A Generalization of
Fermat’s Last Theorem:
The Beal Conjecture and
R. Daniel Mauldin
ndrew Beal is a Dallas banker who The prize. Andrew Beal is very generously of-
has a general interest in mathemat- fering a prize of $5,000 for the solution of this
ics and its status within our culture. problem. The value of the prize will increase by
He also has a personal interest in the $5,000 per year up to $50,000 until it is solved. The
discipline. In fact, he has formulated prize committee consists of Charles Fefferman,
a conjecture in number theory on which he has Ron Graham, and R. Daniel Mauldin, who will act
been working for several years. It is remarkable that as the chair of the committee. All proposed solu-
occasionally someone working in isolation and tions and inquiries about the prize should be sent
with no connections to the mathematical world for-
mulates a problem so close to current research ac-
The abc conjecture. During the 1980s a con-
jectured diophantine inequality, the “abc conjec-
The Beal Conjecture ture”, with many applications was formulated by
Masser, Oesterle, and Szpiro. A survey of this idea
Let A, B, C, x, y , and z be positive integers with has been given by Lang  and an elementary dis-
x, y, z > 2. If Ax + B y = C z , then A, B, and C have cussion by Goldfeld . This inequality can be
a common factor. stated in very simple terms, and it can be applied
Or, slightly restated: to Beal’s problem. To state the abc conjecture, let
The equation Ax + B y = C z has no solution in us say that if a, b, and c are positive integers, then
positive integers A, B, C, x, y , and z with x, y, and N(a,b,c) denotes the square free part of the prod-
z at least 3 and A, B, and C coprime. uct abc . In other words, N(a,b,c) is the product of
It turns out that very similar conjectures have the prime divisors of a, b, and c with each divisor
been made over the years. In fact, Brun in his 1914 counted only once. The abc conjecture can be for-
paper states several similar problems . How- mulated as follows:
ever, it is very timely that this problem be raised
now, since Fermat’s Last Theorem has just recently For each > 0 , there is a constant µ > 1 such that
been proved (or re-proved) by Wiles . Some of if a and b are relatively prime (or coprime) and c
the significant advances made on some problems = a+b, then
closely related to the prize problem by Darmon and
Granville  are indicated below. Darmon and max(|a|, |b|, |c|) ≤ µN(a, b, c)1+ .
Granville in their article also discuss some related
Now let us show that if the abc conjecture holds,
conjectures along this line and provide many rel-
then there are no solutions to the prize problem
when the exponents are large enough.
Let k = log µ/log 2 + (3 + 3 ) . Let min(x, y, z) >
R. Daniel Mauldin is Regents Professor of mathematics at
the University of North Texas, Denton, TX. His e-mail ad- k . Assume A, B, and C are positive integers with
dresses are firstname.lastname@example.org and mauldin@ A and B relatively prime and such that
dynamics.math.unt.edu. Ax + B y = C z . Setting a = Ax and b = B y , we have
1436 NOTICES OF THE AMS VOLUME 44, NUMBER 11
c = a + b = C z . From the abc conjecture and the 657 , 92623 + 153122832 = 1137 , 438 + 962223 =
fact that N(Ax , B y , C z ) ≤ ABC , we have 300429072 , 338 + 15490342 = 156133 . The last
max(Ax , B y , C z ) ≤ µ(ABC)1+ . five big solutions were found by Beukers and Zagier.
Recently Darmon and Merel have shown that
If max(A, B, C) = A , then we would have there are no coprime solutions with exponents
Ax ≤ µA3+3 (x, x, 3) with x ≥ 3 .
Acknowledgment. Since I am not an expert in
or log µ this field, I would like to thank Andrew Granville
x≤ + 3 + 3 ≤ k,
log A and Richard Guy for their expert help in prepar-
ing this note.
which is not the case. A similar argument for the
other two possibilities for the maximum shows that References
our original assumption is impossible.  V. Brun, Über hypothesenbildung, Arc. Math. Naturv-
Next let us give an explicit version of the abc con- idenskab 34 (1914), 1–14.
jecture: If a and b are coprime positive integers and  H. Darmon and A. Granville, On the equations
c = a+b, then c ≤ (N(a, b, c))2. Let us see what this z m = F(x, y) and Axp + By q = cZ r , Bull. London
Math. Soc. 27 (1995), 513–543.
implies for the prize problem. Suppose  H. Darmon and L. Merel, Winding quotients and
Ax + B y = C z , with x ≤ y ≤ z. Again, since Ax and some variants of Fermat’s Last Theorem, preprint.
