Scribd
Upload
65 views25 pages

ELEC3846 Week 1 Lecture 2 (2024)

This document discusses numerical analysis concepts including errors, Lipschitz conditions, and well-posed problems. It defines a well-posed problem as one that has a unique solution, and where the solution depends continuously on the initial conditions. The Lipschitz condition guarantees a unique solution if the derivative of the function is bounded. Examples show how to determine the Lipschitz constant and check if initial value problems are well-posed based on the Lipschitz condition.

Uploaded by

KT Chan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
65 views25 pages

ELEC3846 Week 1 Lecture 2 (2024)

This document discusses numerical analysis concepts including errors, Lipschitz conditions, and well-posed problems. It defines a well-posed problem as one that has a unique solution, and where the solution depends continuously on the initial conditions. The Lipschitz condition guarantees a unique solution if the derivative of the function is bounded. Examples show how to determine the Lipschitz constant and check if initial value problems are well-posed based on the Lipschitz condition.

Uploaded by

KT Chan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 25

ELEC 3846 Numerical

Methods and Optimization


Week 1: Some definitions of numerical analysis: errors, lipschitz,
well-posed problem

Dr. Zhongrui Wang

Department of Electrical and Electronic Engineering


Content
Existence, Uniqueness, and Stability for Ordinary Differential Equations (ODEs)

• Does a unique solution exist?


• If there are more than one solution, we don’t know which solution the numerical
method will lead us towards.)

• Is the problem well-posed?


• If the problem is not well-posed, a small error in the system can lead to a very different
result. The final result will be very unstable.)
Mean Value Theorem
If f(x) is defined and continuous on the interval
[a, b] and differentiable on (a, b), then there is
at least one number c in the interval (a, b) (that
is a<c<b) such that

There exists a point

Such that the tangent line is parallel to the line


passing through (a, f(a)) and (b, f(b)).
Lipschitz Condition
A function f is Lipschitz if and only if

𝑓𝑓 𝑥𝑥, 𝑦𝑦1 − 𝑓𝑓 𝑥𝑥, 𝑦𝑦2 ≤ 𝐿𝐿 𝑦𝑦1 − 𝑦𝑦2

𝐿𝐿 is the Lipschitz constant. If 𝑓𝑓(𝑥𝑥,𝑦𝑦) differentiable w.r.t. to 𝑦𝑦 then by mean value


theorem

𝜕𝜕𝜕𝜕
𝑓𝑓 𝑥𝑥, 𝑦𝑦1 − 𝑓𝑓 𝑥𝑥, 𝑦𝑦2 = � 𝑦𝑦1 − 𝑦𝑦2
𝜕𝜕𝜕𝜕 𝑦𝑦=𝜀𝜀
Lipschitz Condition and Unique Solution

Suppose that
𝐷𝐷 = 𝑥𝑥, 𝑦𝑦 |𝑎𝑎 ≤ 𝑥𝑥 ≤ 𝑏𝑏, −∞ < 𝑦𝑦 < ∞

and that 𝑓𝑓(𝑥𝑥,𝑦𝑦) is continuous on 𝐷𝐷. If f satisfies a Lipschitz condition on 𝐷𝐷 in terms of


the variable y, then the initial-value problem

𝑑𝑑𝑑𝑑 𝑥𝑥
= 𝑓𝑓 𝑥𝑥, 𝑦𝑦 𝑥𝑥 , 𝑎𝑎 ≤ 𝑥𝑥 ≤ 𝑏𝑏, 𝑦𝑦 𝑎𝑎 = 𝛼𝛼
𝑑𝑑𝑑𝑑

has a unique solution 𝑦𝑦 𝑥𝑥 for 𝑎𝑎 ≤ 𝑥𝑥 ≤ 𝑏𝑏.


Example 1
Consider the initial value problem. Does it have a unique solution? If so, what’s
the Lipchitz constant?

Solution:
Example 1

f satisfies the Lipschitz condition in the variable y with Lipschitz constant 4.

Since, additionally, 𝑓𝑓(𝑥𝑥,𝑦𝑦) is continuous when 0 ≤ 𝑥𝑥 ≤ 2 and −∞ < 𝑦𝑦 < ∞, the


theorem implies that a unique solution exists to the initial-value problem.
Example 2
Determine Lipschitz constant of the function

Solution: In order to obtain the maximum


Find the Lipschitz constant w. r. t. y. value, 𝑥𝑥 = 5, and 𝑦𝑦 = 10
Example 3
Determine the Lipschitz constants of the function

Given 𝐷𝐷 = 𝑥𝑥, 𝑦𝑦 |1 ≤ 𝑥𝑥 ≤ 2, −3 ≤ 𝑦𝑦 < 0 𝑜𝑜𝑜𝑜 0 < 𝑦𝑦 ≤ 4

Solution:
For each point pair (𝑥𝑥, 𝑦𝑦1 ) and (𝑥𝑥, 𝑦𝑦1 ) in
Quiz
Determine Lipschitz constants for the function
Quiz
Solution:

Partial derivative w. r. t. y, and the supremum of the partial derivative.


Quiz
Solution:

Partial derivative w. r. t. y, and the supremum of the partial derivative.


Content
Existence, Uniqueness, and Stability for Ordinary Differential Equations (ODEs)

• Does a unique solution exist?

• If there are more than one solution, we don’t know which solution the numerical
method will lead us towards.)

• Is the problem well-posed?

• If the problem is not well-posed, a small error in the system can lead to a very
different result. The final result will be very unstable.)
Well-posed differential equation
The differential equation is well-posed if
• a solution exists (existence)
• the solution is unique (uniqueness)
• the solution depends on continuously on the initial condition / boundary
condition. Or there are positive constants 𝜀𝜀 and 𝜅𝜅 that a unique solution z(x)
exists to the following ODE. (stability)

𝑑𝑑𝑑𝑑 𝑥𝑥 𝑑𝑑𝑑𝑑 𝑥𝑥
= 𝑓𝑓 𝑥𝑥, 𝑦𝑦 𝑥𝑥 = 𝑓𝑓 𝑥𝑥, 𝑧𝑧 𝑥𝑥 + 𝛿𝛿 𝑥𝑥
𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
𝑧𝑧 𝑥𝑥 − 𝑦𝑦 𝑥𝑥 ≤ 𝜅𝜅𝜅𝜅 whenever 𝛿𝛿 𝑥𝑥 ≤ 𝜀𝜀
Basically, it’s saying that if the differential equation is well-posed, then the
approximated solution is accurate when 𝛿𝛿 𝑥𝑥 and 𝜀𝜀 are small.
Example 1
Consider a rocket is launching up and going straight up into
the sky.
Without wind: altitude of rocket is 𝑦𝑦 𝑥𝑥 and the vertical
speed is given by. (Here 𝒙𝒙 denotes time!)
𝑑𝑑𝑑𝑑 𝑥𝑥
= 𝑓𝑓 𝑥𝑥, 𝑦𝑦 𝑥𝑥
𝑑𝑑𝑑𝑑
With wind: altitude of rocket is 𝑧𝑧 𝑥𝑥 and the vertical speed
is given by.

𝑑𝑑𝑑𝑑 𝑥𝑥
= 𝑓𝑓 𝑥𝑥, 𝑧𝑧 𝑥𝑥 + 𝛿𝛿 𝑥𝑥
𝑑𝑑𝑑𝑑
Example 1

𝑧𝑧 𝑡𝑡 − 𝑦𝑦 𝑡𝑡 ≤ 𝜅𝜅𝜅𝜅 whenever 𝛿𝛿 𝑡𝑡 ≤ 𝜀𝜀

This translates to:


Difference between the altitude of rocket with
and without wind is bounded if the wind speed
is bounded.
Lipschitz condition and well-posed problem
The differential equation is well-posed if (1) a unique solution exists (there is
one and only one solution)
𝑑𝑑𝑑𝑑 𝑥𝑥
= 𝑓𝑓 𝑥𝑥, 𝑦𝑦 𝑥𝑥 𝑓𝑓 𝑥𝑥, 𝑦𝑦1 − 𝑓𝑓 𝑥𝑥, 𝑦𝑦2 ≤ 𝐿𝐿 𝑦𝑦1 − 𝑦𝑦2
𝑑𝑑𝑑𝑑
(2) there are positive constants 𝜖𝜖 and 𝜅𝜅 with the property that a unique solution
z(x) exists to the problem
𝑑𝑑𝑑𝑑 𝑥𝑥
= 𝑓𝑓 𝑥𝑥, 𝑧𝑧 𝑥𝑥 + 𝛿𝛿 𝑥𝑥 𝑓𝑓 𝑥𝑥, 𝑧𝑧1 + 𝛿𝛿 𝑥𝑥 − 𝑓𝑓 𝑥𝑥, 𝑧𝑧2 − 𝛿𝛿 𝑥𝑥 ≤ 𝐿𝐿 𝑧𝑧1 − 𝑧𝑧2
𝑑𝑑𝑑𝑑
with 𝑧𝑧 𝑥𝑥 − 𝑦𝑦 𝑥𝑥 ≤ 𝜅𝜅𝜅𝜅 whenever 𝛿𝛿 𝑥𝑥 ≤ 𝜖𝜖.

Will it be good to say: If 𝑓𝑓 is Lipschitz, then the initial value problem is well-
posed?
Lipschitz condition and well-posed problem
Rewrite the problem as
𝑦𝑦 ′ = 𝑓𝑓 𝑥𝑥, 𝑦𝑦
𝑧𝑧 ′ = 𝑓𝑓 𝑥𝑥, 𝑧𝑧 + 𝛿𝛿 𝑥𝑥

Let 𝜖𝜖 𝑥𝑥 = 𝑧𝑧 𝑥𝑥 − 𝑦𝑦 𝑥𝑥 and let 𝑓𝑓 satisfy the Lipschitz condition


𝑓𝑓 𝑥𝑥, 𝑦𝑦1 − 𝑓𝑓 𝑥𝑥, 𝑦𝑦2 ≤ 𝐿𝐿 𝑦𝑦1 − 𝑦𝑦2 . Then we have
Lipschitz condition and well-posed problem

Since 𝛿𝛿 𝑥𝑥 ≤ 𝜀𝜀, so 𝛿𝛿 𝑥𝑥 cannot go beyond 𝜀𝜀 (positive number). Therefore, we get,

𝜖𝜖𝜖 𝑥𝑥 ≤ 𝑓𝑓 𝑥𝑥, 𝑧𝑧 − 𝑓𝑓 𝑥𝑥, 𝑧𝑧 + 𝜀𝜀


𝜖𝜖𝜖 𝑥𝑥 ≤ 𝐿𝐿 𝜖𝜖 𝑥𝑥 + 𝜀𝜀

the above inequality integrated to give (remember we just need to prove|𝑧𝑧 𝑥𝑥 −


𝑦𝑦 𝑥𝑥 | ≤ 𝜅𝜅𝜅𝜅
Example 1
Determine if the following initial-value problem is well-posed.

Solution:
Let’s find the Lipschitz constant w. r. t. y.

One can prove that by solving this ODE


under the initial condition y(0)=1

Well-posed!
Example 2
Determine if the following initial-value problem is well-posed.

Solution:

The Lipschitz constant w. r. t. y is not bounded!


Not well-posed, solution is not unique
Example 2

One can prove that by solving this ODE

Apply the initial condition y(0)=0, which gives 2 solutions


Example 2
Furthermore, a small perturbation in the initial condition, say 𝑦𝑦(0) = 0.1
instead of 𝑦𝑦(0) = 0 results in a significantly different solution over time.
Example 3
Determine if the following initial-value problem is well-posed.

Solution:

One can prove that by solving this ODE


under the initial condition y(0)=1

Well-posed!
Example 4
Determine if the following initial-value problem is well-posed.

Solution:

One can prove that by solving this ODE


under the initial condition y(0)=0.5

Well-posed! 2
𝑦𝑦 𝑡𝑡 = 𝑡𝑡 + 1 − 0.5𝑒𝑒 𝑡𝑡

You might also like