346 CHAPTER 9 CALCULUS
other words, complete "local" information yields complete global information! This
is worth pondering.
In practice, it is not always necessary to use (8). As long as you know (or suspect)
that the series exists, you can opportunistically extract terms of a series.
Expand - 2 _
1
- into a power series about x = o.
x + 1
Solution: We simply use the geometric series tool (see page 133):
1
and thus
(^1 246)
--= 1 - x +x -x + ....
x^2 + 1
Example 9.4.4 Expand � into a power series about x = o.
Solution: Just substitute t = x^2 into the familiar series
t^2 t^3
et = l+t +-+-+···
2! 3!
You may wonder about these last two examples, asking, "Yes, we got a power
series, but how do we know that we actually got the Taylor series that we would have
gotten from (8)?" Once again, don't worry, for the power series expansion is unique.
The essential reason is just a generalization of the "derivative is the best linear approxi
mation" idea mentioned on page 330. For example, let P 2 (X) denote the second-degree
Taylor polynomial for f(x) about x = a. We claim that P 2 (X) is the best quadratic
approximation to f(x) for the simple reason that
lim
f(x) -P 2 (X)
= 0,
x----> a (x - a)^2
while if Q(x) -I-P 2 (X) is any other quadratic polynomial, then
lim
f(x) - Q(x)
-I-O.
x ----> a (x - a?
(9)
Thus P 2 (X) is the only quadratic function that agrees with f(x),f'(x), and f"(x) when
x=a.
This is one reason why power series are so important. Not only are they easy to
manipulate, but they provide "ideal" information about the way functions grow.
Eulerian Mathematics
In the last few pages, we have been deliberately cavalier about rigor, partly because
the technical issues involved are quite difficult, but mostly because we feel that too
much attention to rigor and technical issues can inhibit creative thinking, especially at
two times: