Math 2280 - Lecture 20
Dylan Zwick
Spring 2013
Today we’ll learn about a method for solving systems of differential
equations, the method of elimination, that is very similar to the elimination
methods we learned about in linear algebra. We’ll extend this analogy
further by learning about polynomial differential operators, and how we
can apply analogues of Cramer’s rule using these differential operators to
solving systems of differential equations.
This lectures corresponds with section 4.2 of the textbook. The assigned problems for this section are:
Section 4.2 - 1, 10, 19, 28, 30
The Method of Elimination
For systems of differential equations, particularly linear systems, we can
combine equations like we do in linear algebra to eliminate dependent
variables. This simplifies the equation, and eventually can be used to reduce the equation to a single linear differential equations (usually not of
first-order) that we can then solve using the methods from chapter 3.
Example - Use the method of elimination to find the solution to the initial value problem:
x′ = −3x + 2y
y ′ = −3x + 4y;
x(0) = 1, y(0) = −1
1
More room for the example problem.
x(
)+Y
-Yv+
I
yy/y 7
q\I
y’ Zy
+
/
1
V
/
j2j-€kye4fI{
l-zy
.__7
‘c+h
?y
t3
5e)
/
?y’E_+Zy
1
L
-1(yzy)
fl-y I_
I-r-
7
7
(r-3)()
()
((
y()
Cg
e
--
y’( 3c-ç
•D) _3) cie 4
-3/
+
(,
3+
x()
-
e
5r r—L
-7
—
y(o)
CtL
Polynomial Differential Operators
A polynomial differential operator is a map from functions to functions
the form:
L
=
’+
a_
D
aD + 1
...
,
0
D+a
1
+a
where D represents the derivative operator. So, for example, the differ
ential operator
2D_3
LzD
+
2
when applied to the function
2
7
f (x) = x2 + 4x − 5
yields
L(f ) = D 2 (x2 + 4x − 5) + 2D(x2 + 4x − 5) − 3(x2 + 4x − 5)
= (2) + (4x + 8) − (3x2 + 12x − 15) = −3x2 − 8x + 25.
Polynomial differential operators commute. So, if we have two differential operators, L1 and L2 , then L2 L1 (f ) = L1 L2 (f ). Note that this is not
generally the case for all differential operators. It’s not even necessarily
the case for polynomial operators with variable coefficients.
Any system of two linear differential equations with constant coefficients can be written in the form
L1 x + L2 y = f1 (t)
L3 x + L4 y = f2 (t)
If we act on the top row with the operator L3 , and on the bottom row
with the operator L1 we get
L3 L1 x + L3 L2 y = L3 f1 (t)
.
L1 L3 x + L1 L4 y = L1 f2 (t)
If we then subtract the first equation from the second, using the fact
that the operators commute, we get
(L1 L4 − L2 L3 )y = L1 f2 − L3 f1 ,
in the single variable y. Alternatively, we could have eliminated y in a
like manner from the original system and obtained the equation
(L1 L4 − L2 L3 )x = L4 f1 − L2 f2 .
3
), appears on the left hand
3
2
L
4
L
1
Note that the same operato (L
the operational determinant:
is
called
This
operator
side of both equations.
—
I
L
1
3
L
L
2
4
1
L
4
L
—
,
3
9
L
and the above two equalities are the operational versions of Cramer’s
rule:
L
1
3
L
:
L L
1
2
3 L
L
4
—
L
1
3
L
f
f2
Note that if the operational determinant is identically zero there can be
either no solutions, or infinitely many solutions, to the ODE. Also, this ap
proach that we’ve applied here for systems with two variables generalizes
to systems with an arbitrary numbers of variables, although the computa
tions involved for Cramer’s rule become more and more narsty.
Example Show that the following system is degenerate, and then de
termine whether it has infinitely many or no solutions.
-
(D+2)x +
(D+3)x +
(D + 2)y
(D + 3)y
=
t
3
e
=
t
2
e
T)(Of (PtD3)
c
e
eyI4
ct)
z
yet’
-z_ ; 4
1
ft .9L f(fi(Ci7
cq 4d
If
5
y()
4
)fl 1)11/
$0
Ci(
(C1,41r
M,j
Lii
Hc
H
tN
*—
-H--
cS
H-
-4-
\
(
—,Tho
In
—
I
++
;_ ;_
1+
I
I
CD
-p
CD
CD
C
CD
I;,)
CD
C
C
C
C1I
(D
(D
rI