Search arXiv⌕ Search

arXiv subjects

Xinwei Yu

Publications and source records attributed to Xinwei Yu.

16 recordsLinked to original sources
arXiv2026

Towards Autonomous Formulaic Alpha Discovery: An Evolutionary Computation Perspective

Automated formulaic alpha discovery aims to generate predictive and interpretable trading signals from large symbolic factor spaces. Its effectiveness is constrained by noisy fitness estimates, market nonstationarity, costly backtesting, semantic redundancy, and conflicting practical objectives. Existing studies employ diverse techniques, including genetic programming (GP), evolutionary algorithms (EAs), reinforcement learning (RL), generative flow networks (GFlowNets), Monte Carlo tree search (MCTS), large language models (LLMs), and agentic workflows, but generally examine them as separate algorithmic families. This article introduces, for the first time, a unified evolutionary computation (EC) perspective on automated formulaic alpha discovery, formulating it as a noisy, dynamic, and multiobjective symbolic evolutionary optimization problem. A six-component analytical framework is developed to characterize existing methods through representation, variation, fitness evaluation, selection, memory, and adaptation. Furthermore, an eight-dimensional, autonomy-oriented evaluation framework is proposed, covering search efficiency, fitness reliability, residual alpha quality, economic diversity, tradability, evolutionary autonomy, robustness to nonstationarity, and reproducibility. Together, these frameworks provide a systematic foundation for unifying heterogeneous approaches, diagnosing component-level limitations, and guiding the development of reliable, adaptive, interpretable, and reproducible autonomous alpha discovery systems.

cs.NE↗
arXiv2022

Numerical Analysis for Real-time Nonlinear Model Predictive Control of Ethanol Steam Reformers

The utilization of renewable energy technologies, particularly hydrogen, has seen a boom in interest and has spread throughout the world. Ethanol steam reformation is one of the primary methods capable of producing hydrogen efficiently and reliably. This paper provides an in-depth study of the reformulated system both theoretically and numerically, as well as a plan to explore the possibility of converting the system into its conservation form. Lastly, we offer an overview of several numerical approaches for solving the general first-order quasi-linear hyperbolic equation to the particular model for ethanol steam reforming (ESR). We conclude by presenting some results that would enable the usage of these ODE/PDE solvers to be used in non-linear model predictive control (NMPC) algorithms and discuss the limitations of our approach and directions for future work.

math.AP↗
arXiv2022

(Fusionformer):Exploiting the Joint Motion Synergy with Fusion Network Based On Transformer for 3D Human Pose Estimation

For the current 3D human pose estimation task, a group of methods mainly learn the rules of 2D-3D projection from spatial and temporal correlation. However, earlier methods model the global features of the entire body joint in the time domain, but ignore the motion trajectory of individual joint. The recent work [29] considers that there are differences in motion between different joints and deals with the temporal relationship of each joint separately. However, we found that different joints show the same movement trends under some specific actions. Therefore, our proposed Fusionformer method introduces a self-trajectory module and a mutual-trajectory module based on the spatio-temporal module .After that, the global spatio-temporal features and local joint trajectory features are fused through a linear network in a parallel manner. To eliminate the influence of bad 2D poses on 3D projections, finally we also introduce a pose refinement network to balance the consistency of 3D projections. In addition, we evaluate the proposed method on two benchmark datasets (Human3.6M, MPI-INF-3DHP). Comparing our method with the baseline method poseformer, the results show an improvement of 2.4% MPJPE and 4.3% P-MPJPE on the Human3.6M dataset, respectively.

cs.CV↗
arXiv2021

Fast deep learning correspondence for neuron tracking and identification in C.elegans using synthetic training

We present an automated method to track and identify neurons in C. elegans, called "fast Deep Learning Correspondence" or fDLC, based on the transformer network architecture. The model is trained once on empirically derived synthetic data and then predicts neural correspondence across held-out real animals via transfer learning. The same pre-trained model both tracks neurons across time and identifies corresponding neurons across individuals. Performance is evaluated against hand-annotated datasets, including NeuroPAL [1]. Using only position information, the method achieves 80.0% accuracy at tracking neurons within an individual and 65.8% accuracy at identifying neurons across individuals. Accuracy is even higher on a published dataset [2]. Accuracy reaches 76.5% when using color information from NeuroPAL. Unlike previous methods, fDLC does not require straightening or transforming the animal into a canonical coordinate system. The method is fast and predicts correspondence in 10 ms making it suitable for future real-time applications.

q-bio.QM↗
arXiv2020

Universal Adversarial Attacks with Natural Triggers for Text Classification

Recent work has demonstrated the vulnerability of modern text classifiers to universal adversarial attacks, which are input-agnostic sequences of words added to text processed by classifiers. Despite being successful, the word sequences produced in such attacks are often ungrammatical and can be easily distinguished from natural text. We develop adversarial attacks that appear closer to natural English phrases and yet confuse classification systems when added to benign inputs. We leverage an adversarially regularized autoencoder (ARAE) to generate triggers and propose a gradient-based search that aims to maximize the downstream classifier's prediction loss. Our attacks effectively reduce model accuracy on classification tasks while being less identifiable than prior models as per automatic detection metrics and human-subject studies. Our aim is to demonstrate that adversarial attacks can be made harder to detect than previously thought and to enable the development of appropriate defenses.

cs.CL↗
arXiv2020

NOMA Design with Power-Outage Tradeoff for Two-User Systems

This letter proposes a modified non-orthogonal multiple-access (NOMA) scheme for systems with a multi-antenna base station (BS) and two single-antenna users, where NOMA transmissions are conducted only when the absolute correlation coefficient (CC) between the user channels exceeds a threshold and the BS uses matched-filter (MF) precoding along the user with the stronger average channel gain. We derive the average minimal transmit power to guarantee the signal-to-interference-plus-noise-ratio (SINR) levels of both users. Our results show that the average minimal power grows logarithmically in the reciprocal of the CC threshold and a non-zero threshold is necessary for the modified NOMA scheme to have finite average minimal transmit power. Further, for the massive MIMO scenario, we derive the scaling laws of the average transmit power and outage probability with respect to the antenna numbers, as well as their tradeoff law. Simulation results are shown to validate our theoretical results.

eess.SP↗
arXiv2019

New Regularity Criteria for the Navier-Stokes Equations in Terms of Pressure

In this paper, we generalize the main results of [1] and [31] to Lorentz spaces, using a simple procedure. The main results are the following. Let $n\geq 3$ and let $u$ be a Leray-Hopf solution to the $n$-dimensional Navier-Stokes equations with viscosity $\nu$ and divergence free initial condition $u_0\in L^2(\mathbb{R}^n)\cap L^{k}(\mathbb{R}^n)$ (where $k=k(s)$ is sufficiently large). Then there exists a constant $c>0$ such that if \begin{equation} \|p\|_{L^{r,\infty}(0,\infty;L^{s,\infty}(\mathbb{R}^n))} \frac{n}{2} \end{equation} or \begin{equation} \|\nabla p\|_{L^{r,\infty}(0,\infty;L^{s,\infty}(\mathbb{R}^n))} \frac{n}{3} \end{equation} then $u$ is smooth on $(0, \infty) \times \mathbb{R}^n$. Partial results in the case $n=3$ were obtained in [32], [33] and then recently extended to all appropriate pairs of $r,s$ in [14]. Our results present a unified proof which works for all dimensions $n\geq 3$ and the full range or admissible pairs, $(s,r)$.

math.AP↗
arXiv2015

On Stationary Solutions of the 2D Doi-Onsager Model

We study the 2D Doi--Onsager models with general potential kernel, with special emphasis on the classical Onsager kernel. Through application of topological methods from nonlinear functional analysis, in particular the Leray--Schauder degree theory, we obtain the uniqueness of the trivial solution for low temperatures as well as the local bifurcation structure of the solutions.

math.AP↗
arXiv2013

On Global Regularity of 2D Generalized Magnetohydrodynamic Equations

In this article we study the global regularity of 2D generalized magnetohydrodynamic equations (2D GMHD), in which the dissipation terms are $- \nu (- \triangle)^{\alpha} u$ and $- \kappa (-\triangle)^{\beta} b$. We show that smooth solutions are global in the following three cases: $\alpha \geqslant 1 / 2, \beta \geqslant 1$; $0 \leqslant \alpha < 1 / 2, 2 \alpha + \beta > 2$; $\alpha \geqslant 2, \beta = 0$. We also show that in the inviscid case $\nu = 0$, if $\beta > 1$, then smooth solutions are global as long as the direction of the magnetic field remains smooth enough.

math.AP↗
arXiv2012

On the Euler-Poincar\'e equation with non-zero dispersion

We consider the Euler-Poincar\'e equation on $\mathbb R^d$, $d\ge 2$. For a large class of smooth initial data we prove that the corresponding solution blows up in finite time. This settles an open problem raised by Chae and Liu \cite{Chae Liu}. Our analysis exhibits some new concentration mechanism and hidden monotonicity formula associated with the Euler-Poincar\'e flow. In particular we show the abundance of blowups emanating from smooth initial data with certain sign properties. No size restrictions are imposed on the data. We also showcase a class of initial data for which the corresponding solution exists globally in time.

math.AP↗
arXiv2012

On some Liouville Type Theorems for the Compressible Navier-Stokes Equations

We prove several Liouville type results for stationary solutions of the $d$-dimensional compressible Navier-Stokes equations. In particular, we show that when the dimension $d \geqslant 4$, the natural requirements $\rho \in L^{\infty} (\mathbbm{R}^d)$, $v \in \dot{H}^1 (\mathbbm{R}^d)$ suffice to guarantee that the solution is trivial. For dimensions $d=2,3$, we assume the extra condition $v \in L^{\frac{3d}{d-1}}(\mathbb R^d)$. This improves a recent result of Chae (2012).

math.AP↗
arXiv2009

On Singularity Formation of a Nonlinear Nonlocal System

We investigate the singularity formation of a nonlinear nonlocal system. This nonlocal system is a simplified one-dimensional system of the 3D model that was recently proposed by Hou and Lei in [13] for axisymmetric 3D incompressible Navier-Stokes equations with swirl. The main difference between the 3D model of Hou and Lei and the reformulated 3D Navier-Stokes equations is that the convection term is neglected in the 3D model. In the nonlocal system we consider in this paper, we replace the Riesz operator in the 3D model by the Hilbert transform. One of the main results of this paper is that we prove rigorously the finite time singularity formation of the nonlocal system for a large class of smooth initial data with finite energy. We also prove the global regularity for a class of smooth initial data. Numerical results will be presented to demonstrate the asymptotically self-similar blow-up of the solution. The blowup rate of the self-similar singularity of the nonlocal system is similar to that of the 3D model.

math.AP↗
arXiv2006

Remarks on the Global Regularity for the Super-Critical 2D Dissipative Quasi-Geostrophic Equation

In this article we apply the method used in the recent elegant proof by Kiselev, Nazarov and Volberg of the well-posedness of critically dissipative 2D quasi-geostrophic equation to the super-critical case. We prove that if the initial value is smooth and periodic, and $\left\| \nabla θ_0 \right\|_{L^{\infty}}^{1 - 2 s} \left\| θ_0 \right\|_{L^{\infty}}^{2 s}$ is small, where $s$ is the power of the fractional Laplacian, then no finite time singularity will occur for the super-critically dissipative 2D quasi-geostrophic equation.

math.AP↗
arXiv2006

Level Set Dynamics and the Non-blowup of the 2D Quasi-geostrophic Equation

In this article we apply the technique proposed in Deng-Hou-Yu (Comm. PDE, 2005) to study the level set dynamics of the 2D quasi-geostrophic equation. Under certain assumptions on the local geometric regularity of the level sets of $θ$, we obtain global regularity results with improved growth estimate on $| \nabla^{\bot} θ|$. We further perform numerical simulations to study the local geometric properties of the level sets near the region of maximum $| \nabla^{\bot} θ|$. The numerical results indicate that the assumptions on the local geometric regularity of the level sets of $θ$ in our theorems are satisfied. Therefore these theorems provide a good explanation of the double exponential growth of $| \nabla^{\bot} θ|$ observed in this and past numerical simulations.

math.AP↗
arXiv2004

Geometric properties and non-blowup of 3-D incompressible Euler flow

By exploring a local geometric property of the vorticity field along a vortex filament, we establish a sharp relationship between the geometric properties of the vorticity field and the maximum vortex stretching. This new understanding leads to an improved result of the global existence of the 3-D Euler equation under mild assumptions that are consistent with the observations from recent numerical computations.

math-ph↗