Processing math: 0%

16.5 Power Series Methods

Printed Page 1089

1089

OBJECTIVES

When you finish this section, you should be able to:

  1. Use power series to solve a linear differential equation (p. 1089)

Up to now, we have concentrated on first-order differential equations for which exact solutions could be found. But many first-order differential equations lead to integrals that cannot be expressed in terms of elementary functions. Moreover, most higher-order differential equations cannot be solved exactly. For these reasons, considerable emphasis is placed on methods for approximating solutions of differential equations. One such method, power series, is introduced here.

1 Use Power Series to Solve a Linear Differential Equation

Printed Page 1089

Power series methods assume that a solution y=y(x) of a given differential equation has a power series expansion of the form \bbox[5px, border:1px solid black, #F9F7ED]{\bbox[#FAF8ED,5pt] {y=y( x) =\sum\limits_{k=0}^{\infty }a_{k}x^{k}}}

NEED TO REVIEW?

Power series are discussed in Section 8.8, pp. 600-609.

A power series method consists of finding power series for the terms in the differential equation, as well as the power series for y, y^{\prime} , y^{\prime \prime}, and so on. These power series expansions are substituted into the differential equation to obtain relationships among the coefficients.

We begin with a simple example to outline the basic idea.

EXAMPLE 1Using Power Series to Solve a Linear Differential Equation

Use power series to solve the differential equation y^{\prime} =y.

Solution We assume that the solution of the differential equation can be expressed as the power series \begin{equation} y( x) =\sum\limits_{k=0}^{\infty }a_{k}x^{k} \tag{1} \end{equation}

Then \begin{equation*} y^{\prime} ( x) =\sum\limits_{k=1}^{\infty }ka_{k}x^{k-1} \end{equation*}

Since y^{\prime} =y, this leads to \begin{equation*} \sum\limits_{k=1}^{\infty }ka_{k}x^{k-1}=\sum\limits_{k=0}^{\infty}a_{k}x^{k} \end{equation*}

To obtain relationships among the coefficients, we write out the terms. \begin{equation*} a_{1}x^{0}+2a_{2}x+3a_{3}x^{2}+4a_{4}x^{3}+\cdots =a_{0}x^{0}+a_{1}x+a_{2}x^{2}+a_{3}x^{3}+\cdots \end{equation*}

Because the coefficients of corresponding powers of x are equal, we have \begin{equation*} a_{1}=a_{0}\qquad 2a_{2}=a_{1}\qquad 3a_{3}=a_{2}\qquad 4a_{4}=a_{3}\ \ldots\ na_{n}=a_{n-1}\ldots \end{equation*}

We can express these relationships recursively. \begin{eqnarray*}{{\hspace{-5pc}}{\hspace{-5pc}}} & \enspace a_{1}=a_{0}\,\,\,\,\qquad a_{2}=\dfrac{1}{2}a_{1}=\dfrac{1}{2!}a_{0} &\hspace{-56pt}a_{3}=\dfrac{1}{3}a_{2}=\dfrac{1}{3\cdot 2}a_{0}=\dfrac{1}{3!}a_{0}\\[4pt] & \qquad \qquad \qquad \qquad a_{4} =\dfrac{1}{4}a_{3}=\dfrac{1}{4!}a_{0}\,\,\,\,\, \ldots\,\, \,\,\, a_{n}=\dfrac{1}{n}a_{n-1}=\dfrac{1}{n!}a_{0}\,\,\,\,\,\,\cdots \end{eqnarray*}

1090

The power series (1) takes the form y( x) =\sum\limits_{k=0}^{\infty }a_{k}x^{k}=\sum\limits_{k=0}^{\infty }\dfrac{1}{k!}a_{0}x^{k}=a_{0}\sum \limits_{k=0}^{\infty }\dfrac{x^{k}}{k!}

which we recognize as y( x) =a_{0}e^{x}.

RECALL

As we learned in Section 8.9, e^{x}=\sum \limits_{k=0}^{\infty }\dfrac{x^{k}}{k!}

NOW WORK

Problem 1.

