-
Theoretical Insights into Line Graph Transformation on Graph Learning
Authors:
Fan Yang,
Xingyue Huang
Abstract:
Line graph transformation has been widely studied in graph theory, where each node in a line graph corresponds to an edge in the original graph. This has inspired a series of graph neural networks (GNNs) applied to transformed line graphs, which have proven effective in various graph representation learning tasks. However, there is limited theoretical study on how line graph transformation affects…
▽ More
Line graph transformation has been widely studied in graph theory, where each node in a line graph corresponds to an edge in the original graph. This has inspired a series of graph neural networks (GNNs) applied to transformed line graphs, which have proven effective in various graph representation learning tasks. However, there is limited theoretical study on how line graph transformation affects the expressivity of GNN models. In this study, we focus on two types of graphs known to be challenging to the Weisfeiler-Leman (WL) tests: Cai-Fürer-Immerman (CFI) graphs and strongly regular graphs, and show that applying line graph transformation helps exclude these challenging graph properties, thus potentially assist WL tests in distinguishing these graphs. We empirically validate our findings by conducting a series of experiments that compare the accuracy and efficiency of graph isomorphism tests and GNNs on both line-transformed and original graphs across these graph structure types.
△ Less
Submitted 21 October, 2024;
originally announced October 2024.
-
Decoupled finite element methods for a fourth-order exterior differential equation
Authors:
Xuewei Cui,
Xuehai Huang
Abstract:
This paper focuses on decoupled finite element methods for the fourth-order exterior differential equation. Based on differential complexes and the Helmholtz decomposition, the fourth-order exterior differential equation is decomposed into two second-order exterior differential equations and one generalized Stokes equation. A family of conforming finite element methods are developed for the decoup…
▽ More
This paper focuses on decoupled finite element methods for the fourth-order exterior differential equation. Based on differential complexes and the Helmholtz decomposition, the fourth-order exterior differential equation is decomposed into two second-order exterior differential equations and one generalized Stokes equation. A family of conforming finite element methods are developed for the decoupled formulation. Numerical results are provided for verifying the decoupled finite element methods of the biharmonic equation in three dimensions.
△ Less
Submitted 12 October, 2024;
originally announced October 2024.
-
A Mathematics-Inspired Learning-to-Optimize Framework for Decentralized Optimization
Authors:
Yutong He,
Qiulin Shang,
Xinmeng Huang,
Jialin Liu,
Kun Yuan
Abstract:
Most decentralized optimization algorithms are handcrafted. While endowed with strong theoretical guarantees, these algorithms generally target a broad class of problems, thereby not being adaptive or customized to specific problem features. This paper studies data-driven decentralized algorithms trained to exploit problem features to boost convergence. Existing learning-to-optimize methods typica…
▽ More
Most decentralized optimization algorithms are handcrafted. While endowed with strong theoretical guarantees, these algorithms generally target a broad class of problems, thereby not being adaptive or customized to specific problem features. This paper studies data-driven decentralized algorithms trained to exploit problem features to boost convergence. Existing learning-to-optimize methods typically suffer from poor generalization or prohibitively vast search spaces. In addition, the vast search space of communicating choices and final goal to reach the global solution via limited neighboring communication cast more challenges in decentralized settings. To resolve these challenges, this paper first derives the necessary conditions that successful decentralized algorithmic rules need to satisfy to achieve both optimality and consensus. Based on these conditions, we propose a novel Mathematics-inspired Learning-to-optimize framework for Decentralized optimization (MiLoDo). Empirical results demonstrate that MiLoDo-trained algorithms outperform handcrafted algorithms and exhibit strong generalizations. Algorithms learned via MiLoDo in 100 iterations perform robustly when running 100,000 iterations during inferences. Moreover, MiLoDo-trained algorithms on synthetic datasets perform well on problems involving real data, higher dimensions, and different loss functions.
△ Less
Submitted 2 October, 2024;
originally announced October 2024.
-
Concentration of information on discrete groups
Authors:
Jonathan Hermon,
Xiangying Huang,
Francesco Pedrotti,
Justin Salez
Abstract:
Motivated by the Asymptotic Equipartition Property and its recently discovered role in the cutoff phenomenon, we initiate the systematic study of varentropy on discrete groups. Our main result is an approximate tensorization inequality which asserts that the varentropy of any conjugacy-invariant random walk is, up to a universal multiplicative constant, at most that of the free Abelian random walk…
▽ More
Motivated by the Asymptotic Equipartition Property and its recently discovered role in the cutoff phenomenon, we initiate the systematic study of varentropy on discrete groups. Our main result is an approximate tensorization inequality which asserts that the varentropy of any conjugacy-invariant random walk is, up to a universal multiplicative constant, at most that of the free Abelian random walk with the same jump rates. In particular, it is always bounded by the number d of generators, uniformly in time and in the size of the group. This universal estimate is sharp and can be seen as a discrete analogue of a celebrated result of Bobkov and Madiman concerning random d-dimensional vectors with a log-concave density (AOP 2011). A key ingredient in our proof is the fact that conjugacy-invariant random walks have non-negative Bakry-Émery curvature, a result which seems new and of independent interest.
△ Less
Submitted 25 September, 2024;
originally announced September 2024.
-
Higher order Hardy-Rellich identities
Authors:
Xia Huang,
Dong Ye
Abstract:
In this paper, we show Hardy-Rellich identities for polyharmonic operators $Δ^m$ and radial Laplacian $Δ_r^m$ in $\mathbb{R}^n$ with Hardy-Hénon weight $|x|^α$ for all $m, n\in \mathbb{N}, α\in \mathbb{R}$. Moreover, the iterative method is applied to give Hardy-Rellich equalities with general weights on Riemannian manifolds. These identities provide naturally an alternative approach to obtain and…
▽ More
In this paper, we show Hardy-Rellich identities for polyharmonic operators $Δ^m$ and radial Laplacian $Δ_r^m$ in $\mathbb{R}^n$ with Hardy-Hénon weight $|x|^α$ for all $m, n\in \mathbb{N}, α\in \mathbb{R}$. Moreover, the iterative method is applied to give Hardy-Rellich equalities with general weights on Riemannian manifolds. These identities provide naturally an alternative approach to obtain and improve Hardy-Rellich type inequalities. As example of application, we extend several Rellich inequalities of Tertikas-Zographopoulos (Adv. Math. 2007) to the weighted case; using equality with weights involving logarithmic, we show another new weighted Rellich estimate between integrals of $Δu$ and $|\nabla u|$; we establish also a Rellich identity involving the Laplace-Beltrami operator $Δ_\mathbb{H}$ and the radial Laplacian $Δ_{ρ, \mathbb{H}}$ of the hyperbolic space $\mathbb{H}^n$, which yields in particular brand-new Rellich inequalities for $\|Δ_\mathbb{H} u\|$ in $\mathbb{H}^3$ and $\mathbb{H}^4$.
△ Less
Submitted 19 September, 2024;
originally announced September 2024.
-
Bounding smooth Levi-flat hypersurfaces in a Stein manifold
Authors:
Hanlong Fang,
Xiaojun Huang,
Wanke Yin,
Zhengyi Zhou
Abstract:
This paper is concerned with the problem of constructing a smooth Levi-flat hypersurface locally or globally attached to a real codimension two submanifold in $\mathbb C^{n+1}$, or more generally in a Stein manifold, with elliptic CR singularities, a research direction originated from a fundamental and classical paper of E. Bishop. Earlier works along these lines include those by many prominent ma…
▽ More
This paper is concerned with the problem of constructing a smooth Levi-flat hypersurface locally or globally attached to a real codimension two submanifold in $\mathbb C^{n+1}$, or more generally in a Stein manifold, with elliptic CR singularities, a research direction originated from a fundamental and classical paper of E. Bishop. Earlier works along these lines include those by many prominent mathematicians working both on complex analysis and geometry. We prove that a compact smooth (or, real analytic) real codimension two submanifold $M$, that is contained in the boundary of a smoothly bounded strongly pseudoconvex domain, with a natural and necessary condition called CR non-minimal condition at CR points and with two elliptic CR singular points bounds a smooth-up-to-boundary (real analytic-up-to-boundary, respectively) Levi-flat hypersurface $\widehat{M}$. This answers a well-known question left open from the work of Dolbeault-Tomassini-Zaitsev, or a generalized version of a problem already asked by Bishop in 1965. Our study here reveals an intricate interaction of several complex analysis with other fields such as symplectic geometry and foliation theory.
△ Less
Submitted 12 September, 2024;
originally announced September 2024.
-
Weyl laws for Schrödinger operators on compact manifolds with boundary
Authors:
Xiaoqi Huang,
Xing Wang,
Cheng Zhang
Abstract:
We prove Weyl laws for Schrödinger operators with critically singular potentials on compact manifolds with boundary. We also improve the Weyl remainder estimates under the condition that the set of all periodic geodesic billiards has measure 0. These extend the classical results by Seeley, Ivrii and Melrose. The proof uses the Gaussian heat kernel bounds for short times and a perturbation argument…
▽ More
We prove Weyl laws for Schrödinger operators with critically singular potentials on compact manifolds with boundary. We also improve the Weyl remainder estimates under the condition that the set of all periodic geodesic billiards has measure 0. These extend the classical results by Seeley, Ivrii and Melrose. The proof uses the Gaussian heat kernel bounds for short times and a perturbation argument involving the wave equation.
△ Less
Submitted 8 September, 2024;
originally announced September 2024.
-
Finite time blowup of strong solutions to the two dimensional MHD equations
Authors:
Xiangdi Huang,
Zhouping Xin,
Wei Yan
Abstract:
Whether the smooth solution of the multi-dimensional viscous compressible fluids will blow-up in finite time has always been a chanllenging problem. In the recent work\cite{FM}, Merle et al. proved that there are smooth solutions to the 2D radially symmetric compressible Navier-Stokes equations which will inevitably form shell singularities in finite time.\\ \indent In this article, we first prove…
▽ More
Whether the smooth solution of the multi-dimensional viscous compressible fluids will blow-up in finite time has always been a chanllenging problem. In the recent work\cite{FM}, Merle et al. proved that there are smooth solutions to the 2D radially symmetric compressible Navier-Stokes equations which will inevitably form shell singularities in finite time.\\ \indent In this article, we first prove the existence of local strong solutions that allow vacuum for the two-dimensional viscous compressible MHD equations on bounded domains without magnetic diffusion. Furthermore, it is shown that if the initial data are radial symmetric and its vacuum set contains a ball centered at the origin where the total magnetic field is non-trivial, then the radial symmetric strong solution to the initial boundary value problem will definitely blow up in finite time. This is the first example for the formation of finite time singularity of strong solutions that allows interior vacuum of a viscous compressible fluid.
△ Less
Submitted 5 September, 2024;
originally announced September 2024.
-
Uniform in Time Propagation of Chaos for Mean Field Particle System with Interacting Noise and Partially Dissipative Drifts
Authors:
Xing Huang
Abstract:
In this paper, long time quantitative propagation of chaos in $L^1$-Wasserstein distance for mean field interacting particle system is derived, where the diffusion coefficient is allowed to be interacting and the drift is assumed to be partially dissipative. The main tool relies on reflection coupling, the gradient estimate of the decoupled SDEs, and the Duhamel formula for two semigroups associat…
▽ More
In this paper, long time quantitative propagation of chaos in $L^1$-Wasserstein distance for mean field interacting particle system is derived, where the diffusion coefficient is allowed to be interacting and the drift is assumed to be partially dissipative. The main tool relies on reflection coupling, the gradient estimate of the decoupled SDEs, and the Duhamel formula for two semigroups associated to two time-inhomogeneous diffusion processes on $(\mathbb{R}^d)^N$. Moreover, the long time quantitative propagation of chaos in $L^η$($η\in(0,1)$)-Wasserstein distance is also obtained.
△ Less
Submitted 3 September, 2024;
originally announced September 2024.
-
Solving Integrated Process Planning and Scheduling Problem via Graph Neural Network Based Deep Reinforcement Learning
Authors:
Hongpei Li,
Han Zhang,
Ziyan He,
Yunkai Jia,
Bo Jiang,
Xiang Huang,
Dongdong Ge
Abstract:
The Integrated Process Planning and Scheduling (IPPS) problem combines process route planning and shop scheduling to achieve high efficiency in manufacturing and maximize resource utilization, which is crucial for modern manufacturing systems. Traditional methods using Mixed Integer Linear Programming (MILP) and heuristic algorithms can not well balance solution quality and speed when solving IPPS…
▽ More
The Integrated Process Planning and Scheduling (IPPS) problem combines process route planning and shop scheduling to achieve high efficiency in manufacturing and maximize resource utilization, which is crucial for modern manufacturing systems. Traditional methods using Mixed Integer Linear Programming (MILP) and heuristic algorithms can not well balance solution quality and speed when solving IPPS. In this paper, we propose a novel end-to-end Deep Reinforcement Learning (DRL) method. We model the IPPS problem as a Markov Decision Process (MDP) and employ a Heterogeneous Graph Neural Network (GNN) to capture the complex relationships among operations, machines, and jobs. To optimize the scheduling strategy, we use Proximal Policy Optimization (PPO). Experimental results show that, compared to traditional methods, our approach significantly improves solution efficiency and quality in large-scale IPPS instances, providing superior scheduling strategies for modern intelligent manufacturing systems.
△ Less
Submitted 2 September, 2024;
originally announced September 2024.
-
Free boundary value problem for the radial symmetric compressible isentropic Navier-Stokes equations with density-dependent viscosity
Authors:
Xiangdi Huang,
Weili Meng,
Anchun Ni
Abstract:
This paper is devoted to the study of free-boundary-value problem of the compressible Naiver-Stokes system with density-dependent viscosities $μ=const>0,λ=ρ^β$ which was first introduced by Vaigant-Kazhikhov \cite{1995 Vaigant-Kazhikhov-SMJ} in 1995. By assuming the endpoint case $β=1$ in the radially spherical symmetric setting, we prove the (a priori) expanding rate of the free boundary is algeb…
▽ More
This paper is devoted to the study of free-boundary-value problem of the compressible Naiver-Stokes system with density-dependent viscosities $μ=const>0,λ=ρ^β$ which was first introduced by Vaigant-Kazhikhov \cite{1995 Vaigant-Kazhikhov-SMJ} in 1995. By assuming the endpoint case $β=1$ in the radially spherical symmetric setting, we prove the (a priori) expanding rate of the free boundary is algebraic for multi-dimensional flow, and particularly establish the global existence of strong solution of the two-dimensional system for any large initial data. This also improves the previous work of Li-Zhang \cite{2016 Li-Zhang-JDE} where they proved the similar result for $β>1$. The main ingredients of this article is making full use of the geometric advantange of domain as well as the critical space dimension two.
△ Less
Submitted 31 August, 2024;
originally announced September 2024.
-
Symmetry of positive solutions to biharmonic Lane-Emden equation with singular set
Authors:
Xia Huang,
Yuan Li,
Xianmei Zhou
Abstract:
In this paper, we are devoted to studying the positive weak, punctured or distributional solutions to the biharmonic Lane-Emden equation
\begin{equation*}
Δ^{2} u=u^{p}
\quad
\quad \text{in} \ \mathbb{R}^{N}\setminus Z,
\end{equation*}
where $N\geq5$, $1<p\leq\frac{N+4}{N-4}$, and the singular set $Z$ represents a closed and proper subset of $ \left\lbrace x_{1}=0\right\rbrace $. The s…
▽ More
In this paper, we are devoted to studying the positive weak, punctured or distributional solutions to the biharmonic Lane-Emden equation
\begin{equation*}
Δ^{2} u=u^{p}
\quad
\quad \text{in} \ \mathbb{R}^{N}\setminus Z,
\end{equation*}
where $N\geq5$, $1<p\leq\frac{N+4}{N-4}$, and the singular set $Z$ represents a closed and proper subset of $ \left\lbrace x_{1}=0\right\rbrace $. The symmetry and monotonicity properties of the singular solutions will be given by taking advantage of the moving plane method and the approach of moving spheres.
△ Less
Submitted 13 August, 2024;
originally announced August 2024.
-
Global large strong solutions to the compressible Navier-Stokes equations with density-dependent viscosities, case I: isentropic flows
Authors:
Xiangdi Huang,
Jiaxu Li,
Rong Zhang
Abstract:
In this paper, we consider the Cauchy problem for the three-dimensional barotropic compressible Navier-Stokes equations with density-dependent viscosities. By considering the system as an elliptic-dominated structure and defining suitable energy functionals, after the elaborate index analysis, we establish the global existence of strong solutions as long as the initial data is large enough. This i…
▽ More
In this paper, we consider the Cauchy problem for the three-dimensional barotropic compressible Navier-Stokes equations with density-dependent viscosities. By considering the system as an elliptic-dominated structure and defining suitable energy functionals, after the elaborate index analysis, we establish the global existence of strong solutions as long as the initial data is large enough. This is a big contrast to the classical results where the initial data is a small perturbation of some resting states.
△ Less
Submitted 8 August, 2024;
originally announced August 2024.
-
Restriction of Schrödinger eigenfunctions to submanifolds
Authors:
Xiaoqi Huang,
Xing Wang,
Cheng Zhang
Abstract:
Burq-Gérard-Tzvetkov and Hu established $L^p$ estimates for the restriction of Laplace-Beltrami eigenfunctions to submanifolds. We investigate the eigenfunctions of the Schrödinger operators with critically singular potentials, and estimate the $L^p$ norms and period integrals for their restriction to submanifolds. Recently, Blair-Sire-Sogge obtained global $L^p$ bounds for Schrödinger eigenfuncti…
▽ More
Burq-Gérard-Tzvetkov and Hu established $L^p$ estimates for the restriction of Laplace-Beltrami eigenfunctions to submanifolds. We investigate the eigenfunctions of the Schrödinger operators with critically singular potentials, and estimate the $L^p$ norms and period integrals for their restriction to submanifolds. Recently, Blair-Sire-Sogge obtained global $L^p$ bounds for Schrödinger eigenfunctions by the resolvent method. Due to the Sobolev trace inequalities, the resolvent method cannot work for submanifolds of all dimensions. We get around this difficulty and establish spectral projection bounds by the wave kernel techniques and the bootstrap argument involving an induction on the dimensions of the submanifolds.
△ Less
Submitted 4 August, 2024;
originally announced August 2024.
-
Global large strong solution of the 3D inhomogeneous Navier-Stokes equations with density-dependent viscosity
Authors:
Xiangdi Huang,
Jiaxu Li,
Rong Zhang
Abstract:
This paper concerns the Dirichlet problem of three-dimensional inhomogeneous Navier-Stokes equations with density-dependent viscosity. When the viscosity coefficient $μ(ρ)$ is a power function of the density ($μ(ρ)=μρ^α$ with $α>1$), it is proved that the system will admit a unique global strong solution as long as the initial data are sufficiently large. This is the first result concerning the ex…
▽ More
This paper concerns the Dirichlet problem of three-dimensional inhomogeneous Navier-Stokes equations with density-dependent viscosity. When the viscosity coefficient $μ(ρ)$ is a power function of the density ($μ(ρ)=μρ^α$ with $α>1$), it is proved that the system will admit a unique global strong solution as long as the initial data are sufficiently large. This is the first result concerning the existence of large strong solution for the inhomogeneous Navier-Stokes equations in three dimensions.
△ Less
Submitted 1 August, 2024;
originally announced August 2024.
-
On the measure-theoretic pressure associated with Feldman-Katok metric and max-mean metric
Authors:
Zhongxuan Yang,
Xiaojun Huang,
Jiajun Zhang
Abstract:
In this manuscript, we present modified Feldman-katok metric and modified max-mean metric, then apply them to examine the measure-theoretic pressure and topological pressure.
In this manuscript, we present modified Feldman-katok metric and modified max-mean metric, then apply them to examine the measure-theoretic pressure and topological pressure.
△ Less
Submitted 23 July, 2024;
originally announced July 2024.
-
Strichartz estimates for the Schrödinger equation on compact manifolds with nonpositive sectional curvature
Authors:
Xiaoqi Huang,
Christopher D. Sogge
Abstract:
We obtain improved Strichartz estimates for solutions of the Schrödinger equation on compact manifolds with nonpositive sectional curvatures which are related to the classical universal results of Burq, Gérard and Tzvetkov [11]. More explicitly, we are able refine the arguments in the recent work of Blair and the authors [3] to obtain no-loss $L^p_tL^{q}_{x}$-estimates on intervals of length…
▽ More
We obtain improved Strichartz estimates for solutions of the Schrödinger equation on compact manifolds with nonpositive sectional curvatures which are related to the classical universal results of Burq, Gérard and Tzvetkov [11]. More explicitly, we are able refine the arguments in the recent work of Blair and the authors [3] to obtain no-loss $L^p_tL^{q}_{x}$-estimates on intervals of length $\log λ\cdot λ^{-1} $ for all {\em admissible} pairs $(p,q)$ when the initial data have frequencies comparable to $λ$, which, given the role of the Ehrenfest time, is the natural analog in this setting of the universal results in [11]. We achieve this log-gain over the universal estimates by applying the Keel-Tao theorem along with improved global kernel estimates for microlocalized operators which exploit the geometric assumptions.
△ Less
Submitted 17 July, 2024;
originally announced July 2024.
-
Proof of Lew's conjecture on the spectral gaps of simplicial complexes
Authors:
Xiongfeng Zhan,
Xueyi Huang,
Huiqiu Lin
Abstract:
As a generalization of graph Laplacians to higher dimensions, the combinatorial Laplacians of simplicial complexes have garnered increasing attention. Let $X$ be a simplicial complex on vertex set $V$ of size $n$, and let $X(k)$ denote the set of all $k$-dimensional simplices of $X$. The $k$-th spectral gap $μ_k(X)$ is the smallest eigenvalue of the reduced $k$-dimensional Laplacian of $X$. For an…
▽ More
As a generalization of graph Laplacians to higher dimensions, the combinatorial Laplacians of simplicial complexes have garnered increasing attention. Let $X$ be a simplicial complex on vertex set $V$ of size $n$, and let $X(k)$ denote the set of all $k$-dimensional simplices of $X$. The $k$-th spectral gap $μ_k(X)$ is the smallest eigenvalue of the reduced $k$-dimensional Laplacian of $X$. For any $k\geq -1$, Lew [J. Combin. Theory Ser. A 169 (2020) 105127] established a lower bound for $μ_k(X)$: $$μ_k(X)\geq (d+1)\left(\min_{σ\in X(k)}°_X(σ)+k+1\right)-dn\geq (d+1)(k+1)-dn,$$ where $°_X(σ)$ and $d$ denote the degree of $σ$ in $X$ and the maximal dimension of a missing face of $X$, respectively. In this paper, we identify the unique simplicial complex that achieves the lower bound of the $k$-th spectral gap, $(d+1)(k+1)-dn$, for some $k$, thereby confirming a conjecture proposed by Lew.
△ Less
Submitted 29 July, 2024; v1 submitted 14 July, 2024;
originally announced July 2024.
-
On distance-regular Cayley graphs over abelian groups of rank $2$
Authors:
Xiongfeng Zhan,
Xueyi Huang,
Lu Lu
Abstract:
The problem of constructing or characterizing strongly regular Cayley graphs (or equivalently, regular partial difference sets) has garnered significant attention over the past half-century. In [European J. Combin. 24 (2023) 777--784], Miklavič and Potočnik expanded upon this field, achieving a complete characterization of distance-regular Cayley graphs over cyclic groups by employing the method o…
▽ More
The problem of constructing or characterizing strongly regular Cayley graphs (or equivalently, regular partial difference sets) has garnered significant attention over the past half-century. In [European J. Combin. 24 (2023) 777--784], Miklavič and Potočnik expanded upon this field, achieving a complete characterization of distance-regular Cayley graphs over cyclic groups by employing the method of Schur rings. In [J. Combin. Theory Ser. B 97 (2007) 14--33], they formally proposed the problem of characterizing distance-regular Cayley graphs for any given class of groups. In this context, abelian groups are of particular importance, as a considerable number of distance-regular graphs with classical parameters are precisely Cayley graphs over abelian groups. In this paper, we address the Miklavič-Potočnik problem for abelian groups of rank $2$. More specifically, we identify all distance-regular Cayley graphs over the group $\mathbb{Z}_n \oplus \mathbb{Z}_p$, where $p$ is an odd prime. Our proof utilizes novel tools such as polynomial addition sets, Desarguesian affine planes, and the duality of Schur rings over abelian groups.
△ Less
Submitted 31 July, 2024; v1 submitted 10 July, 2024;
originally announced July 2024.
-
PDEformer-1: A Foundation Model for One-Dimensional Partial Differential Equations
Authors:
Zhanhong Ye,
Xiang Huang,
Leheng Chen,
Zining Liu,
Bingyang Wu,
Hongsheng Liu,
Zidong Wang,
Bin Dong
Abstract:
This paper introduces PDEformer-1, a versatile neural solver capable of simultaneously addressing various partial differential equations (PDEs). With the PDE represented as a computational graph, we facilitate the seamless integration of symbolic and numeric information inherent in a PDE. A graph Transformer and an implicit neural representation (INR) are employed subsequently to generate mesh-fre…
▽ More
This paper introduces PDEformer-1, a versatile neural solver capable of simultaneously addressing various partial differential equations (PDEs). With the PDE represented as a computational graph, we facilitate the seamless integration of symbolic and numeric information inherent in a PDE. A graph Transformer and an implicit neural representation (INR) are employed subsequently to generate mesh-free predicted solutions. We generated a dataset with up to three million samples involving diverse one-dimensional PDEs to pretrain our model. Compared with baseline models trained specifically on benchmark datasets, our pretrained model achieves comparable accuracy via zero-shot inference, and the advantage expands after finetuning. For PDEs new or unseen in the pretraining stage, our model can adapt quickly by finetuning on a relatively small set of examples from the target equation. Additionally, PDEformer-1 demonstrates promising results in the inverse problem of PDE scalar coefficient recovery and coefficient field recovery.
△ Less
Submitted 9 July, 2024;
originally announced July 2024.
-
Bubble solution for the critical Hartree equation in pierced domain
Authors:
Marco Ghimenti,
Xiaomeng Huang,
Angela Pistoia
Abstract:
In this article, we establish the existence of solutions to the following critical Hartree equation \begin{align*} \begin{cases} -Δu=\left(\int_{Ω_\varepsilon}\frac{u^{2_μ^*}}{|x-y|^μ}dy\right)u^{2_μ^*-1}, &\text{ in } Ω_\varepsilon, \\ u=0, &\text{ on } \partialΩ_\varepsilon, \end{cases} \end{align*} where $2_μ^*=\frac{2N-μ}{N-2}$ is the upper critical exponent in the sense of the Hardy-Littlewoo…
▽ More
In this article, we establish the existence of solutions to the following critical Hartree equation \begin{align*} \begin{cases} -Δu=\left(\int_{Ω_\varepsilon}\frac{u^{2_μ^*}}{|x-y|^μ}dy\right)u^{2_μ^*-1}, &\text{ in } Ω_\varepsilon, \\ u=0, &\text{ on } \partialΩ_\varepsilon, \end{cases} \end{align*} where $2_μ^*=\frac{2N-μ}{N-2}$ is the upper critical exponent in the sense of the Hardy-Littlewood-Sobolev inequality, $N\geq 5$, $0<μ<4$ with $μ$ sufficiently close to $0$, $Ω_\varepsilon:=Ω\backslash B(0,\varepsilon)$ and $Ω$ is a bounded smooth domain in $\mathbb{R}^N$, which contains the origin, and $\varepsilon$ is a positive parameter. As $\varepsilon$ goes to zero, we construct bubble solution which blows up at the origin.
△ Less
Submitted 2 July, 2024;
originally announced July 2024.
-
Convergence rates for random feature neural network approximation in molecular dynamics
Authors:
Xin Huang,
Petr Plechac,
Mattias Sandberg,
Anders Szepessy
Abstract:
Random feature neural network approximations of the potential in Hamiltonian systems yield approximations of molecular dynamics correlation observables that have the expected error $\mathcal{O}\big((K^{-1}+J^{-1/2})^{\frac{1}{2}}\big)$, for networks with $K$ nodes using $J$ data points, provided the Hessians of the potential and the observables are bounded. The loss function is based on the least…
▽ More
Random feature neural network approximations of the potential in Hamiltonian systems yield approximations of molecular dynamics correlation observables that have the expected error $\mathcal{O}\big((K^{-1}+J^{-1/2})^{\frac{1}{2}}\big)$, for networks with $K$ nodes using $J$ data points, provided the Hessians of the potential and the observables are bounded. The loss function is based on the least squares error of the potential and regularizations, with the data points sampled from the Gibbs density. The proof uses an elementary new derivation of the generalization error for random feature networks that does not apply the Rademacher or related complexities.
△ Less
Submitted 20 June, 2024;
originally announced June 2024.
-
On Morita equivalences with endopermutation source and isotypies
Authors:
Xin Huang
Abstract:
We introduce a new type of equivalence between blocks of finite group algebras called an almost isotypy. An almost isotypy restricts to a weak isotypy in Broué's original definition, and it is slightly weaker than Linckelmann's version. We show that a bimodule of two block algebras of finite groups - which has an endopermutation module as a source and which induces a Morita equivalence - gives ris…
▽ More
We introduce a new type of equivalence between blocks of finite group algebras called an almost isotypy. An almost isotypy restricts to a weak isotypy in Broué's original definition, and it is slightly weaker than Linckelmann's version. We show that a bimodule of two block algebras of finite groups - which has an endopermutation module as a source and which induces a Morita equivalence - gives rise to an almost isotypy if and only if the character values of a (hence any) source are rational integers. This generalises a previous result of Huang and Zhou. Consequently, if two blocks are Morita equivalent via a bimodule with endopermutation source, then they are almost isotypic.
△ Less
Submitted 20 June, 2024; v1 submitted 30 May, 2024;
originally announced May 2024.
-
One-Shot Safety Alignment for Large Language Models via Optimal Dualization
Authors:
Xinmeng Huang,
Shuo Li,
Edgar Dobriban,
Osbert Bastani,
Hamed Hassani,
Dongsheng Ding
Abstract:
The growing safety concerns surrounding Large Language Models (LLMs) raise an urgent need to align them with diverse human preferences to simultaneously enhance their helpfulness and safety. A promising approach is to enforce safety constraints through Reinforcement Learning from Human Feedback (RLHF). For such constrained RLHF, common Lagrangian-based primal-dual policy optimization methods are c…
▽ More
The growing safety concerns surrounding Large Language Models (LLMs) raise an urgent need to align them with diverse human preferences to simultaneously enhance their helpfulness and safety. A promising approach is to enforce safety constraints through Reinforcement Learning from Human Feedback (RLHF). For such constrained RLHF, common Lagrangian-based primal-dual policy optimization methods are computationally expensive and often unstable. This paper presents a dualization perspective that reduces constrained alignment to an equivalent unconstrained alignment problem. We do so by pre-optimizing a smooth and convex dual function that has a closed form. This shortcut eliminates the need for cumbersome primal-dual policy iterations, thus greatly reducing the computational burden and improving training stability. Our strategy leads to two practical algorithms in model-based and preference-based scenarios (MoCAN and PeCAN, respectively). A broad range of experiments demonstrate the effectiveness of our methods.
△ Less
Submitted 15 September, 2024; v1 submitted 29 May, 2024;
originally announced May 2024.
-
Distributed Bilevel Optimization with Communication Compression
Authors:
Yutong He,
Jie Hu,
Xinmeng Huang,
Songtao Lu,
Bin Wang,
Kun Yuan
Abstract:
Stochastic bilevel optimization tackles challenges involving nested optimization structures. Its fast-growing scale nowadays necessitates efficient distributed algorithms. In conventional distributed bilevel methods, each worker must transmit full-dimensional stochastic gradients to the server every iteration, leading to significant communication overhead and thus hindering efficiency and scalabil…
▽ More
Stochastic bilevel optimization tackles challenges involving nested optimization structures. Its fast-growing scale nowadays necessitates efficient distributed algorithms. In conventional distributed bilevel methods, each worker must transmit full-dimensional stochastic gradients to the server every iteration, leading to significant communication overhead and thus hindering efficiency and scalability. To resolve this issue, we introduce the first family of distributed bilevel algorithms with communication compression. The primary challenge in algorithmic development is mitigating bias in hypergradient estimation caused by the nested structure. We first propose C-SOBA, a simple yet effective approach with unbiased compression and provable linear speedup convergence. However, it relies on strong assumptions on bounded gradients. To address this limitation, we explore the use of moving average, error feedback, and multi-step compression in bilevel optimization, resulting in a series of advanced algorithms with relaxed assumptions and improved convergence properties. Numerical experiments show that our compressed bilevel algorithms can achieve $10\times$ reduction in communication overhead without severe performance degradation.
△ Less
Submitted 29 May, 2024;
originally announced May 2024.
-
Quantum vertex algebra associated to quantum toroidal $\mathfrak{gl}_N$
Authors:
Fulin Chen,
Xin Huang,
Fei Kong,
Shaobin Tan
Abstract:
In this paper, we associate the quantum toroidal algebra $\mathcal{E}_N$ of type $\mathfrak{gl}_N$ with quantum vertex algebra through equivariant $φ$-coordinated quasi modules. More precisely, for every $\ell\in \mathbb{C}$, by deforming the universal affine vertex algebra of $\mathfrak{sl}_\infty$, we construct an $\hbar$-adic quantum $\Z$-vertex algebra…
▽ More
In this paper, we associate the quantum toroidal algebra $\mathcal{E}_N$ of type $\mathfrak{gl}_N$ with quantum vertex algebra through equivariant $φ$-coordinated quasi modules. More precisely, for every $\ell\in \mathbb{C}$, by deforming the universal affine vertex algebra of $\mathfrak{sl}_\infty$, we construct an $\hbar$-adic quantum $\Z$-vertex algebra $V_{\widehat{\mathfrak{sl}}_{\infty},\hbar}(\ell,0)$. Then we prove that the category of restricted $\mathcal{E}_N$-modules of level $\ell$ is canonically isomorphic to that of equivariant $φ$-coordinated quasi $V_{\widehat{\mathfrak{sl}}_{\infty},\hbar}(\ell,0)$-modules.
△ Less
Submitted 15 May, 2024;
originally announced May 2024.
-
Optimal asymptotic volume ratio for noncompact 3-manifolds with asymptotically nonnegative Ricci curvature and a uniformly positive scalar curvature lower bound
Authors:
Xian-Tao Huang,
Shuai Liu
Abstract:
In this paper, we study 3-dimensional complete non-compact Riemannian manifolds with asymptotically nonnegative Ricci curvature and a uniformly positive scalar curvature lower bound. Our main result is that, if this manifold has $k$ ends and finite first Betti number, then it has at most linear volume growth, and furthermore, if the negative part of Ricci curvature decays sufficiently fast at infi…
▽ More
In this paper, we study 3-dimensional complete non-compact Riemannian manifolds with asymptotically nonnegative Ricci curvature and a uniformly positive scalar curvature lower bound. Our main result is that, if this manifold has $k$ ends and finite first Betti number, then it has at most linear volume growth, and furthermore, if the negative part of Ricci curvature decays sufficiently fast at infinity, then we have an optimal asymptotic volume ratio $\limsup_{r\rightarrow\infty}\frac{\mathrm{Vol}(B(p, r))}{r}\leq4kπ$. In particular, our results apply to 3-dimensional complete non-compact Riemannian manifolds with nonnegative Ricci curvature and a uniformly positive scalar curvature lower bound.
△ Less
Submitted 5 June, 2024; v1 submitted 15 May, 2024;
originally announced May 2024.
-
Asymptotic behavior toward viscous shock for impermeable wall and inflow problem of barotropic Navier-Stokes equations
Authors:
Xushan Huang,
Moon-Jin Kang,
Jeongho Kim,
Hobin Lee
Abstract:
We consider the compressible barotropic Navier-Stokes equations in a half-line and study the time-asymptotic behavior toward the outgoing viscous shock wave. Precisely, we consider the two boundary problems: impermeable wall and inflow problems, where the velocity at the boundary is given as a constant state. For both problems, when the asymptotic profile determined by the prescribed constant stat…
▽ More
We consider the compressible barotropic Navier-Stokes equations in a half-line and study the time-asymptotic behavior toward the outgoing viscous shock wave. Precisely, we consider the two boundary problems: impermeable wall and inflow problems, where the velocity at the boundary is given as a constant state. For both problems, when the asymptotic profile determined by the prescribed constant states at the boundary and far-fields is a viscous shock, we show that the solution asymptotically converges to the shifted viscous shock profiles uniformly in space, under the condition that initial perturbation is small enough in H1 norm. We do not impose the zero mass condition on initial data, which improves the previous results by Matsumura and Mei [20] for impermeable case, and by Huang, Matsumura and Shi [8] for inflow case. Moreover, for the inflow case, we remove the assumption in [8]. Our results are based on the method of a-contraction with shifts, as the first extension of the method to the boundary value problems.
△ Less
Submitted 6 May, 2024; v1 submitted 6 May, 2024;
originally announced May 2024.
-
Quasimode concentration on compact space forms
Authors:
Xiaoqi Huang,
Christopher D. Sogge
Abstract:
We show that the upper bounds for the $L^2$-norms of $L^1$-normalized quasimodes that we obtained in [9] are always sharp on any compact space form. This allows us to characterize compact manifolds of constant sectional curvature using the decay rates of lower bounds of $L^1$-norms of $L^2$-normalized log-quasimodes fully resolving a problem initiated by the second author and Zelditch [15]. We are…
▽ More
We show that the upper bounds for the $L^2$-norms of $L^1$-normalized quasimodes that we obtained in [9] are always sharp on any compact space form. This allows us to characterize compact manifolds of constant sectional curvature using the decay rates of lower bounds of $L^1$-norms of $L^2$-normalized log-quasimodes fully resolving a problem initiated by the second author and Zelditch [15]. We are also able to characterize such manifolds by the concentration of quasimodes near periodic geodesics as measured by $L^2$-norms over thin geodesic tubes.
△ Less
Submitted 21 April, 2024;
originally announced April 2024.
-
Curvature and sharp growth rates of log-quasimodes on compact manifolds
Authors:
Xiaoqi Huang,
Christopher D. Sogge
Abstract:
We obtain new optimal estimates for the $L^2(M)\to L^q(M)$, $q\in (2,q_c]$, $q_c=2(n+1)/(n-1)$, operator norms of spectral projection operators associated with spectral windows $[λ,λ+δ(λ)]$, with $δ(λ)=O((\logλ)^{-1})$ on compact Riemannian manifolds $(M,g)$ of dimension $n\ge2$ all of whose sectional curvatures are nonpositive or negative. We show that these two different types of estimates are s…
▽ More
We obtain new optimal estimates for the $L^2(M)\to L^q(M)$, $q\in (2,q_c]$, $q_c=2(n+1)/(n-1)$, operator norms of spectral projection operators associated with spectral windows $[λ,λ+δ(λ)]$, with $δ(λ)=O((\logλ)^{-1})$ on compact Riemannian manifolds $(M,g)$ of dimension $n\ge2$ all of whose sectional curvatures are nonpositive or negative. We show that these two different types of estimates are saturated on flat manifolds or manifolds all of whose sectional curvatures are negative. This allows us to classify compact space forms in terms of the size of $L^q$-norms of quasimodes for each Lebesgue exponent $q\in (2,q_c]$, even though it is impossible to distinguish between ones of negative or zero curvature sectional curvature for any $q>q_c$.
△ Less
Submitted 21 April, 2024;
originally announced April 2024.
-
Long Time $\W_0$-$\widetilde{\W}_1$ type Propagation of Chaos for Mean Field Interacting Particle System
Authors:
Xing Huang,
Fen-Fen Yang,
Chenggui Yuan
Abstract:
In this paper, a general result on the long time $\W_0$-$\widetilde{\W}_1$ type propagation of chaos, propagation of chaos with regularization effect, for mean field interacting particle system driven by Lévy noise is derived, where $\W_0$ is one half of the total variation distance while $\widetilde{\W}_1$ is the $L^1$-Wasserstein distance. By using the method of coupling, the general result is a…
▽ More
In this paper, a general result on the long time $\W_0$-$\widetilde{\W}_1$ type propagation of chaos, propagation of chaos with regularization effect, for mean field interacting particle system driven by Lévy noise is derived, where $\W_0$ is one half of the total variation distance while $\widetilde{\W}_1$ is the $L^1$-Wasserstein distance. By using the method of coupling, the general result is applied to mean field interacting particle system driven by multiplicative Brownian motion and additive $α(α>1)$-stable noise respectively, where the non-interacting drift is assumed to be dissipative in long distance and the initial distribution of interacting particle system converges to that of the limit equation in $\widetilde{\W}_1$.
△ Less
Submitted 17 August, 2024; v1 submitted 2 April, 2024;
originally announced April 2024.
-
Cutoff for random Cayley graphs of nilpotent groups
Authors:
Jonathan Hermon,
Xiangying Huang
Abstract:
We consider the random Cayley graphs of a sequence of finite nilpotent groups of diverging sizes $G=G(n)$, whose ranks and nilpotency classes are uniformly bounded. For some $k=k(n)$ such that $1\ll\log k \ll \log |G|$, we pick a random set of generators $S=S(n)$ by sampling $k$ elements $Z_1,\ldots,Z_k$ from $G$ uniformly at random with replacement, and set $S:=\{Z_j^{\pm 1}:1 \le j\le k \}$. We…
▽ More
We consider the random Cayley graphs of a sequence of finite nilpotent groups of diverging sizes $G=G(n)$, whose ranks and nilpotency classes are uniformly bounded. For some $k=k(n)$ such that $1\ll\log k \ll \log |G|$, we pick a random set of generators $S=S(n)$ by sampling $k$ elements $Z_1,\ldots,Z_k$ from $G$ uniformly at random with replacement, and set $S:=\{Z_j^{\pm 1}:1 \le j\le k \}$. We show that the simple random walk on Cay$(G,S)$ exhibits cutoff with high probability.
Some of our results apply to a general set of generators. Namely, we show that there is a constant $c>0$, depending only on the rank and the nilpotency class of $G$, such that for all symmetric sets of generators $S$ of size at most $ \frac{c\log |G|}{\log \log |G|}$, the spectral gap and the $\varepsilon$-mixing time of the simple random walk $X=(X_t)_{t\geq 0}$ on Cay$(G,S)$ are asymptotically the same as those of the projection of $X$ to the abelianization of $G$, given by $[G,G]X_t$. In particular, $X$ exhibits cutoff if and only if its projection does.
△ Less
Submitted 18 March, 2024;
originally announced March 2024.
