报告时间:2024年6月7日(星期五)8:30-17:00
报告地点:翡翠湖校区科教楼B座1710室
举办单位:数学学院
学术报告信息(一)
报告题目:Computation and analysis for the long time dynamics of (nonlinear) Schrödinger equations
报告时间:2024年6月7日(星期五)8:30-9:30
报 告 人:蔡勇勇 教授
工作单位:北京师范大学
报告简介:
Dispersive PDEs, such as linear/nonlinear Schrödinger equation (NLSE), nonlinear Klein-Gordon equation, nonlinear Dirac equation arise from many different areas, e.g. computational chemistry, plasma physics, quantum mechanics. Recently, the long-time dynamics of such dispersive equations have received much attention. The long time NLSE with small initial data is equivalent to an oscillatory NLSE with $O(1)$ initial data, and such oscillatory PDE is computational expensive. Here we report recent advances on the numerical methods and analysis for the long time NLSE. In particular, an improved uniform error bound for the time-splitting methods for the long-time NLSE is established. Extensions to other dispersive PDEs will be presented.
报告人简介:
蔡勇勇,北京师范大学教授,本硕就读于北京大学,2012年在新加坡国立大学获得博士学位,2016年入选海外高层次人才引进计划青年项目。他先后在威斯康辛大学麦迪逊分校、马里兰大学帕克分校和普渡大学从事博士后研究工作,从2016年至2019年在北京计算科学研究中心任特聘研究员。蔡勇勇博士的研究兴趣是偏微分方程的数值方法及其在量子力学等领域中的应用,在Mathematics of Computation, Journal of Computational Physics和SIAM系列等期刊上发表论文60余篇,多次受邀参加学术会议并做相关报告。
学术报告信息(二)
报告题目: Optimal zero-padding of kernel truncation method
报告时间:2024年6月7日(星期五)9:30-10:30
报 告 人:张勇 教授
工作单位:天津大学
报告简介:
The kernel truncation method (KTM) is a commonly-used algorithm to compute the convolution-type nonlocal potential, where the convolution kernel might be singular at the origin and/or far-field and the density is smooth and fast-decaying. In KTM, in order to capture the Fourier integrand's oscillations that is brought by the kernel truncation, one needs to carry out a zero-padding of the density, which means a larger physical computation domain and a finer mesh in the Fourier space by duality. The empirical fourfold zero-padding [ Vico et al. J. Comput. Phys. (2016) ] puts a heavy burden on memory requirement especially for higher dimension problems. In this paper, we derive the optimal zero-padding factor, that is, \sqrt{d}+1, for the first time together with a rigorous proof. The memory cost is greatly reduced to a small fraction, i.e., (\frac{\sqrt{d}+1}{4})^d, of what is needed in the original fourfold algorithm. For example, in the precomputation step, a double-precision computation on a 256^3 grid requires a minimum $3.4$ Gb memory with the optimal threefold zero-padding, while the fourfold algorithm requires around 8 Gb where the reduction factor is around 60%. Then, we present the error estimates of the potential and density in d space dimension. Next, we re-investigate the optimal zero-padding factor for the anisotropic density. Finally, extensive numerical results are provided to confirm the accuracy, efficiency, optimal zero-padding factor for the anisotropic density, together with some applications to different types of nonlocal potential, including the 1D/2D/3D Poisson, 2D Coulomb, quasi-2D/3D Dipole-Dipole Interaction and 3D quadrupolar potential.
报告人简介:
张勇,天津大学教授。2007年本科毕业于天津大学数学系,2012年在清华大学获得博士学位,曾先后在奥地利维也纳大学,法国雷恩一大和美国纽约大学克朗所从事博士后研究工作。2015年获奥地利自然科学基金委支持的薛定谔基金,2018年入选国家高层次人才计划。张勇博士的研究兴趣主要是偏微分方程的数值计算和分析工作,尤其是快速算法的设计和应用,迄今发表论文20余篇,主要发表在包括SIAM Journal on Scientific Computing, SIAM journal on Applied Mathematics, Multiscale Modeling and Simulation, Mathematics of Computation, Journal of Computational Physics, Computer Physics Communication等计算数学顶尖杂志。
学术报告信息(三)
报告题目: Some AI methods with normalized DNN for stationary and evolutional problems
报告时间:2024年6月7日(星期五)10:30-11:30
报 告 人:赵晓飞 教授
工作单位:武汉大学
报告简介:
报告将首先回顾现有基于深度神经网络求解微分方程的几类机器学习方法,它们基本都是从方程到优化,而第一型原理通常可直接给出优化问题。基于此,我们将介绍两类基于归一化深度网络的算法,分别应用于薛定谔方程的稳态求解与波动方程的初终值问题上,旨在规避方程得到从物理到优化的直接实现。
报告人简介:
赵晓飞,武汉大学教授。2010年本科毕业于北京师范大学,2014年博士毕业于新加坡国立大学,之后在法国INRIA从事博士后研究,2019年入选国家青年人才计划并入职武汉大学。赵晓飞博士的研究兴趣为色散类与动理学模型的数值计算方法与误差分析,相关成果发表在Mathematics of Computation, Journal of Computational Physics及SIAM系列期刊。
学术报告信息(四)
报告题目: Parametric finite element methods for curvature-driven interface evolution with axisymmetric geometry
报告时间:2024年6月7日(星期五)14:00-15:00
报 告 人:赵泉 特任研究员
工作单位:中国科学技术大学
报告简介:
In this talk, I will discuss parametric approximations for axisymmetric geometric evolution equations. We consider two different possible approximations of the curvature, which leads to either an unconditional stability or an asymptotic equal mesh distribution. An exact volume preservation for the discrete solutions can be further strengthened with an appropriate suitable approximation to a combined term of the radial distance and the interface normal. Moreover, we generalize the parametric approximations to the two-phase flow, where the interface evolution is influenced by the bulk quantity. We propose several front-tracking approximations which combines the parametric finite element method for the interface equations with the standard finite element methods for the bulk equations.
报告人简介:
赵泉,中国科学技术大学特任研究员。2017年博士毕业于新加坡国立大学数学系,2021年获洪堡博士后奖学金,2023年入职中国科学技术大学。赵泉博士的研究兴趣为界面演化问题的数值计算与模拟,如材料学中的相关几何偏微分方程以及流体力学中多相流问题。相关工作发布在SIAM系列期刊,Journal of Computational Physics, IMA Journal of Numerical Analysis以及Computer Methods in Applied Mechanics and Engineering等。
学术报告信息(五)
报告题目: A new computational approach for orthogonal spline collocation method
报告时间:2024年6月7日(星期五)15:00-16:00
报 告 人:廖锋 副教授
工作单位:常熟理工学院
报告简介:
This talk is concerned with the numerical solutions of Schrödinger-Boussinesq (SBq) system by an orthogonal spline collocation (OSC) discretization in space and Crank-Nicolson (CN) type approximation in time. By using the mathematical induction argument and standard energy method, the proposed CN+OSC scheme is proved to be unconditionally convergent at the order with mesh-size and time step in the discrete-norm. We devise a new computation method based on the orthogonal diagonalization techniques (ODT) to realize the proposed CN+OSC scheme. In order to compare the performance of ODT, we devise an alternating direction implicit (ADI) method to compute the CN+OSC scheme for high spatial dimension SBq system. As an alternative implementation, the new method ODT not only exhibits more accurate numerical results, but also demonstrates stronger invariance preserving ability. Numerical results are reported to verify the error estimates and the discrete conservation laws.
报告人简介:
廖锋,常熟理工学院副教授。2018年南京航空航天大学取得博士学位后入职常熟理工学院。廖锋博士主要从事偏微分方程保结构算法的相关研究,在Journal of Computational and Applied Mathematics, Applied Numerical Mathematics, Calcolo, Communications in Nonlinear Science and Numerical Simulation 等刊物发表论文20余篇。
学术报告信息(六)
报告题目: 一类时间高频振荡问题的方法研究
报告时间:2024年6月7日(星期五)16:00-17:00
报 告 人:周旋旋 博士
工作单位:南京理工大学
报告简介:
这个报告中我们将首先回顾一类时间高振荡问题的研究现状,依据对该类问题所具有的时间高振荡现象处理技巧的不同,按照时间脉络梳理已有的各种方法的特性以及使用场景。然后介绍嵌套皮卡迭代方法对这一问题的处理方法及其数值实验表现。
报告人简介:
周旋旋,南京理工大学讲师。2019 年博士毕业于南京航空航天大学, 之后分别在北京计算科学研究中心、北京师范大学从事博士后研究,2023 年 6 月入职南京理工大学。周旋旋博士主要从事时间高振荡问题的高分辨率算法以及非光滑势函数薛定谔方程数值算法的相关研究,在Journal of Scientific Computing, Numerical Algorithms 等国外学术期刊发表 SCI 学术论文 7 篇。作为核心成员参与国家自然科学基金(面上项目)“高振荡薛定谔型方程(组)的高分辨率快速算法" 等。