Professional Documents
Culture Documents
Diophantine Equationsandthe Pythagorean Theorem
Diophantine Equationsandthe Pythagorean Theorem
net/publication/331399814
CITATIONS READS
0 647
1 author:
Ray Adams
University of Cambridge
45 PUBLICATIONS 339 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Ray Adams on 28 February 2019.
Abstract
This note explores the role of Diophantine equations in order to illuminate the
nature of the Pythagorean Theorem and to clarify why some integer triples are
Pythagorean triples whilst are some are not. The Pythagorean Theorem is based on
a set of Diophantine equations of degree two of the form x2 + y2 = z2. In this form of
equation, x2, y2 and z2 can each represent a Diophantine equation of degree two,
specifically when these Diophantine equations have a numerical value equal to a
squared integer. Different types of Diophantine equations are explored, including
‘complete’ Diophantine equations that always have a numerical value equal to a
squared integer and generated by the Binomial expansion of degree two. ‘Mutilated’
Diophantine equations are produced by cutting one or more terms from a complete
Diophantine equation. ‘Modified’ Diophantine equations are produced by modifying
specific terms of a complete Diophantine equations. Both of the latter two types of
equation lose the relationship with the Binomial expression and very few of them
generate Pythagorean triple. Finally, the conditions under which Pythagorean
triples are achieved. Further work will explore Diophantine equations of higher
degree.
Ray Adams
Introduction
The well known Pythagorean Theorem is based on a set of Diophantine equations of
the form x2 + y2 = z2. In particular, the present focus in on the Pythagorean triples,
where x, y and z are natural numbers (N) such that x.y.z ≠ 0. A set of triples are said
to be primitive when the three numbers are coprime i.e. their greatest common
divisor (gcd) is 1 i.e. gcd(x, y, z) = 1. [1]
Given its long history, it is not surprising that here are many proofs of the
Pythagorean Theorem, (more than 350). [2] As Charles Frohman has remarked in a
paper that motivated and then proved a generalized Pythagorean theorem for
parallelepipeds in Euclidean space that “The Playfair proof of the Pythagorean
theorem is easy to explain, but somehow mysterious.” [3]
The fame of the Pythagorean Theorem should not obscure its importance. As
Stewart and Tall have commented the “Pythagorean theorem is important, not
because it applies to one particular right – angled triangle, but because it applies to
all of them. It thereby expresses a property of the set of all right – angled triangles.”
[4]
Second, both the Pythagorean Theorem and Fermat’s Last Theorem have their feet
firmly planted in highly similar Diophantine equations. For the Pythagorean
Theorem, the set of relevant Diophantine equations are based on x 2 + y2, whilst
Fermat’s Last Theorem invokes the very similar Diophantine equations of the form
xn + yn, where n ≥ 3.
Yet the former theorem proves an equality that x2 + y2 = z2, under specific
conditions, whilst the latter proves an inequality that xn + y ≠ zn, under all conditions.
There is clearly an unexplained rift between the two theorems, one being positive
but the other being negative. This rift does not appear to have been explained by
any current number theories. But the contrast gets still more dramatic, since an
examination of the extant proofs of the Pythagorean Theorem shows that none of
them appear to have any applicability to a proof of Fermat’s Last Theorem.
Conversely, Andrew Wiles’ spectacular proof of Fermat’s Last Theorem apparently
has no implications for a proof of the Pythagorean Theorem, depending instead
upon the relationship between elliptic curves, modular forms and Frey’s curve. [4]
[7] [8] [9]
The present note sets out to provide a new proof of the Pythagorean Theorem. But
why do we need yet another proof of the Pythagorean Theorem? The purpose of
the present proof is to provide one which not only proves the Pythagorean Theorem
but also provides a basis for considering new proofs of Fermat’s Last Theorem.
Whilst the above expansions may be familiar to many, they are provided here to
underline that fact that such expansions are defined by three key characteristics; (a)
they can all be identified as binomial expansions where r =2, (b) such expressions
can only possess specific numbers of terms and (c) they have the numerical value
equal to some integer squared. These three characteristics are important, as will be
used below to explore why some triples of numbers are Pythagorean triples, whilst
others are not.
Of course, it is possible to create other types of Diophantine equations that are not
complete Diophantine equations in the sense used here. For present purposes, it is
useful to name two other groups. First, a mutilated Diophantine equation is one in
which a terms is cut from a complete Diophantine equation. For example, a2 + b2 +
2ab could become b2 + 2ab, thus destroying its relationship to the Binomial
expansion. Such equations cannot be guaranteed to have a numerical value equal
to an integer square. Second, a modified Diophantine is one in which one or more
terms are modified. For example, a2 + b2 + 2ab could become 2a2 + b2 + 6ab, thus
destroying its relationship to the Binomial expansion. Again, such equations
cannot be guaranteed to have a numerical value equal to an integer square.
