Math Methods 1

Lia Vas

Series Solutions of Ordinary Differential Equations

Let us consider a linear differential equation of the second order

a(x)y 00 + b(x)y 0 + c(x)y = g(x)
where a, b, c, and g are some real-valued functions.
Recall that the equation a(x)y 00 + b(x)y 0 + c(x)y = 0 is the homogeneous part of the above rela-
tion. If yh is the solution of the homogeneous part, the particular solution yp of the non-homogeneous
equation a(x)y 00 +b(x)y 0 +c(x)y = g(x) can be found using the variation of parameters method (or the
undetermined coefficients method) and the general solution of the equation has the form y = yh + yp .
Since these methods are covered extensively in a differential equations course, we shall concentrate
on methods for finding the solution of the homogeneous equation
a(x)y 00 + b(x)y 0 + c(x)y = 0.
Also, since the case when a(x), b(x) and c(x) are all constant is covered in a differential equations
course, we shall concentrate on the case when the functions a(x), b(x) or c(x) are not constant.
To produce the general solution, recall that one needs to find two independent (i.e. not constant
multiple of each other) solutions y1 and y2 , called the fundamental solutions. Then, the general
solution is of the form
y = c1 y1 + c2 y2
where c1 and c2 are two constants.
There are certain methods for solving some non-constant coefficient equations in some particular
cases as the following examples illustrate.
• Legendre’s linear equation a(px + q)2 y 00 + b(px + q)y 0 + cy = 0 where a, b, c, p and q are
constants. This equation can be reduced to an equation with constant coefficients using the
substitution px + q = et . If p = 1 and q = 0, the equation ax2 y 00 + bxy 0 + cy = 0 is called
Euler’s equation.
• Exact equations. If the left hand side of the equation a(x)y 00 + b(x)y 0 + c(x)y = 0 is a
derivative of another equation, it is said to be exact. It can be shown that an equation is
exact if c(x) − b0 (x) + a00 (x) = 0. Note that if a(x) = 0, we can write the equation in the form
b(x)dy = −c(x)ydx and denote Q = b and P = −cy, then the condition c(x) − b0 (x) = 0 is
equivalent to Py = Qx which is the criterion of exactness for the first order differential equation.
• Partially known complementary functions. If one solution y1 of the homogeneous equa-
tion a(x)y 00 +b(x)y 0 +c(x)y = 0 is known, to find the second solution y2 and the general solution
y = c1 y1 + c2 y2 as well, assume that it has the form y2 = vy1 where v is an unknown function.
Substituting the derivatives of y2 in the equation yields a separable differential equation in v 0 .
Solving for v 0 first and then finding v produces the second complementary function y2 of the
general solution.

1
 However, none of the above methods are general. Our next goal is to present a general method
produces the solution of the equation a(x)y 00 + b(x)y 0 + c(x)y = 0 in a form of a power series
that P
y= ∞ n
n=0 an (x − x0 ) . Having a solution in the form of a power series should not be considered as a
downside – most commonly occurring functions such as polynomials, exponential and trigonometric
functions have power series extension.
Let us start by assuming that a(x) 6= 0 (so that the equation is really of the second order) and
by dividing by a(x) to obtain the form
y 00 + p(x)y 0 + q(x)y = 0
Although we denote the independent variable as x suggesting it is a real variable, our following
discussion and the methods presented are valid for functions of complex variables as well.

Regular, Singular and Regular-Singular Points. Let us consider the equation y 00 + p(x)y 0 +
q(x)y = 0 and let us assume that the function p(x) and q(x) are analytic that is, that they have
power series expansions
∞
X ∞
X
n
p(x) = pn (x − x0 ) and q(x) = qn (x − x0 )n
n=0 n=0
centered at a point x = x0 which are convergent on some interval containing x0 . In this case, the
point x = x0 is said to be a regular (or ordinary) point.
If the functions p(x) or q(x) are not analytic at x = x0 , then the point x = x0 is said to be a
singular point. In this case when it is possible to write the equation in the form
(x − x0 )2 y 00 + p(x)(x − x0 )y 0 + q(x)y = 0
where the functions p(x) and q(x) are analytic at x = x0 , the point x = x0 is said to be a regular-
singular point. If it is not possible to represent the equation in this way, the point x = x0 is said
to be irregular or essential singularity.
The following examples showcase some equations that appear in applications in physics, and the
classifications of their relevant singularities.
Equation singular points type

Legendre (1 − x2 )y 00 − 2xy 0 + k(k + 1)y = 0 ±1 regular-singular

Chebyshev (type 1) (1 − x2 )y 00 − xy 0 + p2 y = 0 ±1 regular-singular

Chebyshev (type 2) (1 − x2 )y 00 − 3xy 0 + k(k + 2)y = 0 ±1 regular-singular

Bessel x2 y 00 + xy 0 + (x2 − a2 )y = 0 0 regular-singular

Laguerre xy 00 + (1 − x)y 0 + ay = 0 0 regular-singular

Hermite y 00 − 2xy 0 + 2ay = 0 none all pts regular

Simple harmonic oscillator y 00 + a2 y = 0 none all pts regular

2
 The interval of convergence of solutions. The radius of convergence of the power series
of y is the distance from the center x = x0 to the next nearest singularity of p and q. If R is this
radius (possibly infinite), then the interval of convergence is (x0 − R, x0 + R).
Example 1. Consider the equation y 00 − x−1 1
y = 0. Show that x = 0 is a regular point of this
equation and determine the interval of convergence of the solution.
−1 1
Solutions. For this equation, p = 0 and q = x−1 = 1−x . Both of these functions are analytic:
the first because it is a polynomial and the second since it is a rational function, so both functions
have continuous derivatives of any order, p at any point and q at any point different than 1. Thus,
the point x = 0 is regular. The distance from the center x = 0 to the first singularity x = 1 is 1. So,
the solution converges on the interval (−1, 1).
Alternatively, the functions p and q are analytic since they have convergent powerPseries expan-
1
sions at x = 0. The expansion of p is 0 + 0x + 0x2 + . . . . The expansion of q is 1−x = ∞ n
n=0 x . This
last series is convergent on (−1, 1) and so the power series solution of this equation is convergent on
(−1, 1).

The method of finding series solutions about a regular point. If x = x0 is a regular point
of y 00 + p(x)y 0 + q(x)y = 0, then every solution of this equation can be represented as a power series
centered at x0 .
X∞
y= an (x − x0 )n
n=0

Finding derivatives, one obtains

∞
X ∞
X
0 n−1 00
y = nan (x − x0 ) and y = n(n − 1)an (x − x0 )n−2 .
n=0 n=0

In some cases, these derivatives can be plugged into the equation as they are. For some equations,
they may need to be written in a form so the n-th term has n as the power of (x − x0 ). To achieve
this for y 0 , note first that the first term P
of y 0 (term with n = 0) is 0a0 (x − x0 )−1 = 0, so that the sum
can be started with n = 1 so that y 0 = ∞ n=1 nan (x − x0 )
n−1
. Then, one can replace every n by n + 1
and get
∞
X ∞
X ∞
X
y0 = nan (x − x0 )n−1 = (n + 1)an+1 (x − x0 )n+1−1 = (n + 1)an+1 (x − x0 )n .
n=1 n+1=1 n=0
P∞
Similarly, since the first two terms of y 00 are zero, y 00 = n=2 n(n − 1)an (x − x0 )n−2 . Shifting this last
series from n to n + 2 produces the following.
∞
X ∞
X ∞
X
y 00 = n(n−1)an (x−x0 )n−2 = (n+2)(n+2−1)an+2 (x−x0 )n+2−2 = (n+2)(n+1)an+2 (x−x0 )n
n=2 n+2=2 n=0

The solution y can be obtained following the steps below.

1. Substitute y, y 0 and y 00 in the equation and shift the series you obtain if necessary so that all
the sums have the same power of (x − x0 ).

3
 2. Combine all the series into single one so that the left hand side of the equation is a single power
series.

3. Equate the coefficients of the series you obtained to zero. This produces a recursive equation
that computes the terms of the sequence an .

4. If possible, obtain the sequence an as a function of n. Then obtain power series for two funda-
mental solutions y1 and y2 from the recursive sequence, usually by choosing some convenient
values of a0 and a1 . The “standard” choice of a0 and a1 is that a0 = 1 and a1 = 0 for one and
a0 = 0 and a1 = 1 for another solution. Any “nonstandard” choice of a0 and a1 would be given
to you in problems in this course.
In each case, once you determine the sequence a0 , a1 , a2 . . . , you can substitute these values
back in ∞
X
y= an x n = a0 + a1 x + a2 x 2 + a3 x 3 + . . .
n=0

One choice of a0 and a1 produces one fundamental solution y1 and another produces y2 . The
general solution is y = c1 y1 + c2 y2 where c1 , c2 are two constants.

5. If you can find sums of those series and express y1 and y2 as an elementary function, then it is
said that the solution is in the closed form. The closed form will not always be possible to
obtain.
When finding the closed form of the solutions, the power series expansions of the following
elementary functions may be useful:
∞ ∞
x
X 1 n 1 X
e = x = xn
n=0
n! 1 − x n=0

∞ ∞
X (−1)n X (−1)n 2n
sin x = x2n+1 cos x = x
n=0
(2n + 1)! n=0
(2n)!
P∞
P∞ n=0 an (x−n
Let us review the process of finding the sum of a given series: for a given power series
n
a) , we are finding an elementary function f (x) equal to its sum. The equality f (x) = n=0 an (x−a)
is valid when the series is convergent and the domain of the function f (x) is equal to the convergence
interval of the series.
Example 2. Find the sum of the following series.
∞ ∞ ∞
X (−1)n n
X (−1)n n+2
X 1 2n
(a) x (b) x (c) x
n=0
n! n=0
n! n=0
n!

∞ ∞ ∞
X
n n
X X (−1)n
(d) 2 x (e) (−1)n x2n (f) xn
n=0 n=0 n=0
(2n)!
1
Solution. Have the four “basic” expansions (ex , 1−x , sin x, and cos x) in front when you are
finding the sum.

4
 (a) This series matches the expansion for ex pretty closely: only (−1)n term is “off” but you can
group it with xn to have (−x)n . Thus, the series is ∞ 1 n
P
n=0 n! (−x) . Note that this series is exactly
x
the series obtained by replacing x by −x in the formula for e . Hence,
∞ ∞
X (−1)n X 1
x = n
(−x)n = e−x .
n=0
n! n=0
n!

(b) Write xn+2 as xn x2 so that x2 can be factored in front of the series. The remaining part
matches the previous sum. Hence,
∞ ∞ ∞
X (−1)n X (−1)n X 1
x n+2
=x 2
x =x n
(−x)n = x2 e−x .
2

n=0
n! n=0
n! n=0
n!

(c) Since x2n = (x2 )n , this series is exactly the series obtained by replacing x by x2 in the
expansion of ex . Hence,
∞ ∞
X 1 2n X 1 2 n 2
x = (x ) = ex .
n=0
n! n=0
n!
Note that each of the three series in (a),(b), and (c) converges for every x and so their three
closed forms are
P equal to the initial series for every value of x.
(d) Since ∞ n n
P∞ n 1
n=0 2 x = n=0 (2x) , you can use the formula for 1−x but replace x by 2x. Hence,

∞ ∞
X X 1
2n xn = (2x)n = .
n=0 n=0
1 − 2x
−1
Note that the interval of convergence corresponds to |2x| < 1 ⇒ −1 < x < 1 ⇒ 2
< x < 12 .
(e) Since (−1)n x2n = (−1)n (x2 )n = (−x2 )n ,
∞ ∞
X
n 2n
X 1 1
(−1) x = (−x2 )n = 2
= .
n=0 n=0
1 − (−x ) 1 + x2

Note that the interval of convergence corresponds to | − x2 | < 1 ⇒ |x|2 < 1 ⇒ −1 < x < 1.
(f) The given series matches the expansion for cos x up to the power of x which is even for the
√ 2 √ √
expansion of cosine. Hence, write xn as ( x )n = ( x)2n and replace x by x in the expansion for
cosine to obtain that ∞ ∞
X (−1)n n X (−1)n √ 2n √
x = ( x) = cos x.
n=0
(2n)! n=0
(2n)!
The series converges for every x so the interval of convergence is (−∞, ∞).
To illustrate the entire method, we start by an example that can be solved using much simpler
approach. Thus, we stress that this example is here just to illustrate how the method works, not to
illustrate any real need for the method.
Example 3 (“fake” example). Consider the equation y 00 − 2y 0 + y = 0. Show that x = 0 is a
regular point of this equation and determine the interval of convergence of the solution. Then find
the series solution at x = 0 and express your answers in the closed form by choosing a0 = a1 = 1 for
one and a0 = 0, a1 = 1 for the other fundamental solution.

5
 Solution. For this equation, p = −2 and q = 1. The functions p and q are analytic since they
are polynomials and polynomials have continuous derivatives of any order. In particular, p0 = p00 =
p000 = . . . 0 and q 0 = q 00 = . . . = 0. The interval of convergence is (−∞, ∞) because polynomials have
no singularities.
Alternatively, you can say that p and q are analytic since their power series expansions at x = 0
are p = −2 + 0x + 0x2 + . . . and q = 1 + 0x + 0x2 + . . . and they convergent at every value of x
(including x = 0) since just finitely many terms are nonzero. Thus, the point x = 0 is regular and
the solutions are convergent on (−∞, ∞).
The solution of the equation can be found in the form
∞
X
y= an x n .
n=0

Find the derivatives to be

∞
X ∞
X ∞
X
0 n−1 n−1
y = nan (x − x0 ) = nan (x − x0 ) = (n + 1)an+1 (x − x0 )n and
n=0 n=1 n=0

∞
X ∞
X ∞
X
00 n−2 n−2
y = n(n − 1)an (x − x0 ) = n(n − 1)an (x − x0 ) = (n + 2)(n + 1)an+2 (x − x0 )n
n=0 n=2 n=0
P∞ P∞
and
P∞ substitute them into the equation to get n=0 (n + 2)(n + 1)an+2 xn − 2 n=0 (n + 1)an+1 xn +
n
n=0 an x = 0 ⇒
∞
X
[(n + 2)(n + 1)an+2 − 2(n + 1)an+1 + an ]xn = 0
n=0

Since the series is identically equal to zero, all its terms have to be equal to zero. Thus,

2(n + 1)an+1 − an
(n + 2)(n + 1)an+2 − 2(n + 1)an+1 + an = 0 ⇒ an+2 = for all n = 0, 1, . . .
(n + 1)(n + 2)

The above formula represents a recursive equation which computes the coefficients of the sequence
an . Note that it depends on two initial conditions a0 and a1 . By choosing a0 = a1 = 1 as instructed,
you have that

2−1 1 1 2(2) 21 − 1 1 2(3) 3!1 − 1

2 2−1 1 1
a2 = = = , a3 = = , a4 = = = ,..., an = .
1·2 2 2! 3·2 3! 4·3 4·3·2 4! n!
This produces the solution
∞
1 1 X 1
y 1 = 1 + x + x2 + x3 + . . . = x n = ex .
2! 3! n=0
n!

By taking a0 = 0 and a1 = 1, you obtain that

2 2(2) − 1 1 2(3) 12 − 1 2 1 1 8 16 − 12 5 1
a2 = = 1, a3 = = , a4 = = = = , a5 = = = .
2 3·2 2 4·3 4·3 2·3 3! 5·4 5! 4!

6
 1 1
This indicates that an = (n−1)!
so an+1 = n!
. Thus, the second fundamental solution is
∞ ∞
1 3 1 4
2
X 1 n+1 X 1 n
y2 = x + x + x + x + . . . = x =x x = xex .
2! 3! n=0
n! n=0
n!

Hence, the general solution is y = c1 xex + c2 ex .

As we pointed out, the equation y 00 − 2y 0 + y = 0 has constant coefficients so it could be solved by
considering the characteristic equation r2 − 2r + 1 = 0 ⇒ (r − 1)(r − 1) = 0 and finding the solution
in the form y = c1 ex + c2 xex . So, we turn to more relevant examples – examples of equations that
do not have constant coefficients.
Example 3 (“real” example). Consider the equation (1 − x)2 y 00 − 2y = 0. Show that x = 0
is a regular point of this equation and determine the radius of the convergence of the solution. Find
the series solution at x = 0. Express your solution in the closed form by choosing a0 = 1, a1 = −2
for one and a0 = a1 = 1 for the other fundamental solution.
−2
Solution. Dividing by (1 − x)2 we obtain the form y 00 − (1−x) 2
2 y = 0 so that p = 0 and q = (1−x)2 .

Both of these functions are analytic at x = 0 : the first because it is a polynomial and the second
since it is a rational function with x = 1 as the only singularity. So, both have continuous derivatives
of any order, p at any point and q at any point different than 1. Thus, the point x = 0 is regular.
Alternatively, you can argue that p and q are analytic since they have convergent power series
2
expansions at x = 0. The expansion of p is 0 + 0x + 0xP + . . . . The expansion of q can be obtained
differentiating −2 1−x = −2 n=0 x . So, (1−x)2 = −2 ∞
1 ∞ n −2 n−1
P
n=0 nx and this series is convergent at
x = 0.
Since 1 is a singularity of q and p has no singularity, the solutions converge in an interval of length
1 around x = 0. So, the interval of convergence is (−1, 1).
The solution can be found in the form
∞
X ∞
X ∞
X
y= an xn so that y 0 = nan xn−1 and y 00 = n(n − 1)an xn−2
n=0 n=0 n=0

Substitute the function and its derivatives into the equation.

∞
X ∞
X
2 n−2
(1 − 2x + x ) n(n − 1)an x −2 an x n = 0 ⇒
n=0 n=0

∞
X ∞
X ∞
X ∞
X
n−2 n−1 n
n(n − 1)an x −2 n(n − 1)an x n(n − 1)an x − 2 an xn = 0.
n=0 n=0 n=0 n=0
P∞ P∞
Note that the first two sums are equal to n=2 n(n − 1)an xn−2 and n=1 n(n − 1)an xn−1 so that the
first one can be shifted by 2 (replace each n by P n + 2) and the second one by 1 (replace each n by
n + 1) to obtain ∞ n ∞ n
P
n=0 (n + 2)(n + 1)a n+2 x and n=0 (n + 1)nan+1 x . Hence, we have the following.

∞
X ∞
X ∞
X ∞
X
n n n
(n + 2)(n + 1)an+2 x − 2 (n + 1)nan+1 x + n(n − 1)an x − 2 an x n = 0
n=0 n=0 n=0 n=0

7
Combine all four sums into a single one as follows.
∞
X
[(n + 2)(n + 1)an+2 − 2(n + 1)nan+1 + n(n − 1)an − 2an ]xn = 0.
n=0

Since the series is identically equal to zero, all its terms have to be equal to zero. Thus,

(n + 2)(n + 1)an+2 − 2(n + 1)nan+1 + n(n − 1)an − 2an = 0 for all n = 0, 1, . . .

The equation can be simplified as follows.