-
An Improved Analysis of Langevin Algorithms with Prior Diffusion for Non-Log-Concave Sampling
Authors:
Xunpeng Huang,
Hanze Dong,
Difan Zou,
Tong Zhang
Abstract:
Understanding the dimension dependency of computational complexity in high-dimensional sampling problem is a fundamental problem, both from a practical and theoretical perspective. Compared with samplers with unbiased stationary distribution, e.g., Metropolis-adjusted Langevin algorithm (MALA), biased samplers, e.g., Underdamped Langevin Dynamics (ULD), perform better in low-accuracy cases just be…
▽ More
Understanding the dimension dependency of computational complexity in high-dimensional sampling problem is a fundamental problem, both from a practical and theoretical perspective. Compared with samplers with unbiased stationary distribution, e.g., Metropolis-adjusted Langevin algorithm (MALA), biased samplers, e.g., Underdamped Langevin Dynamics (ULD), perform better in low-accuracy cases just because a lower dimension dependency in their complexities. Along this line, Freund et al. (2022) suggest that the modified Langevin algorithm with prior diffusion is able to converge dimension independently for strongly log-concave target distributions. Nonetheless, it remains open whether such property establishes for more general cases. In this paper, we investigate the prior diffusion technique for the target distributions satisfying log-Sobolev inequality (LSI), which covers a much broader class of distributions compared to the strongly log-concave ones. In particular, we prove that the modified Langevin algorithm can also obtain the dimension-independent convergence of KL divergence with different step size schedules. The core of our proof technique is a novel construction of an interpolating SDE, which significantly helps to conduct a more accurate characterization of the discrete updates of the overdamped Langevin dynamics. Our theoretical analysis demonstrates the benefits of prior diffusion for a broader class of target distributions and provides new insights into developing faster sampling algorithms.
△ Less
Submitted 10 March, 2024;
originally announced March 2024.
-
Revisit on the Slepian concentration problem of the spherical Fourier-Bessel setup
Authors:
Xinpeng Huang
Abstract:
In this paper, we will revisit the Slepian spatiospectral concentration problem for the spherical Fourier-Bessel band-limited spaces introduced for 3-D domain, and discuss its general form in $\mathbb{R}^d$, $d\geq 2$. In particular, we investigate the bimodal distribution of eigenvalues of the concentration operators and give an asymptotic characterization of the Shannon number given a linear rel…
▽ More
In this paper, we will revisit the Slepian spatiospectral concentration problem for the spherical Fourier-Bessel band-limited spaces introduced for 3-D domain, and discuss its general form in $\mathbb{R}^d$, $d\geq 2$. In particular, we investigate the bimodal distribution of eigenvalues of the concentration operators and give an asymptotic characterization of the Shannon number given a linear relation between the Bessel bandwidth and spherical harmonic bandwidth.
△ Less
Submitted 16 September, 2024; v1 submitted 27 February, 2024;
originally announced February 2024.
-
On the spectral extremal problem of planar graphs
Authors:
Xiaolong Wang,
Xueyi Huang,
Huiqiu Lin
Abstract:
The spectral extremal problem of planar graphs has aroused a lot of interest over the past three decades. In 1991, Boots and Royle [Geogr. Anal. 23(3) (1991) 276--282] (and Cao and Vince [Linear Algebra Appl. 187 (1993) 251--257] independently) conjectured that $K_2 + P_{n-2}$ is the unique graph attaining the maximum spectral radius among all planar graphs on $n$ vertices, where $K_2 + P_{n-2}$ i…
▽ More
The spectral extremal problem of planar graphs has aroused a lot of interest over the past three decades. In 1991, Boots and Royle [Geogr. Anal. 23(3) (1991) 276--282] (and Cao and Vince [Linear Algebra Appl. 187 (1993) 251--257] independently) conjectured that $K_2 + P_{n-2}$ is the unique graph attaining the maximum spectral radius among all planar graphs on $n$ vertices, where $K_2 + P_{n-2}$ is the graph obtained from $K_2\cup P_{n-2}$ by adding all possible edges between $K_2$ and $P_{n-2}$. In 2017, Tait and Tobin [J. Combin. Theory Ser. B 126 (2017) 137--161] confirmed this conjecture for all sufficiently large $n$. In this paper, we consider the spectral extremal problem for planar graphs without specified subgraphs. For a fixed graph $F$, let $\mathrm{SPEX}_{\mathcal{P}}(n,F)$ denote the set of graphs attaining the maximum spectral radius among all $F$-free planar graphs on $n$ vertices. We describe a rough sturcture for the connected extremal graphs in $\mathrm{SPEX}_{\mathcal{P}}(n,F)$ when $F$ is a planar graph not contained in $K_{2,n-2}$. As applications, we determine the extremal graphs in $\mathrm{SPEX}_{\mathcal{P}}(n,W_k)$, $\mathrm{SPEX}_{\mathcal{P}}(n,F_k)$ and $\mathrm{SPEX}_{\mathcal{P}}(n,(k+1)K_2)$ for all sufficiently large $n$, where $W_k$, $F_k$ and $(k+1)K_2$ are the wheel graph of order $k$, the friendship graph of order $2k+1$ and the disjoint union of $k+1$ copies of $K_2$, respectively.
△ Less
Submitted 26 February, 2024;
originally announced February 2024.
-
Quantitative Propagation of Chaos in $L^η(η\in(0,1])$-Wasserstein distance for Mean Field Interacting Particle System
Authors:
Xing Huang
Abstract:
In this paper, quantitative propagation of chaos in $L^η$($η\in(0,1]$)-Wasserstein distance for mean field interacting particle system is derived, where the diffusion coefficient is allowed to be interacting and the initial distribution of interacting particle system converges to that of the limit equation in $L^1$-Wasserstein distance. The non-degenerate and second order system are investigated r…
▽ More
In this paper, quantitative propagation of chaos in $L^η$($η\in(0,1]$)-Wasserstein distance for mean field interacting particle system is derived, where the diffusion coefficient is allowed to be interacting and the initial distribution of interacting particle system converges to that of the limit equation in $L^1$-Wasserstein distance. The non-degenerate and second order system are investigated respectively and the main tool relies on the gradient estimate of the decoupled SDEs.
△ Less
Submitted 29 August, 2024; v1 submitted 26 February, 2024;
originally announced February 2024.
-
PDEformer: Towards a Foundation Model for One-Dimensional Partial Differential Equations
Authors:
Zhanhong Ye,
Xiang Huang,
Leheng Chen,
Hongsheng Liu,
Zidong Wang,
Bin Dong
Abstract:
This paper introduces PDEformer, a neural solver for partial differential equations (PDEs) capable of simultaneously addressing various types of PDEs. We propose to represent the PDE in the form of a computational graph, facilitating the seamless integration of both symbolic and numerical information inherent in a PDE. A graph Transformer and an implicit neural representation (INR) are employed to…
▽ More
This paper introduces PDEformer, a neural solver for partial differential equations (PDEs) capable of simultaneously addressing various types of PDEs. We propose to represent the PDE in the form of a computational graph, facilitating the seamless integration of both symbolic and numerical information inherent in a PDE. A graph Transformer and an implicit neural representation (INR) are employed to generate mesh-free predicted solutions. Following pretraining on data exhibiting a certain level of diversity, our model achieves zero-shot accuracies on benchmark datasets that is comparable to those of specifically trained expert models. Additionally, PDEformer demonstrates promising results in the inverse problem of PDE coefficient recovery.
△ Less
Submitted 30 April, 2024; v1 submitted 19 February, 2024;
originally announced February 2024.
-
Decentralized Bilevel Optimization over Graphs: Loopless Algorithmic Update and Transient Iteration Complexity
Authors:
Boao Kong,
Shuchen Zhu,
Songtao Lu,
Xinmeng Huang,
Kun Yuan
Abstract:
Stochastic bilevel optimization (SBO) is becoming increasingly essential in machine learning due to its versatility in handling nested structures. To address large-scale SBO, decentralized approaches have emerged as effective paradigms in which nodes communicate with immediate neighbors without a central server, thereby improving communication efficiency and enhancing algorithmic robustness. Howev…
▽ More
Stochastic bilevel optimization (SBO) is becoming increasingly essential in machine learning due to its versatility in handling nested structures. To address large-scale SBO, decentralized approaches have emerged as effective paradigms in which nodes communicate with immediate neighbors without a central server, thereby improving communication efficiency and enhancing algorithmic robustness. However, current decentralized SBO algorithms face challenges, including expensive inner-loop updates and unclear understanding of the influence of network topology, data heterogeneity, and the nested bilevel algorithmic structures. In this paper, we introduce a single-loop decentralized SBO (D-SOBA) algorithm and establish its transient iteration complexity, which, for the first time, clarifies the joint influence of network topology and data heterogeneity on decentralized bilevel algorithms. D-SOBA achieves the state-of-the-art asymptotic rate, asymptotic gradient/Hessian complexity, and transient iteration complexity under more relaxed assumptions compared to existing methods. Numerical experiments validate our theoretical findings.
△ Less
Submitted 26 February, 2024; v1 submitted 5 February, 2024;
originally announced February 2024.
