperforms private linear and neural-network baselines under moderate-to-strong privacy, shows negligible membership leakage, maintains well-calibrated predictions under strong privacy, and reduces dataset fitting time by 10,000 times, requiring only a single forward pass. By combining formal privacy, speed, and easy of use, PrivTab brings recent advances in AI to applications where sensitive individual-level data have limited their adoption.

cs.LG↗

arXiv2026

The Shape of Speed: Impacts of Partition Geometry and Rank Density in Distributed Quantum Circuit Simulations

In distributed quantum circuit simulation, a poorly shaped partition can halve performance before computation begins. Evaluation on Fugaku across 764 validated configurations (twelve algorithms, thirteen torus partition geometries, and six rank densities for 39-qubit simulations on 1,024 nodes) shows that partition geometry dominates runtime. All twelve algorithms run 1.73-2.31x slower on flat partitions than on near-cubic ones despite identical data transfer, proving the slowdown stems from network delivery rather than communication volume. This penalty scales with the 3D torus partition aspect ratio (runtime $\propto a^{0.39}$, $r = 0.72$). Rank density is secondary, cutting runtime by 11% at 16 ranks per node only on compact geometries. Ultimately, requesting a near-cubic partition with 16 ranks per node roughly halves time-to-solution relative to flat partitions, which also consume 1.82x more energy. A simulator-free all-to-all microbenchmark confirms a similar geometry penalty for collective-dominated workloads.

cs.ET↗

arXiv2026

Rufus-Air: An Open LLM Post-Training Recipe

Rufus-Air is an open and reproducible post-training recipe on GLM-4.5-Air-Base (106B-A12B), organized as a serial pipeline of eight stages: SFT, Reasoning RL, Coding RL, Instruction-Following RL, General Agent, Coding Agent, Search Agent, and RLHF. We document the data, reward design, infrastructure, stage order, and stagewise results needed to reproduce the recipe. Stages progress from basic to advanced capabilities and from hard, verifiable rewards to softer judge-based signals. Training builds on open-source components and public data, much of it used as released, without new human annotation or an in-house distillation teacher. Our main findings are that (i) diverse, high-quality SFT establishes a strong capability floor; (ii) difficulty filtering keeps RL prompts within a productive learning range; (iii) reward reliability provides a practical principle for ordering stages; and (iv) infrastructure and engineering choices are part of the recipe, not just an implementation detail. Rufus-Air improves over the official GLM-4.5-Air post-trained release and is competitive with similarly sized open models.

cs.CL↗

arXiv2026

Neutron Double-Differential Cross Sections for Spallation Reactions from an ANN Model

In this paper, we present a data-driven artificial neural network (ANN) model for describing the double differential cross sections (DDCS) of neutron emission in nuclear spallation reactions. The ANN model is found to be precise, flexible, and efficient in predicting differential cross sections of nuclear reactions and in learning the complex dependence of neutron DDCS on the projectile energy ($T_p$), target nucleus ($A$ and $Z$), neutron energy ($T_n$), and neutron emission angle ($θ_n$). The model is trained on replicas of experimental data that incorporate uncertainties. Several regularization schemes are examined during ANN training. The input variables of the constructed ANN framework are also investigated, and the following six key variables are selected for the input layer of the ANN model: $θ_{n}$, ${\rm log}(T_n/T_p)$, $T_n/T_p$, ${\rm log}(T_p)$, $A^{2/3}$, and $N/Z$. The ANN predictions are compared with training data provided by various experimental collaborations, showing excellent agreement. The resulting model is further tested on test data with projectile energies, target nuclei, and neutron emission angles different from those in the training data, indicating strong predictive power and generalization capability of the ANN framework. As an illustration, the neutron DDCS as functions of $T_n$, $θ_n$, and projectile energy $T_p$ are predicted and presented for copper target. The proposed high-precision ANN model is expected to be beneficial for accelerator-driven system (ADS) design and many other applications in nuclear physics, astrophysics, and nuclear technology development.

nucl-th↗

arXiv2026

The classification of some polynomial maps in dimension three

In the paper, we classify all polynomial maps of the form $H=(u(x,y),\allowbreak v(x,y,z),h(x,y,z))$ in the case that $JH$ is nilpotent and $°_zv\geq 2°_zh$. Then we give the structure of $H=(u(x,y),v(x,y,z),h(x,y,z))$ if $JH$ is nilpotent and $°_zh\leq 3$.

math.AC↗

arXiv2026

PIMID: A Full-System Simulator with Intricacy and Diversity for Processing-in-Memory

Processing-in-Memory addresses the memory wall by co-locating computation with memory, but because real PIM hardware remains scarce, simulation is the primary way to explore the PIM design space. Yet existing PIM simulators each cover only part of that space: they typically model a single memory technology, fix processing elements at one level of the memory hierarchy, support a single execution model, and stop at the device boundary. We therefore present PIMID, an execution- and trace-driven full-system simulator that closes these gaps in one tool. PIMID supports both the shared-memory and message-passing execution models, running annotated parallel code in OpenMP and MPI side by side across eleven memory technologies (seven DRAM standards, SRAM, and three non-volatile memories); it places PEs anywhere from subarrays to logic dies, sweeps PE count and core-model fidelity, and prices the in-memory network per technology from measured congestion. Its single-process host-device co-simulation resolves an end-to-end time and energy breakdown (host preparation, device compute, and explicit boundary charges) that device-only tools cannot produce. Across the resulting dual-execution-model dataset, PIMID shows that the memory technology alone moves execution time by more than an order of magnitude and that the best host main memory is not the best PIM substrate; that regular kernels scale superlinearly with PE count as in-memory bandwidth co-scales with compute; that graph traversal under message-passing hits a collective-communication wall absent under shared memory; and that at full-system scope the offload trades time for energy only on the bandwidth-class memory: shared-memory offload saves energy on HBM3 while a 16-core host keeps every end-to-end time win. PIMID's plugin interfaces let new engines and models be added through standardized YAML specifications as PIM technology evolves.

cs.AR↗

arXiv2026

Reciprocity formulas for certain generalized Hardy-Berndt sums

In this paper, we introduce the generalized Hardy-Berndt sums. We establish some formulas of products of the Bernoulli and Euler functions by using the Fourier series technique and some properties of the periodic-zeta and Lerch-zeta functions. As applications of the results, we give some reciprocity formulas for the generalized Hardy-Berndt sums. It turns out that Hardy's reciprocity formulas follow as special cases and that Goldberg's three-term reciprocity formulas are recovered.

math.NT↗

arXiv2026

A unification of Euler-Maclaurin and Euler-Boole summation formulas

In this paper, we establish a summation formula associated with the generalized Apostol-Bernoulli functions. This formula unifies the Euler-Maclaurin summation formula, the Euler-Boole summation formula, and the character analogues of these two formulas. We also apply it to obtain the general power sum formula and the special values of Berndt's generalized $L$-function at non-positive integers.

math.NT↗

arXiv2026

On the special values of Berndt's generalized $L$-function

In this paper, we study the generalized $L$-function considered by Berndt (1975). We introduce the generalized Apostol-Bernoulli polynomials and the generalized Apostol-Bernoulli functions, and establish some properties for them, including the Fourier series for these functions. We show that the values of Berndt's generalized $L$-function at integers are explicitly evaluated in terms of the generalized Apostol-Bernoulli functions.

math.NT↗

arXiv2026

On Effectiveness and Efficiency of Agentic Tool-calling and RL Training

Tool-calling is a central component of modern large language model (LLM) agents, equipping them with skills beyond their parametric knowledge. This paper studies tool-calling along two complementary axes: effectiveness, i.e., how this capability is measured, and efficiency, i.e., how it is learned. On effectiveness, we systematically analyze tool-calling evaluation pipelines and show that results can be highly sensitive to seemingly minor, often undocumented implementation choices including the random seed, system prompt, multi-turn template construction, and how prior interaction/reasoning history is carried forward. These choices can lead to substantial differences in reported performance, especially in multi-turn settings where without rigorous standardization, leaderboard rankings are unreliable. On efficiency, we examine standard reinforcement learning (RL) for tool-calling and identify two sources of computational waste: (i) during rollouts, many prompts produce no learning signal, and (ii) during policy updates, optimization incurs high computational cost. Guided by these findings, we introduce two techniques that accelerate RL-based tool-calling training, achieving substantial wall-clock speedup without degrading performance.

cs.LG↗

arXiv2026

Loong: Synthesize Long Chain-of-Thoughts at Scale through Verifiers

Recent advances in Large Language Models (LLMs) have shown that their reasoning capabilities can be significantly improved through Reinforcement Learning with Verifiable Reward (RLVR), particularly in domains like mathematics and programming, where ground-truth correctness can be automatically evaluated. However, extending this success to other reasoning-intensive domains remains challenging due to the scarcity of high-quality, verifiable datasets and the high cost of human supervision. In this work, we introduce the Loong Project: an open-source framework for scalable synthetic data generation and verification across a diverse range of reasoning-intensive domains. The framework consists of two key components: (1) LoongBench, a curated seed dataset containing 8,729 human-vetted examples across 12 domains (e.g., Advanced Mathematics, Chemistry, Logic), each paired with executable code and rich metadata; and (2) LoongEnv, a modular synthetic data generation environment that supports multiple prompting strategies to produce new question-answer-code triples. Together, these components form an agent-environment loop that enables reinforcement learning, where an LLM-based agent is rewarded for generating Chain-of-Thought (CoT) solutions that align with code-executed answers. Empirically, we benchmark LoongBench on a broad suite of both open-source and proprietary LLMs to evaluate domain coverage and reveal performance bottlenecks. In addition, we conduct a comprehensive analysis of synthetic data generated by LoongEnv, examining correctness, difficulty, and diversity. Code and documentation are available at https://github.com/camel-ai/loong.

cs.LG↗

arXiv2026

Spontaneous patterning of cell size on curved surfaces

Tissue surfaces exhibit complex curvature during embryogenesis and oncogenesis. Evidence shows that cells can actively sense curvature to regulate behavior and fate, yet the underlying mechanism remains unclear. Here, we develop a vertex model for arbitrary curved surfaces and uncover spontaneous cell size patterning on ellipsoidal surfaces: cells in high-curvature regions are consistently larger than those in low-curvature regions. This non-uniformity arises from a mechanical competition encoded in Riemannian geometry: positive Gaussian curvature reduces the perimeter-to-area ratio of polygonal cells, relaxing cell-edge tension in high-curvature regions, which is compensated by area expansion to maintain global force balance. This area pattern is robust against variations in model parameters and matches observations in biological systems. The perimeter pattern, in contrast, is governed by competition between the intrinsic geometric tendency and the deformation required by force balance, and undergoes reversal beyond a critical shape index. Together, these findings establish self-organized spatial variations in cell size as a potential physical mechanism for curvature sensing.

physics.bio-ph↗

arXiv2026

Structured Labeling Enables Faster Vision-Language Models for End-to-End Autonomous Driving

Vision-Language Models (VLMs) offer a promising approach to end-to-end autonomous driving due to their human-like reasoning capabilities. However, troublesome gaps remains between current VLMs and real-world autonomous driving applications. One major limitation is that existing datasets with loosely formatted language descriptions are not machine-friendly and may introduce redundancy. Additionally, high computational cost and massive scale of VLMs hinder the inference speed and real-world deployment. To bridge the gap, this paper introduces a structured and concise benchmark dataset, NuScenes-S, which is derived from the NuScenes dataset and contains machine-friendly structured representations. Moreover, we present FastDrive, a compact VLM baseline with 0.9B parameters. In contrast to existing VLMs with over 7B parameters and unstructured language processing(e.g., LLaVA-1.5), FastDrive understands structured and concise descriptions and generates machine-friendly driving decisions with high efficiency. Extensive experiments show that FastDrive achieves competitive performance on structured dataset, with approximately 20% accuracy improvement on decision-making tasks, while surpassing massive parameter baseline in inference speed with over 10x speedup. Additionally, ablation studies further focus on the impact of scene annotations (e.g., weather, time of day) on decision-making tasks, demonstrating their importance on decision-making tasks in autonomous driving.

cs.CV↗