Search All of the Math Forum:

Views expressed in these public forums are not endorsed by NCTM or The Math Forum.

Notice: We are no longer accepting new posts, but the forums will continue to be readable.

Topic: definitions of exp/log
Replies: 0

 John Hagood Posts: 8 Registered: 12/6/04
definitions of exp/log
Posted: Oct 18, 2001 2:30 PM

Here is another way to definite exp(x), appropriate for first year calculus
or early in advanced calculus (prior to developing the derivative).

(1) Start by extending a^x from the rationals to a continuous function on
the reals.

(2) Consider the functions x/a^x for constant a>1. In first year calculus,
graphical evidence suggests that each such function has exactly one
maximum, and that location of the maximum decreases as x increases. Take e
to be the value of a such that the maximum occurs at x = 1.

(3) It is now an easy and entertaining exercise to obtain quite rigorously
the derivative of e^x. Furthermore, the proof of the traditional limit
lim_{n- > infinity}(1+1/n)^n = e is pleasantly quick as well.

This will appear in the November 2001 issue of College Mathematics Journal.

For a more advance class - i.e., Introduction to Analysis or Advance
Calculus - the details of (2) can be filled in as follows (these are
reasonable exercises for students)
(2a) For each a>1, x -> x/a^x attains a maximum. The extreme value theorem
will get this, but if that is not yet in hand, it is an easy exercise to
show that for each a>1 here is a cut (L,R) of the rationals such that x/a^x
increases strictly on L and decreases strictly on R . This approach also
shows that the location of the maximum is unique.
(2b) The maximum of x/a^{kx} occurs at x_a/k where x_a is the location of
the maximum of x/a^x.
(2c) There is a unique number e such that x/e^x has a maximum at x =
1. (The uniqueness can be proved using (2b) and the fact that for a given
a>1, x->a^x maps onto the positive reals. The intermediate value theorem
makes this immediate, but it is not needed in light of elementary
properties of a^x over the rationals (x->a^x is strictly increasing and
unbounded above, a^t-a^s <= a^{t+1}(t-s) for s<t, and the extension in (1)
) if one prefers to avoid it.

An alternative approach not mentioned already in this thread appears in
(John Kemeny, The exponential function, Amer. Math. Monthly, 64 (1957),
158-160).

John W. Hagood
Department of Mathematics and Statistics
PO Box 5717
Northern Arizona University
Flagstaff, AZ 86011-5717
Phone: 520-523-6879
Fax: 520-523-5847

------------------------------------------------------------

-HOW TO UNSUBSCRIBE

To UNSUBSCRIBE from the calc-reform mailing list,
send mail to:

majordomo@ams.org

with the following in the message body: