Research Article Volume 1 Issue 3
University of Athens, Greece
Correspondence: A Mazaris, Gnosis Mathematical Research Institute, 25, Agiou Antoniou str., 15235 Vrilissia, Greece, Tel +3069 4002 5916
Received: May 05, 2018 | Published: June 6, 2018
Citation: Mazaris A. A new proof of fermat’s last theorem. Open Acc J Math Theor Phy. 2018;1(3):107-113. DOI: 10.15406/oajmtp.2018.01.00016
For centuries, the proof of Fermat's last theorem using mathematical theory and techniques of his time has been a mystery for the entire mathematical community. Over the years, a large number of attempts have been made without so far a widely accepted proof that is based purely on mathematics that Fermat could have used himself. In the present work, we provide a novel approach to prove Fermat’s famous Theorem. We assume that the theorem states a true proposition and we end up in a contradiction, thus concluding that the initial assumption cannot hold, emphasizing the value of a classical method of proof, such as the proof by contradiction. Our method is based on the concept of the limit, a well known topic even at Fermat’s time, upon which the modern mathematical edifice was built.
Keywords: fermat, fermat’s last theorem, diophantine equations
Fermat’s Last Theorem (1637) is perhaps the single most famous mathematical problem of all times. Despite the extensive literature devoted to its proof, and its numerous applications in other field of science, it remained unsolved for nearly 4 centuries. Although finally proved in 1995 by A Wiles1 the Theorem never stopped being a challenge for the broader mathematical community, mainly because Wiles’ proof was based on an extensive mathematical background that was not nearly available in Fermat’s era. Over the years, a large number of attempts were unsuccessful, giving reasonable raise to concerns questioning whether Fermat had indeed come up with a general and concise proof, as he has claimed in his writings. Could Fermat have proven his Last Theorem? This is a question of both historical and mathematical value, regardless of the unambiguous significance of Wiles’ modern proof.
In this study, we attempt to prove Fermat’s Last Theorem using a novel approach that is based on the concept of limit, which was known in considerable depth back in the 16th century, suggesting that it could have been close to Fermat’s unrevealed proof. The work employs the well-know technique of the proof by contradiction, and is structured in 2 parts, leading to the final result.
The Theorem states that: if a, b, c is positive integers then there is no natural integer n>2.n>2. such that:
an+bn=cnan+bn=cn (1)
We will assume that Fermat’s Theorem is valid, that is we accept that there are positive integers a, b, c and a natural integer n>2 for which equation (1) is satisfied. Let us consider the relationship:
xn+yn=zn xn+yn=zn where x is a positive real number, y a positive integer, and z a positive real number. Also, let us assume that x>ax>a andy>b.y>b.
We will prove that the above relationship is valid: Let y be a positive integer such thaty>by>b andnn a natural integer such thatn>2.n>2. Then the number ynyn is a positive integer number. Also, let xx be a positive real number such thatx>ax>a andna natural integer such thatn>2.n>2. Then the numberxnxn is positive real number. If xn+yn=λxn+yn=λ Thenλλ is a positive real number as the sum of a positive integer number and a positive real number. As a positive real number,λλ can be written in the form λ=(n√λ)nλ=(n√λ)n wherenis an integer number and n>2n>2 Assigning z=n√λz=n√λ see Appendix A yields: xn+yn=zn xn+yn=zn where zz is a positive real number Forznzn it holds that zn>cnzn>cn because: Since y>by>b it follows that
yn>bnyn>bn (2)
Similarly,x>ax>a x>ax>a implies thatxn>anxn>an (3) Adding relations (2) and (3) yields: xn+yn>an+bnxn+yn>an+bn
zn>cnzn>cn
In addition to equation (1) that we assume to be true, we have now proved that:xn+yn=zn where n>2xn+yn=znwhere n>2 (4)
Subtracting (1) from (4) we obtain a new equation:xn−an+yn−bn=zn−cn where n>2xn−an+yn−bn=zn−cnwhere n>2 (5)
As produced by equations (1) and (4), equation (5) applies only if equation (1) holds, given that equation (4) was proved valid.
Notice that we must not accept all the triples of the form: (x,y,z)=(ka,kb,kc) (x,y,z)=(ka,kb,kc) where kk is any positive integer and kk If we accept these triples, from equation (5) we will have:
(ka)n−an+(kb)n−bn=(kc)n−cn(ka)n−an+(kb)n−bn=(kc)n−cn
where nn any positive integer numberknan−an+knbn−bn=kncn−cnknan−an+knbn−bn=kncn−cn
(kn−1)an+(kn−1)bn=(kn−1)cn(kn−1)an+(kn−1)bn=(kn−1)cn
If kn−1≠0kn−1≠0 wherek≠1.k≠1. any positive integer number, thenkn−1kn−1 Dividing by kn−1kn−1 yields an+bn=cn.an+bn=cn. Therefore, the use of the equation: xn+yn=zn xn+yn=zn becomes unavailing. Note that we must also exclude all the Pythagorean triples, because if (a,b,c)(a,b,c) is a Pythagorean triple then all triples (ka,kb,kc)(ka,kb,kc) wherekk any positive integer, are also Pythagorean. We suppose that equation (5) is true because it results from equation (1) (Fermat’s Theorem) which we have assumed it is true. We will show that equation (5) always leads to a mathematical contradiction, thus proving that the equation initially assumed to be true (in our case Fermat’s Theorem) is false.
In the following section, we will show that equation (5) cannot be true, because it leads to a contradiction. This automatically means that equation (1) is a false statement, because equation (4) is true and equation (5) has resulted from (1) and (4). We follow a 2-part proofing process, with the final outcome presented in the results section.
Part 1
In this section we will start from equation (5) (i.e. we assume equation (1) is valid) and we will show that
e1yn−bn=limx→a(zn−cnyn−bn)1xn−an e1yn−bn=limx→a(zn−cnyn−bn)1xn−an
Equation (5) states that
xn−an+yn−bn=zn−cnxn−an+yn−bn=zn−cn
Sincey>by>b we obtain yn>bnyn>bn and yn−bn>0.yn−bn>0. Dividing by xn−anyn−bn+1=zn−cnyn−bn xn−anyn−bn+1=zn−cnyn−bn yields: x>a,x>a, and as xn>an ⇒ xn−an>0 ⇒ 1 xn−an>0xn>an⇒xn−an>0⇒1 xn−an>0 we can further proceed as: <(xn−anyn−bn+1)1xn−an=(zn−cnyn−bn)1xn−an(xn−anyn−bn+1)1xn−an=(zn−cnyn−bn)1xn−an Therefore (1+xn−anyn−bn)1xn−an=(zn−cnyn−bn)1xn−an(1+xn−anyn−bn)1xn−an=(zn−cnyn−bn)1xn−an (1+1yn−bn . 11xn−an)1xn−an=(zn−cnyn−bn)1xn−an(1+1yn−bn . 11xn−an)1xn−an=(zn−cnyn−bn)1xn−anThe last equation can be written as:
(1+1yn−bn . 11xn−an)1xn−an=(zn−cnyn−bn)1xn−an(1+1yn−bn . 11xn−an)1xn−an=(zn−cnyn−bn)1xn−an (6) Both sides of equation (6) can be viewed as functions of variable limx→a(1+1yn−bn . 11xn−an)1xn−an= limx→a(zn−cnyn−bn)1xn−anlimx→a(1+1yn−bn . 11xn−an)1xn−an= limx→a(zn−cnyn−bn)1xn−an (note that: zn−cn=xn−an+yn−bnzn−cn=xn−an+yn−bn ) Since ais accumulation point for both sides of equation (6), it follows that:
limx→a(1+1yn−bn . 11xn−an)1xn−an= limx→a(zn−cnyn−bn)1xn−anlimx→a(1+1yn−bn . 11xn−an)1xn−an= limx→a(zn−cnyn−bn)1xn−an
In equation (7), the term1yn−bn is a positive rational number. Also, it holds that limx→a1xn−an=+∞ because always xn−an>0. However, the definition of the exponential function:eω=limν→∞(1+ωv)v
Implies that eω=limx→a(1+ωf(x))f(x) Wheref(x) is a function of variable x and limx→af(x)=+∞ Let ω=1yn−bn andf(x)=1xn−an so that the above definition can be written as: eω=limx→a(1+ωf(x))f(x) with D(f)=(a, +∞)
Substituting the above into equation (7) yields:
e1yn−bn=limx→a(zn−cnyn−bn)1xn−an (8)
Part 2
In this section we will start again from equation (5) (i.e. we assume equation (1) is valid) and we will show that e1=limx→a(zn−cnyn−bn)1 xn−an Beginning from equation (5): xn−an+yn−bn=zn−cn (with n>2) Since xn−an>0:
1+yn−bnxn−an=zn−cnxn−an (9)
In the last equation, the quantities yn−bnxn−an andzn−cnxn−an are positive real numbers so we can consider their natural logarithm lnyn−bnxn−an andlnzn−cnxn−an Also, it holds that:limx→ayn−bnxn−an=+∞ andlimx→azn−cnxn−an=+∞. The quantities yn−bnxn−an andzn−cnxn−an are considered as functions of variablex andais accumulation point for both of them. Equation (9) now becomes:1+elnyn−bnxn−an=elnzn−cnxn−an (10)
Since xn−an>0 implies that 1 xn−an>0, equation (10), and therefore equation (9), can be written as:>1+e(xn−an)ln(yn−bnxn−an)1 xn−an=e(xn−an)ln(zn−cnxn−an)1 xn−an (11)
while as previously stated: limx→ayn−bnxn−an=+∞ andlimx→azn−cnxn−an=+∞ At the limit of x®a equation (11) becomes:limx→ae(xn−an)ln(yn−bnxn−an)1 xn−an=limx→ayn−bnxn−an=+∞ (12)
andlimx→ae(xn−an)ln(zn−cnxn−an)1 xn−an=limx→azn−cnxn−an=+∞ (13)
According to the well known property of exponential function: limx→ae(xn−an)ln(yn−bnxn−an)1 xn−an=elimx→a[(xn−an)ln(yn−bnxn−an)1 xn−an] Using the properties of limits we further get:
elimx→a[(xn−an)ln(yn−bnxn−an)1 xn−an]=elimx→a(xn−an).limx→aln(yn−bnxn−an)1 xn−an (14)
The limx→aln(yn−bnxn−an)1 xn−an is defined because a is accumulation point for the function ln(yn−bnxn−an)1 xn−an (considered as function of variable). Similarly we have:elimx→a[(xn−an)ln(zn−cnxn−an)1 xn−an]=elimx→a(xn−an).limx→aln(zn−cnxn−an)1 xn−an (15)
Note thelimx→aln(zn−cnxn−an)1 xn−an is defined because a is an accumulation point for the function ln(zn−cnxn−an)1 xn−an (considered as a function of variable x).Finally, applying the property of exponentiation: exy=(ex)y we have:
limx→ae(xn−an)ln(yn−bnxn−an)1 xn−an=elimx→a[(xn−an)ln(yn−bnxn−an)1 xn−an]=[elimx→a(xn−an)]limx→aln(yn−bnxn−an)1 xn−an (16)
andlimx→ae(xn−an)ln(zn−cnxn−an)1 xn−an=elimx→a[(xn−an)ln(zn−cnxn−an)1 xn−an]=[elimx→a(xn−an)]limx→aln(zn−cnxn−an)1 xn−an (17)
Recall that through equations (12) and (13) we showed that the limit of equations and (17) is +∞. We will proceed by proving that limx→aln(yn−bnxn−an)1 xn−an=+∞ It holds that: <limx→ayn−bnxn−an=+∞ (18) Also, it holds that limx→a(xn−an)=0 therefore limx→a1xn−an=+∞ (19) because xn−an is always positive. Equation (18) implies that: limx→alnyn−bnxn−an=+∞ (20) see Appendix B By multiplying equations (19) and (20) we obtain: (limx→a1xn−an).(limx→alnyn−bnxn−an)=(+∞).(+∞) limx→a(1xn−an.lnyn−bnxn−an)=+∞limx→aln(yn−bnxn−an)1 xn−an=+∞
Similarly, we get: limx→aln(zn−cnxn−an)1 xn−an=+∞It holds that: elimx→a(xn−an)=e0=1
Thus,
equation (16) results 1+∞
(in our case it holds that 1+∞=+∞
) and equation (17) then results1+∞
(in our case it holds that1+∞=+∞
).
Generally 1+∞
is an indeterminate form, but here we showed that
To facilitate the rest of the work, we write the outcome of equation (16) as:
1limx→aln(yn−bnxn−an)1 xn−an
(21)
where
limx→aln(yn−bnxn−an)1 xn−an=+∞
and the outcome of equation (17) as:
1limx→aln(zn−cnxn−an)1 xn−an
(22)
Where
limx→aln(zn−cnxn−an)1 xn−an=+∞
Therefore, equation (11) now gives:
1+e(xn−an)ln(yn−bnxn−an)1 xn−an=e(xn−an)ln(yn−bnxn−an)1 xn−an
and thus limx→a[1+e(xn−an)ln(yn−bnxn−an)1 xn−an]=limx→ae(xn−an)ln(yn−bnxn−an)1 xn−an
The both limits exist because a is an accumulation point of the corresponding functions.
Finally:
limx→a1+limx→a[e(xn−an)ln(yn−bnxn−an)1 xn−an]=limx→ae(xn−an)ln(yn−bnxn−an)1 xn−an
Using equations 1+[elimx→a(xn−an)]limx→aln(yn−bnxn−an)1 xn−an= [elimx→a(xn−an)]limx→aln(zn−cnxn−an)1 xn−an
and
Using equations (21) and (22):
1+1limx→aln(yn−bnxn−an)1 xn−an=1limx→aln(zn−cnxn−an)1 xn−an
(23)
where
limx→aln(yn−bnxn−an)1 xn−an=+∞
and limx→aln(zn−cnxn−an)1 xn−an=+∞
Now, we will prove that any quantity of the form
where 1limx→af(x)
and limx→af(x)=+∞
can always get the valueelimx→af(x).
More specifically, we will show that:
1limx→af(x)=elimx→af(x)
where
limx→af(x)=+∞
and 1limx→af(x)=+∞
We begin from the definition of the exponential function
e=limx→a(1+1f(x))f(x)
and sincelimx→af(x)=+∞
we can write:
11limx→af(x)=1+1limx→af(x)
or 11limx→af(x)=limx→a[1+1f(x)]
If we raise both sides of the last equation to the power of [limx→af(x)]2,
we have:
[11limx→af(x)][limx→af(x)]2={limx→a[1+1f(x)]}[limx→af(x)]2
For the left-hand side of the last equation, the operation
1limx→af(x).[limx→af(x)]2=[limx→af(x)]2limx→af(x)=limx→af(x)2limx→af(x)=limx→af(x)2f(x)=limx→af(x)
is permitable according to the properties of limits.
Hence:
1limx→af(x)=limx→a[1+1f(x)][limx→af(x)]2
1limx→af(x)= [limx→a[1+1f(x)]limx→af(x)]limx→af(x)
(24)
However,
limx→a[1+1f(x)]f(x)=limx→aeln[1+1f(x)]f(x)=elimx→aln[1+1f(x)]f(x)=elimx→a[f(x).ln[1+1f(x)]]=
=elimx→af(x).limx→aln[1+1f(x)] =elimx→af(x).ln{limx→a[1+1f(x)]}=eln{limx→a[1+1f(x)]}limx→af(x)={limx→a[1+1f(x)]}limx→af(x)
Therefore:
limx→a[1+1f(x)]f(x)={limx→a[1+1f(x)]}limx→af(x)=e
Substituting into (24) yields:
1limx→af(x)=elimx→af(x)
(25)
where
limx→af(x)=+∞
and 1limx→af(x)=+∞
which we intended to show.
In equations (23) and (25) we consider asln(yn−bnxn−an)1 xn−an
the quantities ln(zn−cnxn−an)1 xn−an
and ln(zn−cnxn−an)1 xn−an
which are functions of variable only (for ln(zn−cnxn−an)1 xn−an
recall that: zn−cn=xn−an+yn−bn
)
and
limx→aln(yn−bnxn−an)1 xn−an=+∞ ,
limx→aln(zn−cnxn−an)1 xn−an=+∞
Also, it holds that
1limx→aln(yn−bnxn−an)1 xn−an=+∞ ,
1limx→aln(zn−cnxn−an)1 xn−an=+∞
Hence, equation (23) according to equation (25) gives:
1+1limx→aln(yn−bnxn−an)1 xn−an=1limx→aln(zn−cnxn−an)1 xn−an
1+elimx→aln(yn−bnxn−an)1 xn−an=elimx→aln(zn−cnxn−an)1 xn−an
(26)
From the definition of the exponential function we have:
ey=limx→a[1+yf(x)]f(x) where limx→af(x)=+∞
since is for an accumulation point,
as a result:
limx→aeyf(x)=limx→a[1+yf(x)] where limx→af(x)=+∞
or
elimx→ayf(x)=1+limx→ayf(x) where limx→af(x)=+∞
The following equation:
elimx→ayf(x)=1+limx→ayf(x) where limx→af(x)=+∞
(27)
is true in two cases:
Hence can take any value, provided that any of the 2 conditions stated above is preserved.
