Numerical Analysis (I)
Fall 1998
Jinn-Liang Liu
http://www.math.nctu.edu.tw/~jinnliu
Projects
-
(Due 10/30/98) Newton's Method (I): Demo
-
(Due 11/13/98) Matrix Computations (I): Demo
-
Read the example (7.2.12) in Patel p.180 and solve it by adaptc++
-
Solve C.8 in Patel p.186 by adaptc++
-
CParabProb (I): 1D Convection-Diffusion-Reaction (Parabolic) Problems
-
CParabProb (II): 1D Convection-Diffusion-Reaction (Parabolic) Problems
-
Lecture note writing: (Due 1/15/99)
Exams
-
Midterm Exam on 11/23/98
-
Final Exam 1/18/99
Lectures
-
Computer Arithmetic
(Chap 1) 10/12
-
Newton's Method
(Chap 2) 10/19
-
Convergence Properties
-
Taylor's Theorem
-
Scientific Computing
-- An Over View
-
Nature -- Biology, Chemistry, Physics*, and Sociology
-
Law of Mass Conservation
-
Law of Energy Conservation
-
Law of Momentum Conservation
-
Mathematical Modeling
-
Algebraic Equations
-
Ordinary Differential Equations (ODEs)
-
Partial Differential Equations (PDEs)*
-
Time Dependent Problems*
-
Heat Equation* (Parabolic Type)
-
Wave Equation (Hyperbolic Type)
-
Time Independent Problems
-
Laplace's Equation* (Elliptic Type)
-
Integral Equations
-
Mixed Equations
-
Differential Algebraic Equations
-
Integral Differential Equations
-
Numerical Computations
-
Numerical Differentiation
-
Numerical Integration
-
Numerical Linear Algebra* (Ax=b)
-
Numerical Methods for Nonlinear Equations*
-
Newton's Method
-
Monotone Method
-
Numerical Methods for ODEs
-
Numerical Methods for PDEs*
-
Time Discretization
-
Forward Difference Method -- Explicit Method
-
Central Difference Method -- Richardson's Method
-
Backward Difference Method -- Implicit Method
-
The Crank-Nicolson Method
-
Spatial Discretization
-
Finite Difference Method (FDM)
-
Finite Element Method (FEM)
-
Finite Volume Method (FVM)
-
A Mathematical Model
-
Convection-Diffusion-Reaction
(CDR) Model
-
Convection: Together + Carry -- Transfer by movement.
-
Diffusion: Apart + Flow -- Spreading out in every
direction.
-
Reaction: Response to a stimulus or influence.
-
Applications
-
Heat Conduction
-
Highway Traffic
-
River Pollution
-
Semiconductor
-
Modeling: Conservation
of Mass & Mean
Value Theorem
-
Space variable: x (unit L), time variable: t (unit
T)
-
Substance: Heat, cars, pollutants, algae etc. (unit
M)
-
u=u(x,t): The density or concentration of a substance
at x and at time t in units of mass per unit distance (M/L).
-
q=q(x,y): Its rate of flow past x at time t in units
of mass per unit time (M/T).
-
f=f(x,t): The rate at which the desity of mass is
changing due to source, sink, or reaction (M/(LT)).
-
v=v(x,t): The velocity of the flow in whch the substance
moves (L/T).
-
Convection: q=vu
-
Diffusion: q=- c (du/dx), c>0, a constant of proportionality.
-
Reaction: f.
-
Equation: du/dt
= c {d^2u/dx^2} - {d(vu)/dx} + f
-
The Finite Didderence Method for
the Heat Equation
-
Problem
-
Heat Equation
-
Boundary Conditions (BCs)
-
Initial Condition (IC)
-
Numerical Methods
-
Time Discretization by FDM
-
The Forward Difference Method
-
The Backward Difference Method
-
The Central Difference (Crank-Nicalson) Method
-
Space Discretization by FDM
-
The Central Difference Method => Stiffness
Matrix S
-
Patel p.139 Table 5.3.1
-
Full Discretization in Matrix Form
-
Impose BCs and IC => Ax=b
-
The Finite Didderence Method for
Convection-Diffusion-Reaction Problems
-
Space Discretization by FDM
-
Stiffness Matrix S
-
Load Vector F
-
Full Discretization in Matrix Form
-
Impose BCs and IC => Ax=b
-
The Finite Element Method for the
Heat Equation
-
Weak Formulation
-
Integration by Parts
-
Test and Trial Functions
-
Function Spaces
-
Finite Element Spaces
-
Semi-Discrete Finite Element Approximation
Text
Vithal A. Patel, Numerical Analysis, Saunders College Publishing,
1994.
Syllabus
Chapters 1, 2, 7, 11, 12, 13, 14, 15