SESSION 10: ORDINARY DIFFERENTIAL EQUATION

Second-order linear differential equations

• Theorem 1: The set of all solutions of an nth-order ordinary linear homogeneous
differential equation forms an n-dimensional vector space. It means that
𝑦 = 𝑐1 𝑦1 + 𝑐2 𝑦2 (1)
is a solution of
𝑑2𝑦 𝑑𝑦
+ 𝑝 (𝑥 ) + 𝑞 (𝑥 )𝑦 = 0 (2)
𝑑𝑥 2 𝑑𝑥
• If 𝑦1 and 𝑦2 are two linearly independent solutions of the above differential equation and
𝑐1 and 𝑐2 are the arbitrary constants.
• Theorem 2: A necessary and sufficient condition that solutions 𝑦1 and 𝑦2 of a second-
order linear differential equation be linearly independent is that a special determinant of
these solutions be different from zero. That is to say,
𝑦1 𝑦2
|𝑦 ′ 𝑦 ′ | ≠ 0 (3)
1 2

• The above determinant is called the Wronskian of 𝑦1 and 𝑦2 denoted by 𝑊 (𝑦1 , 𝑦2 ), it

will play an important role in connection with Green’s function.

• Homogeneous differential equations with constant coefficients

▪ The standard form for the general second-order homogeneous differential
equation with constant coefficients 𝑝0 and 𝑞0 is
𝑦 " + 𝑝0 𝑦 ′ + 𝑞0 𝑦 = 0 (4)
𝑑
▪ Using the linear operator 𝐷 defined by 𝐷 ≡ , we may write Eqn. (4) as,
𝑑𝑥

(𝐷 2 + 𝑝0 𝐷 + 𝑞0 )𝑦 = 0 (5)
▪ 𝐷 2 + 𝑝0 𝐷 + 𝑞0 is called the auxiliary or characteristic equation. The roots of the
characteristics may be (a) real and unequal (b) real and equal (c)a complex-
conjugate pair.
 ▪ Case 1: The roots of the characteristic equation are real and unequal. Assume the
roots of the characteristic equation are 𝑎 and 𝑏. We may therefore write the
original differential equation Eqn. (4) in the following factored form:
(𝐷 − 𝑎 ) 𝑢 = 0 (6)
where 𝑢 = (𝐷 − 𝑏)𝑦.
▪ Eqn. (6) may be solved by letting 𝑢 = 𝑐𝑒 𝑎𝑥 and 𝑦 = 𝑐𝑒 𝑏𝑥 ∫ 𝑒 (𝑎−𝑏)𝑥 𝑑𝑥 + 𝑐2𝑒 𝑏𝑥 .
In this case, the solution of Eqn. (6) is,
𝑦 = 𝑐1 𝑦1 + 𝑐2 𝑦2 = 𝑐1 𝑒 𝑎𝑥 + 𝑐2 𝑒 𝑏𝑥 (7)
Example 1
Solve
𝑦 " + 𝑦 ′ − 2𝑦 = 0
Solution
The corresponding characteristic equation is
𝐷2 + 𝐷 − 2 = 0
(𝐷 − 1)(𝐷 + 2) = 0
with roots equal to 1 and −2. By the use of case 1, the general solution of the differential
equation to be solved is
𝑦 = 𝑐1𝑒 𝑥 + 𝑐2 𝑒 −2𝑥

▪ Case 2: The roots of the characteristic equation are real and equal. Since the roots
are real and equal 𝑎 = 𝑏, we may write the original differential equation in the
following factored form:
(𝐷 − 𝑎 ) 𝑢 = 0
where 𝑢 = (𝐷 − 𝑎)𝑦.
▪ On solving the differential equation, we obtain
𝑦(𝑥 ) = 𝑐1 𝑦1 + 𝑐2 𝑦2 = 𝑐1 𝑥𝑒 𝑎𝑥 + 𝑐2 𝑒 𝑏𝑥 (8)

Josef Hoëné de Wronski (1778-1853), Polish mathematician known for work that involved applying philosophy to mathematics. He
developed a series expansion whose coefficients were determinants, now known as Wronskians.
Example 2
Solve
𝑦 " − 2𝑦 ′ + 𝑦 = 0
Solution
The characteristic equation may be factored as follows
(𝐷 − 1)(𝐷 − 1) = 0
with solution 𝑎 = 𝑏 = 1. The general solution is
𝑦(𝑥 ) = (𝑐1 𝑥 + 𝑐2)𝑒 𝑥
Example 3
Solve
𝑦 " + 9𝑦 ′ = 0
Solution
The characteristic equation is
(𝐷 2 + 9) = 0
with roots ±3𝑖 . The general solution is
𝑦(𝑥 ) = 𝑐1 𝑒 3𝑖𝑥 + 𝑐2𝑒 −3𝑖𝑥

Example 4. The classical harmonic oscillator.

Consider the motion of a mass 𝑚 attached to the end of a spring with spring constant 𝑘. See
figure below. For applied force with magnitude 𝐹 and restoring force with magnitude −𝑘𝑥, the
equation of motion of this mass when released is Hooke’s law, 𝐹 = −𝑘𝑥 or 𝑚𝑥̈ + 𝑘𝑥 = 0. (a)
Find 𝑥 (𝑡) subject to the initial conditions 𝑥(0) = 𝑋0 and 𝑥̇ (0) = 0. Condition 𝑥(0) = 𝑋0 means
that initially the mass is a distance 𝑋0 from the equilibrium position; this maximum distance
from the equilibrium is called the amplitude. The condition 𝑥̇ (0) = 0 means the mass is initially
at rest. (b) Find the period of the motion.
Solution
(a) The standard form of the equation of motion is
𝑥̈ + 𝜔2 𝑥 = 0
𝑘
where 𝜔2 = . The corresponding characteristic equation is
𝑚

𝐷 2 + 𝜔2 = 0
with solutions ±𝑖𝜔. The general solution of the equation of motion is therefore
𝑥 (𝑡) = 𝑐1𝑒 𝑖𝜔𝑡 + 𝑐2 𝑒 −𝑖𝜔𝑡
𝑥(𝑡) = 𝑐1 (cos 𝜔𝑡 + 𝑖 sin 𝜔𝑡) + 𝑐2 (cos 𝜔𝑡 − 𝑖 sin 𝜔𝑡)
𝑥 (𝑡) = 𝐴 cos 𝜔𝑡 + 𝐵 sin 𝜔𝑡
Applying the initial conditions 𝑥 (0) = 𝑋0 = 𝐴 and 𝑥̇ (0) = 0 = 𝐵𝜔 or 𝐵 = 0 since 𝜔 ≠ 0, we
obtain the following solution for the original equation of motion
𝑥(𝑡) = 𝑋0 cos 𝜔𝑡
(b) The period of motion is the time required for the mass to repeat itself, it is determined as
follows:
𝑥 (𝑡 ) = 𝑥 (𝑡 + 𝑇 )
where 𝑇 is the period.
𝑋0 cos 𝜔𝑡 = 𝑋0 cos 𝜔(𝑡 + 𝑇)

𝑋0 cos 𝜔(𝑡 + 𝑇) = 𝑋0 (cos 𝜔𝑡 cos 𝜔𝑇 − sin 𝜔𝑡 sin 𝜔𝑇)

𝜔𝑇 = 2𝜋𝑛
where 𝑛 = 0, 1, 2, 3 … The period is given by
2𝜋𝑛 𝑚
𝑇= = 2𝜋𝑛√
𝜔 𝑘
𝑛 is the mode of oscillation.
Example 5.
Consider the motion of a particle of mass 𝑚 initially at rest and subject to a restoring force of
−𝑘𝑥 and a damping force of −𝑎𝑥̇ . The equation of motion of this particle is
𝑚𝑥̈ = −𝑘𝑥 − 𝑎𝑥̇
where 𝑥(0) = 𝑋0 and 𝑥̇ (0) = 0. The equation of motion in standard form is
𝑥̈ + 2𝛿𝑥̇ + 𝜔2 𝑥 = 0 ,
𝑘 𝑎
where 𝜔2 = 𝑚 and 2𝛿 = 𝑚. The factor 2 in the equation of motion is used for convenience. Find

the solution of the equation of motion for the following cases (a) 𝛿 = 0 (no damping), (b) 𝛿 = 𝜔
(critical damping), (c) 𝛿 < 𝜔 (light damping) and (d) 𝛿 > 𝜔 (heavy damping).
Solution
(a) The equation of motion for 𝛿 = 0 reduces to
𝑥̈ + 𝜔2 𝑥 = 0
the solution of the differential equation is
𝑥(𝑡) = 𝑋0 cos 𝜔𝑡
for 𝑥 (0) = 𝑋0 and 𝑥̇ (0) = 0. The motion in this case is oscillatory and periodic with constant
amplitude 𝑋0 .

(b) For 𝛿 = 𝜔 ≠ 0, the solutions of the corresponding characteristics equation are real and
equal, −𝛿. The solution of the equation of motion for critical damping is
𝑥(𝑡) = (𝑐1 𝑡 + 𝑐2 )𝑒 −𝛿𝑡 = 𝑋0 (𝛿𝑡 + 1)𝑒 −𝛿𝑡
for 𝑥 (0) = 𝑋0 and 𝑥̇ (0) = 0. The motion in this case is not oscillatory and approaches
equilibrium at a rapid rate.

(c) For light damping 𝛿 < 𝜔 the solutions of the corresponding characteristics equation are
−𝛿 ± 𝑖∆ where ∆= √𝜔 2 + 𝛿 2 . The solution of the equation of motion is
𝛿
𝑥 (𝑡) = 𝑋0 (cos ∆𝑡 + sin ∆𝑡) 𝑒 −𝛿𝑡
∆
for 𝑥 (0) = 𝑋0 and 𝑥̇ (0) = 0. In this case, the motion is oscillatory with decreasing amplitude,
𝑋0 𝑒 −𝛿𝑡 , and is not periodic.

(d) For heavy damping 𝛿 > 𝜔, the solutions of the characteristic equation are −𝛿 ± ∆′ where
∆′= √𝜔 2 − 𝛿 2 . The solution of the equation of motion foe heavy damping subject to
initial conditions 𝑥 (0) = 𝑋0 and 𝑥̇ (0) = 0 is
𝛿 + ∆′ (𝛿+∆′ )𝑡
𝛿 − ∆′ ′
( )
𝑥 𝑡 =( ′
) 𝑋0 𝑒 +( ′
) 𝑋0 𝑒 −(𝛿+∆ )𝑡 .
2∆ 2∆
The motion is not oscillatory and approaches equilibrium at a rate less rapid than for critical
damping.

Nonhomogeneous differential equations with constant coefficient

• The standard form for second-order nonhomogeneous differential equations with constant
coefficients is
 𝑦 " + 𝑝0 𝑦 ′ + 𝑞0 𝑦 = 𝑓(𝑥 ) (1)

• The two widely used methods for solving differential equations in the above category are
1. successive integration and
• undetermined coefficients, 𝑦 = 𝑦ℎ + 𝑦𝑝 where 𝑦ℎ is the solution of the corresponding
homogeneous differential equation and 𝑦𝑝 is any solution of the nonhomogeneous
differential equation.

• Method of successive integration

▪ If 𝑎 and 𝑏 are the roots of the characteristic equation corresponding to Eqn. (1),
then Eqn. (1) may be written as,
(𝐷 − 𝑎 ) 𝑢 = 𝑓 (𝑥 )
(2)

where 𝑢 = (𝐷 − 𝑏)𝑦. The differential equation may be solved using the general
formula, we obtain,

𝑢 = 𝑒 𝑎𝑥 ∫ 𝑓 (𝑥 )𝑒 −𝑎𝑥 𝑑𝑥 + 𝑐1 𝑒 𝑎𝑥 ≡ 𝑄̅ (𝑥 ) (3)