Although this example led to a power series solution we recognized, most solutions will not. The next examples will illustrate typical power series solutions.

EXAMPLE 2Using Power Series to Solve a Linear Differential Equation

Use power series to solve the differential equation y^{\prime} -2y=e^{-x}.

Solution Assume that a solution y=y( x) of the differential equation can be expressed as the power series y(x)=\sum\limits_{k=0}^{\infty }a_{k}x^{k}. Then y^{\prime}(x)=\sum\limits_{k=1}^{\infty }ka_{k}x^{k-1}. Since e^{-x}=\sum \limits_{k=0}^{\infty }\dfrac{(-1)^{k}}{k!}x^{k}, the differential equation can be written as \begin{equation} \underset{\color{#0066A7}{\hbox{\(y^{\prime}\)}}}{\underbrace{\sum\limits_{k=1}^{\infty }ka_{k}x^{k-1}}}-\underset{\color{#0066A7}{\hbox{\(y\)}}}{2\underbrace{\sum\limits_{k=0}^{\infty }a_{k}x^{k}}}=\underset{\color{#0066A7}{\hbox{\(e^{-x}\)}}}{\underbrace{\sum\limits_{k=0}^{\infty }\frac{(-1)^{k}}{k!}x^{k}}} \tag{2} \end{equation}

To obtain relationships among the coefficients, we write out the terms of the power series. \begin{eqnarray*} &&\hspace{-.75pc}\big( a_{1}\,{+}\,2a_{2}x\,{+}\,3a_{3}x^{2}\,{+}\,4a_{4}x^{3}\,{+}\,5a_{5}x^{4}\,{+}\,\cdots \big)\!\,{-}\,2\big(a_{0}\,{+}\,a_{1}x\,{+}\,a_{2}x^{2}\,{+}\,a_{3}x^{3}\,{+}\,a_{4}x^{4}\,{+}\,\cdots \big)\\[4pt] && \enspace \quad =1-x+\frac{1}{2!}x^{2}-\frac{1}{3!}x^{3}+\dfrac{1}{4!}x^{4}+\cdots \end{eqnarray*}

Now we combine the coefficients of corresponding powers of x to get \begin{eqnarray*} &&(a_{1}-2a_{0})+(2a_{2}-2a_{1})x+(3a_{3}-2a_{2})x^{2}+(4a_{4}-2a_{3})x^{3}+ \cdots\\[4pt] &&\qquad =1-x+\frac{1}{2!}x^{2}-\frac{1}{3!}x^{3}+\dfrac{1}{4!}x^{4}-\cdots \end{eqnarray*}

Since the coefficients of corresponding powers of x are equal, we have \begin{equation*} \begin{array}{rcl@{\quad}l@{\quad}rcl@{\;}l} a_{1}-2a_{0}&=&1 & \hbox{or equivalently,} & a_{1}&=&2a_{0}+1 \\ 2a_{2}-2a_{1}&=&-1 & \hbox{or equivalently,} & a_{2}&=&\dfrac{2a_{1}-1}{2}=\dfrac{ 2\left( 2a_{0}+1\right) -1}{2}=2a_{0}+\dfrac{1}{2} \\ 3a_{3}-2a_{2}&=&\dfrac{1}{2!} & \hbox{or equivalently,} & a_{3}&=&\dfrac{2a_{2}+ \dfrac{1}{2!}}{3}=\dfrac{2}{3}\left( 2a_{0}+\dfrac{1}{2}\right) +\dfrac{1}{ 3\cdot 2!}\\ &&&&&=&\dfrac{4}{3}a_{0}+\dfrac{3}{3!} \\ 4a_{4}-2a_{3}&=&-\dfrac{1}{3!} & \hbox{or equivalently,} & a_{4}&=&\dfrac{2a_{3}- \dfrac{1}{3!}}{4}=\dfrac{1}{2}a_{3}-\dfrac{1}{4!}\\ &&&&& =&\dfrac{1}{2}\left( \dfrac{4 }{3}a_{0}+\dfrac{3}{3!}\right)-\dfrac{1}{4!}=\dfrac{2}{3}a_{0}+\dfrac{5}{4!}\\ 5a_{5}-2a_{4}&=&\dfrac{1}{4!} & \hbox{or equivalently,} & a_{5}&=&\dfrac{2a_{4}+ \dfrac{1}{4!}}{5}=\dfrac{2}{5}a_{4}+\dfrac{1}{5!}\\ &&&&&=&\dfrac{2}{5}\left( \dfrac{2 }{3}a_{0}+\dfrac{5}{4!}\right) +\dfrac{1}{5!}=\dfrac{4}{15}a_{0}+\dfrac{11}{ 5!} \end{array} \end{equation*}

1091

This recursive formula ( n+1) a_{n+1}=2a_{n}+( -1) ^{n}\dfrac{1}{( n) !} can be used to find a_{n} in terms of a_{0,} as we did for a_{1}, a_{2}, a_{3}, a_{4}, and a_{5}. The power series representation of the general solution of the equation is \begin{eqnarray*} y(x) =\sum\limits_{k=0}^{\infty }a_{k}x^{k}&=&a_{0}+(2a_{0}+1)x+\left( 2a_{0}+\frac{1}{2!}\right) x^{2}+\left( \frac{4}{3}a_{0}+\frac{3}{3!}\right) x^{3}\\[4pt] &&+\left( \frac{2}{3}a_{0}+\frac{5}{4!}\right)\! x^{4}+\!\left( \dfrac{4}{15} a_{0}+\dfrac{11}{5!}\right)\! x^{5}+\cdots \\[4pt] & =&a_{0}\!\left( 1+2x+2x^{2}+\frac{4}{3}x^{3}+\frac{2}{3}x^{4}+\dfrac{4}{15} x^{5}+\cdots \right) \\[4pt] &&+\,x+\frac{1}{2!}x^{2}+\frac{3}{3!}x^{3}+\frac{5}{4!} x^{4}+\dfrac{11}{5!}x^{5}+\cdots \end{eqnarray*}

where a_{0} is arbitrary.

There is a more direct method for obtaining the recursion formula in Example 2. If we change the index of summation from k to k+1 in the first power series on the left side of (2) and distribute the -2 inside the second power series, we have \sum_{k=0}^{\infty }(k+1)a_{k+1}x^{k}+\sum_{k=0}^{\infty }(-2) a_{k}x^{k}=\sum_{k=0}^{\infty }\frac{(-1)^{k}}{k!}x^{k}

Now add the two series on the left. \sum_{k=0}^{\infty }[(k+1)a_{k+1}-2a_{k}]x^{k}=\sum_{k=0}^{n}\frac{(-1)^{k}}{ k!}x^{k}

Then ( n+1) a_{n+1}-2a_{n}=\dfrac{( -1) ^{n}}{n!}, as before.

EXAMPLE 3Using Power Series to Solve a Linear Differential Equation

Use power series to find the general solution of the linear differential equation \begin{equation*} y^{\prime \prime} +xy^{\prime} +2y=0 \end{equation*}

Solution If y(x)=\sum\limits_{k=0}^{\infty }a_{k}x^{k} is a solution to the differential equation, then y^{\prime} (x)=\sum\limits_{k=1}^{\infty }ka_{k}x^{k-1}\qquad \hbox{and}\qquad y^{\prime \prime} (x)=\sum\limits_{k=2}^{\infty }k(k-1)a_{k}x^{k-2}

Now we substitute these power series into the differential equation. \begin{eqnarray*} \sum\limits_{k=2}^{\infty }k(k-1)a_{k}x^{k-2}\,{+}\,x\!\sum\limits_{k=1}^{\infty}ka_{k}x^{k-1}+2\sum\limits_{k=0}^{\infty }a_{k}x^{k}\!&=&0\quad {\color{#0066A7}{y^{\prime \prime} +xy^{\prime} +2y=0}} \\[4pt] \sum\limits_{k=2}^{\infty }k(k-1)a_{k}x^{k-2}+\sum\limits_{k=1}^{\infty }ka_{k}x^{k}+\sum\limits_{k=0}^{\infty }2a_{k}x^{k}\!&=&0 \quad {\color{#0066A7}{\hbox{Move }x\hbox{ and } 2\hbox{ into the summation.}}} \end{eqnarray*}

Next we adjust the indexes of summation so that x^{k} appears in each series. Here, only the first series needs modification. If we replace k with k+2 in the first series, we obtain \begin{equation*} \underset{\color{#0066A7}{\hbox{Replace }k\hbox{ with }k+2}}{\underbrace{\sum\limits_{k=0}^{ \infty }( k+2) (k+1)a_{k+2}x^{k}}}+\sum\limits_{k=1}^{\infty }ka_{k}x^{k}+\sum\limits_{k=0}^{\infty }2a_{k}x^{k}=0 \end{equation*}

1092

The index in the first and third series begins at k=0, and the index in the second series begins at k=1. We write out the k=0 term of the first and third series separately. Then each summation starts at k=1. \begin{eqnarray*} \left( 2\right) \left( 1\right) a_{2}x^{0}+2a_{0}x^{0}+\sum\limits_{k=1}^{\infty }( k+2) (k+1)a_{k+2}x^{k}+\sum\limits_{k=1}^{\infty }ka_{k}x^{k}+\sum\limits_{k=1}^{\infty }2a_{k}x^{k}& =&0 \\[4pt] 2a_{2}+2a_{0}+\sum\limits_{k=1}^{\infty }\left[ (k+2)(k+1)a_{k+2}+(k+2) a_{k}\right] x^{k}& =&0 \end{eqnarray*}

Equating the coefficients of corresponding powers of x (the coefficients on the right side are all 0), we have \begin{eqnarray*} \hspace{-5pt}2a_{2}+2a_{0}&=&0 \hspace{20pt}(n+2)(n+1)a_{n+2}+(n+2)a_{n}=0 & \\[2pt] a_{2}&=&-a_{0}\quad\hspace{140pt} a_{n+2}=-\dfrac{a_{n}}{(n+1)}\quad n=1,2,3,\ldots \end{eqnarray*}

If n=1, we find a_3=-\frac{a_1}{2}. Then we use the recursion formula on the left to obtain all of the coefficients in terms of a_{0} and a_{1}. That is, \begin{equation*} \begin{array}{@{}l@{\quad\ \ \ }l@{\quad\ \ \ }l@{\quad\ \ \ }l@{}} a_{2}= -a_{0} & a_{4}=-\dfrac{a_{2}}{3}=\dfrac{a_{0}}{3} & a_{6}=-\dfrac{a_{4}}{5}=-\dfrac{a_{0}}{3\cdot 5} & a_{8}=-\dfrac{a_{6}}{7}=\dfrac{a_{0}}{3\cdot 5\cdot 7}\\ a_{3}=-\dfrac{a_{1}}{2} & a_{5}=-\dfrac{a_{3}}{4}=\dfrac{a_{1}}{2\cdot 4} & a_{7}=-\dfrac{a_{5}}{6}=-\dfrac{a_{1}}{2\cdot 4\cdot 6} \end{array} \end{equation*}

and so on. Since a_{0} and a_{1} can be chosen arbitrarily, the power series representation for the general solution is \begin{eqnarray*} y(x)&=&a_{0}\!\left( 1-x^{2}+\frac{x^{4}}{3}-\frac{x^{6}}{3\cdot 5}+\frac{x^{8}}{3\cdot 5\cdot 7}-\cdots \right)\\[5pt] &&+\,a_{1}\!\left( x-\frac{x^{3}}{2}+\frac{x^{5}}{2\cdot 4}-\frac{x^{7}}{2\cdot 4\cdot 6}+\cdots \right) \end{eqnarray*}

NOW WORK

Problem 7.

