Browse

Discrete Partial Differential Equations on Graphs and Applications of Ricci Curvature
그래프 위에서의 이산 편미분 방정식과 리치 곡률의 응용

Cited 0 time in Web of Science Cited 0 time in Scopus
Authors
이가람
Advisor
이기암
Major
자연과학대학 수리과학부
Issue Date
2017-08
Publisher
서울대학교 대학원
Keywords
GraphsDiscrete partial differential equationsLaplacian operatorEigenvaluesRicci curvatureLi-Yau gradient estimateHarnack inequalityPorous medium equation
Description
학위논문 (석사)-- 서울대학교 대학원 자연과학대학 수리과학부, 2017. 8. 이기암.
Abstract
In this paper, we mainly discuss how partial differential equations can be modeled in a graph setting. Most discussions are based on previous studies from the references. In Section 1, we first introduce graph theoretic notions to establish discrete analogues of calculus on graphs. In particular, the operator of our main interest is the discrete Laplace operator. In Section 2, we study discrete versions of linear PDEs such as the Laplace equation, the heat equation, and the wave equation. We also give their properties and show uniqueness via the energy method. In Section 3, we provide a generalization of lower Ricci curvature bound in the discrete framework due to Bakry and Emery. With this, we can get an estimate for the eigenvalue of the Laplace operator on graphs. In Section 4, using the results from Section 3, we investigate various gradient estimates for positive solutions to the heat equation such as Li-Yau estimate and Hamilton estimate on graphs. As an application of the Li-Yau estimate, a Harnack type inequality will be proved. In Section 5, we show the L1-contraction, uniqueness and existence of solutions to the Porous medium equation and discuss the way of interpreting discrete versions of them on graphs.
Language
English
URI
https://hdl.handle.net/10371/138080
Files in This Item:
Appears in Collections:
College of Natural Sciences (자연과학대학)Dept. of Mathematical Sciences (수리과학부)Theses (Master's Degree_수리과학부)
  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Browse