If we assume y=[f(x)]2
we have:
limx→ayf(x)=limx→a[f(x)]2f(x)=limx→af(x)=+∞
So the value is acceptable and the equation (27) now becomes:
elimx→af(x)=1+limx→af(x)
(28)
Where limx→af(x)=+∞
This equation results directly from the definition of the exponential function and is applicable to any function that satisfies
We will now apply equation (28) to equation (26)
Note that we consider ln(yn−bnxn−an)1 xn−an
and theln(zn−cnxn−an)1 xn−an
as functions of x and let us also point out that limx→aln(yn−bnxn−an)1 xn−an=+∞
and limx→aln(zn−cnxn−an)1 xn−an=+∞
From equation (26) we obtain:
1+elimx→aln(yn−bnxn−an)1 xn−an=elimx→aln(zn−cnxn−an)1 xn−an
and in accordance with equation (28)
1+1+limx→aln(yn−bnxn−an)1 xn−an=1+limx→aln(zn−cnxn−an)1 xn−an
1=limx→aln(zn−cnxn−an)1 xn−an−limx→aln(yn−bnxn−an)1 xn−an
1=limx→a[ln(zn−cnxn−an)1 xn−an−ln(yn−bnxn−an)1 xn−an]
1=limx→a[ln(zn−cnyn−bn)1 xn−an]
by applying a property of logarithms,
1=lnlimx→a(zn−cnyn−bn)1 xn−an
and finally
e1=limx→a(zn−cnyn−bn)1 xn−an
(29)
After a sequence of logical steps, we have now reached a contradiction. More specifically: In Part 1, we have assumed that equation (5) is true and therefore equation (1) is satisfied. Then we showed that: e1yn−bn=limx→a(zn−cnyn−bn)1xn−an Similarly, in Part 2 we have assumed that equation (5) is true, and therefore equation (1) is satisfied, and we showed that: e1=limx→a(zn−cnyn−bn)1 xn−an Equations (8) and (29) imply that: e1yn−bn=e1 because the limit of a function is unique. Thus: yn−bn=1 Or yn=bn+1 for n>2 Since y>b (by choise) there exist a positive integer k such as:which leads to (b+k)n=bn+1 or (b+k)⋅(b+k)⋅… ⋅(b+k)↔=bn+1 n-terms The left-hand side of the above equation produces the termand The equality necessarily holds for so we have: (b+1)n=bn+1 This equation applies only for n=1 which is a contradiction because we started with the supposition that We have thus proved the following: We have accepted that if are positive integers then there is natural integer such that: an+bn=cn and we ended up in a proven relation that contradicts our initial supposition, forcing us to conclude that it is wrong, meaning that ifa, b, c are positive integers then there is no natural integer n>2 such that: an+bn=cn as the Theorem states.
Combining the fundamental concept of the limit with a classical proofing method, as proof by contradiction is, we have provided a novel proof of Fermat’s Last Theorem. Assuming that the equation introduced by the Theorem holds, and following a series of logical steps presented in 2 parts, we were able to reach a contradiction and thus prove the Theorem’s proposition. The main novelty of our approach resides in the use of pre-modern mathematical concepts and techniques that have been known since Fermat’s initial conjecture. As also stated in the introduction, there is an historic significance linked to a classic proof of the Theorem that we consider equally important to the well known existing proof.
To further illustrate the broad applicability of the Theorem, we select and present below 3 of its classic, direct applications.
This means that there is no anglesuch thatwithnatural number such thatandcongruent numbers.
Proof: If withpositive integers anddividing by we have:
wherecongruent number withand alsocongruent number with
By assigningand we have:
Then, according to Fermat’s Last Theorem, there is no anglesatisfying the above equation for
Sinusoidal functions are commonly used in Physics (in waves, oscillations and alternating current) as well as in Astrophysics (particle motion in combined magnetic-electric fields).
This means that there is no elliptic curve Frey witch the form is
Elliptic curves2 are widely applied in elliptic curve cryptography (ECC) and integer factorization.
For it folows that it is impossible to have a cube with a dimention (edge length) that is a natural number split into two cubes, each with natural number edges, and preserve its initial volume. This constraint is taken into account in industrial design, 3D printing and related software and can be viewed as the inability to expand the Pythagorean Theorem that revolutionized planar geometry, in 3D applications.
None.
The author declares that there is no conflict of interest.
©2018 Mazaris. This is an open access article distributed under the terms of the, which permits unrestricted use, distribution, and build upon your work non-commercially.