NEED TO REVIEW?

Maclaurin series are discussed in Section 8.9, pp. 611-614

A second type of series solution method involves a differential equation with initial conditions and makes use of a Maclaurin series.

EXAMPLE 4Using a Maclaurin Series to Solve a Linear Differential Equation

  1. (a) Use a Maclaurin series to find the solution of
    \begin{equation} y^{\prime \prime} =x^{2}y+e^{x}y^{\prime} \tag{3} \end{equation}
    given the initial conditions y(0)=1 and y^{\prime} (0)=1.
  2. (b) Use the first five terms of the series to approximate values of y=y( x) for 0\leq x\leq 1.
  3. (c) Use a numeric differential equation solver with 10,000 equally spaced numbers in the interval [ 0,1] to solve the differential equation in (a). Then construct a table showing the values of y for x=0,0.1,0.2,\ldots,0.9,1.

Solution (a) We assume that the solution of the differential equation is given by the Maclaurin series y(x)=\sum_{k=0}^{\infty }\frac{y^{(k)}(0)}{k!}x^{n}=y(0)+y^{\prime} ( 0) x+\frac{y^{\prime \prime} (0)}{2!}x^{2}+\frac{y^{\prime \prime \prime} (0)}{3!}x^{3}+\cdots

1093

We substitute the initial conditions, y(0)=1 and y^{\prime} (0)=1, into (3). Then \begin{equation*} y^{\prime \prime} ( 0) =0^{2}\cdot 1+e^{0}\cdot 1=1\qquad {\color{#0066A7}{\hbox{\(x=0; y(0) =1; y^{\prime} (0) =1; y^{\prime \prime}(x)= x^{2}y + e^{x}y^{\prime}\)}}} \end{equation*}

Now we differentiate y^{\prime \prime} =x^{2}y+e^{x}y^{\prime} with respect to x to find y^{\prime \prime \prime} ( 0) . \begin{eqnarray*} y^{\prime \prime \prime} ( x) &=&( x^{2}y^{\prime} +2xy) +( e^{x}y^{\prime \prime} +e^{x}y^{\prime}) \\[3pt] y^{\prime \prime \prime} ( 0) &=& ( 0^{2}\cdot 1+2\cdot 0\cdot 1) +( e^{0}\cdot 1+e^{0}\cdot 1) =2 \end{eqnarray*}

We continue differentiating and evaluating the derivative at x=0. \begin{eqnarray*} y^{(4)}( x) & =&x^{2}y^{\prime \prime} +4xy^{\prime} +2y+e^{x}y^{\prime \prime \prime} +2e^{x}y^{\prime \prime} +e^{x}y^{\prime} \\[3pt] y^{(4)}(0)& =&0^{2}\cdot 1+4\cdot 0\cdot 1+2\cdot 1+e^{0}\cdot 2+2\cdot e^{0}\cdot 1+e^{0}\cdot 1=7 \end{eqnarray*}

and so on. The Maclaurin series then becomes \begin{equation*} y(x)=1+x+\frac{1}{2!}x^{2}+\frac{2}{3!}x^{3}+\frac{7}{4!}x^{4}+\cdots \end{equation*}

(b) We construct Table 1 that uses the first five terms of the series to approximate select values of y in the interval 0\leq x\leq 1.

Table 1: TABLE 1 Maclaurin Series Approximation of y Using Five Terms of the Series
x 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
y 1.0 1.1054 1.2231 1.3564 1.5088 1.6849 1.8898 2.1294 2.4101 2.7394 3.125

(c) Using every thousandth term from the numeric solution, we construct Table 2.

Table 2: TABLE 2 Select Values of y Using a Numeric Equation Solver
x 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
y 1.0 1.1054 1.2232 \ 1.3569 1.5114 1.6933 1.9127 2.1836 2.5271 2.9747 3.5752

From the results of Tables 1 and 2, we can see how the five-term approximation to the series solution of the differential equation deteriorates as we move away from the center of convergence 0.

NOW WORK

Problem 15.