▪ On substituting 𝑢 into Eqn. (2) and solving for 𝑦, we obtain,

𝑦(𝑥 ) = 𝑒 𝑏𝑥 ∫ 𝑄̅ (𝑥 )𝑒 −𝑏𝑥 𝑑𝑥 + 𝑐2 𝑒 𝑏𝑥 (4)

Example 1
By use of the method of successive integration, solve
𝑦 " − 2𝑦 ′ + 𝑦 = 2 cos 𝑥
Solution
The roots of the characteristic equation of the corresponding homogeneous differential equation
are real and equal, 1. The equation for characteristic equation is 𝑄̅ (𝑥 ) reduces to

𝑄̅ (𝑥 ) = 𝑒 𝑎𝑥 ∫ 𝑓(𝑥 )𝑒 −𝑎𝑥 𝑑𝑥 + 𝑐1𝑒 𝑎𝑥

𝑄̅ (𝑥 ) = 2𝑒 𝑥 ∫ 𝑒 −𝑥 cos 𝑥 𝑑𝑥 + 𝑐1 𝑒 𝑥

𝑄̅ (𝑥 ) = sin 𝑥 − cos 𝑥 + 𝑐1𝑒 𝑥 .

On substituting the above expression for 𝑄̅ (𝑥 ) into Eqn. (4), the general solution of the original
differential equation becomes,

𝑦(𝑥 ) = 𝑒 𝑏𝑥 ∫ 𝑄̅ (𝑥 )𝑒 −𝑏𝑥 𝑑𝑥 + 𝑐2𝑒 𝑏𝑥

𝑦(𝑥 ) = 𝑒 𝑥 ∫(sin 𝑥 − cos 𝑥 + 𝑐1𝑒 𝑥 )𝑒 𝑥 𝑑𝑥 + 𝑐2 𝑒 𝑥

𝑦(𝑥 ) = − sin 𝑥 + (𝑐1 𝑥 + 𝑐2)𝑒 𝑥

• Method of undetermined coefficients

1. 𝑓 (𝑥 ) is a polynomial of degree 𝑛 ≥ 0.
(a) If 0 is not a root of the characteristic equation, then assume
𝑦𝑝 = 𝐴0 + 𝐴1 𝑥 + ⋯ + 𝐴𝑛 𝑥 𝑛
(b) If 0 is a single root of the characteristic equation, then assume
𝑦𝑝 = 𝑥 (𝐴0 + 𝐴1 𝑥 + ⋯ + 𝐴𝑛 𝑥 𝑛 )
(c) If 0 is a double root of the characteristic equation, then assume
𝑦𝑝 = 𝑥 2 (𝐴0 + 𝐴1 𝑥 + ⋯ + 𝐴𝑛 𝑥 𝑛 )
2. 𝑓 (𝑥 ) is of the form 𝐶𝑒 𝑘𝑥
(a) If 𝑘 is not a root of the characteristic equation, then assume 𝑦𝑝 = 𝐴𝑒 𝑘𝑥
(b) If 𝑘 is a single root of the characteristic equation, then assume 𝑦𝑝 = 𝐴𝑥𝑒 𝑘𝑥
(c) If 𝑘 is a double root of the characteristic equation, then assume 𝑦𝑝 = 𝐴𝑥 2 𝑒 𝑘𝑥

3. 𝑓 (𝑥 ) is of the form sin 𝑘𝑥, cos 𝑘𝑥, or sin 𝑘𝑥 + cos 𝑘𝑥

(a) If 𝑖𝑘 is not a root of the characteristic equation, then assume
𝑦𝑝 = 𝐴 cos 𝑘𝑥 + 𝐵 sin 𝑘𝑥
(b) If 𝑖𝑘 is a single root of the characteristic equation, then assume
𝑦𝑝 = 𝐴𝑥 cos 𝑘𝑥 + 𝐵𝑥 sin 𝑘𝑥
Example 2
The equation of motion for a mass attached to the end of a vertical spring fixed at the other end is
𝑦̈ + 𝜔2 𝑦 = −𝑔
where 𝑔 is the acceleration due to gravity. Find 𝑦(𝑡) subject to 𝑦(0) = 𝑌0 and 𝑦̇ (0) = 0.
Solution
The general solution of the homogeneous equation 𝑦̈ + 𝜔2 𝑦 = 0 is 𝑦ℎ = 𝐴 cos 𝜔𝑡 + 𝐵 sin 𝜔𝑡
since the roots of the characteristic equation are ±𝑖𝜔. The nonhomogeneous term is a constant
(polynomial form), and we assume that 𝑦𝑝 = 𝐴0 . On substituting 𝑦𝑝 into the original differential
𝑔 𝑔
equation, we obtain 𝐴0 = − 𝜔2 or 𝑦𝑝 = − 𝜔2. The general solution (𝑦 = 𝑦ℎ + 𝑦𝑝 ) of the original

differential equation is
𝑔
𝑦(𝑡) = 𝐴 cos 𝜔𝑡 + 𝐵 sin 𝜔𝑡 −
𝜔2
𝑔
Initial condition 𝑦(0) = 𝑌0 leads to 𝐴 = 𝑌0 − 𝜔2 and initial condition 𝑦̇ (0) = 0

yields 𝐵 = 0. The particular solution, therefore, becomes

𝑔
𝑦(𝑡) = (𝑌0 + 2 ) cos 𝜔𝑡
𝜔
Example 3
By the use of method of undetermined coefficient, solve
𝑦 " − 2𝑦 ′ + 𝑦 = 2 cos 𝑥
Solution
Since the roots of the characteristic equation are both 1 (double root, real and equal) the solution
of the corresponding homogeneous differential equation is
𝑦ℎ = (𝑐1 𝑥 + 𝑐2)𝑒 𝑥 .
The roots of the characteristic equation are not equal 𝑖𝑘 = 𝑖 (𝑘 = 1) and the assumed form for 𝑦𝑝
is
𝑦𝑝 = 𝐴 cos 𝑘𝑥 + 𝐵 sin 𝑘𝑥 .
On substituting the above equation for 𝑦𝑝 into the original differential equation, we obtain
−(𝐴 cos 𝑘𝑥 + 𝐵 sin 𝑘𝑥 ) + 2(𝐴 sin 𝑘𝑥 − 𝐵 cos 𝑘𝑥 ) + 𝐴 cos 𝑘𝑥 + 𝐵 sin 𝑘𝑥 = 2 cos 𝑥 .
From the above equation, it is clear that 𝐴 = 0 and 𝐵 = −1; hence, 𝑦𝑝 = − sin 𝑥. The general
solution of the original equation is therefore,
𝑦(𝑥 ) = (𝑐1 𝑥 + 𝑐2 )𝑒 𝑥 − sin 𝑥.
Homogeneous differential equations with variable coefficients
• The standard form for homogeneous differential equations with variable coefficient is
𝑦 " + 𝑝 (𝑥 )𝑦 ′ + 𝑞 (𝑥 )𝑦 = 𝑓 (𝑥 ) (1)

• The usual procedure for solving differential equation of the form givenin Eqn. (1) is the
power series method due to Frobenius and Fuchs.
• Frobenius yields the following two types of information concerning the solution of Eqn.
(1):
(a) form of the solution as a result of the nature of 𝑝(𝑥 ) and 𝑞(𝑥 )
(b) form of the solution as indicated by the nature of the solution of the indicial equation.
• We will only use the second type of information.
1. If 𝑝(𝑥 ) and 𝑞(𝑥 ) are regular at 𝑥 = 0, then Eqn. (1) possesses two distinct solutions
of the form
∞

𝑦(𝑥 ) = ∑ 𝑎𝜆 𝑥 𝜆 (2)
𝜆=0

(𝑎𝜆 ≠ 0).
2. If 𝑝(𝑥 ) and 𝑞(𝑥 ) are singular at 𝑥 = 0 but 𝑥𝑝(𝑥 ) and 𝑥 2 𝑞(𝑥 ) are regular at 𝑥 = 0,
then there will always be at least one solution of Eqn. (1) of the form
∞

𝑦(𝑥 ) = ∑ 𝑎𝜆 𝑥 𝜆+𝑘 (3)

𝜆=0

(𝑎𝜆 ≠ 0).
3. If 𝑝(𝑥 ) and 𝑞(𝑥 ) are irregular singular points at 𝑥 = 0 and 𝑥𝑝(𝑥 ) and 𝑥 2 𝑞 (𝑥 ) are
singular at 𝑥 = 0, then regular solution of Eqn. (1) may not exist. In this case, no
general method for solving the equation is known.

Example 1
Consider the differential equation
𝑥𝑦 " + 2𝑦 ′ + 𝑥𝑦 = 0
By use of the power series method obtain the (a) indicial equation and its two solutions (b)
recursion formula and (c) general solution of the differential equation, 𝑦(𝑥 ).
Solution
The standard form of this differential equation is,
2
𝑦" + 𝑦′ + 𝑦 = 0
𝑥
2
Note that 𝑝(𝑥 ) → 𝑥 and 𝑞(𝑥 ) → 1; hence the differential equation is of the type 2 form and the

form the solution is

∞

𝑦(𝑥 ) = ∑ 𝑎𝜆 𝑥 𝜆+𝑘
𝜆=0

(𝑎𝜆 ≠ 0). On substituting the above equation into the original differential equation, one obtains
∞ ∞

∑ 𝑎𝜆 (𝜆 + 𝑘 + 1)(𝜆 + 𝑘 )𝑥 𝜆+𝑘−2 + ∑ 𝑎𝜆 𝑥 𝜆+𝑘 = 0

𝜆=0 𝜆=0

The basic plan at this stage is to write the above equation using a single sum. On replacing 𝜆
with 𝜆′ + 2 in the first sum, the power of 𝑥 in the first sum becomes the same as that in the
second sum. The above equation now becomes
∞ ∞

∑ 𝑎𝜆′ +2 (𝜆′ + 𝑘 + 3)(𝜆′ + 𝑘 + 2)𝑥 𝜆′ +𝑘

+ ∑ 𝑎𝜆 𝑥 𝜆+𝑘 = 0
𝜆′ =−2 𝜆=0
′
Since the sum is independent of the index, 𝜆 in the first sum can be replaced with 𝜆, and we may
write the equation as,
∞

𝑎0 𝑘(𝑘 + 1)𝑥 𝑘−2 + 𝑎1 (𝑘 + 1)(𝑘 + 2)𝑥 𝑘−1 + ∑{𝑎𝜆+2 (𝜆 + 𝑘 + 3)(𝜆 + 𝑘 + 2) + 𝑎𝜆 }𝑥 𝜆+𝑘 = 0

𝜆=0

Terms in the above equation are linearly independent and we required that
𝑎0 𝑘 (𝑘 + 1) = 0 ,
this is called the indicial equation. The indicial equation results from equating the coefficient of
the lowest power of the variable to zero. We also required
𝑎1 (𝑘 + 1)(𝑘 + 2) = 0
and
𝑎𝜆+2 (𝜆 + 𝑘 + 3)(𝜆 + 𝑘 + 2)
which is the recursion formula. The recursion formula is used to evaluate the coefficient for th
remaining powers of the variable. In this case, the solutions of the indicial equation are 𝑘 = 0 and
𝑘 = −1. When 𝑘 = 0, 𝑎1 = 0, because of equation between the indicial and recursion formula.
The coefficient 𝑎1 is arbitrary when 𝑘 = −1, and two independent solutions of the original
differential equation may be obtained by use of 𝑘 = −1 since 𝑎0 is arbitrary by hypothesis. The
form of the solution of the original differential equation becomes,
∞

𝑦(𝑥 ) = ∑ 𝑎𝜆 𝑥 𝜆−1
𝜆=0

Values of the coefficients in the above equation are obtained from the recursion formula using 𝑘 =
−1. The general expressions for even and odd expansion coefficients respectively are
(−1)𝑗 𝑎0
𝑎2𝑗 =
(2𝑗)!
and
(−1)𝑗 𝑎1
𝑎2𝑗+1 =
(2𝑗 + 1)!
for 𝑗 = 0, 1, 2, 3, . ..
The general solution of the original differential equation is obtained by substituting the above
coefficients into our 𝑦(𝑥 ),
∞ ∞
(−1)𝑗 2𝑗 (−1)𝑗 𝑥 2𝑗+1
𝑦(𝑥 ) = 𝑎0 ∑ 𝑥 + 𝑎1 ∑ .
(2𝑗)! (2𝑗 + 1)!
𝑗=0 𝑗=0

Nonhomogeneous differential equations with variable coefficients

• Variation of parameters and Green’s functions methods are normally used to solved
nonhomogeneous linear differential equations with variable coefficients that occur in
mathematical physics. The standard form for these differential equations is given in Eqn.
(1).
• The method of variation of parameters due to Lagrange will now be used to solve Eqn.
(1) subject to the conditions given below. Assume the solution has the form
𝑦(𝑥 ) = 𝑣1 (𝑥 )𝑦1 + 𝑣2 (𝑥 )𝑦2 (4)
 • In the above equation, 𝑦1 and 𝑦2 are two linearly independent solutions of the
corresponding homogeneous differential equation; they can be determined by use of the
power series method.
• Functions 𝑣1 and 𝑣2 are unknown parameters to be determined. We assume that 𝑓 (𝑥 ) is
continuous in some region of interest 𝑎 ≤ 𝑥 ≤ 𝑏 and that
𝑣 ′1 𝑦1 + 𝑣 ′2 𝑦2 = 0 (5)
Note that
𝑦 ′ = 𝑣1 𝑦 ′1 + 𝑣2 𝑦 ′2 (6)
and
𝑦 " = 𝑣 ′1 𝑦 ′1 + 𝑣1 𝑦 "1 + 𝑣 ′2 𝑦 ′2 + 𝑣2 𝑦 " 2 (7)
Substituting Eqns. (4), (6) and (7) into Eqn. (1), we obtain
𝑣 ′ 1 𝑦 ′ 1 + 𝑣 ′ 2 𝑦 ′ 2 = 𝑓 (𝑥 ) (8)
Solving 𝑣 ′1 and 𝑣 ′2 using Eqn. (5) and (8), we obtain
0 𝑦2
| |
𝑓 (𝑥 ) 𝑦2′ 𝑦2 𝑓 (𝑥 ) (9)
𝑣 ′1 = =−
∆ ∆
𝑦1 𝑦2
where ∆= |𝑦 ′ 𝑦2′ | and
1
𝑦 0
| 1′ |
𝑦1 𝑓 (𝑥 ) 𝑦1 𝑓 (𝑥 ) (10)
𝑣 ′1 = =−
∆ ∆
• The quantity ∆ is just the Wronkian 𝑊 (𝑦1 , 𝑦2 ) of 𝑦1 and 𝑦2 ; it is not equal to zero since
𝑦1 and 𝑦2 are linearly independent by hypothesis.
• Solving the above equations for 𝑣1 and 𝑣2 and substituting Eqns. (9) and (10) into Eqn.
(4), we obtain the solution of Eqn. (1) subject to the assumption in Eqn. (5); the general
solution of Eqn. (1) may be written as,
𝑓 (𝑥 )𝑦2 𝑑𝑥 𝑓(𝑥 )𝑦1 𝑑𝑥
𝑦(𝑥 ) = −𝑦1 ∫ + 𝑦2 ∫ (10)
𝑊 (𝑦1 , 𝑦2 ) 𝑊 (𝑦1 , 𝑦2 )
Example 2
By use of the variation of parameters method, solve
𝑥 2 𝑦 ” − 2𝑥𝑦 ′ + 2𝑦 = 𝑥 ln 𝑥
(𝑥 ≠ 0).
Solution
The standard form of this differential equation is,
2 2 ln 𝑥
𝑦” − 𝑦′ + 2 𝑦 =
𝑥 𝑥 𝑥
2 2 ln 𝑥
On comparing the above equation with Eqn. (1) we find that 𝑝 = − 𝑥, 𝑞 = 𝑥2, and 𝑓 (𝑥 ) = .
𝑥

Take 𝑦1 = 𝑥 and 𝑦2 = 𝑥 2 where 𝑊 (𝑦1 , 𝑦2 ) = 𝑥 2 . The general solution of the original

differential equation is,
ln 𝑥 ln 𝑥
𝑥 2 ( 𝑥 ) 𝑑𝑥 𝑥 ( 𝑥 ) 𝑑𝑥
𝑦(𝑥 ) = −𝑥 ∫ + 𝑥2 ∫
𝑥2 𝑥2

ln 𝑥 𝑑𝑥 ln 𝑥 𝑑𝑥
𝑦(𝑥 ) = −𝑥 ∫ + 𝑥2 ∫
𝑥 𝑥2

1
𝑦(𝑥 ) = −𝑥 [ (ln 𝑥 )2 + ln 𝑥 + 1] − 𝑐1 𝑥 + 𝑐2 𝑥 2
2
• Eqn. (10) will now be put in the form of a definite integral that is useful in solving initial
or boundary value problems.
• Let 𝑥 be a point in the closed interval [𝑎, 𝑏] such that the first term in Eqn. (10) is
replaced by a definite integral from 𝑏 to 𝑥 and the second term is replaced by a definite
integral from 𝑎 to 𝑥. In terms of the two indicated integrals,
𝑥 𝑏 𝑏
𝑦1 (𝑡)𝑦2 (𝑥 )𝑑𝑡 𝑦1 (𝑥 )𝑦2 (𝑡)𝑑𝑡
𝑦 (𝑥 ) = ∫ +∫ = ∫ 𝐺 (𝑥, 𝑡)𝑓(𝑡)𝑑𝑡 (11)
𝑎 𝑊 (𝑡 ) 𝑥 𝑊 (𝑡 ) 𝑎

• The function 𝐺 (𝑥, 𝑡) in the above equation is called the Green’s function for Eqn. (10)
subject to the appropriate boundary conditions. The Green’s function is defined by
𝑦 (𝑡)𝑦2 (𝑥 )⁄𝑊 (𝑡) ≡ 𝐺1 𝑓𝑜𝑟 𝑎 < 𝑡 < 𝑥
𝐺 (𝑥, 𝑡) = { 1 (12)
𝑦1 (𝑥 )𝑦2 (𝑡)⁄𝑊 (𝑡) ≡ 𝐺1 𝑓𝑜𝑟 𝑥 < 𝑡 < 𝑏
• Note that the Green’s function depends only 𝑦1 , 𝑦2 and the Wronskian. The quantity
𝑊 (𝑡) means 𝑊 [𝑦1 (𝑡), 𝑦2 (𝑡)]. The value of the Green’s function approach is related to
the fact that initial or boundary conditions are incorporated in the formulation of the
problem in a natural manner. At 𝑡 = 𝑎, 𝐺1 (𝑥, 𝑡) satisfies the boundary condition imposed
on 𝑦(𝑥 ) and 𝐺2 (𝑥, 𝑡) satisfies the boundary condition for 𝑦(𝑥 ) at 𝑡 = 𝑏.
Example 2
By use of the Green’s function method, find the solution of
𝑦 ” = 6𝑥
where 𝑦 ′(0) = 𝑦(1) = 0, 𝑦1 = 𝑥 and 𝑦2 = 𝑥 − 1.
Solution
Here the Wronskian equals 1 and the Green’s functions become
𝐺1 (𝑥, 𝑡) = 𝑡(𝑥 − 1)
for 0 ≤ 𝑡 ≤ 𝑥, and
𝐺2 (𝑥, 𝑡) = 𝑥 (𝑡 − 1)
for 𝑥 ≤ 𝑡 ≤ 1. The solution of the differential equation is,
1
𝑦(𝑥 ) = ∫ 𝐺 (𝑥, 𝑡) = 𝑥 3 .
0