arXiv subjects

Yifei Wu

Publications and source records attributed to Yifei Wu.

At least 37 records · Page 2Linked to original sources

arXiv2025

Regularization for the Schrödinger equation with rough potential: high-dimensional case

In this work, we investigate the regularization mechanisms of the Schrödinger equation with a spatial potential $$ i\partial_t u+Δu+ηu =0, $$ where $η$ denotes a given spatial potential. The regularity of solutions constitutes one of the central problems in the theory of dispersive equations. Recent works \cite{Bai-Lian-Wu-2024, M-Wu-Z24} have established the sharp regularization mechanisms for this model in the whole space $\mathbb{R}$ and on the torus $\mathbb{T}$, with $η$ being a rough potential. The present paper extends the line of research to the high-dimensional setting with rough potentials $η\in L_x^r+L_x^{\infty}$. More precisely, we first show that when $1\leq r <\frac d2$, there exists some $η\in L_x^r+L_x^{\infty}$ such that the equation is ill-posed in $H_x^γ$ for any $γ\in \mathbb{R}$. Conversely, when $\frac d2 \leq r \leq \infty$, the expected optimal regularity is given by $$H_x^{γ_*}, \quad γ_*=\mbox{min}\{2+\frac d2-\frac dr, 2\}.$$ We establish a comprehensive characterization of the regularity, with the exception of two dimensional endpoint case $d=2, r=1$. Our novel theoretical framework combines several fundamental ingredients: the construction of counterexamples, the proposal of splitting normal form method, and the iterative Duhamel construction. Furthermore, we briefly discuss the effect of the interaction between rough potentials and nonlinear terms on the regularity of solutions.

math.AP↗

arXiv2025

Ming-UniAudio: Speech LLM for Joint Understanding, Generation and Editing with Unified Representation

Existing speech models suffer from competing requirements on token representations by understanding and generation tasks. This discrepancy in representation prevents speech language models from performing instruction-based free-form editing. To solve this challenge, we introduce a novel framework that unifies speech understanding, generation, and editing. The core of our unified model is a unified continuous speech tokenizer MingTok-Audio, the first continuous tokenizer to effectively integrate semantic and acoustic features, which makes it suitable for both understanding and generation tasks. Based on this unified continuous audio tokenizer, we developed the speech language model Ming-UniAudio, which achieved a balance between generation and understanding capabilities. Ming-UniAudio sets new state-of-the-art (SOTA) records on 8 out of 12 metrics on the ContextASR benchmark. Notably, for Chinese voice cloning, it achieves a highly competitive Seed-TTS-WER of 0.95. Leveraging this foundational model, we further trained a dedicated speech editing model Ming-UniAudio-Edit, the first speech language model that enables universal, free-form speech editing guided solely by natural language instructions, handling both semantic and acoustic modifications without timestamp condition. To rigorously assess the editing capability and establish a foundation for future research, we introduce Ming-Freeform-Audio-Edit, the first comprehensive benchmark tailored for instruction-based free-form speech editing, featuring diverse scenarios and evaluation dimensions spanning semantic correctness, acoustic quality, and instruction alignment. We open-sourced the continuous audio tokenizer, the unified foundational model, and the free-form instruction-based editing model to facilitate the development of unified audio understanding, generation, and manipulation.

cs.CL↗

arXiv2025

ScenGAN: Attention-Intensive Generative Model for Uncertainty-Aware Renewable Scenario Forecasting

To address the intermittency of renewable energy source (RES) generation, scenario forecasting offers a series of stochastic realizations for predictive objects with superior flexibility and direct views. Based on a long time-series perspective, this paper explores uncertainties in the realms of renewable power and deep learning. Then, an uncertainty-aware model is meticulously designed for renewable scenario forecasting, which leverages an attention mechanism and generative adversarial networks (GANs) to precisely capture complex spatial-temporal dynamics. To improve the interpretability of uncertain behavior in RES generation, Bayesian deep learning and adaptive instance normalization (AdaIN) are incorporated to simulate typical patterns and variations. Additionally, the integration of meteorological information, forecasts, and historical trajectories in the processing layer improves the synergistic forecasting capability for multiscale periodic regularities. Numerical experiments and case analyses demonstrate that the proposed approach provides an appropriate interpretation for renewable uncertainty representation, including both aleatoric and epistemic uncertainties, and shows superior performance over state-of-the-art methods.

cs.LG↗

arXiv2025

System-Level Performance and Communication Tradeoff in Networked Control with Predictions

Distributed control of large-scale systems is challenging due to the need for scalable and localized communication and computation. In this work, we introduce a Predictive System-Level Synthesis PredSLS framework that designs controllers by jointly integrating communication constraints and local disturbance predictions into an affine feedback structure. Rather than focusing on the worst-case uncertainty, PredSLS leverages both current state feedback and future system disturbance predictions to achieve distributed control of networked systems. In particular, PredSLS enables a unified system synthesis of the optimal $κ$-localized controller, therefore outperforms approaches with post hoc communication truncation, as was commonly seen in the literature. The PredSLS framework can be naturally decomposed into spatial and temporal components for efficient and parallelizable computation across the network, yielding a regret upper bound that explicitly depends on the prediction error and communication range. Our regret analysis not only reveals a non-monotonic trade-off between control performance and communication range when prediction errors are present, but also guides the identification of an optimal size for local communication neighborhoods, thereby enabling the co-design of controller and its underlying communication topology.

eess.SY↗

arXiv2025

Globally Optimal Movable Antenna-Enhanced multiuser Communication: Discrete Antenna Positioning, Motion Power Consumption, and Imperfect CSI

Movable antennas (MAs) represent a promising paradigm to enhance the spatial degrees of freedom of conventional multi-antenna systems by dynamically adapting the positions of antenna elements within a designated transmit area. In particular, by employing electro-mechanical MA drivers, the positions of the MA elements can be adjusted to shape a favorable spatial correlation for improving system performance. Although preliminary research has explored beamforming designs for MA systems, the intricacies of the power consumption and the precise positioning of MA elements are not well understood. Moreover, the assumption of perfect CSI adopted in the literature is impractical due to the significant pilot overhead and the extensive time to acquire perfect CSI. To address these challenges, we model the motion of MA elements through discrete steps and quantify the associated power consumption as a function of these movements. Furthermore, by leveraging the properties of the MA channel model, we introduce a novel CSI error model tailored for MA systems that facilitates robust resource allocation design. In particular, we optimize the beamforming and the MA positions at the BS to minimize the total BS power consumption, encompassing both radiated and MA motion power while guaranteeing a minimum required SINR for each user. To this end, novel algorithms exploiting the branch and bound (BnB) method are developed to obtain the optimal solution for perfect and imperfect CSI. Moreover, to support practical implementation, we propose low-complexity algorithms with guaranteed convergence by leveraging successive convex approximation (SCA). Our numerical results validate the optimality of the proposed BnB-based algorithms. Furthermore, we unveil that both proposed SCA-based algorithms approach the optimal performance within a few iterations, thus highlighting their practical advantages.

eess.SP↗

arXiv2025

Computing rough solutions of the KdV equation below ${\bf L^2}$

We establish a novel numerical and analytical framework for solving the Korteweg--de Vries (KdV) equation in the negative Sobolev spaces, where classical numerical methods fail due to their reliance on high regularity and inability to control nonlinear interactions at low regularities. Numerical analysis is established by combining a continuous reformulation of the numerical scheme, the Bourgain-space estimates for the continuous reformulation, and a rescaling strategy that reduces the reformulated problem to a small initial value problem, which allow us to bridge a critical gap between numerical analysis and theoretical well-posedness by designing the first numerical method capable of solving the KdV equation in the negative Sobolev spaces. The numerical scheme is proved to have nearly optimal-order convergence with respect to the spatial degrees of freedom in the $H^{-\frac{1}{2}}$ norm for initial data in $H^s$, with $-\frac{1}{2} < s \leq 0$, a result unattainable by existing numerical methods.

math.NA↗

arXiv2025

Ming-Omni: A Unified Multimodal Model for Perception and Generation

We propose Ming-Omni, a unified multimodal model capable of processing images, text, audio, and video, while demonstrating strong proficiency in both speech and image generation. Ming-Omni employs dedicated encoders to extract tokens from different modalities, which are then processed by Ling, an MoE architecture equipped with newly proposed modality-specific routers. This design enables a single model to efficiently process and fuse multimodal inputs within a unified framework, thereby facilitating diverse tasks without requiring separate models, task-specific fine-tuning, or structural redesign. Importantly, Ming-Omni extends beyond conventional multimodal models by supporting audio and image generation. This is achieved through the integration of an advanced audio decoder for natural-sounding speech and Ming-Lite-Uni for high-quality image generation, which also allow the model to engage in context-aware chatting, perform text-to-speech conversion, and conduct versatile image editing. Our experimental results showcase Ming-Omni offers a powerful solution for unified perception and generation across all modalities. Notably, our proposed Ming-Omni is the first open-source model we are aware of to match GPT-4o in modality support, and we release all code and model weights to encourage further research and development in the community.

