0% found this document useful (0 votes)
21 views7 pages

MATH219 Lecture 5

Uploaded by

wpaul2860
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
0% found this document useful (0 votes)
21 views7 pages

MATH219 Lecture 5

Uploaded by

wpaul2860
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/ 7

MATH 219

Fall 2024
Lecture 5
Lecture notes by Özgür Kişisel

Content: Exact equations and integrating factors (section 2.6).


Suggested Problems: (Boyce, Di Prima, 10th edition)
§2.6: 3, 5, 9, 11, 13, 18, 20, 24, 30, 32

1 Exact equations
Let us consider the case of an arbitrary first order ODE once again. Suppose that
our independent variable is x. Let us assume for a moment that we found the
solutions of the equation and that they can be written in the implicit form
F (x, y) = c.
by leaving the constant c alone. Of course, when c changes, the solution curve will
change; we view the solution curves as level curves of the function F . We can
easily express dy/dx in terms of x as follows: Take the derivative of both sides with
respect to x. By the chain rule,
∂F ∂F dy
+ = 0
∂x ∂y dx
dy ∂F/∂x
= −
dx ∂F/∂y
The question is whether we can reverse this process. Namely, given the ODE, can
we recover such a function F (x, y)? One important remark at this point is that,
even if we could, given the ODE we just know the ratio of ∂F/∂x to ∂F/∂y, but
F and the values of the partial derivatives themselves are not uniquely determined
at all. In some favourable cases, the functions appearing in the particular way we
write the ODE will a priori be equal to the derivatives of a certain function F . To
better understand this, let us write the ODE in a more symmetric form:
dy
M (x, y) + N (x, y) = 0.
dx
1
An equivalent way of writing this equation is

M (x, y)dx + N (x, y)dy = 0.

Derivatives with respect to x are suppressed in this notation. Instead of d/dx we


write d etc. 1 With this notation, we have
∂F ∂F
d(F (x, y)) = dx + dy.
∂x ∂y

Definition 1.1 The equation M (x, y)dx + N (x, y)dy = 0 is called exact on a do-
main D if there exists a differentiable function F (x, y) on D such that the left hand
side of the equation can be written as d(F (x, y)) on D. A function F (x, y) satisfying
this condition is called a potential for this equation.

By the chain rule, the condition on F (x, y) is equivalent to the pair of equations

∂F ∂F
= M, =N
∂x ∂y

We emphasize that the potential F (x, y) must be defined as a single valued function
on the whole domain D.

Example 1.1 The equation ydx+(x+2y)dy = 0 is exact on R2 . Indeed, if F (x, y) =


xy + y 2 then

d(F (x, y)) = d(xy + y 2 )


= ydx + (x + 2y)dy

Remark 1.1 Recall from multivariable calculus that a vector field ⟨M (x, y), N (x, y)⟩
is called conservative if it can be written in the form ∇F for some function F (x, y).
It is clear that the equation M dx + N dy = 0 is exact if and only if the vector field
⟨M, N ⟩ is conservative.
1
From a more advanced perspective, this is an equality of differential 1-forms on R2 when
restricted to a solution curve. We will not pursue this viewpoint here.

2
If the equation M dx + N dy = 0 is exact with potential F , then it can be rewritten
as dF = 0. Consequently, the equations F (x, y) = c for arbitrary values of c give us
all solutions of the ODE in an implicit form.
Example: Solve the initial value problem ydx + (x + 2y)dy = 0, y(1) = 5.
Solution: We saw above that the equation is exact with F (x, y) = xy+y 2 a potential.
Therefore the solutions of the equation are xy + y 2 = c. Using the intial condition,
we find that c = 1 × 5 + 52 = 30. Hence the solution is xy + y 2 = 30 (in implicit
form). □
Suppose that M (x, y) and N (x, y) are themselves continuously differentiable on a
common domain D. As in the case of conservative vector fields, a necessary condition
for the existence of a potential function is

∂M ∂F ∂F ∂N
= = =
∂y ∂x∂y ∂y∂x ∂x

This condition is not always sufficient for the existence of a potential F . However, if
the domain is simply connected, then it is sufficient. A simply connected domain,
intuitively, is a domain with no interior holes. An example of a simply connected
domain is the interior of a rectangle. We will formulate and prove the claim made
in this paragraph in the particular case of a rectangular domain.

Theorem 1.1 (Test for exactness) Suppose that M, N, ∂M/∂y and ∂N/∂x are con-
tinuous on a rectangular domain R = (a, b) × (c, d). Then M dx + N dy = 0 is exact
if and only if ∂M/∂y = ∂N/∂x at each point of R.

Proof: Fix x0 ∈ R. The functions F (x, y) satisfying the equation ∂F/∂x = M can
be found by integrating M along a line segment from (x0 , y) to (x, y), since each
such line segment remains in the rectangle R:
Z (x,y)
F (x, y) = M (s, y)ds
(x0 ,y)

The result is any antiderivative of M with respect to x plus a function of y to be


determined. Namely, it is of the form F (x, y) = R(x, y) + h(y) where ∂R/∂x = M .
The question is whether or not we can always choose h(y) so that the equation

3
∂F/∂y = N is also satisfied. We need
∂F ∂R
= + h′ (y) = N (x, y)
∂y ∂y
∂R
h′ (y) = N (x, y) −
∂y
The last equation has a solution for h′ (y) (and consequently for h(y)) if and only if
its right hand side is independent of x. In order to test whether this is true or not,
let us look at its partial derivative with respect to x:
∂ 2R
 
∂ ∂R ∂N
N (x, y) − = −
∂x ∂y ∂x ∂x∂y
∂N ∂ 2R
= −
∂x ∂y∂x
∂N ∂M
= −
∂x ∂y
= 0.
Therefore we can solve for h(y), and the equation is exact. This completes the proof.

Example: Find the value of the constant a for which the ODE
3ey dx + (2y + axey )dy = 0
is exact on R2 . Solve the equation for this value of a.
Solution: M (x, y) = 3ey and N (x, y) = 2y + axey . Since
∂M ∂N
= 3ey = aey
∂y ∂x
the equality ∂M/∂y = ∂N/∂x holds if and only if a = 3. Since M, N and their
partial derivatives are all continuous on R2 , and R2 is simply connected, we can
conclude that the ODE is exact for a = 3. Now,
∂F
= 3ey
∂x
F (x, y) = 3xey + h(y)
∂F
= 3xey + h′ (y) = 2y + 3xey
∂y

h (y) = 2y.

4
Therefore h(y) = y 2 is a solution and F (x, y) = 3xey + y 2 is a potential. The
solutions of the ODE are
3xey + y 2 = c
where c ∈ R is a constant. □
Example: Solve the initial value problem
 
2xy
2
− 2x dx − (2 − ln(x2 + 1))dy = 0, y(5) = 0.
x +1
Determine the largest interval on which the solution is valid.
2xy
Solution: M (x, y) = − 2x and N (x, y) = −2 + ln(x2 + 1). We compute
x2 + 1
∂M 2x ∂N
= 2 =
∂y x +1 ∂x
Both M, N and their partial derivatives are continuous on R2 , and R2 is simply
connected. Therefore, by the test for exactness, the equation is exact. Let us find a
potential:
∂F 2xy
= 2 − 2x
∂x x +1
F (x, y) = y ln(x2 + 1) − x2 + h(y)
∂F
= ln(x2 + 1) + h′ (y) = −2 + ln(x2 + 1)
∂y
h′ (y) = −2
h(y) = −2y.
We deduce that F (x, y) = y ln(x2 + 1) − x2 − 2y is a potential. All solutions of the
ODE are y ln(x2 + 1) − x2 − 2y = c. Using the initial condition y(5) = 0, we find
that 0 ln(52 + 1) − 52 − 2 × 0 = c, therefore c = −25. So,
y ln(x2 + 1) − x2 − 2y = −25
y(ln(x2 + 1) − 2) = x2 − 25
x2 − 25
y= .
ln(x2 + 1) − 2
This function is defined if and only if ln(x2 + 1) − 2 ̸= 0, namely for x2 + 1 ̸= e2 .
The interval
√ of definition,
√ which √must be a connected
√ interval, could then be either
2 2 2 2
of (−∞, − e − 1), (− e − 1, e − 1) or (√ e − 1, ∞) but since the initial point
x = 5 belongs to the last one, the answer is ( e2 − 1, ∞). □

5
2 Integrating Factors
Recall from the lecture on first order linear equations that an ODE of the form
y ′ + p(t)y = q(t) can be solved by multiplying the equation by an integrating factor
µ(t). In this case, the equation
R for µ(t) turned out to be easy to solve and we even
got a formula µ(t) = exp( p(t)dt).
Let us now suppose that we have an ODE of the form M (x, y)dx + N (x, y)dy = 0.
If the equation is exact, then we know how to proceed. If it is not exact, we may
try to find an integrating factor µ(x, y) such that, after multiplication with µ, the
new ODE
µM dx + µN dy = 0
is exact. For simplicity, let us assume that all of these functions and their partial
derivatives are continuous on a rectangle R, so that we can use the test for exactness.
Then the new equation is exact if and only if
∂(µM ) ∂(µN )
=
∂y ∂x
∂µ ∂M ∂µ ∂N
M +µ = N +µ .
∂y ∂y ∂x ∂x
The problem that we encounter here is that this new differential equation for µ is
terribly difficult to solve. It is not even an ODE; it is a PDE. Therefore, finding
an integrating factor in this very general setting is a hopelessly difficult task. Only
when there is some additional information that tells us something about the form of
the integrating factor, this method could be useful. It is unclear what these “special
cases” are, although there are several classical textbooks on differential equations
which tabulate some of them.
Example: Show that µ(x, y) = (x2 + y 2 )−1 is an integrating factor for the ODE

(3x2 + x + 3y 2 )dx + (7x2 + y + 7y 2 )dy = 0

and use it to find all solutions of this ODE.


Solution: The original equation is not exact (please check this yourself). If we
multiply the ODE throughout by the given function µ(x, y), we obtain
   
x y
3+ dx + 7 + dy = 0.
x + y2
2 x + y2
2

6
The functions 3 + x/(x2 + y 2 ) and 7 + y/(x2 + y 2 ) are defined on R2 − {(0, 0)}. Since
this set is not a simply connected domain, we cannot use the test for exactness in
this case. Instead, we should try to directly show that a potential function exists:
∂F x
= 3+
∂x x2 + y2
1
F (x, y) = 3x + ln(x2 + y 2 ) + h(y)
2
∂F y y
= 2 2
+ h′ (y) = 7 + 2
∂y x +y x + y2

h (y) = 7
h(y) = 7y.

Therefore the function F (x, y) = 3x + 7y + ln(x2 + y 2 )/2 is a potential. Existence


of a potential implies that µ(x, y) is indeed an integrating factor. The solutions of
the ODE are
1
3x + 7y + ln(x2 + y 2 ) = c
2
where c ∈ R is a constant. □

You might also like