International

Baccalaureate
Extended essay cover
Candidates must complete this page and then give this cover and their final version of the extended essay to their supervisor.
Candidate session number
Candidate
name
School number
School name
Examination session (May or November) Year
Diploma Programme subject in which this extended essay is registered:
2.0l2
(For an extended essay in the area of languages, state the language and whether it is group 1 or group 2.)
e===
Lo-st- - 1 heoreJVt ?
Candidate's declaration
If this declaration is not signed by the candidate the extended essay will not be assessed.
The extended essay I am submitting is my own work (apart from guidance allowed by the International
Baccalaureate).
I have acknowledged each use of the words, graphics or ideas of another person, whether written, oral or
visual.
I am aware that the word limit for all extended essays is 4000 words and that examiners are not required
to read beyond this limit
This is the final version of my extended essay.
Candidate's signature: Date: OI/J2/7CJtl
IB Cardiff use only: A: B:
- ---
International Baccalaureate, Peterson House,
Ma lthnli<:.P A.vPOIIP r.arrliff r,::dp r.arr!iff F ? ~ Rr,t W a lo:>c:. llnitPr! Kinnr!nm
--- ----
Supervisor's report
The supervisor must complete the report below and then give the final version of the extended essay, with this
cover attached, to the Diploma Programme coordinator. The supervisor must sign this report; otherwise the
extended essay will not be assessed and may be returned to the school.
Name of supervisor (CAPITAL letters)____,
Comments
Please comment, as appropriate, on the candidate's performance, the context in which the candidate undertook
the research for the extended essay, any difficulties encountered and how these were overcome (see page 13 of
the extended essay guide). The concluding interview (viva voce) may provide useful mformation. These
comments can help the examiner award a level for criterion K (holistic judgment). Do not comment on any
adverse personal circumstances that may have affected the candidate If the amount of time spent with the
candidate was zero, you must explain this, in particular how it was then possible to authenticate the essay as the
candidate's own work. You may attach an additional sheet if there is insufficient space here.
---:;r /....,_
5
+ L.-<--JZ- d .r ovd! -r-
( -"-5 e.-.--e-1- o! / .A"1(!.-/ J.,-j I cJ
12
u av h d
f ; I r4 !ft'rlf f [ srJ'v v,-s- JcJY\#
Q'(Z o- \o..-1 I (_;,',/' c_..c (/ / c=- --41 a- 7
J j c->S......e e ./6 -
!o
1
f c c./ rv jJ .
1 s _.._ rvl&./ L _,(&!
If ' /f; rT
/-rL If tJ."' flU
r (, n I ' L L '--> p _QsLJ
q_.( l} ifLv }._ "--'--'--0>.- h J ' ( rfL
f f '- rC "--'' i L [J
I have read the final version of the extended essay that will be submitted to the examiner.
To the best of my knowledge, the extended essay is the authentic work of the candidate.
I spent I I 0 I hours with the candidate discussing the progress of the extended essay.
Supervisor's signatur
o2 .)2....
Assessment form (for examiner use only)
Candidate session number
Achievement level
First Second
examiner maximum examiner
Assessment criteria A research question
rn
2
D
B introduction
@]
2
D
c investigation
[1]
4
D
D knowledge and understanding
CD
4
D
E reasoned argument
@]
4
D
F analysis and evaluation
[]
4
D
G use of subject language

4
D
H conclusion
[I] 2
D
formal presentation
m
4
D
J abstract
[I]
2
D
K holistic judgment
[I]
4
D
Total out of 36
CJ
of first examiner. _
TAL letters)
of second examiner ---------------- ---
TAL letters)
Examiner number:
Exammer number: ___ _
Mathematics Extended Essay
TO WHAT EXTENT WAS THE
METHOD OF INFINITE DESCENT
CONCLUSIVE IN PROVING FERMAT'S
LAST THEOREM?
Candidate Number
December 2011
Session: May 2012
Word Count: 3,704
Essay Supervisor:
1
Abstract
The research question for this essay is "To what extent was t he
method of infinite descent conclusive in proving Fermat's Last
Theorem?" Fermat's Last Theorem is one of the greatest problems
ever encountered in mathematics, and it became a real fascination of
mine after reading about it. I discovered that the method of infinite
I,\ . ?
descent was a common appearance among these books, so I decided to ..., -......v
investigate how conclusive it was in proving Fermat's Last Theorem and {)
- - I I U \- \o,.e UtP\
thus emerged my research question. f1. 4 ,.. ... tv,r
This essay focuses on the proofs used by Euler, Dirichlet and Kum- \-
mer, and how the method of infinite descent is used in each of them. The w(T"fcA:w.
proofs themselves are not focused on in detail, only a brief summary of rt <' c\...
how the proof works can be given seeing as how the proofs are exten-
sive, alt hough beautifully elegant. Euler 's proof of n = 3 was the first
looked at , as it presents the template for the method of infinite descent.
Dirichlet's and Kummer's proofs were then investigated afterwards, ob-
serving how the meLhod of infinite descent evolved to work for different
exponenLs of Fermat's Last Theorem.


The method that Ernst Kummer used to prove Fermat's Last The-
orem for specific exponents was at first made for regular primes, but
was then adapted for irregular primes. This led to Fermat's Last The-
orem being proved for all prime exponent s up to 4 million with the aiel
of computers. This might seem conclusive enough, since 4 million is a
relatively large number , however no one knows for sure that it might not
work for prime numbers above 5 million. Therefore the conclusion is that
the method of infinite descent is useful in giving,.an idea as to whether
Fermat's Last Theorem is true or not, will(neY;p D. J
_ '"" "w CQN. \)-V. kc.. :;) , (..v _..,._ \..t ci
Word Count: 295
Contents
Abstract
1 Introduction to Fermat's Last Theorem (FLT)
1.1 Pythagoras Theorem . . . . . . . . . . . .
1.2 The beginning of Fermat's Last Theorem .
2 Proofs of Fermat's Last Theorem
2.1 Euler and the method of infinite descent
2.1.1 Euler's proof for n = 3 ..... .
2.2 Dirichlet, Sophie Germain's Theorem, proof by contradic-
1
1
1
2
2
3
tion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2. 1 Dirichlet's proof for n = 5 . . . . . . . . . . . . 6
2.3 Lame and Kwnrncr: Cyclotomic Integers to prove FLT 7
2.3.1 Lame's idea of a final proof . . . . 8
2.3.2 Liouville's discovery of Lame's fiaw 9
2.3.3 Kummer and cyclotomic integers 9
3 Conclusion 13
4 Bibliography
15
ll
1 Introduction to Fermat's Last Theorem
(FLT)
1.1 Pythagoras Theorem
Fermat's Last Theorem is lmown to be one of the greatest mathematical
problems the world has ever encountered. Its simple look is deceiving,
troubling mathematicians for 350 years untH Andrew Wiles cracked it in
1995. The histmy of t his problem starts in the sixth century B.C. with
Pythagoras of Samos. During t his essay I shall be referring to Fermat's
Last Theorem as FLT.
As Simon Singh said in his book 'Fermat's Enigma', "Usually half the
difficulty in a mathematics problem is underst anding the question, but
in this case it was straightforward- ... "
1
The problem of Fermat's Last
Theorem looks very familiar to most people as it is based on Pythagoras'
Thf>..orem, a theorem engraved in millions of people's brain: (
"u].t \S . ?
( x2 +y2 = z2 / JC l' :::>
Pythagoras of Samos and his brotherhood in Croton, Italy, managed
to find a very elegant proof for tllis, one of their biggest successes, su bse-
q uently leading t o one of the gr$test mathematical problem of all t ime.
\. ?
1.2 The beginning of Fermat's Last Theorem
Inevitably, t his led to mathematicians asking themselves what would
happen if t he power in the equation was changed from ' 2' to '3' so t hat.
it looked like this:
I Singh J 998, p.6
1
. .
No one knew that they had unleashed a monster of an equation, and
although finding to Py}hagoras thcore1p, also called Pythagorean
triples, was relatively easy, finding solutions to this new modified ver-
sion to be imposib]e. If the power in t he equation is changed
to an even higher number, finding solutions appears to be equally im-
possible. For centuries, mathematicians tried to find solutions to these
modifications of Pythagoras' theorem with no success. Tbis led the great
seventeenth-century French mathematician Pierre de Fermat to believe
that the reason nobody could find any solutions was because there were
no solutions. In the margin of bis copy of Diophantus' Arithmetica, he
noted his observations:

1 It is impossible for a cube to be written as a sum of two
'a power to be written as the sum of two fmtrth powers
or, in general, for any numbe1 which is a power greater than the second
to be written as a sum of two like powers. r\ v
This gave birth to an adventure that would last 350 years to
Fermat's Last Theorem with some successes but many failures.
success would come with Leonhard Euler in the 18th century when he
discovered a proof for the case where the power of the equatiou is 3.
s
2 Proofs of Fermat's Last Theorem (
2.1 Euler and the method of infinite descent
The method of infinite descent was the first method used to try and solve
Fermat's Last Theorem for specific exponents, the first ones being n = 4
and later on n = 3. This method is a particular form of proof by con-
tradiction (sec further) and it is seen in Fermat's jottings in Arithmetica
by Diophantus. Fermat used this method to prove the case for n = 4,
- - 1J
2
and this is the most complete calculation by Fermat he ever committed
j ,1 . \-'- vr. Ci..-, M \.,. <- \
The method of infinite descent is very simple to understand. You
begin by assuming that there is a solution to Fermat's Last Theorem for
n=4:


r.o tT ,
After examining the properties of this solution, you can that if
this solution does exist, t hen there must be a smaller solution (X
2
, Y
2
, Z
2
).
If you then examine this solution, you can find an even smaller solution
(X
3
, Y3 , Z
3
). This can be done infinitely many times, finding infinitely
many smaller solutions. However, the solutions to Fermat's Last Theo-
rem must be whole numbers, therefore you cannot have infinitely many
smaller solutions tha whole numbers so you reject the asswnpt ion
that there is a {l Fermat's Last :Theorem for n = 4.
Leouhard Euler saw this proof by Fermat and used this as his starting
point for finding a general proof to prove all other cases of FLT. Euler
started by attempting to prove FLT for n = 3. He adapted thic; method
of infinite descent used by Fermat and was able to prove it . This was the
first major breakthrough on FLT since Fermat himself, and it motivated
more mathematicians to start working on it. In the following section
I will show how Euler proved that there were no solutions to Fermat's
Last Theorem for n = 3 using the method of infinite descent.
2.1.1 Euler's proof for n = 3
2
The first thing Euler did was to assume there was a solution for Fermat's
Last theorem for the case where n = 3.
An important part of the whole proof is showing that different num-
2
Eel wards 2000
3
r,J-
bers are coprime:) The first time he does tlus is when he shows t hat
y, z are coprime. To do this he proved that if an integer d divides 2
numbers in Fermat's Last Theorem, then divides the nth power of
t he third. After that he proved that if dn divides xn then d divides t hat
number x. By doing this he showed that if 2 of the numbers in FLT have
a greatest common divisor larger than 1, then this number also divides
the 3rd number. Therefore you can divide all of them by that number
and keep doing this until they are coprime.
Euler then went on to show that if x, y, z are coprime, then there
exist two integers p, q such that:
1. gcd(p, q) = 1
2. p, q are positive
3. p, q have opposite parity (one is odd, one is even)
4. 2p(p
2
+ 3q
2
) is a cube
He later proved that t he greatest common divisor of 2p, p
2
+ 3q
2
can
only be 1 or 3. Euler did this by showing that the greatest common
divisor cannot be 2 because p
2
+ 3q
2
is odd and it can't be any prime
larger than 3 by showing that it would divide both p and q, going against
p aud q being coprime.
By doing this, Euler cau show that (2a)(a- 3b)(a + 3b) is a cube
because (2a)(a- 3b)(a+ 3b) = 2a
3
- 18ab
2
= 2p (2p is a cube). Again he
shows that 2a, a - 3b, a+ 3b are coprime so that each of them is a cube.
And thus he found a new solution to Fermat 's Last Theorem for n = 3
since A
3
= 2a = (a + 3b) + (a - 3b) = B
2
+ 0
2
. He then showed that this
new solution is smaller than the previous solution. This argument can
be done infinitely many times and so there is a case of infinite descent .
Since the solutions to Fermat's Last Theorem must be whole numbers,
this is contradictory so he rejects the initial assumption that there exists
a solution to Fermat's Last Theorem for n = 3.
3Coprime: Two integers a, bare said to be coprime if t heir greatest common divisor
is 1 (tlley have no common positive divisor ot her than 1)
4
In tbjs proof we see t he fundamental principles of the method of infi-
nite descent. Euler assumed a solution existed, and tluough some
number theory, (greatest common divisors, and proving two numbers are
coprime) Euler was able to prove that if a solution did exist, then an-
other solution must exist . This solution 2o. , a- 3b, a+ 3b is shown to be
smaller than the first solutions and this process can be repeated infinitely
many times, which would make no sense. Therefore there is no solution
to FLT for n = 3. The original proof for n = 3 is much ..t:(lan
the condensed version I have presented, as I have only shown the mosJ,
--
important parts of the argument leading to the final conclusion. This
method was adapted from Pierre de Fermat's proof of n = 4, and it was
the first major breakt hrough of Fermat's Last Theorem after Pierre de
Fermat hin1Self. Mathematicians went back to work, and the method of
infinite descent seemed promising to give a final proof for all exponents
of Fermat's Last Theorem.
2.2 Dirichlet, Sophie Germain's Theorem, proof by
contradiction
Proof by contradiction is a very popular form of proof in the world of
mathematics, and can be seen in many cases, for example proving the
irrationality of ..;2. It is very similar to the method of infinite descent.
seen previously, a.c:; that itself is a type of proof by contradiction. To
explain how proof by contradiction works, I will usc t he example of
proving the irrationality of ..;2.
You start by assuming that something is true, in thic:; ca.c:;e you assume
that ..j2 is rational. If ..j2 is rational, then it can be written as a fraction
By doing some calculations we can t hen find that t his fraction can be
q
simplified:
1. ..j2 = E
q 2
2. Square bot h sides: 2 =
5
. ' I

M\-
\o..H dtA
\-

3. Mult iply by q
2
: 2q
2
= p
2
4. Ftom this we can sec that p
2
must be even, so p must also be even.
Therefor e, we can substitute 2m for p: 2q
2
= (2m)
2
=4m
2
5. Divide both sides by 2: q
2
=2m
2
6. Here we have the same situation as before, q
2
is even therefore q
must b e even. We can t hen say that q = 2n. From t his we have found
that .J2 =
2
m = !!!
2n n
7. We now have a fraction t hat is simpler than T!. which is
q n
8. Tllis argument call be repeated over and over again to find simpler
fract ions. However we know that fractions cannot be simplified forever
t herefore we must rej ect our asswnption t hat /2 is rational.
We can see that this proof by contradiction, is actually another case
---
of t he method of infinite descent. Johann Dirichlet used thls method to
a-aeiiipt to prove t he case for n = 5. Dirichlet completed part of the
proof for n = 5, and t he whole proof was then completed by Adrien-
Marie Legendre. This proof uses Sophle Germain's Theorem, named
after Sophie Germain that deals wi th t he divisibility of the solutions of
FLT, which I will explain in the proof for n = 5 in the following section.
2.2.1 Dirichlet's proof for n = 5
4
Just like Euler did, Dirichlet started by assuming there was a solut ion
to Fermat's Last Theorem and proving t hat the solutions :c;, y, z were
coprime. Dirichlet t hen made t he assumption that yare odd and z is
even, because there can only be at most one even number since they are
all coprime but there must be at least one even because odd+odd:fodd.
Dirichlet used Sophie Germain's Theorem to help hlm prove tllis case
of Fermat's Last Theorem. Sophie Germain's t heorem said that if Fer-
mat's Last Theorem is t rue for any prim )2: 3 and if 2f, )t- 1 is a prime,
? rust divide the product x yz. He used this theorem to sho\ that
1
E wards 2000 V...O \}v.t \ Q V\ 1
6 v'Ci
either 5 divides z or it divides x, y.
I
.. ~
Dirichlet. then assumed that 5 divides z and showed a case of infinite
descent by showing that if there is a solution, there must be a smaller
solution. Therefore, if 5 divides z, there are no integer solutions. He
also showed that. if 5 divides x or y (doesn't matter which since they are
symmetric) tl1en t here are also no integer solutions by using the method
of infinite descent once again.
The proof that Dirichlet used to prove Fermat's Last Theorem for
n = 5 showed lots of similarities to the previous proof by Euler for n = 3,
t he main one being that they both used the method of infinite descent
to prove that there couldn't be any solutions. A considerable amount
of number t heory was again used, the concept of coprime numbers and
--
greatest common divisors being used constantly to develop the argument
-
further. The method of infinite descent was the key to proving Fermat's
Last Theorem for n = 5, however it needed an extra bit of help from
Sophie Germain's theorem. For n = 3,4 the method of infinite descent
alone was enough to prove Fermat's Last Theorem, but for this case
something else was needed to come to the final conclusion. Doubts about
the method of infinite descent being used to prove Fermat's Last Theorem
completely started appearing, but people had faith in it, and so continued
to use it to carry on proving specific exponents.
2.3 Lame and Kummer: Cyclotomic Integers to prove
FLT
In 1847 the French academy of sciences set up an award and offered
prizes, of which one was a gold medal and 3,000 francs, to whoever
could prove Fermat's Last Theorem once and for all. Mathematicians
were given an extra motivation to go and prove FLT, as apart from the
personal satisfaction of proving it, t here was a respectable sum of money
involved as well. Various rumours were running around France as to who
7
was using which methods and how close people were to actually proving
it. The big shock came on the 1st of March 184 7, in the hands of Gabriel
Lame.
2.3.1 Lame's idea of a final proof
Gabriel Lame had proved FLT for the case n = 7 and was now stepping
up in front of the meeting of the French acade1ny of sciences and made
it known that he was on the verge of proving Fermat's Last Theorem.
Lame's idea was very simple and could potentially work if it were not
for t he flaw in his logic that Liouville and Kummer pointed out later .
Lame realized that :in the previous proofs for the cases n = 3, 4, 5, 7, a lot
depended on an algebraic factori zation of some sort . An example would
be in the case for n = 3, where x
3
+ y
3
is factorized into (x + y)(x
2
-
xy + y
2
) . Lame noted that a.c:; n becomes very large, it becomes harder to
factorize as t he degree of the polynomial becomes very large. Therefore,
Lame thought of using complex numbers to factorize x
11
+ yn complet ely
into linear factors. The only way this can be done is by inputting a
complex mm1ber a such that d' = 1 where(a =I= :J The equation would
then look like this: \...,. \r
xn + yn = (x + y)(x + ay)(x + a
2
y) ... (x + a
11
-
1
y)
Once Lame had this equation, all that was left for him Lo do was prove
that all the linear factors are coprime, i.e. their greatest common divisor
is 1. This would mean, as seen in t he other proofs earlier, t hat each
linear factor is an nth power and from tills he would then demonstrate
a case of infinite descent which would prove FLT. It seemed as t hough
with the help of complex numbers, the method of infinite descent would
prove Fermat's Last Theorem once and for all. It had been used for
n = 3, 4, 5, 7, and had worked perfectly so people were becoming more
convinced that t his would finally give t he solution everyone was looking
8
for.
However, Lame missed out a minor detail, but a detail that would
ultimately make all his work up to then useless. After Lame's presen-
tation, Liouville came up on the podium, and showed everyone Lam6's
unfortunate flaw in his proof.
2.3.2 Liouville's discovery of Lame's flaw
We all lmow that integers can only be fully factorized in one way, for
example the number 76 is factorized to 76 = 2
2
* 19 and it can't be
factorized in any other way. In other words, "there is only one possible
combination of primes that will multiply together to give any part icular
integer greater than l " r,. Lame's proposed proof depended on this theo-
rem, however he had failed to consider if complex numbers could also be
factorized uniquely and Liouville was there to point this out. This didn't
stop Lame though, a.c; he realized that the law for integers also worked
for complex numbers when n = 5. He was determined to carry on with
his work.
However, later on, Liouville read a letter from Ernst Kummer, a
German mathematician, stating that Liouville was correct when he was
quest ioning Lame's usc of unique factorization on complex numbers. Ap-
parently, Kummer had proved this in a memoir he had published three
years earlier. After this, Lame deserted his attempts io prove Fermat's
Last Theorem and Kummer continued this work, trying to find an alter-
native.
2.3.3 Kummer and cyclotomic integers
The problem which Kummer posed was the breaking up of numbers
built up from a by repeated addition, multiplication and subtraction
- ,..... -
into prime factors. The numbers look like this:
5
Singh 1998, p .l14
9
77
I
In tllis number , a
1
, a
2
, ... , a>.-J are integers. Kummer used t he lett er
A to represent a prime number and say that a>. = 1, (a #(1}- These
complex numbers are known as cyclotomic integers. Since a>- = 1, Kum-
mer reduced all the powers of the equation by saying t hat a>-+
1
= a, 1
a>-+
2
= a
2
and so on. I\ {1. ., v I Q \( "' t
1
W(i'l h c:AeGl41'r '
An interesting property of cyclot omic integers that will be needed
later to prove another property is that "representations of cyclotomic
integers in the form as seen above are not tmique"
6
. An example of
this would be that 1 +a+ a
2
+ ... + a>.-J = a>. +a+ a2 + ... + a>.- 1 =
a ( 1+ a+ o
2
+ ... + a'-
1
). This implies that either 1+ o + a
2
+ ... +o'-
1
0
or a = 1. Kummer had already assumed that a # 1 therefore t he former )
must be t rue.