cs.AI↗

arXiv2025

FedDAG: Federated Domain Adversarial Generation Towards Generalizable Medical Image Analysis

Federated domain generalization aims to train a global model from multiple source domains and ensure its generalization ability to unseen target domains. Due to the target domain being with unknown domain shifts, attempting to approximate these gaps by source domains may be the key to improving model generalization capability. Existing works mainly focus on sharing and recombining local domain-specific attributes to increase data diversity and simulate potential domain shifts. However, these methods may be insufficient since only the local attribute recombination can be hard to touch the out-of-distribution of global data. In this paper, we propose a simple-yet-efficient framework named Federated Domain Adversarial Generation (FedDAG). It aims to simulate the domain shift and improve the model generalization by adversarially generating novel domains different from local and global source domains. Specifically, it generates novel-style images by maximizing the instance-level feature discrepancy between original and generated images and trains a generalizable task model by minimizing their feature discrepancy. Further, we observed that FedDAG could cause different performance improvements for local models. It may be due to inherent data isolation and heterogeneity among clients, exacerbating the imbalance in their generalization contributions to the global model. Ignoring this imbalance can lead the global model's generalization ability to be sub-optimal, further limiting the novel domain generation procedure. Thus, to mitigate this imbalance, FedDAG hierarchically aggregates local models at the within-client and across-client levels by using the sharpness concept to evaluate client model generalization contributions. Extensive experiments across four medical benchmarks demonstrate FedDAG's ability to enhance generalization in federated medical scenarios.

cs.CV↗

arXiv2025

Introducing Resilience to IRS-Assisted Secure Wireless Systems

Intelligent reflecting surfaces (IRSs) are envisioned as a transformative method for enhancing the physical layer security of future communication networks. However, current IRS-assisted secure wireless system designs incur high time cost for the joint optimization of beamforming vectors and IRS phase shifts and are susceptible to unforeseen disruption. Therefore, introducing resilience in IRS-assisted secure systems is essential. In this paper, we first quantify the resilience performance as the combination of the absorption and adaptation performance. In particular, the absorption performance measures the system's robustness when facing failure while the adaptation performance reflects the system's ability to recover from failure. Then, we propose a two-timescale transmission protocol aiming to enhance system resilience while limiting maximum information leakage to eavesdroppers. Specifically, in the initialization phase, the base station (BS) beamforming vectors and IRS phase shifts are optimized iteratively using an alternating optimization (AO) approach to improve the absorption performance of the system, i.e., the worst-case achievable rate, given a predefined maximum tolerable leakage rate, by accounting for the impact of long-term channel variations. When the system detects an outage, a fast system recovery mechanism is activated to restore operation by adjusting only the BS beamforming vectors, which enhances the adaptation performance of the system. Our numerical results reveal that the proposed algorithm significantly improves system resilience, including system robustness and recovery performance. Moreover, the proposed design guarantees outage-free transmission over a long time interval for moderate-size IRSs.

eess.SP↗

arXiv2024

Cross-organ Deployment of EOS Detection AI without Retraining: Feasibility and Limitation

Chronic rhinosinusitis (CRS) is characterized by persistent inflammation in the paranasal sinuses, leading to typical symptoms of nasal congestion, facial pressure, olfactory dysfunction, and discolored nasal drainage, which can significantly impact quality-of-life. Eosinophils (Eos), a crucial component in the mucosal immune response, have been linked to disease severity in CRS. The diagnosis of eosinophilic CRS typically uses a threshold of 10-20 eos per high-power field (HPF). However, manually counting Eos in histological samples is laborious and time-intensive, making the use of AI-driven methods for automated evaluations highly desirable. Interestingly, eosinophils are predominantly located in the gastrointestinal (GI) tract, which has prompted the release of numerous deep learning models trained on GI data. This study leverages a CircleSnake model initially trained on upper-GI data to segment Eos cells in whole slide images (WSIs) of nasal tissues. It aims to determine the extent to which Eos segmentation models developed for the GI tract can be adapted to nasal applications without retraining. The experimental results show promising accuracy in some WSIs, although, unsurprisingly, the performance varies across cases. This paper details these performance outcomes, delves into the reasons for such variations, and aims to provide insights that could guide future development of deep learning models for eosinophilic CRS.

eess.IV↗

arXiv2024

Global well-posedness and scattering of 3D defocusing, cubic Schrödinger equation

In this paper, we study the global well-posedness and scattering of 3D defocusing, cubic Schrödinger equation. Recently, Dodson [arXiv:2004.09618] studied the global well-posedness in a critical Sobolev space $\dot{W}^{11/7,7/6}$. In this paper, we aim to show that if the initial data belongs to $\dot H^\frac12$ to guarantee the local existence, then some extra weak space which is subcritical, is sufficient to prove the global well-posedness. More precisely, we prove that if the initial data belongs to $\dot{H}^{1/2}\cap \dot{W}^{s,1}$ for $12/13<s \leqslant 1$, then the corresponding solution exists globally and scatters.

math.AP↗

arXiv2024

Global well-posedness for the defocusing 3D quadratic NLS in the sharp critical space

In this paper, we prove the global well-posedness of defocusing 3D quadratic nonlinear Schrödinger equation \begin{align*} i\partial_t u + \frac12Δu = |u| u, \end{align*} in its sharp critical weighted space $\mathcal F \dot H_x^{1/2}$ for radial data. Killip, Masaki, Murphy, and Visan [2017, NoDEA] have proved its global well-posedness and scattering, if the $\mathcal F \dot H_x^{1/2}$-norm of the solution is bounded in the maximal lifespan. Now, we remove this a priori assumption for the global well-posedness statement in the radial case. Our method is based on the almost conservation of pseudo conformal energy. This energy scales like $\dot H_x^{-1}$, which is supercritical. We are still able to derive the global well-posedness using this monotone quantity. The main observation is that we can establish the local solution in supercritical weighted space when the initial time is away from the origin.

math.AP↗

arXiv2024

Globally Optimal Resource Allocation Design for Discrete Phase Shift IRS-Assisted Multiuser Networks with Perfect and Imperfect CSI

Intelligent reflecting surfaces (IRSs) are a promising low-cost solution for achieving high spectral and energy efficiency in future communication systems by enabling the customization of wireless propagation environments. Despite the plethora of research on resource allocation design for IRS-assisted multiuser wireless communication systems, the optimal design and the corresponding performance upper bound are still not fully understood. To bridge this gap in knowledge, in this paper, we investigate the optimal resource allocation design for IRS-assisted multiuser multiple-input single-output systems employing practical discrete IRS phase shifters. In particular, we jointly optimize the beamforming vector at the base station and the discrete IRS phase shifts to minimize the total transmit power for the cases of perfect and imperfect channel state information (CSI) knowledge. To this end, two novel algorithms based on the generalized Benders decomposition (GBD) method are developed to obtain the globally optimal solution for perfect and imperfect CSI, respectively. Moreover, to facilitate practical implementation, we propose two corresponding low-complexity suboptimal algorithms with guaranteed convergence by capitalizing on successive convex approximation (SCA). In particular, for imperfect CSI, we adopt a bounded error model to characterize the CSI uncertainty and propose a new transformation to convexify the robust quality-of-service constraints. Our numerical results confirm the optimality of the proposed GBD-based algorithms for the considered system for both perfect and imperfect CSI. Furthermore, we unveil that both proposed SCA-based algorithms can attain a locally optimal solution within a few iterations. Moreover, compared with the state-of-the-art solution based on alternating optimization, the proposed low-complexity SCA-based schemes achieve a significant performance gain.

cs.IT↗

arXiv2024

The cubic nonlinear Schrödinger equation with rough potential

We consider the cubic nonlinear Schrödinger equation with a spatially rough potential, a key equation in the mathematical setup for nonlinear Anderson localization. Our study comprises two main parts: new optimal results on the well-posedness analysis on the PDE level, and subsequently a new efficient numerical method, its convergence analysis and simulations that illustrate our analytical results. In the analysis part, our results focus on understanding how the regularity of the solution is influenced by the regularity of the potential, where we provide quantitative and explicit characterizations. Ill-posedness results are also established to demonstrate the sharpness of the obtained regularity characterizations and to indicate the minimum regularity required from the potential for the NLS to be solvable. Building upon the obtained regularity results, we design an appropriate numerical discretization for the model and establish its convergence with an optimal error bound. The numerical experiments in the end not only verify the theoretical regularity results, but also confirm the established convergence rate of the proposed scheme. Additionally, a comparison with other existing schemes is conducted to demonstrate the better accuracy of our new scheme in the case of a rough potential.