B y are coprime,  D. Goldfeld, Beyond the Last Theorem, Math Hori-
C z ≤ (N(Ax B y C z ))2 ≤ (ABC)2 < C 2(z/x+z/y+1) . zons (September 1996), 26–31, 34.
 S. Lang, Old and new conjectured diophantine in-
So 1/2 < 1/x + 1/y + 1/z . Since x, y , and z are equalities, Bull. Amer. Math. Soc. 23 (1990), 37–75.
greater than 2, we have the following possibilities  A. Wiles, Modular elliptic curves and Fermat’s Last
Theorem, Ann. Math. 141 (1995), 443–551.
for (x,y,z): (3, 3, z > 3), (3, 4, z ≥ 4), (3, 5, z ≥ 5),
(3, 6, z ≥ 7), (4, 4, z ≥ 5) , and a finite list of other
There are only finitely many possible solu-
tions. In 1995 Darmon and Granville  showed
that if the positive integers x, y, and z are such that Andrew Beal is a number theory enthusiast re-
1/x + 1/y + 1/z < 1 , then there are only finitely siding in Dallas, Texas. He grew up in Lansing,
many triples of coprime integers A, B, C satisfying Michigan, and attended Michigan State Uni-
Ax + B y = C z . Since each of x, y, and z is greater versity. He has a particular interest in some of
than 2, then 1/x + 1/y + 1/z < 1 unless Fermat’s work and has spent many, many
x = y = z = 3. But Euler and possibly Fermat knew hours thinking about Fermat’s Last Theorem.
there are no solutions in this case. So for each He believes that Fermat did possess a relatively
triple x, y, and z , all greater than 2, there can be simple non-geometry-based proof for FLT, and
he continues to search for it. He also believes
only finitely many solutions to the diophantine
that Fermat had a method of solution for Pell’s
equation Ax + B y = C z .
equation that remains unknown and that was
Related problems. What happens if it is only re-
a function of the squares whose sum equals the
quired that x, y, and z be ≥ 2 and at least one of
them is greater than 2 and A, B, and C are coprime?
Andrew is forty-four years old. He and his
There is a detailed analysis in  of those cases
wife, Simona, have five children. He is the
where x, y, z ≥ 2 and 1/x + 1/y + 1/z > 1 .
founder/chairman/owner of Beal Bank, Dal-
What happens if we require only that
las’s largest locally owned bank. He is also the
1/x + 1/y + 1/z < 1 and A, B, and C are coprime?
recent founder/CEO/owner of Beal Aerospace,
This problem is also discussed by Darmon and
which is designing and building a next-gener-
Granville. In fact, they have formulated
ation rocket for launching satellites into earth
The Fermat-Catalan Conjecture. There are only orbits.
finitely many triples of coprime integer powers Beal Bank, Toyota, and the Dallas Morning
xp , y q , z r for which News are the primary sponsors of the Dallas
1 1 1 Regional Science and Engineering Fair. Beal
xp + y q = z r with + + < 1.
p q r Bank is also a primary sponsor of the Dallas
Area Odyssey of the Mind Competition. An-
So far, as mentioned in , ten solutions have been drew Beal has been a major benefactor for the
found. The first five are small solutions. They are mathematics program at the University of
1 + 23 = 32 , 25 + 72 = 34 , 73 + 132 = 29 , 27 + 173 = North Texas through his substantial scholar-
712 , 35 + 114 = 1222 . ships for graduate students and for students in
Also five large solutions have been found: the Texas Academy of Mathematics and Science.
177 + 762713 = 210639282 , 14143 + 22134592 =
DECEMBER 1997 NOTICES OF THE AMS 1437