(n + 2)(n + 1)an+2 − 2(n + 1)nan+1 + (n2 − n − 2)an = 0 ⇒

(n + 2)(n + 1)an+2 − 2(n + 1)nan+1 + (n − 2)(n + 1)an = 0

Note that every term has (n + 1) factor, so you can simplify this further as follows and then solve
for an+2 in terms of an and an+1 .

2nan+1 − (n − 2)an
(n + 2)an+2 − 2nan+1 + (n − 2)an = 0 ⇒ an+2 =
(n + 2)

Note that a2 = 2a20 = a0 , a3 = 2a03+a1 , a4 = 4a43 = a3 , a5 = 6a35−a3 = a3 , a6 = 8a3 −2a

6
3
= a3 . . .
Choosing a0 = 1 and a1 = −2 produces a2 = 1 and 0 = a3 = a4 = a5 = . . . . So the only non-zero
1, a1 = −2 and a2 = 1. This yields a polynomial solution. Substitute these
coefficients are a0 =P
values back to y1 = ∞ n
n=0 an x to get your first solution

∞
X
y1 = an xn = a0 + a1 x + a2 x2 + a3 x3 + . . . = 1 − 2x + 1x2 + 0x3 + 0x4 + . . . = 1 − 2x + x2 = (1 − x)2 .
n=0

2a0 +a0
Choosing a1 = a0 = 1, we obtain a2 = 1, a3 = 3
= 1, a4 = a3 = 1, . . . an = 1. This produces
∞ ∞
X X 1
y2 = an x n = 1 + x + x 2 + x 3 + x 4 + . . . = xn = .
n=0 n=0
1−x

Note that the radius of convergence of this series is 1 so it converges for −1 < x < 1 just as we
concluded earlier. The two solutions y1 and y2 are obviously independent (since one is a polynomial
and the other is not) so a general solution is
1
y = c1 (1 − x)2 + c2 .
1−x
Practice Problems.

1. Consider the differential equation y 00 −y = 0. Show that x = 0 is a regular point of this equation
and determine the interval of convergence of the solution. Find the series solutions of the given
equation about x = 0. Express your answers in the closed form by choosing a0 = a1 = 1 for
one and a0 = −a1 = 1 for the other fundamental solution.

8
2. Consider Hermite equation y 00 − 2xy 0 + 4y = 0. Show that x = 0 is a regular point of this
equation and determine the interval of convergence of the solution. Find the series solutions of
the given equation about x = 0. Find the closed form of one solution and list a few terms of
the second solution.

Solutions.

1. For this equation, p = 0 and q = −1. The functions p and q are analytic since they are
polynomials and polynomials have continuous derivatives of any order. In particular, p0 = p00 =
p000 = q 0 = q 00 = . . . 0. Thus, x = 0 is a regular point. The interval of convergence is (−∞, ∞)
because polynomials have no singularities.
Alternatively, you can say that p and q are analytic since their power series expansions at x = 0
are p = 0 + 0x + 0x2 + . . . and q = −1 + 0x + 0x2 + . . . and they convergent at every value of x
(including x = 0) since just finitely many terms are nonzero. Thus, the point x = 0 is regular
and the solutions are convergent on (−∞, ∞).
The solution can be found in the form y = ∞ n
P
n=0 an x . Since the series of p and q converge for
all values of x, the interval of convergence of solutions is (−∞, ∞).
Find the derivatives to be
∞
X ∞
X ∞
X ∞
X
y0 = nan xn−1 = (n + 1)an+1 xn and y 00 = n(n − 1)an xn−2 = (n + 2)(n + 1)an+2 xn
n=0 n=0 n=0 n=0

and substitute them into the equation.

∞
X ∞
X ∞
X
n n
(n + 2)(n + 1)an+2 x − an x = 0 ⇒ [(n + 2)(n + 1)an+2 − an ]xn = 0
n=0 n=0 n=0

Since the series is identically equal to zero, all its terms have to be equal to zero. Thus,
an
(n + 2)(n + 1)an+2 − an = 0 for all n = 0, 1, . . . ⇒ an+2 = for all n = 0, 1, . . .
(n + 1)(n + 2)
1 1 a1 1 a2 1 1
Choosing a0 P
= a1 = 1, we have that a2 = 1·2
= 2!
, a3 = 2·3
= 3!
, a4 = 3·4
= 4!
⇒ an = n!
.
Hence, y1 = ∞ 1 n x
n=0 n! x = e .
Choosing a0 = −a1 = 1, we obtain a2 = 1·2 1 a1
= 2!1 , a3 = 2·3 = −1
3!
a2
, a4 = 3·4 a3
= 4!1 , a5 = 4·5 =
(−1) n ∞ (−1) n ∞ (−x) n
−1
⇒ an = n! . Hence, y2 = n=0 n! xn = n=0 n! = e−x and y = c1 ex + c2 e−x is a
P P
5!
general solution. Note that the radius of the convergence is infinity because the equation has
no singular points.

2. For this equation, p = −2x and q = 4. The functions p and q are analytic since they are
polynomials and polynomials have continuous derivatives of any order. In particular, p0 = −2
and p00 = p000 = q 0 = q 00 = . . . 0. Thus, x = 0 is a regular point. The interval of convergence is
(−∞, ∞) because polynomials have no singularities.
Alternatively, you can say that p and q are analytic since their power series expansions at x = 0
are p = 0 − 2x + 0x2 + 0x3 + . . . and q = 4 + 0x + 0x2 + . . . and they convergent at every value

9
of x (including x = 0) since just finitely many terms are nonzero. Thus, the point x = 0 is
regular and the solutions are convergent on (−∞, ∞).
The solution can be found in the form y = ∞ n
P
n=0 an x . Since the series of p and q converge for
all values of x, the interval of convergence of solutions is (−∞, ∞).
∞
X ∞
X ∞
X ∞
X
n 0 n−1 00 n−2
y= an x ⇒ y = nan x and y = n(n − 1)an x = (n + 2)(n + 1)an+2 xn
n=0 n=0 n=0 n=0

Substitute these into the equation and get

∞
X ∞
X ∞
X
n n
(n + 2)(n + 1)an+2 x − 2 nan x + 4 an x n = 0 ⇒
n=0 n=0 n=0

∞
X
[(n + 2)(n + 1)an+2 − 2nan + 4an ] xn = 0.
n=0

Since the series is identically equal to zero, all its terms have to be equal to zero. Thus,

(2n − 4)an
(n + 2)(n + 1)an+2 − 2nan + 4an for all n = 0, 1, . . . ⇒ an+2 = for all n = 0, 1, . . .
(n + 1)(n + 2)

Thus the even-indexed coefficients depends on a0 and the odd-indexed coefficients on a1 . We

can obtain two linearly independent solutions by taking the “standard” choice of a0 and a1 :
a0 = 1 and a1 = 0 for y1 and a0 = 0 and a1 = 1 for y2 .
If a0 = 1 and a1 = 0, then a1 = a3 = a5 = a7 . . . = 0. Since a0 = 1, a2 = −4
2
= −2 and a4 = 0.
Thus, a4 = a6 = a8 = a10 = . . . = 0. So, the only two nonzero coefficients are a0 = 1 and
a2 = −2. Hence, y1 is a polynomial

y1 = a0 + a2 x2 = 1 − 2x2 .

−2 −2
If a0 = 0 and a1 = 1, then a0 = a2 = a4 = a6 . . . = 0. Since a1 = 1, a3 = 2·3
= 3!
,
−2·2 2 2 3
a5 = 2·3·4·5 = −2
5!
, a7 = −6·2
5!·6·7
= −3·2
7!
. . . Thus

2 3 22 5 3 · 23 7
y2 = x − x − x − x ....
3! 5! 7!
The general solution y = c1 y1 + c2 y2 is convergent on (−∞, ∞).

10