L'Hopital's Rule in a Poisson Derivation

Author(s): William P. Cooke

Source: The American Mathematical Monthly, Vol. 95, No. 3 (Mar., 1988), pp. 253-254
Published by: Mathematical Association of America
Stable URL: http://www.jstor.org/stable/2323629
Accessed: 20-06-2016 15:30 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

Mathematical Association of America is collaborating with JSTOR to digitize, preserve and extend access
to The American Mathematical Monthly

This content downloaded from 128.206.9.138 on Mon, 20 Jun 2016 15:30:52 UTC
All use subject to http://about.jstor.org/terms
 The Teaching of Mathematics

EDITED BY JOAN P. HUTCHINSON AND STAN WAGON

L'Hopital's Rule in a Poisson Derivation

WILLIAM P. COOKE
Departmenit of Mathemnatics and Physical Sciences, West Texas State Universitv, Canvon, TX 79016

Elementary derivations of the Poisson distribution have two basic forms, which
we might call direct and indirect. Given assumptions of independence and negligi-
bility of probabilities for more than one arrival in a small time interval, the
assumption that the probability of an arrival in a small time interval is proportional
to the length of the interval implies a direct derivation (cf. [1]). When we just assume
that the arrival probability depends on the length of the interval, then we must first
develop the exponential density of interarrival times-the indirect approach (cf. [4]).
Here L'Hopital's rule (see [3, p. 121]) will be used in the indirect approach to yield a
simple alternative to the usual derivation as shown in Wagner [4].
The direct derivation is more suitable for a first course in probability, where
discrete distributions are usually discussed early. In a course on queuing theory or
operations research, however, this alternate indirect derivation could simplify the
teaching of the topic.
From the assumptions in Wagner [4] the exponential density of interarrival times
follows:

af(t) = exp(-t), X > 0, t > 0 (1)

f) elsewhere.

Define Pn(t) as

Pn(t) = Prob [ n arrivals in the interval (O, t)].

Our independence and negligibility assumptions give

Pn(t + h) = Pn(t)PO(h) + Pn-1(t)P1(h), n = 1,2, ... (2)

if h > 0 is regarded as being very small. Subtracting Pn(t) from both sides of (2),
dividing by h, and taking the limit gives the derivative

P,(t) = lim [{ Pn(t)[Po(h) - 1] + Pn-1(t)PI(h)}/h], n = 1,2,. (3)

h-0O

From (1) we obtain

Po(h) = f Xexp(-Xt) dt = exp(-Xh), (4)

and our negligibility assumption gives

P1(h) = 1 - Po(h) = 1 - exp(-Xh).

253

This content downloaded from 128.206.9.138 on Mon, 20 Jun 2016 15:30:52 UTC
All use subject to http://about.jstor.org/terms
 254 NEAL KOBLITZ [March

Then (3) becomes

P,'(t) = P,(t) lim { [exp(-Xh) - ]/h} (5)

+?Pn-1(t) lim { [1 -exp(-Xh)]/h}, nn=1 2.

Each of the two limits in (5) has the indeterminate form 0/0, but one application
of L'Hopital's rule to each limit produces

Pn'(t) = -XPn(t) + XPn-l(t) n = 1, 2,...

Then we can use either mathematical induction or the probability generating

function (cf. [2]), along with (4), to show that

P (t) = [(Xt)nexp(-Xt)]/n!, n = 0,1,2, ... .

the Poisson distribution with pairameter Xt.

REFERENCES

1. J. E. Freund, Mathematical Statistics, 2nd ed., p. 85, Prentice-Hall, 1971.

2. H. A. Taha, Operations Research, 3rd ed., pp. 634-635, Macmillan, 1982.
3. A. E. Taylor, Advanced Calculus, Blaisdell, 1955.
4. H. M. Wagner, Principles of Operations Research, pp. 846-847, Prentice-Hall, 1969.

Problems that Teach the Obvious but Difficult

NEAL KOBLITZ
Department of Mathematics, University of Washington, Seattle, WA 98195

Most of our students are supposed to be learning calculus and other subjects in
order to be able to recognize and apply the concepts in contexts which may be far
removed from the settings that typify textbook problems. The transition is likely to
present obstacles for them. Some simple matters which we take for granted and
generally ignore should, in my opinion, be a major consideration in our design of
exercises. I would like to illustrate by discussing four of my favorite calculus word
problems.

Problem 1. Find the centroid of Nevada.

Here we make a slight simplification, giving Nevada a small piece of Arizona, so

that it becomes a perfect trapezoid. We also assume that 1? is a constant distance
throughout Nevada. The student is given the latitude and longitude of the four
corners of Nevada: 42?N, 120?W; 42?N, 114?W; 39?N, 120?W; 35?N, 114?W. This
problem at first stumps many students because of the absence of any equations and
the unfamiliar form in which the coordinates of the corners are given. The first step
in solving the problem is to decide how to set up the xy-axes. One can choose the

This content downloaded from 128.206.9.138 on Mon, 20 Jun 2016 15:30:52 UTC
All use subject to http://about.jstor.org/terms