-
Virtual Morita equivalences and Brauer character bijections
Authors:
Xin Huang
Abstract:
We extend a theorem of Kessar and Linckelmann concerning Morita equivalences and Brauer character bijections between blocks to virtual Morita equivalences. As a corollary, we obtain that Navarro's refinement of Alperin's weight conjecture holds for blocks with cyclic and Klein four defect groups, blocks of symmetric and alternating groups with abelian defect groups, and $p$-blocks of…
▽ More
We extend a theorem of Kessar and Linckelmann concerning Morita equivalences and Brauer character bijections between blocks to virtual Morita equivalences. As a corollary, we obtain that Navarro's refinement of Alperin's weight conjecture holds for blocks with cyclic and Klein four defect groups, blocks of symmetric and alternating groups with abelian defect groups, and $p$-blocks of ${\rm SL}_2(q)$ and ${\rm GL}_2(q)$, where $p|q$.
△ Less
Submitted 26 March, 2024; v1 submitted 24 January, 2024;
originally announced January 2024.
-
Well-Posedness for McKean-Vlasov SDEs Driven by Multiplicative Stable Noises
Authors:
Chang-Song Deng,
Xing Huang
Abstract:
We establish the well-posedness for a class of McKean-Vlasov SDEs driven by symmetric $α$-stable Lévy process ($1/2<α\leq1$), where the drift coefficient is Hölder continuous in space variable, while the noise coefficient is Lipscitz continuous in space variable, and both of them satisfy the Lipschitz condition in distribution variable with respect to Wasserstein distance. If the drift coefficient…
▽ More
We establish the well-posedness for a class of McKean-Vlasov SDEs driven by symmetric $α$-stable Lévy process ($1/2<α\leq1$), where the drift coefficient is Hölder continuous in space variable, while the noise coefficient is Lipscitz continuous in space variable, and both of them satisfy the Lipschitz condition in distribution variable with respect to Wasserstein distance. If the drift coefficient does not depend on distribution variable, our methodology developed in this paper applies to the case $α\in(0,1]$. The main tool relies on heat kernel estimates for (distribution independent) stable SDEs and Banach's fixed point theorem.
△ Less
Submitted 20 January, 2024;
originally announced January 2024.
-
A note on weak Banach mean equicoontinuity
Authors:
Zhongxuan Yang,
Xiaojun Huang
Abstract:
Consider a topological dynamical system $(X, T)$ endowed with the metric $d$. We introduce a novel function as $\overline{BF}(x, y) = \limsup_{n-m \rightarrow +\infty} \inf_{σ\in S_{n,m}} \frac{1}{n-m} \sum_{k=m}^{n-1} d\left(T^{k} x, T^{σ(k)} y\right)$, where the permutation group $S_{n,m}$ is utilized. It is demonstrated that $BF(x, y)$ exists when $x, y \in X$ are uniformly generic points. Leve…
▽ More
Consider a topological dynamical system $(X, T)$ endowed with the metric $d$. We introduce a novel function as $\overline{BF}(x, y) = \limsup_{n-m \rightarrow +\infty} \inf_{σ\in S_{n,m}} \frac{1}{n-m} \sum_{k=m}^{n-1} d\left(T^{k} x, T^{σ(k)} y\right)$, where the permutation group $S_{n,m}$ is utilized. It is demonstrated that $BF(x, y)$ exists when $x, y \in X$ are uniformly generic points. Leveraging this function, we introduce the concept of weak Banach mean equicontinuity and establish that the dynamical system $(X, T)$ exhibits weak Banach mean equicontinuity if and only if the uniform time averages $f_B^{*}(x) = \lim_{n-m \rightarrow +\infty} \frac{1}{n-m} \sum_{k=m}^{n-1} f\left(T^{k} x\right)$ are continuous for all $f \in C(X)$. Finally, we demonstrate that in the case of a transitive system, the equivalence between weak Banach mean equicontinuity and weak mean equicontinuity is established.
△ Less
Submitted 18 January, 2024;
originally announced January 2024.
-
A note on weak mean equicontinuity and strong mean sensitivity
Authors:
Zhongxuan Yang,
Xiaojun Huang
Abstract:
In this paper, we study the weak mean metric and give some properties by replacing the Besicovitch pseudometric with weak mean metric in the definition of mean equicontinuity and mean sensitivity. We study an opposite side of weak mean equicontinuity, strong mean sensitivity and we obtain a version of Auslander-Yorke dichotomies: minimal topological dynamical systems are either weak mean equiconti…
▽ More
In this paper, we study the weak mean metric and give some properties by replacing the Besicovitch pseudometric with weak mean metric in the definition of mean equicontinuity and mean sensitivity. We study an opposite side of weak mean equicontinuity, strong mean sensitivity and we obtain a version of Auslander-Yorke dichotomies: minimal topological dynamical systems are either weak mean equicontinuous or strong mean sensitive, and transitive topological dynamical systemss are either almost weak mean equicontinuous or strong mean sensitive. Furthermore, motivated by the localized idea of sensitivity, we introduce some notions of new version sensitive tuples and study the properties of these sensitive tuples, we show that a transitive dynamical system is strong mean sensitive if and only if it admits a strong mean sensitive tuple. Finally, We introduce the notions of weakly mean equicontinuity of a topological dynamical system respect to a given continuous function $f$, and we show that a topological dynamical system is weakly mean equicontinuity then it is weakly mean equicontinuity with respect to every continuous function.
△ Less
Submitted 18 January, 2024; v1 submitted 18 January, 2024;
originally announced January 2024.
-
New type of solutions for the critical Lane-Emden system
Authors:
Wenjing Chen,
Xiaomeng Huang
Abstract:
In this paper, we consider the critical Lane-Emden system \begin{align*} \begin{cases} -Δu=K_1(y)v^p,\quad y\in \mathbb{R}^N,&\\ -Δv=K_2(y)u^q,\quad y\in \mathbb{R}^N,&\\ u,v>0, \end{cases} \end{align*} where $N\geq 5$, $p,q\in (1,\infty)$ with $\frac{1}{p+1}+\frac{1}{q+1}=\frac{N-2}{N}$, $K_1(y)$ and $K_2(y)$ are positive radial potentials. Under suitable conditions on $K_1(y)$ and $K_2(y)$, we c…
▽ More
In this paper, we consider the critical Lane-Emden system \begin{align*} \begin{cases} -Δu=K_1(y)v^p,\quad y\in \mathbb{R}^N,&\\ -Δv=K_2(y)u^q,\quad y\in \mathbb{R}^N,&\\ u,v>0, \end{cases} \end{align*} where $N\geq 5$, $p,q\in (1,\infty)$ with $\frac{1}{p+1}+\frac{1}{q+1}=\frac{N-2}{N}$, $K_1(y)$ and $K_2(y)$ are positive radial potentials. Under suitable conditions on $K_1(y)$ and $K_2(y)$, we construct a new family of solutions to this system, which are centred at points lying on the top and the bottom circles of a cylinder.
△ Less
Submitted 17 January, 2024;
originally announced January 2024.
-
Faster Sampling without Isoperimetry via Diffusion-based Monte Carlo
Authors:
Xunpeng Huang,
Difan Zou,
Hanze Dong,
Yian Ma,
Tong Zhang
Abstract:
To sample from a general target distribution $p_*\propto e^{-f_*}$ beyond the isoperimetric condition, Huang et al. (2023) proposed to perform sampling through reverse diffusion, giving rise to Diffusion-based Monte Carlo (DMC). Specifically, DMC follows the reverse SDE of a diffusion process that transforms the target distribution to the standard Gaussian, utilizing a non-parametric score estimat…
▽ More
To sample from a general target distribution $p_*\propto e^{-f_*}$ beyond the isoperimetric condition, Huang et al. (2023) proposed to perform sampling through reverse diffusion, giving rise to Diffusion-based Monte Carlo (DMC). Specifically, DMC follows the reverse SDE of a diffusion process that transforms the target distribution to the standard Gaussian, utilizing a non-parametric score estimation. However, the original DMC algorithm encountered high gradient complexity, resulting in an exponential dependency on the error tolerance $ε$ of the obtained samples. In this paper, we demonstrate that the high complexity of DMC originates from its redundant design of score estimation, and proposed a more efficient algorithm, called RS-DMC, based on a novel recursive score estimation method. In particular, we first divide the entire diffusion process into multiple segments and then formulate the score estimation step (at any time step) as a series of interconnected mean estimation and sampling subproblems accordingly, which are correlated in a recursive manner. Importantly, we show that with a proper design of the segment decomposition, all sampling subproblems will only need to tackle a strongly log-concave distribution, which can be very efficient to solve using the Langevin-based samplers with a provably rapid convergence rate. As a result, we prove that the gradient complexity of RS-DMC only has a quasi-polynomial dependency on $ε$, which significantly improves exponential gradient complexity in Huang et al. (2023). Furthermore, under commonly used dissipative conditions, our algorithm is provably much faster than the popular Langevin-based algorithms. Our algorithm design and theoretical framework illuminate a novel direction for addressing sampling problems, which could be of broader applicability in the community.
△ Less
Submitted 11 January, 2024;
originally announced January 2024.
-
Chemical distance for the half-orthant model
Authors:
Nicholas Beaton,
Mark Holmes,
Xin Huang
Abstract:
The half-orthant model is a partially oriented model of a random medium involving a parameter $p\in [0,1]$, for which there is a critical value $p_c(d)$ (depending on the dimension $d$) below which every point is reachable from the origin. We prove a limit theorem for the graph-distance (or "chemical distance") for this model when $p<p_c(2)$, and also when $1-p$ is larger than the critical paramet…
▽ More
The half-orthant model is a partially oriented model of a random medium involving a parameter $p\in [0,1]$, for which there is a critical value $p_c(d)$ (depending on the dimension $d$) below which every point is reachable from the origin. We prove a limit theorem for the graph-distance (or "chemical distance") for this model when $p<p_c(2)$, and also when $1-p$ is larger than the critical parameter for site percolation in $\mathbb{Z}^d$. The proof involves an application of the subadditive ergodic theorem. Novel arguments herein include the method of proving that the expected number of steps to reach any given point is finite, as well as an argument that is used to show that the shape is "non-trivial" in certain directions.
△ Less
Submitted 7 January, 2024;
originally announced January 2024.
-
Coordinating Guidance, Matching, and Charging Station Selection for Electric Vehicle Ride-Hailing Services through Data-Driven Stochastic Optimization
Authors:
Xiaoming Li,
Chun Wang,
Xiao Huang
Abstract:
Electric vehicles (EVs) play a pivotal role in sustainable ride-hailing services primarily due to their potential in reducing carbon emissions and enhancing environmental protection. Despite their significance, current research in the realm of EV batched matching frequently overlooks critical aspects such as rider demand uncertainty and charging station (CS) selection, leading to inefficiencies li…
▽ More
Electric vehicles (EVs) play a pivotal role in sustainable ride-hailing services primarily due to their potential in reducing carbon emissions and enhancing environmental protection. Despite their significance, current research in the realm of EV batched matching frequently overlooks critical aspects such as rider demand uncertainty and charging station (CS) selection, leading to inefficiencies like decreased matching rates and prolonged waiting times for both riders and EV drivers. To fill the research gap, we propose a data-driven optimization framework that incorporates two inter-connected stochastic optimization models to address the challenges. The first model aims to relocate the idle EVs under satisfied conditions to the designated regions based on the probabilistic rider demand forecasting result before the real rider demand is revealed. Taking the solutions of the first model as the input, the second model optimizes the batched matching results by minimizing the rider's average waiting time and EV charging waiting time at CS. This integrated framework not only elevates the matching rate through the incorporation of rider demand uncertainties in the guidance module but also substantially curtails both rider and EV charging waiting times by synergizing guidance with CS selection choices. Empirical validation of our framework was conducted through an extensive case study in New York City, utilizing real-world data sets. The validation results demonstrate that the proposed data-driven optimization framework outperforms the benchmark models in terms of the proposed evaluation metrics. Most importantly, when deploying our framework, the charging waiting time of the EVs with low SOC can be reduced up to 73.6% compared to the benchmark model without CS selection.
△ Less
Submitted 8 January, 2024; v1 submitted 6 January, 2024;
originally announced January 2024.
-
Drinfeld Module and Weil pairing over Dedekind domain of class number two
Authors:
Chuangqiang Hu,
Xiao-Min Huang
Abstract:
The primary objective of this paper is to derive explicit formulas for rank one and rank two Drinfeld modules over a specific domain denoted by A. This domain corresponds to the projective line associated with an infinite place of degree two. To achieve the goals, we construct a pair of standard Drinfeld modules whose coefficients are in the Hilbert class field of A. We demonstrate that the period…
▽ More
The primary objective of this paper is to derive explicit formulas for rank one and rank two Drinfeld modules over a specific domain denoted by A. This domain corresponds to the projective line associated with an infinite place of degree two. To achieve the goals, we construct a pair of standard Drinfeld modules whose coefficients are in the Hilbert class field of A. We demonstrate that the period lattice of the exponential functions corresponding to both modules behaves similarly to the period lattice of the Carlitz module, the standard rank one Drinfeld module defined over rational function field. Moreover, we employ Andersons t-motive to obtain the complete family of rank two Drinfeld modules. This family is parameterized by the invariant J = λ^{q^2+1} which effectively serves as the counterpart of the j-invariant for elliptic curves. Building upon the concepts introduced by van~der~Heiden, particularly with regard to rank two Drinfeld modules, we are able to reformulate the Weil pairing of Drinfeld modules of any rank using a specialized polynomial in multiple variables known as the Weil operator. As an illustrative example, we provide a detailed examination of a more explicit formula for the Weil pairing and the Weil operator of rank two Drinfeld modules over the domain A.
△ Less
Submitted 9 October, 2024; v1 submitted 28 December, 2023;
originally announced December 2023.
-
A New Global Optimization Method Based on Simplex Branching for Solving a Class of Non-Convex QCQP Problems
Authors:
Bo Zhang,
YueLin Gao,
Xia Liu,
XiaoLi Huang
Abstract:
Quadratic constrained quadratic programming problems often occur in various fields such as engineering practice, management science, and network communication. This article mainly studies a non convex quadratic programming problem with convex quadratic constraints. Firstly, based on our existing results, the problem is reconstructed as an equivalent problem with a simple concave quadratic objectiv…
▽ More
Quadratic constrained quadratic programming problems often occur in various fields such as engineering practice, management science, and network communication. This article mainly studies a non convex quadratic programming problem with convex quadratic constraints. Firstly, based on our existing results, the problem is reconstructed as an equivalent problem with a simple concave quadratic objective function in the result space, with a convex feasible domain. A global optimization algorithm for solving equivalent problems is proposed based on a branch and bound framework that can ensure the global optimality of the solution. This algorithm combines effective relaxation processes with branching processes related to new external approximation techniques. Finally, the theoretical feasibility of the algorithm was analyzed.
△ Less
Submitted 23 December, 2023;
originally announced December 2023.
-
Linear stability of inner case of double averaged spatial restricted elliptic three body problem
Authors:
Xiumin Huang,
Yan Luo,
Kaicheng Sheng,
Yiru Ye
Abstract:
We study the secular effects in the motion of an asteroid with negligible mass in a spatial restricted elliptic three body problem with arbitrary inclination. Averaging over mean anomalies of the asteroid and the planet are applied to obtain the double averaged Hamiltonian system. It admits a two-parameter family of orbits corresponding to the motion of the third body in the plane of primaries' mo…
▽ More
We study the secular effects in the motion of an asteroid with negligible mass in a spatial restricted elliptic three body problem with arbitrary inclination. Averaging over mean anomalies of the asteroid and the planet are applied to obtain the double averaged Hamiltonian system. It admits a two-parameter family of orbits corresponding to the motion of the third body in the plane of primaries' motion. The aim of our investigation is to analyze the stability of these orbits in inner case. We show that they are stable in the linear approximation and give descriptions of linear stability with respect to the eccentricity and argument of periapsis of asteroid. Numerical simulations of different types of orbits are performed as well.
△ Less
Submitted 19 September, 2024; v1 submitted 13 December, 2023;
originally announced December 2023.
-
Understanding the Influence of Digraphs on Decentralized Optimization: Effective Metrics, Lower Bound, and Optimal Algorithm
Authors:
Liyuan Liang,
Xinmeng Huang,
Ran Xin,
Kun Yuan
Abstract:
This paper investigates the influence of directed networks on decentralized stochastic non-convex optimization associated with column-stochastic mixing matrices. Surprisingly, we find that the canonical spectral gap, a widely used metric in undirected networks, is insufficient to characterize the impact of directed topology on decentralized algorithms. To overcome this limitation, we introduce a n…
▽ More
This paper investigates the influence of directed networks on decentralized stochastic non-convex optimization associated with column-stochastic mixing matrices. Surprisingly, we find that the canonical spectral gap, a widely used metric in undirected networks, is insufficient to characterize the impact of directed topology on decentralized algorithms. To overcome this limitation, we introduce a novel metric termed equilibrium skewness. This metric, together with the spectral gap, accurately and comprehensively captures the influence of column-stochastic mixing matrices on decentralized stochastic algorithms. With these two metrics, we clarify, for the first time, how the directed network topology influences the performance of prevalent algorithms such as Push-Sum and Push-Diging. Furthermore, we establish the first lower bound of the convergence rate for decentralized stochastic non-convex algorithms over directed networks. Since existing algorithms cannot match our lower bound, we further propose the MG-Push-Diging algorithm, which integrates Push-Diging with a multi-round gossip technique. MG-Push-Diging attains our lower bound up to logarithmic factors, demonstrating its near-optimal performance and the tightness of the lower bound. Numerical experiments verify our theoretical results.
△ Less
Submitted 27 April, 2024; v1 submitted 8 December, 2023;
originally announced December 2023.
