EM701p First-Order Ordinary Differential Equations

Advanced Engineering Mathematics 1.
First-order ODEs 1
1. First-order Ordinary Differential Equations

1.1 Basic concept and ideas
1.2 Geometrical meaning of direction fields
1.3 Separable differential equations
1.4 Exact differential equations and
Integrating factors
1.5 Linear differential equations and
Bernoulli equations
1.6 Orthogonal trajectories of curves
1.7 Existence and uniqueness of solutions
Advanced Engineering Mathematics 1. First-order ODEs 2
1.1 Basic concepts and ideas
V Equations − b ± b2 − 4ac − 1 ± 1 + 48 4
= = 1, -
3y 2 + y - 4 = 0 ⇒ y = ? 2a 6 3
where y is an unknown.
x = -2 , f(x) = -24
V Functions x = -1 , f(x) = -6
f(x) = 2x 3 + 4x , x = 0 , f(x) = 0
where x is a variable. x = 1 , f(x) = 6
: :
V Differential equations
A differential equation is an equation contains one or several derivative
of unknown functions (or dependent variables). For example,
2
d y + x ⎡ dy ⎤ + 2 dy = xy ,
2
e
d x2 ⎢⎣ dx ⎥⎦ x dx ⇐ (ordinary differential equation)
∂ z2 ∂z ∂z ⇐ (partial differential equation)

+ +x + y 3 z = xyz .
∂x∂y ∂x ∂y
V There are several kinds of differential equations

V An ordinary differential equation (ODE) is an equation that contains one
independent variable and one or several derivatives of an unknown
function (or dependent variable), which we call y(x) and we want to
determine from the equation. For example,
dy
= cos x (i.e., y ′ = cos x )
dx
x
x y ′′′ y ′ + 2 e y ′′ = ( x + 2) y
2 2 2
where y is called dependent variable and

x is called independent variable.
V If a differential equation contains one dependent variable and two or
more independent variables, then the equation is a partial differential
equation (PDE).
V If differential equations contain two or more dependent variable and one
independent variable, then the set of equations is called a system of
differential equations.
V Summary
A differential equation contains
(1) one dependent variable and one independent variable Ö
an ordinary differential equation.
(2) one dependent variable and two or more independent variable Ö
a partial differential equation.
(3) Two or more dependent variable and one independent variable Ö
a system of differential equations.
y1’(x) = 2 y1(x) - 4 y2(x) y1(x) = c1 4 ex + c2 e -2x
y2’(x) = y1(x) - 3 y2(x) y2(x) = c1 ex + c2 e -2x
(4) Two or more dependent variable and two or more independent
variable Ö a system of partial differential equations.
(rarely to see)
V What is the purpose of differential equations ?

Many physical laws and relations appear mathematically in the form of
such equations. For example, electronic circuit, falling stone, vibration, etc.
(1) Current I in an RL-circuit I
Resister (R) Electro-motive

LI’ + RI = E.
Inductor (L) force (E)
(2) Falling stone y
y “ = g = constant.
(3) Pendulum
L
Lθ ” + g sinθ = 0.
θ
V Any physical situation involved motion or measure rates of change can be

described by a mathematical model, the model is just a differential
equation.
The transition from the physical problem to a corresponding mathematical
model is called modeling.
In this course, we shall pay our attention to solve differential equations
and don’t care of modeling.
Physical modeling solving

situation DE Solution
That is, the purposes of this course are that

given a differential equation
1. How do we know whether there is a solution ?
2. How many solutions might there be for a DE, and how are they related?
3. How do we find a solution ?
4. If we can’t find a solution, can we approximate one numerically?
V A first-order ODE is an equation involving one dependent variable, one

independent variable, and the first-order derivative. For example,
y’ + xy 2 – 4 x 3 = 0
(y’ )3/2 + x 2 – cos(xy’) = 0.
V A solution of a first-order ODE is a function which satisfies the equation.

For example,
y(x) = e 2x is a solution of y ‘ – 2y = 0.
y(x) = x 2 is a solution of xy ‘ = 2y.
V A solution which appears as an implicit function, given in the form

H(x, y) = 0, is called an implicit solution;
for example x 2 + y 2 –1 = 0 is an implicit solution of DE yy ‘ = -x.
In contrast to an explicit solution with the form of y = f(x);
for example, y = x 2 is an explicit solution of xy ’ = 2y.
V A general solution is a solution containing one arbitrary constant;

for example, y = sin x + c is a general solution of y’ = cos x.
A particular solution is a solution making a specific choice of constant on
the general solution. Usually, the choice is made by some additional
constraints.
For example, y = sin x – 2 is a particular solution of y’ = cos x with the
condition y(0) = - 2.
V A differential equation together with an initial condition is called an initial

value problem. For example,
y’ = f(x, y), y(x0) = y0,
where x0 and y0 are given values.
−1 − 3
V For example, xy ’ = 3y, y (-4) = 16 ⇒ y = cx 3 ⇒ c = ⇒ y= x .
4 4
V Problem of Section 1.1.

1.2 Direction fields

V Purpose
To sketch many solution curves of a given DE without actually solving
the differential equation.
V Method of direction fields

The method applies to any differential equation y ’ = f (x, y).
Assume y(x) is a solution of a given DE.
y (x) has slope y’(x0) = f (x0, y0) at (x0, y0).
(i) draw the curves f(x, y) = k , k is a real constant. These curves are
called isoclines of the original DE.
(ii) along each isocline, draw a number of short line segments (called
lineal element) of slope k to construct the direction field of the original
DE. (That is, the direction field is just the set of all connected lineal
elements.)
V Ex.1. Graph the direction field of the 1st-order DE y’ = xy.

