arXiv subjects

Kai Jiang

Publications and source records attributed to Kai Jiang.

At least 37 records · Page 2Linked to original sources

arXiv2026

A Structure-Exploiting Implicit-Explicit Trust Region Method for Computing Second-Order Stationary Points of the Landau-Brazovskii Model

This work focuses on the reliable computation of second-order stationary points in the high-dimensional nonconvex energy landscape of the Landau-Brazovskii (LB) model, a fundamental model for studying phases and phase transitions. For this purpose, we develop an efficient implicit-explicit trust region (IMEX-TR) method. Trust region (TR) methods can avoid saddle-point stagnation and guarantee convergence to second-order stationary points under appropriate conditions. However, their direct application to the LB model has been impractical because the Hessian is dense if treated directly. The proposed IMEX-TR method overcomes this difficulty by exploiting the Hessian's special structure: the linear interaction part is diagonal in reciprocal space, whereas the nonlinear bulk-energy part is diagonal in physical space. Based on this structure, we design an efficient solver for the TR subproblem that with globally convergent guarantee and enjoys FFT-based acceleration, with $\mathcal O(N \log N)$ complexity per iteration. Existing first-order gradient-based methods for the LB model only guarantee convergence to first-order stationary points and may stagnate at saddle points. In contrast, the proposed IMEX-TR method inherits the theoretical guarantee of converging to second-order stationary points while remaining computationally practical. Numerical experiments verify the theoretical properties of the algorithm and demonstrate its robustness in locating stable phases from different initial conditions. Numerical results also show that IMEX-TR can escape unstable stationary states reached by first-order schemes and converge to physically meaningful second-order stationary points. These results suggest that targeting second-order stationary points provides an effective computational paradigm for exploring complex free-energy landscapes and identifying stable or metastable states.

math.NA↗

arXiv2026

Projected Sobolev Natural Gradient Descent for Efficient Neural Network Solution of the Gross--Pitaevskii Equation

This paper introduces a projected Sobolev natural gradient descent (NGD) method for computing ground states of the Gross--Pitaevskii equation. By projecting a continuous Riemannian Sobolev gradient flow onto the normalized neural network tangent space, we derive a discrete NGD algorithm that preserves the normalization constraint. The numerical implementation employs variational Monte Carlo with a hybrid sampling strategy to accurately account for the normalization constant. To enhance computational efficiency, a matrix-free Nyström-preconditioned conjugate gradient solver is adopted to approximate the NGD operator without explicit matrix assembly. Numerical experiments demonstrate that the proposed method converges significantly faster than physics-informed neural network approaches and exhibits near-constant wall-clock time and memory over the tested high-dimensional range when the hidden architecture is fixed and the implicit Gram operator is used. Moreover, the resulting neural-network solutions provide high-quality initial guesses that substantially accelerate subsequent refinement by traditional high-precision solvers.

math.NA↗

arXiv2026

Vidu S1: A Real-Time Interactive Video Generation Model

We introduce Vidu S1, a real-time interactive video generation model supporting voice control of digital characters. Users can control video generation content at any moment through voice instructions. Vidu S1 supports infinite-length real-time video generation without blurring, drift, or visual distortion. Built with TurboDiffusion and TurboServe, Vidu S1 outputs 540p real-time videos at up to 42 FPS on regular consumer GPUs. Users can upload custom images of real people, anime, and pets, and choose different voice tones for personalized experiences. Experiments show that Vidu S1 achieves the best performance across all test metrics while fully meeting real-time inference requirements. A playable online demo is available at https://vidu.com/vidu-stream.

cs.CV↗

arXiv2026

Miles: Metric Learning with Expandable Subspace for Pre-Trained Model-Based Class-Incremental Learning