Obviously not all triples are Pythagorean triples. The limitations of the Pythagorean
triples are becoming clearer and more systematic. When we add two Diophantine
equations, not all such combinations produce a third complete Diophantine
equation of degree two and with a numerical value equal to a squared integer.
Can mutilated and modified Diophantine equations take a numerical value equal to a
squared integer? First consider mutilated Diophantine equations. For example,
compare a2 + b2 + 2ab with b2 + 2ab, where one of the terms has been cut. Can a 2 +
b2 + 2ab = b2 + 2ab? This equality can only hold when a2 equals zero, in which case
b2 + 2ab reduces to b2. Can b2 + 2ab equal a squared integer, courtesy of the
Pythagorean Theorem? This could only be possible if 2ab = x 2, but this would
require that x = √2.√ab, but as this involves √2, an irrational number, then x cannot
be an integer squared. This argument generalises to the other complete
Diophantine equations discussed above. Consider b 2 + nab where n is any integer.
For such equations to take the Pythagorean form of x 2 + b2, then n must be a square
and ab must also be a square i.e. a=b. But we have already found above that when
a=b, the result cannot be a Pythagorean triple. But what about a 2 + b2? Can this
expression be equal to an integer squared? This is a case of the Pythagorean
theorem and is discussed below.
Next, it should be possible to demonstrate that the sum of two different complete
Diophantine equations of second degree, as defined with above, can produce a third
Diophantine equation of second degree: i.e. a 2 + b2 = c2 but only under specific
circumstances that can be well defined, where a, b and c can be represented by
complete Diophantine equations of second degree [10].
The first and most obvious question is to ask if a Pythagorean triple can be based
on two equal numbers?
Starting with the expression a2 + b2 = c2
Consider what happens if a2 = b2.
Then a2 + b2 = 2a2 or, if you prefer, a2 + b2 = 2b2.
Thus a2 + b2 = 2a2 = c2.
In this case, c2 equals 2a2 and 2b2 and so c equals ±a√2 or ±b√2.
Note that √2 is an irrational number.
If so, then c can only be a rational number if a = b = √2, so that c becomes 2.
In this case a2 + b2 = c2 becomes 2 + 2 = 4 and c = √4 = 2. In this final case, a 2 and b2
are not integer squares whilst c2 is a square.
What if a and b are equal to a multiple of √2, so that a = b = d√2, where d is an
integer constant?
In this case, a2 + b2 = c2 becomes 2d2 + 2d2 = 4d2 and this becomes d2 + d2 = 2d2
Thus, as c2 = 2d2 then c = ±d√2. But √2 is an irrational number and cannot be used as
a basis for a set of integer Pythagorean triples.
Thus, when a = b, it is not possible to generate a Pythagorean triple in which two of
the terms are identical, as any such construction will involve the irrational number
√2 or some simple multiple of √2.
Going back to: a2 + b2 = c2. Perhaps the above result only applies when a and b
share a common component. So it is important to consider two new expressions
that do not share a common component.
In this case, let a = (x + y), b = (w + z) i.e. two demonstrably different numbers
without a common component (see Case Two above, where the two numbers do
have a common component.)
Also c = (t + v)
When a2 + b2 = c2
c2 = (x + y)2 + (w + z)2
= (x2 + 2xy + y2) + (w2 + 2wz + z2)
But c = (t + v)
c2 = (t2 + 2tv + v2) = (x2 + 2xy + y2) + (w2 + 2wz + z2)
So a2 + b2 = c2 if the above equality applies.
In terms of the constituent components, then:
(t2) = (x2) + (w2) and
(v2) = (y2) + (z2) and
(2tv) = (2xy) + (2wz) or
(t2 + v2) = (x2 + y2) + (w2 + z2)
REFERENCES
[1] Gareth A. Jones and J. Mary Jones, Elementary Number Theory, London, Springer – Verlag London, (2008).
[2] John C. Sparks, The Pythagorean Theorem Crown Jewel of Mathematics, Bloomington, Indiana, Author
House (2008).
[3] Charles Frohman, The Full Pythagorean Theorem, arXiv 001.0201v1 [math.HO] Retrieved February 23, 2019
from the arXiv database.
[4] Ian Stewart and David Tall (2015). The Foundations of Mathematics. Oxford. Oxford University Press.
[5] Sparling, George A. J. (2006). "Spacetime is spinorial; new dimensions are timelike". arXiv:gr-qc/0610068.
Retrieved February 23, 2019 from the arXiv database.
[6] Alf van der Poorten (1996), Notes on Fermat’s Last Theorem, New York, NY, John Wiley & Sons
[7] Kenneth Ribet (1990). On modular representations of Gal (Q / Q) arising from modular forms. Invent. Math,
100, 431 – 476.
[8] A. Wiles. Modular elliptic curves and Fermat’s Last Theorem. Ann. Of Math. 141, 443 – 551. (1995)
[9] R. l. Taylor and A. Wiles Ring theoretic properties of certain Hecke algebras, Ann. Of Math, 141 553 – 572.
(1995)