(i) draw the curves (isoclines) xy = …-2, -1, 0, 1, 2, …
y
xy = 3
xy = 2
xy = 1
x
xy = -1
xy = -2
xy = -3
(ii) draw lineal elements on each isocline,
y xy = 3
xy = 2
xy = 1
x
xy = -1
xy = -2
xy = -3
(iii) connect the related lineal elements to form the direction field.
(a) By computer. (b) By hand.

Direction field of y’ = xy.
V Problems of Section 1.2.
1.3 Separable differential equations

V A DE is called separable if it can be written in the form of
g(y) y’ = f(x) or g(y) dy = f(x) dx
To solve the equation by integrate both sides with x,

∫ g ( y )y ' dx = ∫ f ( x )dx + c
dy
⇒ ∫ g(y ) dx = ∫ f ( x )dx + c
dx
⇒ ∫ g ( y )dy = ∫ f ( x )dx + c .
V Ex.1. Solve 9 yy '+4 x = 0.

9 2
⇒ 9 ydy = −4 xdx ⇒ y = −2 x 2 + c '
2
x2 y 2
⇒ + = c with c = c ' .
9 4 18
V Ex.2. Solve y ' = 1 + y 2

dy
⇒ = dx
1+ y 2
1 1 −1⎛ x ⎞
⇒ tan−1( y ) = x + c ∫ dx = tan ⎜ ⎟+c
2
a x+ 2
a ⎝a⎠
⇒ y = tan( x + c ).
V Ex. Initial value problem

y’ + 5x4y 2 = 0 with initial condition y(0)=1.
dy dy
⇒ = −5 x 4 y 2 ⇒ 2 = −5 x 4dx
dx y
1 1
⇒ − = −x5 + c ⇒ y = 5 .
y x −c
1
Since y (0) = 1 and y (0) = ⇒ c = −1
−c
1
⇒ y= 5 .
x +1
V Ex.3. Solve y’ = ky , y(0) = 2.
x
V Ex. Solve y ' = , y (1) = 3.
y
1 2 1 2
⇒ ydy = xdx ⇒ y = x + c ⇒ y = ± x 2 + 2c .
2 2
Since y (1) = 3 = 1 + 2c
⇒ c = 4 ⇒ solution y ( x ) = x 2 + 8.
V Ex.5. Solve y’ = -2xy , y(0) = 1.
V Example of no separable DE (x-1)y’ = 3x2 + y.
V Note: There is no nice test to determine easily whether or not a 1st-order

equation is separable.
Reduction to separable forms

V Certain first-order differential equation are not separable but can be made
separable by a simple change of variables (dependent variable)
The equation of the form y' = g ⎛⎜ ⎞⎟ can be made separable; and the
y
⎝x⎠
form is called the R-1 formula.
y
step 1. Set = u , then y = ux (change of variables).
x
step 2. Differential y’ = u + xu’ (product differentiation formula).
⎛y⎞
step 3. The original DE y ' = g ⎜ ⎟ ⇒ u + xu ' = g (u )
⎝x⎠
du g (u ) − u du dx
⇒ xu ' = g (u ) − u ⇒ = ⇒ = .
dx x g (u ) − u x
step 4. integrate both sides of the equation.
step 5. replace u by y/x.
V Ex.6. Solve 2xyy’ – y 2 + x2 = 0.

Dividing by x2, we have
2
y ⎛y⎞ y
2 y '−⎜ ⎟ + 1 = 0 ⇒ 2u (u + u ' x ) − u 2 + 1 = 0 by setting u =
x ⎝x⎠ x
⇒ 2u 2 + 2uxu '−u 2 + 1 = 0 ⇒ 2uxu '+u 2 + 1 = 0
2udu − dx
⇒ − 2uxu ' = u 2 + 1 ⇒ =
u2 + 1 x f '(x)
∫ dx = ln f ( x ) + c
f (x)
( )
⇒ ln 1 + u 2 = − ln x + c * ⇒ 1 + u 2 =
c
x
2 2
⎛y⎞ c ⎛ c⎞ c2
⇒ 1 + ⎜ ⎟ = ⇒ x + y = cx ⇒ ⎜ x − ⎟ + y = .
2 2 2
⎝x⎠ x ⎝ 2⎠ 4
V Ex. Solve initial value problem
y 2 x 3 cos( x 2 )
y' = + , y ( π ) = 0.
x y
y
Change of variable u =
x
2 x 2 cos( x 2 )
⇒ xu '+u = u + ⇒ uu ' = 2 x cos( x 2 )
u
⇒
u2
2
( ) ( )
= sin x 2 + c ⇒ y = ux = x 2 sin x 2 + 2c .
Since y ( π ) = 0 ⇒ c = 0 ⇒ y = x 2 sin x 2 . ( )
V Ex.7. Solve (2x - 4y + 5) y’ + x - 2y + 3 = 0.

