494
Y
1
01 1 z
Figure 9.263
e3 4 X
correct to 5 or 10 decimal places
Exponential and logarithmic functions
which we can now obtain very easily
by evaluating the integrals. Each
of these formulas actually deter-
mines e, and the second one says
that e is so determined that the area
-4 of the region shaded in Figure
9.263 is 1. Many other formulas
involving e will appear later, and one
of them will enable us to calculate e
with surprisingly little effort. Mean-
while, we can remark that many persons remember the 16D (16-decimal,
but only 15 after the decimal point) approximation to e by mentally
grouping digits in the form
(9.264) e = 2.7 1828 1828 45 90 45
so we can visualize the repeated 1828 followed by 45 and twice 45
and 45.
The number e is the natural base of exponentials and logarithms.
Other bases sometimes appear. The base 10 is used when we give a
number n and say that "there are 10^ atoms in the universe," but not
even Eddington suggested that this should be differentiated with respect
to n. We never differentiate or integrate a is e. If, as never
or rarely happens outside misguided examinations in calculus, we are
called upon to differentiate or integrate a1:1 where a e, we write
(9.265) ak1z = ekta
and take logarithms with base e to obtain kix log a = k2x and hence
k2 = kl log a. We then work with ek'z instead of aI. z.
Our theory of exponentials and logarithms enables us to provide the
promised proof of the power formula which, for convenience of reference,
we put in a theorem.
Theorem 9.27 If n is a constant, integer or not, then
dxn = nxn-1
when x>0.
To prove this theorem, let y = xn and take logarithms to obtain
log y = n log x. Since log y is differentiable, we can, because y= eb0y,
conclude that y itself is differentiable andcan use the chain rule to obtain
(1/y)dy/dx = n/x so dy/dx= nx". This proves Theorem 9.27. We
are now able to prove the last of the basic limit theorems, Theorem 3.288,
which we now restate.