0
? 1."' tA.: '""t. ' <,\T'v ' ! t' J o \- V. ltt 1
Another property of cyclotomic integers that was necessary for Kum-
mer t o t ry to prove Fermat's Last Theorem is the norm of a cyclotomic
integer. The norm of a cyclotomic integeriJ(a) would be written as
,_ -- c
N J(a) ar1 it is defined a.c; "the product of A- 1 conjugates of f(a)"
7
:
. . . 9 ? Go ; w i \.- '
Nf(a) = f(a)f(a
2
) ... f(a>. - l ) , 'r _. \ - ./.w.( .
He then established that the norm of any cyclotomic integer is an
integer itself. The proof for this is fairly simple. We must fir st note
that if we convert a - ai(j = 1, 2, ... , >.- 1) tllis only rearranges t he
factors of N f (a) but does not change t he norm. Thus, we have t hat
N f(a) = Co+ c1a + c2a
2
+ .. . + C>. - la>.-
1
(=j Co+ c1ai + G2a
2
i + ... +
C>._
1
a<>.- J)J. From this we can say t hat eo- eo = Cj - c
1
= 0. Therefore
Cj = cl> (j = 1, 2, 3, ... , A-1) and N J(a) =

... + a>.- J ).From

earlier , we know that 1 +a+ a
2
+ ... + a>-r = 0 and so we know that
6
Edwar ds 2000, p.82
7
Edwards 2000, p.83 J ?

10
\ '::I,
n+a
2
+ ... = - 1. Hence N.f(a) = eo+ct (a+a
2
+ ... = eo-cl
which is an integer .
If a cyclotomic integer is found to have norm 1, this integer is then
called a mut. A cyclotomic integer h(a) which L'3 irreducible (it cannot
be factored into two primes) only has the factorizatious h(a) = f(a)g(a)
where either of them is a unit. If one was talking about ordinary integers
here, they would be tempted to then call h(a) prime. However, when
talldng about cyclotomic integers, just because it is irreducible, does not
mean it is prime. Another factor that has to be taken into account for
a cyclotomic integer to be prime is that "there must exist cyclotomic
integers that it does not divide and if the product of any two cyclQtomic ) \- l.. 1
integers it does not Qj.v..ide, is itself cyclotomic integer it does not divide"
8
. l. e.t"(' ,j f. o. .
--- __.- "
The fact that a cyclotomic integer can be irreducible but not prime
is the main problem that causes t he failure of unique factorization for
cyclotomic integers.
Now Kllllllner had to apply t his to try and prove Fermat's Last The-
orem. His problem now was that he had to factor binomials of the form
x + aiy and also find all possible prime factors to t hose binomials. This
would t hen show if + y)(x + ay)(x + a
2
y) ... (x + aA-
1
y) were rela-
tively prime so that he could prove FLT. Kummer was aware that not
all cyclotomic integers could be factori7.ed in only one way and so he
introduced t he concept of ideal numbers. Kummer's discovery was that
- ..
"the set of all complex integers defined by an nth root of uni ty could
be so enlarged by the introduction of ideal m.unbers that unique factor-
ization into primes would prevail in the enlarged set''
9
. An example of
how ideal nmnbers can help make unique factorization possible is shown
below:
1. 25 can be factorized into 5 * 5 or it can also be factorized into
(4 - 8)(4 + 8) where f) = 3i
8
Edwards 2000, p .8L1
9
Dickson 1917, p.170
11
2. Tbis problem of unique factori zat ion can be solved by the intr(}-
duction of a.,(3, 1 ideal prime numbers such that:
5 = a(3
4-B = o?
