arXiv subjects

Jing Leng

Publications and source records attributed to Jing Leng.

6 recordsLinked to original sources

arXiv2026

Entropic lattice Boltzmann method for general anisotropic advection--diffusion

Many transport processes exhibit direction-dependent diffusion, described macroscopically by the full-tensor anisotropic advection--diffusion equation (ADE). Numerical discretization is demanding when the principal axes are rotated relative to the mesh, since mixed derivatives and oblique fluxes amplify grid-orientation errors under large tensor contrasts. This paper develops a local entropic lattice Boltzmann discretization for the general anisotropic ADE. The non-equilibrium population is split into a first-order flux sector and a residual ghost sector. The diffusion tensor is imposed through local tensorial relaxation of the flux, while higher-order kinetic content is controlled by an ADE-corrected entropic stabilizer with positivity fallback. Chapman--Enskog analysis shows the scheme recovers the target full-tensor equation with a discrete-time diffusivity relation between the physical tensor and the flux-relaxation matrix. The update is local, matrix-free, and applies to rotated, spatially varying, heterogeneous, and dynamically coupled tensor transport. We validate it on 3D benchmarks--advected Gaussian plumes, decay of rotated Fourier modes, and source-driven transport with varying tensors--covering off-diagonal diffusion, high-P\'eclet advection, anisotropy ratios of O(104)O(10^4) O(104), and local contrasts up to $3\times10^4:1$. It is then applied to orientation-induced Taylor dispersion of Brownian rods, quantifying enhancement from shear-driven rotation. Heat-conduction tests include rotated thermal-conductivity measurements and effective conduction in heterogeneous porous media with anisotropy up to $10^4:1. Finally, anisotropic Rayleigh--B\'enard convection is simulated to examine how plume morphology and heat transfer change over seven decades of anisotropy ratios, demonstrating an accurate, stable local solver for strongly anisotropic advection--diffusion.

physics.flu-dyn↗

arXiv2026

HACHIMI: Scalable and Controllable Student Persona Generation via Orchestrated Agents

Student Personas (SPs) are emerging as infrastructure for educational LLMs, yet prior work often relies on ad-hoc prompting or hand-crafted profiles with limited control over educational theory and population distributions. We formalize this as Theory-Aligned and Distribution-Controllable Persona Generation (TAD-PG) and introduce HACHIMI, a multi-agent Propose-Validate-Revise framework that generates theory-aligned, quota-controlled personas. HACHIMI factorizes each persona into a theory-anchored educational schema, enforces developmental and psychological constraints via a neuro-symbolic validator, and combines stratified sampling with semantic deduplication to reduce mode collapse. The resulting HACHIMI-1M corpus comprises 1 million personas for Grades 1-12. Intrinsic evaluation shows near-perfect schema validity, accurate quotas, and substantial diversity, while external evaluation instantiates personas as student agents answering CEPS and PISA 2022 surveys; across 16 cohorts, math and curiosity/growth constructs align strongly between humans and agents, whereas classroom-climate and well-being constructs are only moderately aligned, revealing a fidelity gradient. All personas are generated with Qwen2.5-72B, and HACHIMI provides a standardized synthetic student population for group-level benchmarking and social-science simulations. Resources available at https://github.com/ZeroLoss-Lab/HACHIMI

cs.CL↗

arXiv2025

Algorithms for Stable Roommate with Externalities

In the roommate matching model, given a set of 2n agents and n rooms, we find an assignment of a pair of agents to a room. Although the roommate matching problem is well studied, the study of the model when agents have preference over both rooms and roommates was recently initiated by Chan et al. [11]. We study two types of stable roommate assignments, namely, 4-person stable (4PS) and 2-person stable (2PS) in conjunction with efficiency and strategy-proofness. We design a simple serial dictatorship based algorithm for finding a 4PS assignment that is Pareto optimal and strategy-proof. However, the serial dictatorship algorithm is far from being 2PS. Next, we study top trading cycle (TTC) based algorithms. We show that variations of TTC cannot be strategy-proof or PO. Finally, as Chan et al. (2016) showed that deciding the existence of 2PS assignment is NP-complete, we identify preference structures where a 2PS assignment can be found in polynomial time.

cs.GT↗

arXiv2024

Precise Drive with VLM: First Prize Solution for PRCV 2024 Drive LM challenge

This technical report outlines the methodologies we applied for the PRCV Challenge, focusing on cognition and decision-making in driving scenarios. We employed InternVL-2.0, a pioneering open-source multi-modal model, and enhanced it by refining both the model input and training methodologies. For the input data, we strategically concatenated and formatted the multi-view images. It is worth mentioning that we utilized the coordinates of the original images without transformation. In terms of model training, we initially pre-trained the model on publicly available autonomous driving scenario datasets to bolster its alignment capabilities of the challenge tasks, followed by fine-tuning on the DriveLM-nuscenes Dataset. During the fine-tuning phase, we innovatively modified the loss function to enhance the model's precision in predicting coordinate values. These approaches ensure that our model possesses advanced cognitive and decision-making capabilities in driving scenarios. Consequently, our model achieved a score of 0.6064, securing the first prize on the competition's final results.

cs.CV↗

arXiv2019

Taming Distrust in the Decentralized Internet with PIXIU

Decentralized Internet is booming. People are fascinated by its promise that users can truly own their data. However, in a decentralized Internet, completing a task usually involves multiple nodes with mutual distrust. Such distrust might eventually become a major obstacle for the growth of the decentralized Internet. In this paper, we analyze the distrust using a simple model and highlight the properties required to faithfully accomplish one task in a decentralized Internet. We also introduce our draft solution -- PIXIU, a framework to mitigate the distrust among different nodes. In PIXIU, we design and utilize trust-λ and decentralized executor to achieve the above-needed properties.

cs.CR↗

arXiv2014

An Iterative Minimization Formulation for Saddle-Point Search

This paper proposes and analyzes an iterative minimization formulation for search- ing index-1 saddle points of an energy function. This formulation differs from other eigenvector-following methods by constructing a new objective function near the guess at each iteration step. This leads to a quadratic convergence rate, in comparison to the linear case of the gentlest ascent dynamics (E and Zhou, nonlinearity, vol 24, p1831, 2011) and many other existing methods. We also propose the generalization of the new methodology for saddle points of higher index and for constrained energy functions on manifold.

math.NA↗