3.1 Linear systems and Linearity principles

advertisement
MA282 - DE
3.1 Linear systems and Linearity principles
* Linear systems and 2nd order linear equations are most important in this course.
Review
(1) Linear system with two dependent variables
𝑑π‘₯
= π‘Žπ‘₯ + 𝑏𝑦
𝑑𝑑
{
𝑑𝑦
= 𝑐π‘₯ + 𝑑𝑦
𝑑𝑑
where π‘Ž, 𝑏, 𝑐, 𝑑 are constants.
In vector notaion,
π‘‘π‘Œ
= 𝐹(π‘Œ)
𝑑𝑑
where π‘Œ = (π‘₯, 𝑦).
(2) 2nd order homogeneous linear equation
𝑑2𝑦
𝑑𝑦
+
𝑝
+ π‘žπ‘¦ = 0
𝑑𝑑 2
𝑑𝑑
It can be converted into a linear system:
𝑑𝑦
=𝑣
{ 𝑑𝑑
𝑑𝑣
= −𝑝𝑣 − π‘žπ‘¦
𝑑𝑑
Matrix notation for linear systems
𝑑π‘₯
= π‘Žπ‘₯ + 𝑏
{ 𝑑𝑑
𝑑𝑦
= 𝑐π‘₯ + 𝑑𝑦
𝑑𝑑
Let 𝐴 = (
π‘Ž
𝑐
π‘₯
𝑏
) and π‘Œ = (𝑦); then we can rewrite as
𝑑
𝑑π‘₯
π‘Ž
( 𝑑𝑑 ) = (
𝑑𝑦
𝑐
𝑑𝑑
1
𝑏 π‘₯
)( )
𝑑 𝑦
Or more compactly as
π‘‘π‘Œ
= π΄π‘Œ
𝑑𝑑
where 𝐴 is called the coefficient matrix.
Example 1. Partially decoupled system
𝑑π‘₯
= −π‘₯ + 2𝑦
𝑑𝑑
{
𝑑𝑦
=𝑦
𝑑𝑑
The general solution is
{
π‘₯(𝑑) = 𝑦0 𝑒 𝑑 + (π‘₯0 − 𝑦0 )𝑒 −𝑑
𝑦(𝑑) = 𝑦0 𝑒 𝑑
where (π‘₯(0), 𝑦(0)) = (π‘₯0 , 𝑦0 ).
Written in vector notation, the general solution is
𝑦0
π‘₯ − 𝑦0
π‘Œ(𝑑) = 𝑒 𝑑 (𝑦 ) + 𝑒 −𝑑 ( 0
)
0
0
Example 2. Damped harmonic oscillator
𝑑2𝑦
𝑑𝑦
+
3
+ 2𝑦 = 0
𝑑𝑑 2
𝑑𝑑
The equivalent system is
𝑑𝑦
=𝑣
{ 𝑑𝑑
𝑑𝑣
= −3𝑣 − 2𝑦
𝑑𝑑
We used the guessing technique to find solutions 𝑦1 (𝑑) = 𝑒 −2𝑑 and 𝑦2 (𝑑) = 𝑒 −𝑑 . (The
characteristic equation is 𝑠 2 + 3𝑠 + 2 = 0. ) In vector notation,
−2𝑑
1
π‘Œ1 (𝑑) = ( 𝑒 −2𝑑 ) = 𝑒 −2𝑑 ( )
−2
−2𝑒
−𝑑
1
π‘Œ2 (𝑑) = ( 𝑒 −𝑑 ) = 𝑒 −𝑑 ( )
−1
−𝑒
2
Given a linear system
π‘‘π‘Œ
𝑑𝑑
= π΄π‘Œ, how do we calculate the vector field at any given point π‘Œ0 ?
We do matrix multiplication
𝑏 π‘₯0
)( )
𝑑 𝑦0
π‘Ž
𝑐
π΄π‘Œ0 = (
Example
𝑑π‘₯
=𝑦
{ 𝑑𝑑
𝑑𝑦
= −π‘₯
𝑑𝑑
𝑑π‘₯
0 1 π‘₯
⇔ ( 𝑑𝑑 ) = (
)( )
𝑑𝑦
−1 0 𝑦
𝑑𝑑
The matrix notation is useful for a large number of dependent variables.
How do we calculate the equilibrium points of
π‘‘π‘Œ
𝑑𝑑
= π΄π‘Œ?
Need to solve
π΄π‘Œ = 𝑂 ⇔ (
ο‚·
π‘Ž
𝑐
π‘Žπ‘₯ + 𝑏𝑦 = 0 − − − π‘’π‘ž1
0
𝑏 π‘₯
) (𝑦) = ( ) ⇔
𝑐π‘₯ + 𝑑𝑦 = 0 − − − π‘’π‘ž2
0
𝑑
Clearly, (0, 0) is a solution, so an equilibrium point. This solution is called the trivial
solution.
ο‚·
To find other equilibrium points, from eq1, if π‘Ž ≠ 0,
𝑏
π‘₯=− 𝑦
π‘Ž
Substitute this for π‘₯ in eq2,
𝑏
𝑐 (− 𝑦) + 𝑏𝑦 = 0
π‘Ž
which is equivalent to
(π‘Žπ‘‘ − 𝑏𝑐)𝑦 = 0
Hence either π‘Žπ‘‘ − 𝑏𝑐 = 0 or 𝑦 = 0.
If 𝑦 = 0 then π‘₯ = 0, again we have the trivial solution. Therefore, a linear system has
nontrivial equilibrium points only if π‘Žπ‘‘ − 𝑏𝑐 = 0.
3
This quantity π‘Žπ‘‘ − 𝑏𝑐 is the determinant of the coefficient matrix 𝐴, denoted det𝑨 or |𝑨|
Theorem The origin is always an equilibrium point of a linear system. It is only equilibrium
point if an only det𝐴 ≠ 0.
Linearity Priniciple
Return to Example 1. Partially decoupled system. In matrix notation,
π‘‘π‘Œ
−1
=(
0
𝑑𝑑
2
)π‘Œ
1
Consider three different initial conditions
1
1
2
π‘Œ1 (0) = ( ), π‘Œ2 (0) = ( ), π‘Œ3 (0) = ( )
0
1
1
They corresponds to the three solutions
𝑑
−𝑑
1
1
π‘Œ1 (𝑑) = 𝑒 −𝑑 ( ), π‘Œ2 (𝑑) = 𝑒 𝑑 ( ), π‘Œ3 (𝑑) = (𝑒 +𝑑𝑒 )
0
1
𝑒
since
{
𝑦0
π‘₯(𝑑) = 𝑦0 𝑒 𝑑 + (π‘₯0 − 𝑦0 )𝑒 −𝑑
π‘₯ − 𝑦0
⇔ π‘Œ(𝑑) = 𝑒 𝑑 (𝑦 ) + 𝑒 −𝑑 ( 0
)
𝑑
0
0
𝑦(𝑑) = 𝑦0 𝑒
Graphs:
How are these 3 solutions related?
π‘Œ3 (𝑑) = π‘Œ1 (𝑑) + π‘Œ2 (𝑑)
4
Theorem (Linearity Principle) Suppose that
π‘‘π‘Œ
= π΄π‘Œ
𝑑𝑑
is a linear system of differential equations.
1. If π‘Œ(𝑑) is a solution and k is a constant, then π‘˜π‘Œ(𝑑) is also a solution.
2. If π‘Œ1 (𝑑), π‘Œ2 (𝑑) are 2 solutions, then π‘Œ1 (𝑑) + π‘Œ2 (𝑑) is also a solution.
Therefore the linear combination of π‘Œ1 (𝑑) and π‘Œ2 (𝑑), i.e., π‘˜1 π‘Œ1 (𝑑) + π‘˜2 π‘Œ2 (𝑑) is a solution of a
linear system.
Example Solve
π‘‘π‘Œ
−1
=(
0
𝑑𝑑
2
) π‘Œ,
1
−1
)
−2
π‘Œ(0) = (
Note π‘Œ(𝑑) = π‘˜1 π‘Œ1 (𝑑) + π‘˜2 π‘Œ2 (𝑑) is the solution. Now we can find
1
−1
1
( ) = π‘Œ(0) = π‘˜1 π‘Œ1 (0) + π‘˜2 π‘Œ2 (0) = π‘˜1 ( ) + π‘˜2 ( )
0
−2
1
π‘˜ + π‘˜2
=( 1
)
π‘˜2
⇒ π‘˜2 = −2, π‘˜1 = 1
So the solution to IVP is
−𝑑
𝑑
−𝑑
𝑑
π‘Œ(𝑑) = (𝑒 ) − 2 (𝑒 𝑑 ) = (𝑒 − 2𝑒
𝑑 )
0
𝑒
−2𝑒
Question For an arbitrary linear system
every IVP?
π‘‘π‘Œ
𝑑𝑑
= π΄π‘Œ, how many solutions do we need to solve
Answer 2
Definition Two solutions π‘Œ1 (𝑑), π‘Œ2 (𝑑) of a linear system for which π‘Œ1 (0), π‘Œ2 (0) are linearly
independent are called linearly independent solutions of the linear system.
1
1
Here, π‘Œ1 (0) = ( )and π‘Œ2 (𝑑) = ( ) are linearly independent
0
1
⇔ (1, 0) and (1, 1) do not lie on the same line through the origin.
5
Download