If we set u=y/x, then the equation will become no-separable.
1
One way by setting x - 2y = v. Then y ' = (1 − v ' ).
2
1 5 5
⇒ (2v + 5) (1 − v ' ) + v + 3 = 0 ⇒ v + − vv '− v '+v + 3 = 0
2 2 2
⇒ 2v + 5 − 2vv '−5v '+2v + 6 = 0 ⇒ (2v + 5)v ' = 4v + 11
2v + 5 1 1
⇒ dv = dx ⇒ (1 − )dv = dx
4v + 11 2 4v + 11
1 1
⇒ (1 − )dv = 2dx ⇒ v − ln 4v + 11 = 2 x + c * .
4v + 11 4
Since v = x − 2y
f '(x)
1 ∫ dx = ln f ( x ) + c
⇒ x − 2y − ln 4 x − 8 y + 11 = 2 x + c * f (x)
4
⇒ 4 x − 8 y − ln 4 x − 8 y + 11 = 8 x − c
⇒ 4 x + 8 y + ln 4 x − 8 y + 11 = c.
R-2 formula
V Now we want to handle differential equations of the form
dy ⎛ ax + by + c ⎞
= f⎜ ⎟ , where a, b, c, g, e, and h are constants.
dx ⎝ gx + ey + h ⎠
dy ⎛ a + b ( y / x ) + (c / x ) ⎞
It implies that = f⎜ ⎟,
dx ⎝ g + e ( y / x ) + (h / x ) ⎠
which is R-1 formula when c = h = 0, and
R-2 formula when c ≠ 0 or h ≠ 0.
V There are two ways to solve the equation:

i. R-2 formula ⇒ R-1 formula ⇒ separable or
ii. R-2 formula ⇒ separable (directly).
V Case 1. Suppose that ae – bg ≠ 0.

Change variables x = X + α
y = Y + β to eliminate the effect of c and h ,
where X and Y are two new variables; α and β are two constants.
dy dY
=
dx dX
The differential equation becomes
dY ⎛ a ( X + α ) + b (Y + β ) + c ⎞ ⎛ a X + b Y + (aα + bβ + c ) ⎞
= f⎜ ⎟ = f⎜ ⎟.
dX ⎝ g ( X + α ) + e (Y + β ) + h ⎠ ⎝ g X + e Y + ( g α + eβ + h ) ⎠
Now we choose α and β such that
aα + bβ + c = 0
{ gα + eβ + h = 0
Since ae – bg ≠ 0 , then exist α and β satisfying these equations
⎛ ⎛Y ⎞⎞
⎜ a + b⎜ ⎟ ⎟
dY ⎛ aX + bY ⎞ ⎜ ⎝ X ⎠⎟.
Such that = f⎜ ⎟=f It is a R - 1 formula.
dX ⎝ gX + eY ⎠ ⎜ g + e⎛⎜ Y ⎞⎟ ⎟
⎜ ⎟
⎝ ⎝ X ⎠⎠
V Ex.
2
dy ⎛ 2 x + y − 1⎞
=⎜ ⎟ , where ae – bg = 2 * 0 – 1 * 1 ≠ 0.
dx ⎝ x − 2 ⎠
Let x = X + α and y =Y + β to get
dY ⎛ 2 X + Y + (2α + β − 1) ⎞
2
=⎜ ⎟
dX ⎝ X + (α − 2 ) ⎠
Solving the system of linear equations
2 α + β -1 = 0
{ α - 2 = 0 ⇒ α = 2 and β = -3.
Then the equation becomes
2 2
dY ⎛ 2 X + Y ⎞ ⎛ Y⎞
=⎜ ⎟ = ⎜2 + ⎟ .
dX ⎝ X ⎠ ⎝ X⎠
Let u = Y / X ⇒ Y = Xu .
dY du du
= (2 + u ) ⇒ X
2
=u+ X = u 2 + 3u + 4
dX dX dX
⇒ 2
du dX
⇒
du dX 1 1 −1 x
= = ∫ dx = tan +c.
u + 3u + 4 X ⎛
2
3⎞ 7 X 2
x a + 2
a a
⎜u + ⎟ +
⎝ 2⎠ 4
3
u +
⇒ ln X =
1
tan−1 2 + c ⇒ ln X = 2 tan−1 ⎡ 2 ⎛⎜ u + 3 ⎞⎟⎤ + c .
⎢ 7⎝ 2 ⎠⎥⎦
7 7 7 ⎣
2 2
Since u = Y / X,
1⎡ ⎛ 7 ⎤
ln X =
2 ⎛ 2Y + 3 X ⎞
tan−1⎜ ⎟ + c ⇒ Y ( X ) = ⎢ 7 X tan⎜⎜ (ln X − c )⎞⎟⎟ − 3 X ⎥ .
7 ⎝ 7X ⎠ 2⎣ ⎝ 2 ⎠ ⎦
Since X = x - 2 and Y = y + 3.
1⎡ ⎛ 7 ⎤
y (x ) = ⎢ 7 (x − 2 ) tan⎜⎜ (ln x − 2 − c )⎞⎟⎟ − 3(x − 2)⎥ − 3 .
2⎣ ⎝ 2 ⎠ ⎦
V Case 2. Suppose that ae – bg = 0.

ax + by
Set v = ……………………………………. (1)
a
⇒ av = ax + by ⇒ a(v − x ) = by
bg
Since ae = bg (i.e., a = )
e
bg
⇒ (v − x ) = by ⇒ g (v − x ) = ey
e
gx + ey
⇒ gv = gx + ey ⇒ v = ……………………. (2)
g
dv b dy
by Eq.(1) = 1+
dx a dx
dy a ⎛ dv ⎞
So that = ⎜ − 1⎟ . ……….…………………. (3)
dx b ⎝ dx ⎠
dy ⎛ ax + by + c ⎞
Then the original equation = f⎜ ⎟
dx ⎝ gx + ey + h ⎠
can be derived from Eqs.(1), (2), and (3).
a ⎛ dv ⎞ ⎛ av + c ⎞ dv b ⎛ av + c ⎞
⇒ ⎜ − 1⎟ = f ⎜ ⎟ ⇒ = 1+ f ⎜ ⎟
b ⎝ dx ⎠ ⎝ gv + h ⎠ dx a ⎝ gv + h ⎠
dv
⇒ = dx. It is a separable equation.
b ⎛ av + c ⎞
1+ f ⎜ ⎟
a ⎝ gv + h ⎠
V Ex.
dy 2x + y − 1
= , where ae − bg = 2 * 2 − 4 * 1 = 0.
dx 4 x + 2y − 4
2x + y
Let v = .
2
The differential equation becomes

dv 1 ⎛ 2v − 1 ⎞
= 1+ ⎜ ⎟
dx 2 ⎝ 4v − 4 ⎠
⎛ 36 4 ⎞
⎜ 8v − − ⎟
⎛ 8v − 8 ⎞ 5 5 ⎟dv = dx
⇒ ⎜ ⎟dv = dx ⇒ ⎜
⎝ 10v − 9 ⎠ ⎜⎜ 10v − 9 ⎟⎟
⎝ ⎠
⎛ 4 4 ⎛ 1 ⎞⎞ 4v 2
⇒ ⎜ − ⎜ ⎟ ⎟dv = dx ⇒ − ln 10v − 9 + c = x
⎝ 5 5 ⎝ 10v − 9 ⎠ ⎠ 5 25
2
⇒ (2 x + y ) − 2 ln10 x + 5 y − 9 + c = x
5 25
x 2y 2
⇒ − + − ln 10 x + 5 y − 9 + c = 0 .
5 5 25
1.4 Exact differential equations

V Now we want to consider a DE as
dy - M ( x, y )
= .
dx N ( x, y )
That is, M(x, y)dx + N(x, y)dy = 0.
V The solving principle can be

method 1: transform this equation to be separable or R-1;
method 2: to find a function u(x, y) such that the total differential du is
equal to Mdx + Ndy.
V In the latter strategy, if u exists, then equation Mdx + Ndy = 0 is called

exact, and u(x, y) is called a potential function for this differential
equation.
We know that “du = 0 ⇒ u(x, y) = c” ;
it is just the general solution of the differential equation.
V How to find such an u ?

To find u, u is regarded
∂u ∂u
since du = dx + dy = Mdx + Ndy, as a function of two
∂x ∂y independent variables
∂u ∂u x and y.
= M and = N.
∂x ∂y
step1. to integrate M w.r.t. x or integrate N w.r.t. y to obtain u. Assume

u is obtained by integrating M, then
u(x, y) = ∫Mdx + k(y).
∂u
step2. partial differentiate u w.r.t. y (i.e., ), and to compare with
∂y
N to find k function.
V How to test Mdx + Ndy = 0 is exact or not ?
Proposition (Test for exactness)
∂N ∂M
If M, N, , and are continuous over a rectangular region R,
∂x ∂y
∂M ∂N
then “Mdx + Ndy = 0 is exact for (x, y) in R if and only if = in R ”.
∂y ∂x
V Ex.1. Solve (x3 + 3xy 2)dx + (3x2y + y 3)dy = 0.

1st step: (testing for exactness)
M = x3 + 3xy 2, N = 3x2y + y 3
∂M ∂N
= 6 xy =
∂y ∂x
It implies that the equation is exact.
2nd step:
1 4 3
u = ∫Mdx + k(y) = ∫(x3 +3xy 2)dx + k(y) = x + x 2y 2 + k(y)
4 2
3rd step:
∂u
Since = N ⇒ 3x 2y + k’(y) = 3x 2y + y 3,
∂y
1
k ’(y) = y 3. That is k(y) = y 4 + c*.
4
1 4
Thus u(x, y) = (x + 6x 2y 2 + y 4) + c*.
4
1
The solution is then (x 4 + 6x 2y 2 + y 4) = c.
4
This is an implicit solution to the original DE.
4th step: (checking solution for Mdx + Ndy = 0)

d 1 4 dc
[ ( x + 6 x 2 y 2 + y 4 )] =
dx 4 dx
1
⇒ (4x 3 + 12xy 2 + 12x 2yy’ + 4y 3y’) = 0
4
⇒ (x 3 + 3xy 2) + (3x 2y + y 3)y’ = 0
⇒ (x 3 +3xy 2)dx + (3x 2y + y 3)dy = 0. QED
V Ex.2. Solve (sinx cosh y)dx – (cosx sinh y)dy = 0, y(0) = 3.

e x + e- x e x - e- x
Answer. coshx = , sinhx =
2 2
M = sinx cosh y, N = - cosx sinh y
∂M ∂N
= sin x sinh y = . The DE is exact.
∂y ∂x
If u =∫sinx cosh y dx + k(y) = - cosx cosh y + k(y)
∂u
= − cos x sinh y ⇒ k = constant ⇒ Solution is cosx cosh y = c.
∂y
Since y(0) = 3, cos0 cosh 3 = c ⇒ cos x cosh y = cosh 3.
V Ex.3. (non-exact case)

ydx – xdy = 0
M = y, N = -x
∂M ∂N
step 1: =1 ≠ = -1.
∂y ∂x
If you solve the equation by the same method.
step 2: u = ∫Mdx + k(y) = xy + k(y)
∂u
step 3: = x + k’(y) = N = -x
∂y
⇒ k’(y) = -2x.
Since k(y) depends only on y; we can not find the solution.
Try u = ∫Ndy + k(x) also gets the same contradiction.
Truly, the DE is separable.

Integrating factors
dy - M ( x , y )
V If a DE = (or M(x, y)dx + N(x, y)dy = 0) is not exact,
dx N(x , y )
then we can sometimes find a nonzero function F(x, y) such that
F(x, y)M(x, y)dx + F(x, y)N(x, y)dy = 0 is exact.
We call F(x, y) an integrating factor for Mdx + Ndy = 0.
V Note
1. Integrating factor is not unique.
2. The integrating factor is independent of the solution.
V Ex. Solve ydx – xdy = 0 (non-exact)

1
Assume there is an integrating factor F = 2 , then the original DE
becomes exact, x
y -1
∂( 2 ) ∂ ( )
x = 1 = x
∂y x2 ∂x
1 1 1
There are several differential factors: 2 , xy , 2 , ....
y x + y2
(conclusion: Integrating factor is not unique)
V Ex.5. Solve 2 sin(y 2) dx + xy cos(y 2) dy = 0, Integrating factor F (x, y) = x 3.

FM = 2x 3 sin(y 2)
FN = x 4y cos(y 2)
∂(FM ) ∂(FN )
= 4 x 3 y cos( y 2 ) =
∂y ∂x
Then we can solve the equation by the method of exact equation.
V How to find integrating factors ?

there are no better method than inspection or “try and error”.
V How to “try and error” ?

Since (FM) dx + (FN) dy = 0 is exact,
∂(FM ) ∂(FN ) ∂M ∂F ∂N ∂F
= ; that is, F +M =F +N .......... .... (1)
∂y ∂x ∂y ∂y ∂x ∂x
Let us consider three cases:
Case 1. Suppose F = F(x) or F = F(y)

∂F
i. If F = F(x), then = 0.
∂y
It implies that Eq.(1) becomes
∂M ∂N dF
F =F +N
∂y ∂x dx
∂M ∂N dF
⇒ F( − )=N
∂y ∂x dx
1 ∂M ∂N 1
⇒ ( − )dx = dF .
N ∂y ∂x F
1 ∂M ∂N
If ( − ) is a function of x only,
N ∂y ∂x
then the DE becomes separable.
1 ∂M ∂N
∫ ( − )dx + c * = ln F
N ∂y ∂x
⎡ 1 ∂M ∂N ⎤
⇒ F = exp ⎢ ∫ ( − )dx ⎥ .
⎣ N ∂y ∂x ⎦
∂F
ii. If F = F(y), then = 0.
∂x
It implies that PDE (1) becomes
∂M ∂F ∂N
F +M =F
∂y ∂y ∂x
∂F ∂N ∂M
⇒M = F( − )
∂y ∂x ∂y
1 1 ∂N ∂M
⇒ dF = ( − )dy .
F M ∂x ∂y
1 ∂N ∂M
If ( − ) is a function of y only, then the DE
M ∂x ∂y
⎡ 1 ∂N ∂M ⎤
becomes separable and F ( y ) = exp ⎢ ∫ ( − )dy ⎥ .
⎣ M ∂x ∂y ⎦
Case 2. Suppose F(x, y) = x ay b and attempt to solve coefficients a and b

by substituting F into Eq.(1).
Case 3. If cases 1 and 2 both fail, you may try other possibilities,
such as eax + by, xaeby, eaxyb, and so on.
V Ex. (Example for case 1)

Solve the initial value problem
2xydx + (4y + 3x 2)dy = 0, y(0.2) = -1.5
M = 2xy, N = 4y +3x 2
∂M ∂N
= 2x ≠ = 6 x (non-exact)
∂y ∂x
1 ∂M ∂N
Testing whether ( − ) depends only on x or not.
N ∂y ∂x
1 ⎛ ∂M ∂N ⎞ 1
⎜ − ⎟= ( 2 x − 6 x ) depends on both x and y.
N ⎝ ∂y ∂x ⎠ 4 y + 3 x 2
1 ∂N ∂M
testing whether ( − ) depends only on y or not.
M ∂x ∂y
1 ⎛ ∂N ∂M ⎞ 1 2
⎜ − ⎟= (6 x − 2 x ) = depends only on y.
M ⎝ ∂x ∂y ⎠ 2 xy y
2 2
Thus F(y) = exp ∫ dy = y .
y
The original DE becomes
2xy 3dx + (4y 3 + 3x 2y 2)dy = 0 (exact)
u =∫2xy 3dx + k(y) = x 2y 3 + k(y)
∂u
= 3x 2y 2 + k’(y) = 4y 3 + 3x 2y 2
∂y
⇒ k’(y) = 4y 3
⇒ k(y) = y 4 + c*
⇒ u = x 2y 3 + y 4 + c* = c’
⇒ x 2y 3 + y 4 = c .
Since y(0.2) = -1.5 ⇒ c = 4.9275.
Solution x 2y 3 + y 4 = 4.9275.
V Ex. (Example for case 2)

(2y 2 – 9xy)dx + (3xy – 6x 2)dy = 0 (non-exact)
1 ∂M ∂N 1 y + 3x
( − )= ( 4 y − 9 x − 3 y + 12 x ) = depends on x & y .
N ∂y ∂x 3 xy − 6 x 2 3 x( y − 2 x )
1 ∂N ∂M 1 − 3x − y
( − )= ( 3 y − 12 x − 4 y + 9 x ) = depends on x & y .
M ∂x ∂y 2y 2 − 9 xy y (2y − 9 x )
Take F = x a y b , then
∂(FM ) ∂(FN )
=
∂y ∂x
∂(2y 2 x a y b − 9 xyx a y b ) ∂(3 xyx a y b − 6 x 2 x a y b )
⇒ =
∂y ∂x
⇒ 2(2+b)y b+1x a – 9(b+1)x a+1y b = 3(a+1)x ay b+1 – 6(a+2)x a+1y b
⇒ 2(2+b) = 3(a+1)
9(b+1) = 6(a+2)
⇒ 3a – 2b – 1 = 0
{
6a – 9b + 3 = 0
⇒ a=1
{
b=1
⇒ F(x, y) = xy.
1.5 Linear differential equation and Bernoulli equation

V A first-order DE is said to be linear if it can be written
y’ + p(x)y = r (x).
If r (x) = 0, the linear DE is said to be homogeneous, if r (x) ≠ 0, the
linear DE is said to be nonhomogeneous.
V Solving the DE
(a) For homogeneous equation ( ⇒ separable)
y’ + p(x)y = 0
dy
⇒ = -p(x)y
dx
1
⇒ dy = - p(x)dx
y
⇒ ln|y| = -∫p(x)dx + c*
⇒ y = ce -∫p(x)dx.
(b) For nonhomogeneous equation

(py – r )dx + dy = 0
since 1 ( ∂M − ∂N ) = p is a function of x only,

N ∂y ∂x
we can take an integrating factor
⎡ 1 ∂M ∂N ⎤
F(x) = exp ⎢ ∫ ( − )dx ⎥ = e ∫ pdx
⎣ N ∂y ∂x ⎦
such that the original DE y’ + py = r becomes
e∫pdx(y’ + py) = (e∫pdxy)’ = e∫pdxr
Integrating with respect to x,

e∫pdxy = ∫e∫pdxr dx + c
⇒ y(x) = e -∫pdx [∫e∫pdxrdx + c] .
V The solution of the homogeneous linear DE is a special case of the
solution of the corresponding non-homogeneous linear DE.
V Ex.1. Solve the linear DE

y’ – y = e 2x
Solution.
p = -1, r = e 2x, ∫pdx = -x
y(x) = e x [∫e –x e 2x dx + c]
= e x [∫e xdx + c]
= e 2x + ce x.
V Ex.3. Solve the linear DE

y’ + 2y = e x (3 sin 2x + 2 cos 2x)
Solution.
p = 2, r = e x (3 sin 2x + 2 cos 2x), ∫pdx = 2x
y(x) = e -2x [∫e 2x e x (3 sin 2x + 2 cos 2x) dx + c]
= e -2x [e 3x sin 2x + c]
= c e -2x + e x sin 2x .
Bernoulli equation
V The Bernoulli equation is formed of
y’ + p(x) y = r(x) y a , where a is a real number.
If a = 0 or a = 1, the equation is linear.
V Bernoulli equation can be reduced to a linear form by change of variables.

We set u(x) = [ y(x) ]1-a,
then differentiate the equation and substitute y’ from Bernoulli equation
u’ = (1 - a) y –a y ’ = (1 - a) y -a (r y a - py)
= (1 - a) (r - py1-a)
= (1 - a) (r - pu)
⇒ u’ + (1 - a) pu = (1 - a) r (This is a linear DE of u.)
V Ex.5.
y’ - Ay = - By 2
a = 2, u = y -1
u’ = -y -2 y ’ = -y -2 (-By 2 + Ay) = B – Ay -1 = B – Au
⇒ u’ + Au = B
u = e -∫pdx [ ∫e∫ pdx r dx + c ]
= e -Ax [ ∫Be Ax dx + c ]
= e -Ax [ B/A e Ax + c ]
= B/A + c e -Ax
⇒ y = 1/u = 1/(B/A + ce -Ax)

Riccati equation (problem 44 on page 40)

y’ = p(x) y 2 + q (x) y + r (x) is a Riccati equation.
V Solving strategy
If we can some how (often by observation, guessing, or trial and error)
produce one specific solution y = s(x), then we can obtain a general
solution as follows:
Change variables from y to z by setting
y = s(x) + 1/z
⇒ y’= s’(x) - (1/z2) z’
Substitution into the Riccati equation given us
s’(x) - (1/z2) z’ = [ p(x) s(x)2 + q (x) s(x) + r (x) ] +
[ p(x) (1/z2) + 2p(x)s(x) (1/z) + q (x) (1/z) ]
Since s(x) is a solution of original equation.
⇒ - (1/z2) z’ = p (1/z2) + 2 p s (1/z) + q (1/z)
multiplying through by -z2

⇒ z’ + (2 p s + q) z = -p ,
which is a linear DE for z and can be found the solution.
⇒ z = c/u(x) + [1/u(x) ∫ -p(x)u(x)dx] ,
where u(x) = e ∫[2ps + q] dx
⇒ z = e -∫[2ps + q] dx [ ∫-e ∫(2ps + q) dx p dx + c ]
Then, y = s(x) + 1/z is a general solution of the Riccate equation.
V There are two difficulties for solving Riccati equations:

(1) one must first find a specific solution y = s(x).
(2) one must be able to perform the necessary integrations.
V Ex. y’ = (1/x) y 2 + (1/x) y - 2/x , s(x) = 1.

Solution. y(x) = (2x 3 + c)/(c – x 3).
Summary for 1st order DE

1. Separable f(x) dx = g(y) dy [separated integration]
2. R-1 formula dy/dx = f(y/x) [change variable u = y/x]
3. R-2 dy/dx = f((ax + by + c)/(gx + ey +h)), c ≠ 0 or h ≠ 0.

with two cases i. ae - bg ≠ 0 [x = X+ α, y =Y+ β ⇒ R-1 ⇒ separable]
ii. ae -bg = 0 [v = (ax+by)/a = (gx+ey)/g ⇒ separable]
4. Exact dy/dx = -M(x, y)/N(x, y) [∂M/∂y = ∂N/∂x ⇒ exact]

(M dx + N dy = 0) [deriving u ⇒ du = Mdx + Ndy]
5. Integrating factor dy/dx = -M/N [find F ⇒ (FM)dx + (FN)dy = 0 is exact]

try some factors i. (∂M/∂y - ∂N/∂x)/N = F(x) or (∂N/∂x - ∂M/∂y)/M = F(y)
ii. F = xayb
iii. F = eax+by, xaeby, eaxyb, …
6. Linear 1st-order DE y’ + p(x)y = r(x) integratin g factor F ( x ) = e ∫ pdx

y = e -∫p(x) dx [ ∫ r(x) e ∫p(x) dx dx + c ]
7. Bernoulli equation y’ + p(x) y = r(x) y a

set u(x) = [y(x)] 1-a , ⇒ u’ + (1-a) pu = (1-a) r (linear DE)
8. Riccati equation y’ = p(x)y2 + q(x)y + r(x)

(1) guess a specific solution s(x)
(2) change variable y = s(x) + 1/z
(3) to derive a linear DE z’ + (2ps + q) z = -p .
1.6 Orthogonal trajectories of curves

V Purpose
use differential equation to find curves that intersect given curves at right
angles. The new curves are then called the orthogonal trajectories of the
given curves.
V Example
y
Any blue line is orthogonal

to any pink circle.
V Principle
to represent the original curves by the general solution of a DE y’ = f(x, y),
then replace the slope y’ by its negative reciprocal, 1/y’ , and solve the
new DE -1/y’ = f(x, y).
V Family of curves
If for each fixed value of c the equation F(x, y, c) = 0 represents a curve in
the xy-plane and if for variable c it represents infinitely many curves, then
the totality of these curves is called a one-parameter family of curves, and
c is called the parameter of the family.
V Determination of orthogonal trajectories

step 1. Given a family of curves F(x, y, c) = 0,
to find their DE in the form y’ = f(x, y),
step 2. Find the orthogonal trajectories by solving their DE y’ = -1/f(x, y).
V Ex.
(1) The equation F(x, y, c) = x + y + c = 0 represents a family of
parallel straight lines.
(2) The equation F(x, y, c) = x2 + y2 - c2 = 0 represents a family of

concentric circles of radius c with center at the original.
y y
x x
c=1
c = -2 c=2
c=3
c=0
c=3
V Ex.
(1) differentiating x + y + c = 0, gives the DE y’ = -1.
(2) differentiating x2 + y2 - c2 = 0 ,
gives the DE 2x + 2yy’ = 0 ⇒ y’ = -x/y.
(3) differentiating the family of parabolas y = cx2 ,

gives the DE y’ = 2cx.
since c = y/x2, y’ = 2y/x.
Another method
c = yx -2
⇒ 0 = y’x -2 - 2yx -3
⇒ y’x = 2y
⇒ y’ = 2y/x .
V Ex.1.
Find the orthogonal trajectories of the parabolas y = cx2.
step 1. y’ = 2y/x
step 2. solve 2
x + 2=c
y’ = -x/2y y = cx2 y
2
⇒ 2y dy = -x dx
⇒ y 2 = -x2/2 + c*
2
⇒ x + y 2 = c.
2
V Ex.
Find the orthogonal trajectories of the circles x2 + (y - c)2 = c2.
step 1. Differentiating x2 + (y - c)2 to give 2x + 2 (y – c) y’ = 0
⇒ y’ = x/(c-y) (error)
Correct derivation x2 + (y - c)2 = c2
⇒ x2 + y 2 - 2cy = 0
⇒ x2 y -1 + y = 2c
⇒ 2xy -1 - x2 y –2 y’ + y’ = 0
⇒ 2xy -1 = (x2 y -2 - 1) y’
⇒ 2xy = (x2 - y 2) y’
⇒ y’ = 2xy/(x2 - y 2)
step 2. Solve y’ = (y 2 - x2)/2xy
⇒ y’ = y/2x - x/2y (R-1 formula)
Solution. (x - e)2 + y 2 = e2,
where e is a constant.
1.7 Existence and uniqueness of solutions

V Consider an initial value problem
y’ = f(x, y) , y(x0) = y0
There are three possibilities of solution,

(i) no solution; e.g., |y’| + |y| = 0 , y(0) = 1.
0 is the only solution of the differential equation, the condition
contradicts to the equation;
moreover, y and y ’ are not continuous at x = 0.
1 2
(ii) unique solution; e.g., y’ = x , y(0) = 1, solution y = x +1
2
(iii) infinitely many solution; e.g., xy’ = y – 1, y(0) = 1,
solution y = 1 + cx .
V Problem of existence
Under what conditions does an initial value problem have at least one
solution?
V Problem of uniqueness
Under what conditions does that problem have at most one solution?
V Theorem 1 (Existence theorem)

If f(x, y) is continuous at all points (x, y) in some rectangle
R : |x – x0| < a , |y - y0| < b
and bounded in R: |f(x, y)| ≤ k for all (x, y) in R, then the initial value
problem “y’ = f(x, y), y(x0) = y0 “ has at least one solution y(x).
y
b R
yo
b
a a x
xo
V Theorem 2 (Uniqueness theorem)