4 + () (32
Now each factor of 25, breaks down further into ideal prime numbers
such that 25 = o? (3
2
so that now there is only one ~ to factorize ~
The introduction of ideal nwnbers helped to save unique factoriza-
tion for complex numbers and Kummer was able to prove Fermat's Last
Theorem for all regular primes, but not for t he irregular primes which
occur around 39% of t he time. Kummer's method was later extended to
irregular primes in the 20th centmy and his met hod was then inputted
into computers so that the computers could carry on proving cases for
Fermat's Last Theorem. By 1993, Fermat's Last Theorem had been
proved for all prime numbers up to 4 million. Fermat's Last Theorem
only needs to be proved for prime numbers as every other number is
built up fTom prime numbers therefore it could be re-written with the
exponent as one of the primes. Thanks to Ernst Kummer, the method
of infinite descent prevailed and was used right up until the final proof of
Fermat's Last Theorem to prove specific exponents. Some people might
say t hat Fermat's Last Theorem is true simply because so many cases
have been proved, and it seems as though this will carry on working for
all other exponents. However, mathematicians are never satisfied wi0
the finite, onl.z_ with t ~ infiJP.Le. It could be that Fermat's Last Theorem
is false when the exponent is 5 million. Only when Andrew Wiles proved
Fermat's Last Theorem for all possible exponents, were mathematicians
satisified t hat it was true.
12
j)
3 Conclusion
This wasn't the end of the history of proving Fermat's Last Theorem.
Goro Shimma and Yutaka Taniyama would create the Taniyama-Shimma
conjectme. Frey would then link this to Fermat's Last Theorem and
whoever could prove the Taniyama-Shimura conjecture would then sub-
sequently prove Fermat's Last Theorem and that someone was Andrew
Wiles.
10
The method of infinite descent was the key to the first proofs of
Fermat's Last Theorem. The pme method created by Pierre de Fermat
himself, is seen in both his and Euler's proof of FLT for n = 3, 4 and it
is both simple yet solid. In the proof for n = 5 by Dirichlet, you can see
the method of infinite descent being used along with Sophie Germain's
theorem about the divisibility of solutions to Fermat's Last Theorem to
show that there are no solutions and we can also see many aspects t hat
are similar between Dirichlet's proof and Euler's proof.
Later on it seemed that the method of infinite descent would be the
key to proving Fermat's Last Theorem completely. Lame proposed a way
to do this, by factorizing using roots of unity and cyclotomic integers,
and then using the method of infinite descent to show that there couldn't
possibly be a solution in the first place. Liouville and Kummer however
crushed this proposed proof because a cyclotomic integer does not abide
to the laws of ordinary integers when talking about unique factorization
into prime numbers.
Kummer however, picked this method up again and by introducing
the concept of ideal numbers, he was able to prove Fermat's Last Theo-
rem for regular primes using tllis method of factorizing and then proving
a case of infinite descent. Kummer's method then lived on and was
adapted to be able to prove Fermat's Last Theorem for irregular primes
as well and computers were able to prove it for all prin1es up to 4 million
10
Singb 1998
13
by 1993.
The method of infinite descent was a beautiful and elegant way
to prove Fermat's Last theorem and it is usual for mathematicians to
strongly believe in something elegant. The method works perfectly for
specific exponents, with a few adjustments along the way, the main ones
being Sophie Germain's theorem, the introduction of ideal numbers, and
also an adaption of this method to work for irregular primes. However,
how conclusive was the method of infinite descent in proving Fermat 's
Last Theorem? From a normal person's perspective, it is perfectly ac-
ceptable, as proving a theorem for all prime numbers up to 4 million
seems to be solid evidence. However, from a mathematician's perspec-
tive, the method of infinite descent was nowhere near conclusive in prov-
~
ing Fermat's La.st Theorem. The actual proof is very different, using
~ h m ~ ~ m p l x mathematics concerning elliptic curves and modular
forms. The method of infinite descent helped to motivate mathemati-
cians to carry on working on Fermat's Last Theorem, and other proofs
emerged fr;Q!!l...i.t a.s well such as provin_g the irrationality of ../2. Over-
- .....__ ---- -
all the method of infirrite descent was a gTeat discovery in the world of
mathematics, but was only partly successful in proving Fermat's Last
Theorem from a mathematicians perspective.
14
4 Bibliography
Works cited in this essay:
References
[1] Dickson, L.E. "Fermat's Last Theorem and the origin and natme
of the t heory of algebraic numbers". The Annals of Mathematics,
Second Series, Vol. 18, No. 4 (June 1917), pp. 161-187, Annals of
Mathematics, 1917.
[2] Edwards, Harold M. "Fermat's Last Theorem: A genetic introduc-
tion to algebraic number theory". Graduate Texts in Mathematics,
Springer, 2000
[3] Singh, Simon. "Fermat's Enigma". Anchor nooks, 1998
15