Class Incremental Learning (CIL) aims to learn new concepts consistently from a data stream without forgetting. Unlike typical CIL methods which need to learn a model from scratch, pre-trained model (PTM) can easily adapt to a new task with fine-tuning. However, existing PTM-based CIL methods fail to achieve a trade-off between performance and computational expenditure, i.e., they either adopt the same parameter space so that leading catastrophic forgetting, or expand a new branch for each task but adding more computational cost. To this end, we propose MetrIc Learning with Expandable Subspace (Miles) to harness the prior information within pre-trained knowledge, thereby orchestrating an efficient expansion of the parameter space through guided optimization. Specifically, it decouples the learnable modules with the pre-trained model, exploiting prior information from intermediate features of the backbone network to enable more flexible parameter expansion. Then, a central loss is adopted to guide the new category to cluster towards the corresponding prototype in the new task subspace while incorporating an auxiliary distance regularization term to maintain metric equilibrium across tasks. Extensive experiments on six benchmark datasets demonstrate that Miles achieves state-of-the-art performance in various CIL settings.

cs.CV↗

arXiv2026

MobileSAM2: Lightweight Segment Anything for Spatial Intelligence

The recent large video foundation model, SAM2, enables segment anything in both images and videos, serving as a powerful base model for various applications. However, many of such use cases require to operate on resource-constrained devices like mobile phones and laptops. In this work, we aim to make SAM2 more mobile-friendly by distilling the heavyweight SAM2 into a lightweight model, facilitating segment anything in both images and videos on mobile devices. To this end, we propose Hypergraphical Knowledge Distill (HyperKD), which introduces the idea of hypergraph into knowledge distillation, aiming to effectively model and transfer SAM2's generalizable and comprehensive knowledge. HyperKD consists of Temporal HyperKD and Granularity HyperKD that construct hypergraphs to explicitly model and extract the generalizable temporal knowledge and the comprehensive multi-granularity knowledge from SAM2 respectively, which are then distilled into the lightweight student model by aligning it with the constructed hypergraphs. Besides, we present MobileSAM2, a new family of lightweight SAM2 that balances efficiency and effectiveness via searching the best model architectures with HyperKD during model size reduction. Extensive experiments validate MobileSAM2 across multiple benchmarks and show promising generalization performance on embodied AI tasks.

cs.CV↗

arXiv2026

SpikeATac: A Multimodal Tactile Finger with Taxelized Dynamic Sensing for Dexterous Manipulation

In this work, we introduce SpikeATac, a multimodal tactile finger combining a taxelized and highly sensitive dynamic response (PVDF) with a static transduction method (capacitive) for multimodal touch sensing. Named for its `spiky' response, SpikeATac's 16-taxel PVDF film sampled at 4 kHz provides fast, sensitive dynamic signals to the very onset and breaking of contact. We characterize the sensitivity of the different modalities, and show that SpikeATac provides the ability to stop quickly and delicately when grasping fragile, deformable objects. Beyond parallel grasping, we show that SpikeATac can be used in a learning-based framework to achieve new capabilities on a dexterous multifingered robot hand. We use reinforcement learning from human feedback to fine-tune the behavior of a policy to modulate force. Our hardware platform and learning pipeline together enable a difficult dexterous and contact-rich task that has not previously been achieved: in-hand manipulation of fragile objects. Videos are available at https://roamlab.github.io/spikeatac/ .

cs.RO↗

arXiv2026

High-Mobility and High-Reliability Top-Gate Oxide Semiconductor Transistors by Oxygen Engineering

In this work, we investigate the role of oxygen (O) on the performance of top-gate (TG) atomic-layer-deposited (ALD) oxide semiconductor transistors. The results reveal distinct defect characteristics and positive bias temperature instability (PBTI) degradation mechanisms between oxygen-rich (O-rich) and oxygen-deficient (O-poor) devices. It is found that an O-rich device fabrication process followed by O-free annealing can effectively achieve TG indium-rich (In-rich) oxide semiconductor transistors with high mobility, high reliability and high stability in hydrogen environment because O-rich process can suppress oxygen vacancies and their interaction with hydrogen, while O-free annealing plays a critical role in minimizing the formation of O-rich defects such as oxygen dimers (O-O bonds). Consequently, TG In-rich transistors with high mobility, steep subthreshold slope, and high PBTI reliability at high temperature are demonstrated. The understanding of O-rich defects provides a new insight to overcome the mobility-stability trade-off.

cond-mat.mtrl-sci↗

arXiv2026

Latent Noise Mask for Reducing Visual Redundancy in Multimodal Large Language Models

Multimodal large language models (MLLMs) often fail in fine-grained visual reasoning, as question-relevant visual cues are diluted by dense and redundant image tokens. Recent multimodal reasoning methods usually extend chain-of-thought from language models into visual or latent spaces, seeking to add intermediate reasoning states while overlooking the negative impact of redundant visual tokens. We propose LatEnt Noise maSk (Lens), a question-conditioned visual evidence purification framework that empowers MLLMs to reason with cleaner visual cues in latent space. Lens introduces a lightweight Lens Evidence Token (LET) to score which visual tokens support the current question and preserve them during decoding. Guided by the LET scores, it injects adaptive latent noise into low-relevance tokens, softly suppressing distractors without changing the model backbone or token sequence. With only one temporary learnable control token and a lightweight noise generator, Lens adds minimal overhead while improving the base MLLM by 2.4-6.4 points on most VQA datasets and by 4.1-6.4 points on grounding tasks. These results show that multimodal reasoning can benefit more directly from cleaner question-relevant visual evidence than from simply extending the reasoning trace.

cs.CV↗

arXiv2026

TurboServe: Serving Streaming Video Generation Efficiently and Economically

Streaming video generation is emerging as a new serving workload in which users interact with long-lived sessions that generate video progressively, chunk by chunk. Unlike offline video generation or typical LLM serving, streaming video generation must preserve session state across active and idle periods, repeatedly schedule ongoing sessions, and deliver each chunk under a tight latency target. This creates two key serving challenges in multi-user, multi-GPU environments: session duration heterogeneity, where long-running sessions make placement decisions suboptimal over time, and temporal user-demand heterogeneity, where the number of active sessions fluctuates sharply across bursts and idle periods. We present TurboServe, the first serving system designed specifically for streaming video generation workloads. TurboServe formulates serving as an online scheduling problem that jointly coordinates session placement and GPU provisioning. Its closed-loop scheduling algorithm combines a migration-aware placement controller, which rebalances sessions across GPUs to reduce the maximum per-chunk latency, with a load-driven autoscaling controller, which adapts the GPU budget to workload variation for improved cost efficiency. To support these decisions at runtime, TurboServe implements coalesced chunk processing for batching concurrent active sessions on the same GPU, GPU-CPU offloading for session suspension and resumption, and NCCL-based GPU-GPU migration for online rebalancing. We evaluate TurboServe on real-world production traces from Shengshu Technology across multiple model sizes and GPU clusters with up to 64 NVIDIA B300 GPUs. Compared with baseline serving configurations, TurboServe reduces worst-case per-chunk latency by 37.5% and total GPU operating cost by 37.2% on average. Our code is publicly available at https://github.com/shengshu-ai/TurboServe.

cs.DC↗

arXiv2026

HiGR: Industrial-Scale Hierarchical Generative Slate Recommendation Framework in Tencent

Slate recommendation, which presents users with a ranked item list in a single display, is ubiquitous across mainstream online platforms. While recent generative recommendation methods have shown strong potential in modeling item sequences with semantic IDs, directly applying them to industrial-scale slate recommendation faces a fundamental disconnect: entangled SID spaces confound high-level list planning, fine-grained autoregressive decoding over long sequences limits semantic planning efficiency, and token-level objectives misalign with holistic slate quality. In this paper, we propose HiGR, an industrial-scale hierarchical generative framework for slate recommendation that bridges this disconnect through a co-designed pipeline. First, HiGR learns structured SIDs via a Prefix-Contrastive Residual Quantized VAE (PCRQ-VAE). By enforcing high-level prefixes to capture shared semantics, PCRQ-VAE creates a controllable discrete space that acts as a prerequisite for efficient planning. Leveraging this structured space, our Hierarchical Slate Decoder (HSD) shifts autoregressive modeling from entangled token-level decoding to coarse-grained preference embeddings. This design significantly reduces inference latency while allowing explicit global slate structure planning. Finally, this stable planning space enables an ORPO-based listwise alignment mechanism to optimize triple-objective implicit feedback-ranking fidelity, genuine user interest, and diversity. Extensive offline experiments show that HiGR outperforms state-of-the-art baselines by over 10% in offline recommendation quality while achieving a $5\times$ inference speedup. Online A/B tests on Tencent platforms further improve watch time by 1.22% and video plays by 1.73%. HiGR has been deployed on multiple Tencent platform surfaces, serving hundreds of millions of users and proving its industrial-scale applicability.

cs.IR↗

arXiv2026

Structure and energetics of grain boundaries in self-assembled double-gyroid block copolymer networks

Grain boundaries (GBs) are ubiquitous defects in crystalline materials. However, they remain less explored in block copolymer ordered phases. Here, we develop a self-consistent field theory framework to investigate GB structure and energetics in double-gyroid (DG) diblock copolymer networks. The GB energy landscape is obtained as a function of GB orientation, which reveals multiple local minima representing distinct network-switching GBs. Remarkably, the global minimum is a previously unidentified asymmetric-tilt network-switching GB (ATNS), exhibiting a lower energy than the experimentally observed $(422)$ twin boundary (TB). Comparative analyses of representative low- (ATNS, $(422)$ TB) and high-energy twist ($(0\bar{1}\bar{1})$, $(100)$ TNSs) GBs reveal that, unlike enthalpy-dominated hard matter, GB stability in DG networks is predominantly entropy-driven. Twist-type GBs generate new nodes and disrupt nodal coplanarity, causing chain packing frustration and large entropy penalties. Conversely, the ATNS preserves favorable network connectivity and minimizes conformational constraints on polymer chains, making it the energetically preferred GB.

cond-mat.soft↗

arXiv2026

Bayesian Sparse Regression for Microbiome-Metabolite Data Integration

Numerous studies have shown that microbial metabolites, which represent the products of bacteria in the human gut, play a key role in shaping cancer risk and response to treatment. However, metabolite data typically contain a large proportion of missing values, which may result from either low abundance or technical challenges in data processing. Moreover, given the compositionality of microbiome data, where the observed abundances can only be interpreted on a relative scale, standard variable selection methods are not applicable. In this project, we propose a novel Bayesian regression method to address these challenges in the integration of metabolite and microbiome data. Key features of our proposed model include modeling the two different mechanisms of missingness for the metabolite data and adopting a Bayesian prior designed to address the compositional characteristics of microbiome data. We demonstrate on simulated data that our proposed model can accurately impute the unobserved true metabolite values and correctly select the relevant microbiome predictors. We further illustrate our method using real data from a study focused on understanding the interplay between the microbiome and metabolome in colorectal cancer.

stat.AP↗

arXiv2026

Geometric construction of superintegrable Poisson projection chains via Poisson centralizers

We introduce a geometric framework for constructing superintegrable systems from Poisson centralizers (commutants) in the Lie-Poisson algebra $S(\mathfrak{g})$ of a complex semisimple Lie algebra. Starting from a chain of reductive subgroups, we study the corresponding invariant Poisson subalgebras and their Poisson centers, and formulate superintegrability in terms of a \emph{Poisson projection chain} of affine Poisson varieties. For a maximal torus $T\subset G$, we prove that the inclusions $S(\mathfrak{g})^G\subset S(\mathfrak{g})^T\subset S(\mathfrak{g})$ determine a superintegrable chain and identify the associated quotient maps $\mathfrak{g}\xrightarrow{χ_T}\mathfrak{g}//T\xrightarrowρ\mathfrak{g}//G$. The rank (transcendence degree) computations yield the expected dimension split between commuting Hamiltonians and first integrals, and we describe the corresponding symplectic leaves in the intermediate space. Several examples illustrate how the centralizer generators organize into explicit superintegrable Poisson chains.

math-ph↗

arXiv2026

Poisson Centralisers and Polynomial Superintegrability for Magnetic Geodesic Flows on Reductive Homogeneous Spaces

We provide a method for formulating superintegrable magnetic geodesic flows on reductive homogeneous spaces $M=G/A$, with $G$ a compact semisimple Lie group and $A$ a closed subgroup of $G$. In the twisted cotangent bundle $(T^*M,ω_\varepsilon)$, with $ω_\varepsilon=ω_{\mathrm{can}}+\varepsilon\,π^*ω_{\mathrm{KKS}}$ being the canonical plus Kirillov-Kostant-Souriau (KKS) forms, we build two canonical and commuting families of polynomial first integrals: one pulled back from the Lie algebra $\mathfrak{g}$ of $G$ via the magnetic moment map $P$, and one pulled back from a $\mathrm{Ad}(A)$-invariant affine slice of $\mathfrak{m} \cong T_{eA}M$, where $eA$ is the identity of $G/A$. Their common image generates a reduced Poisson algebra obtained from a fiber tensor product, and the natural multiplication map into a Poisson subalgebra of polynomial functions $\mathcal{O}(T^*M) \subset C^\infty(T^*M)$ is Poisson and injective. The center of this fiber tensor product is contained in the Poisson center of the symmetric algebra of $\mathfrak{g}$. In a dense regular locus, the resulting projection chain realises a superintegrable system. As examples, two $\mathrm{SU}(3)$ cases are studied (regular torus and irregular $\mathrm{S}(\mathrm{U}(2)\times \mathrm{U}(1))$ quotients), which illustrate the construction and produce explicit action-angle coordinates.

math-ph↗

arXiv2026

KernelBenchX: A Comprehensive Benchmark for Evaluating LLM-Generated GPU Kernels

LLM-based Triton kernel generation has attracted significant interest, yet a fundamental empirical question remains unanswered: where does this capability break down, and why? We present KernelBenchX, a benchmark designed to answer this question through category-aware evaluation of correctness and hardware efficiency across 176 tasks in 15 categories. Our systematic comparison of five representative methods yields three main findings. First, task structure determines correctness more than method design. Category explains nearly three times more variance in semantic correctness than method (9.4% vs 3.3% explained deviance), and 72% of Fusion tasks fail across all five methods while Math tasks are solved consistently. Second, iterative refinement improves correctness, but not performance. Across GEAK iterations, compile rate rises from 52.3% to 68.8% while average speedup declines from $1.58\times$ to $1.44\times$; newly rescued kernels consistently underperform persistently correct ones ($1.16\times$ vs $1.58\times$ speedup in round~0$\to$1). Third, correctness does not imply efficiency. 46.6% of correct kernels are slower than the PyTorch eager baseline, and cross-hardware speedup variance reaches $21.4\times$. Besides, quantization remains completely unsolved (0/30 successes) despite non-trivial compilation rates, revealing systematic misunderstanding of numerical computation contracts rather than surface-level syntax errors. These findings suggest that future progress depends on handling global coordination, explicitly modeling numerical precision, and incorporating hardware efficiency into generation. The code is available at https://github.com/BonnieW05/KernelBenchX

cs.LG↗

arXiv2026

Optimizing IMPULSED Acquisition Protocols for Clinical 3T Scanners Through Bayesian Experimental Design

To optimize diffusion MRI acquisition protocols for IMPULSED model at clinical 3T scanner using Bayesian experimental design, enabling accurate cellular-scale parameter estimation under realistic scan time and scanner hardware constraints. Expected Information Gain (EIG) was used as the optimization objective to maximize the information content of acquired measurements for IMPULSED model fitting. Bayesian optimization with Gaussian process surrogates efficiently searched the high-dimensional acquisition parameter space, including pulse types (PGSE, OGSEn1, and OGSEn2), diffusion times, and b-values. Optimized protocols were systematically evaluated against a heuristically designed baseline protocol through simulation studies assessing classification accuracy and parameter estimation performance across SNR levels of 5-40. Robustness to optimization assumptions was examined by varying prior distributions and assumed SNR. In-vivo validation was performed using canine tumor data acquired at 3T. The optimized protocol eliminated OGSEn2 acquisitions, concentrated measurements at high b-values, employing concurrently optimized diffusion timing. Compared to the baseline protocol, the optimized design achieved superior classification accuracy for distinguishing cell populations and reduced parameter estimation error across biologically relevant parameter ranges at various SNRs. Performance advantages were consistent across diverse optimization scenarios, demonstrating robustness to prior knowledge and noise assumptions. In-vivo parameter maps showed substantially improved quality and smoothness. Bayesian optimization substantially improves IMPULSED acquisition design for clinical 3T scanners. This principled, algorithm-agnostic framework enables accurate diffusion MRI cytometry under clinical constraints, with potential applications to tumor characterization and treatment monitoring.

physics.med-ph↗

arXiv2026

Quasiperiodic Elliptic Operators: Projection Method and Convergence Analysis

Quasiperiodic elliptic operators (QEOs) serve as fundamental models in both mathematics and physics, as exemplified by their role in the numerical modeling of one-dimensional photonic quasicrystals. However, distinct from periodic elliptic operators, approximating eigenpairs for QEOs poses significant challenges, particularly in capturing the full spectral structure (notably the continuous spectrum) and deriving convergence guarantees in the absence of compactness. In this paper, we develop a high-accuracy numerical method to compute eigenpairs of QEOs based on the projection method, which embeds quasiperiodic operators into a higher-dimensional periodic torus. To address the non-compactness issue, we construct a directional-derivative Hilbert space along irrational manifolds of a high-dimensional torus and characterize operators equivalent to QEOs within this space. By integrating spectral theory for non-compact operators into the Babuška-Osborn eigenproblem framework, we establish rigorous convergence analysis and prove that our method achieves spectral accuracy. Numerical experiments validate the accuracy and efficiency of the proposed method, including a one-dimensional photonic quasicrystal and two- and three-dimensional QEOs.

math.NA↗

arXiv2026

Framework for Quasiperiodic Interfaces: Proximal Coincidence Point Set and Computation

We present a unified theoretical and computational framework that bridges mathematical quasiperiodicity with classical crystallographic models. Based on a rigorous cut-and-projection construction, the proposed proximal coincidence point set (PCPS) theory extends the classical coincidence site lattice model and further incorporates physically motivated perturbations encoding interfacial atomic mobility as well as visual indistinguishability. Spectral characteristics of PCPS naturally motivate a conserved Landau-Brazovskii model combined with projection method, yielding unified high accuracy in resolving quasiperiodic order across the entire interfacial plane. Representative quasiperiodic features are revealed in our numerical results, including generalized Fibonacci sequences in BCC [110] tilt GBs, as well as repetitive patterns within the interstices of dislocation networks in low-angle BCC [100] twist GBs and phase boundaries between BCC and face-centered cubic crystals. In high-angle BCC [100] twist GBs, 12- and 8-fold quasicrystals emerge, while the PCPS theory combined with cyclotomic field projections further explains their restrictions of non-crystallographic symmetries. This framework not only provides a rigorous theoretical explanation for interface structures but also offers a path toward modeling other types of incommensurate structures.

cond-mat.mtrl-sci↗