If f(x, y) and ∂f/∂y are continuous for all (x, y) in that rectangle R and
bounded,
(a) | f | ≤ k , (b) ∂f ≤ M for all (x, y) in R, then the
∂y
initial value problem has at most one solution y(x).
Hence, by Theorem 1, it has precisely one solution.
V The conditions in the two theorems are sufficient conditions rather than
necessary ones and can be lessened.
∂f
For example, condition ≤ M may be replaced by the weaker
∂y
condition f ( x, y 2) − f ( x, y 1) ≤ M y 2 − y 1 , where y1 and y2 are on the
boundary of the rectangle R. The later formula is known as a Lipschitz
condition. However, continuity of f(x, y) is not enough to guarantee the
uniqueness of the solution.
V Ex. 2. (Nonuniqueness)
The initial value problem
y' = y, y ( 0 ) = 0.
has the two solutions
⎧ x2 / 4 if x≥0
y = 0 and y = ⎨ 2
⎩− x / 4 if x<0
Although f ( x, y ) = y ' = y is continuous for all y. The Lipschitz

condition is violated in any region that include the line y = 0, because for
y1 = 0 and positive y2, we have
f ( x, y 2) − f ( x, y 1) y2 1
= =
y 2 − y1 y2 y2
and this can be made as large as we please by choose y2 sufficiently

small, whereas the Lipschitz condition requires that the quotient on the
left side of the above equation should not exceed a fixed constant M.
Picards’ iteration method

V Picards method gives approximate solutions of an initial value problem
y’ = f(x, y), y(x0) = y0
(i) The initial value problem can be written in the form
y ( x ) = y 0 + ∫x f [t , y (t )] dt .......... .......... .......... .......... .......... ... (1)
x
0
(ii) Take an approximation unknown

y1( x ) = y 0 + ∫x f [t , y 0 (t )] dt
x
0
(iii) Substitute the function y1(x) in the same way to get

y 2 ( x ) = y 0 + ∫x f [t , y1(t )] dt
x
0
(iv) y n ( x ) = y 0 + ∫x0 f [t , y n −1(t )] dt

x
Under some conditions, the sequence will converge to the solution y(x)
of the original initial value problem.
V Ex.3. Find approximate solutions to the initial value problem

y’ = 1 + y 2 , y(0) = 0 .
Solution.
x0 = 0 , y0 = 0 , f(x, y) = 1 + y 2
y1( x ) = 0 + ∫0 [1 + 0] dt = x
x
y 2 ( x ) = y 0 + ∫0 f [t , y1(t )] dt
x
= 0 + ∫0 [1 + t 2 ] dt = x +
x 1 3
x
3
y 3 ( x ) = y 0 + ∫0 f [t , y 2 (t )] dt
x
x⎡ 1 ⎤
= 0 + ∫0 ⎢1 + (t + t 3 )2 ⎥ dt
⎣ 3 ⎦
1 2 5 1 7
= x + x3 + x + x
3 15 63
:
Exact solution y(x) = tan(x).

Laparoscopic surgical simulation

EM701p First-Order Ordinary Differential Equations

Transféré par

Informations du document

Description originale:

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

EM701p First-Order Ordinary Differential Equations

Transféré par

Droits d'auteur :

Formats disponibles

Advanced Engineering Mathematics 1.

1. First-order Ordinary Differential Equations

Advanced Engineering Mathematics 1. First-order ODEs 2

1.1 Basic concepts and ideas

∂ z2 ∂z ∂z ⇐ (partial differential equation)

V There are several kinds of differential equations

where y is called dependent variable and

Advanced Engineering Mathematics 1. First-order ODEs 4

V What is the purpose of differential equations ?

Resister (R) Electro-motive

(2) Falling stone y

Advanced Engineering Mathematics 1. First-order ODEs 6

V Any physical situation involved motion or measure rates of change can be

Physical modeling solving

That is, the purposes of this course are that

V A first-order ODE is an equation involving one dependent variable, one

V A solution of a first-order ODE is a function which satisfies the equation.

V A solution which appears as an implicit function, given in the form

Advanced Engineering Mathematics 1. First-order ODEs 8

V A general solution is a solution containing one arbitrary constant;

V A differential equation together with an initial condition is called an initial

V Problem of Section 1.1.

1.2 Direction fields

V Method of direction fields

Advanced Engineering Mathematics 1. First-order ODEs 10

V Ex.1. Graph the direction field of the 1st-order DE y’ = xy.

(a) By computer. (b) By hand.

V Problems of Section 1.2.

Advanced Engineering Mathematics 1. First-order ODEs 12

1.3 Separable differential equations

To solve the equation by integrate both sides with x,

V Ex.1. Solve 9 yy '+4 x = 0.

V Ex.2. Solve y ' = 1 + y 2

V Ex. Initial value problem

Advanced Engineering Mathematics 1. First-order ODEs 14

V Ex.3. Solve y’ = ky , y(0) = 2.

V Ex.5. Solve y’ = -2xy , y(0) = 1.

V Example of no separable DE (x-1)y’ = 3x2 + y.

V Note: There is no nice test to determine easily whether or not a 1st-order

Reduction to separable forms

step 5. replace u by y/x.

Advanced Engineering Mathematics 1. First-order ODEs 16

V Ex.6. Solve 2xyy’ – y 2 + x2 = 0.

V Ex. Solve initial value problem

Advanced Engineering Mathematics 1. First-order ODEs 18

V Ex.7. Solve (2x - 4y + 5) y’ + x - 2y + 3 = 0.

V There are two ways to solve the equation:

Advanced Engineering Mathematics 1. First-order ODEs 20

V Case 1. Suppose that ae – bg ≠ 0.

Advanced Engineering Mathematics 1. First-order ODEs 22

V Case 2. Suppose that ae – bg = 0.

Advanced Engineering Mathematics 1. First-order ODEs 24

The differential equation becomes

V Problems of Section 1.3.

Advanced Engineering Mathematics 1. First-order ODEs 26

1.4 Exact differential equations

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

V The solving principle can be

V In the latter strategy, if u exists, then equation Mdx + Ndy = 0 is called

V How to find such an u ?

step1. to integrate M w.r.t. x or integrate N w.r.t. y to obtain u. Assume

Advanced Engineering Mathematics 1. First-order ODEs 28