数学学科Seminar第2962讲 一种新的Wasserstein梯度流动态方法

创建时间:  2025/11/17  邵奋芬   浏览次数:   返回

报告题目 (Title):A new flow dynamic approach for Wasserstein gradient flows.

(一种新的Wasserstein梯度流动态方法)

报告人 (Speaker):程青 特聘研究员(同济大学)

报告时间 (Time):2025年11月19日(周三)13:00

报告地点 (Place):校本部F309

邀请人(Inviter):李新祥

主办部门:理学院数学系

报告摘要:We develop a new flow dynamic approach for Wasserstein gradient flows. Motivatied by the classic JKO scheme, we develop a new class of Lagrangian schemes which only need to solve minimization problems with respective to displacement instead of density and velocity. The new approach can effectively capture the movement of the trajectory of meshes, and also can automatically preserve the nice properties of Wasserstein gradient flow structure, for example, positivity-preserving, mass conserving and energy dissipation.Numerical experiments are shown in 1D and 2D for Keller-Segel equations,Focker-Plancker equations, Porous medium equations.

上一条:物理学科Seminar第776讲 低维波导与巨原子耦合体系相干动力学调控及其在量子器件中的应用

下一条:数学学科Seminar第2961讲 科学机器学习中的信息计算


数学学科Seminar第2962讲 一种新的Wasserstein梯度流动态方法

创建时间:  2025/11/17  邵奋芬   浏览次数:   返回

报告题目 (Title):A new flow dynamic approach for Wasserstein gradient flows.

(一种新的Wasserstein梯度流动态方法)

报告人 (Speaker):程青 特聘研究员(同济大学)

报告时间 (Time):2025年11月19日(周三)13:00

报告地点 (Place):校本部F309

邀请人(Inviter):李新祥

主办部门:理学院数学系

报告摘要:We develop a new flow dynamic approach for Wasserstein gradient flows. Motivatied by the classic JKO scheme, we develop a new class of Lagrangian schemes which only need to solve minimization problems with respective to displacement instead of density and velocity. The new approach can effectively capture the movement of the trajectory of meshes, and also can automatically preserve the nice properties of Wasserstein gradient flow structure, for example, positivity-preserving, mass conserving and energy dissipation.Numerical experiments are shown in 1D and 2D for Keller-Segel equations,Focker-Plancker equations, Porous medium equations.

上一条:物理学科Seminar第776讲 低维波导与巨原子耦合体系相干动力学调控及其在量子器件中的应用

下一条:数学学科Seminar第2961讲 科学机器学习中的信息计算