math.NA↗

arXiv2024

PrPSeg: Universal Proposition Learning for Panoramic Renal Pathology Segmentation

Understanding the anatomy of renal pathology is crucial for advancing disease diagnostics, treatment evaluation, and clinical research. The complex kidney system comprises various components across multiple levels, including regions (cortex, medulla), functional units (glomeruli, tubules), and cells (podocytes, mesangial cells in glomerulus). Prior studies have predominantly overlooked the intricate spatial interrelations among objects from clinical knowledge. In this research, we introduce a novel universal proposition learning approach, called panoramic renal pathology segmentation (PrPSeg), designed to segment comprehensively panoramic structures within kidney by integrating extensive knowledge of kidney anatomy. In this paper, we propose (1) the design of a comprehensive universal proposition matrix for renal pathology, facilitating the incorporation of classification and spatial relationships into the segmentation process; (2) a token-based dynamic head single network architecture, with the improvement of the partial label image segmentation and capability for future data enlargement; and (3) an anatomy loss function, quantifying the inter-object relationships across the kidney.

eess.IV↗

arXiv2023

Well-posedness of the discrete nonlinear Schrödinger equations and the Klein-Gordon equations

The primary objective of this paper is to investigate the well-posedness theories associated with the discrete nonlinear Schrödinger equation and Klein-Gordon equation. These theories encompass both local and global well-posedness, as well as the existence of blowing-up solutions for large and irregular initial data. The main results of this paper presented in this paper can be summarized as follows: 1. Discrete Nonlinear Schrödinger Equation: We establish global well-posedness in $l^p_h$ spaces for all $1\leq p\leq \infty$, regardless of whether it is in the defocusing or focusing cases. 2. Discrete Klein-Gordon Equation (including Wave Equation): We demonstrate local well-posedness in $l^p_h$ spaces for all $1\leq p\leq \infty$. Furthermore, in the defocusing case, we establish global well-posedness in $l^p_h$ spaces for any $2\leq p\leq 2σ+2$. In contrast, in the focusing case, we show that solutions with negative energy blow up within a finite time.

math.DS↗

arXiv2023

Demystifying Compiler Unstable Feature Usage and Impacts in the Rust Ecosystem

Rust programming language is gaining popularity rapidly in building reliable and secure systems due to its security guarantees and outstanding performance. To provide extra functionalities, the Rust compiler introduces Rust unstable features (RUF) to extend compiler functionality, syntax, and standard library support. However, these features are unstable and may get removed, introducing compilation failures to dependent packages. Even worse, their impacts propagate through transitive dependencies, causing large-scale failures in the whole ecosystem. Although RUF is widely used in Rust, previous research has primarily concentrated on Rust code safety, with the usage and impacts of RUF from the Rust compiler remaining unexplored. Therefore, we aim to bridge this gap by systematically analyzing the RUF usage and impacts in the Rust ecosystem. We propose novel techniques for extracting RUF precisely, and to assess its impact on the entire ecosystem quantitatively, we accurately resolve package dependencies. We have analyzed the whole Rust ecosystem with 590K package versions and 140M transitive dependencies. Our study shows that the Rust ecosystem uses 1000 different RUF, and at most 44% of package versions are affected by RUF, causing compiling failures for at most 12%. To mitigate wide RUF impacts, we further design and implement a RUF-compilation-failure recovery tool that can recover up to 90% of the failure. We believe our techniques, findings, and tools can help to stabilize the Rust compiler, ultimately enhancing the security and reliability of the Rust ecosystem.

cs.SE↗

arXiv2023

Scattering for defocusing mass sub-critical NLS

In this paper, we consider the $L_x^2$-scattering of defocusing mass sub-critical nonlinear Schrödinger equations with low weighted initial condition. It is known that the scattering holds with $\mathcal{F} H^1$-data, while the continuity of inverse wave operator breaks down with $L^2$-data. Moreover, for large $\mathcal{F} H^s$-data with $s<1$, there only exists the wave operator result, but scattering results are lacking. Our subject is to study the scattering in low weights space. Our results are divided into two parts. Our first result presents a systematic study on the scattering on $\mathcal{F} H^s$ for certain $s<1$, without any restrictions on smallness or radial symmetry. This extends the previous results to spaces with lower weights. Our second result is the almost sure scattering on $L^2$ by introducing a ``narrowed'' Wiener randomization in physical space. For mass subcritical NLS when $d\ge 2$, this result represents the first scattering result without imposing any conditions related to smallness, radial symmetry, or weighted properties on the initial data.

math.AP↗