The first post in this series presented some explanation as to why the abc conjecture seems like a reasonable attempt to mathematically codify a big idea. This idea is that the prime factorization of a sum of two numbers should not really relate to those of the individual numbers. Equivalently, it says that if we see an equation like 3 + 53 = 27, we should think of it as a "rare event" or "coincidence" that big powers of small primes are related in this way. The second post provided some examples and numerical evidence rigorous version of the conjecture. To review, this states that
The abc Conjecture: For any ε > 0, no matter how small, for all but finitely many equations of the form a + b = c where a and b are relatively prime, rad(abc)1 + ε > c.
Again, the radical rad(n) of an integer n is the product of its distinct prime factors. However, none of what has been discussed so far constitutes a mathematical proof that the abc conjecture is true or false.
In 2012, the Japanese mathematician Shinichi Mochizuki shocked the mathematical community by publishing, out of the blue, what he claimed was a proof of the abc conjecture. However, the initial excitement at this announcement was quickly replaced by confusion; almost no one was able to decipher the tools used in the proof, which totaled over 500 pages in length! Mochizuki, working in isolation for years, had built up a brand new mathematical formalism which he called "Inter-Universal Teichmüller Theory" that was bizarre and unfamiliar to other researchers. The language and notation (an sample of which is provided in the screenshot below) seemed alien, even to mathematicians!
Moreover, he refused to publicly lecture on the new material, instead only working with a few close colleagues. The combination of the length and inscrutability of the proof with his unwillingness to elucidate it discouraged people from attempting to understand it. In the years since the proof was published, skepticism has mounted concerning the proof's validity. While a small group of mathematicians defend it, a majority of the mathematical community thinks it is unlikely that the proof is valid. For now, the abc conjecture remains effectively open.
Nevertheless, it is certain that attempts to prove the conjecture will continue. It has a number of useful applications that would solve a myriad of other mathematical problems, should it be true. To illustrate the power of the abc conjecture, we give one famous example of an application: Fermat's Last Theorem.
One of the first equations we considered in this series was x2 + y2 = z2, which relates the side lengths of right triangles. This equation has infinitely many solutions, namely 32 + 42 = 52, 52 + 122 = 132, etc. Fermat's Last Theorem states that if we raise the exponents from 2 to any higher power, there are no solutions in the positive integers. That is, x3 + y3 = z3, x4 + y4 = z4, and so on are not satisfied by any x, y, and z > 0. Famously claimed by Pierre de Fermat in the 17th century, this problem remained unsolved for centuries. In 1985, when the abc conjecture was first stated, it remained open.
So let us assume that we have (somehow) proven the abc conjecture, and were interested in Fermat's Last Theorem. The first thing to note about the equation xn + yn = zn is that if we had a solution for this equation, we could always find one for which xn and yn were relatively prime. This is because if they have a common prime factor, so must zn, and we can cancel this factor (raised to the nth power) from both sides. Therefore, we have arrived at a situation in which we can apply the abc conjecture. The radical of xn, for any n, is at most x since multiplying x by itself does not introduce any more prime factors that were not already there. Hence rad(xnynzn) = rad(xn)rad(yn)rad(zn) ≤ xyz < z3. Therefore, for ε > 0, we have that
rad(xnynzn)1 + ε < (z3)1 + ε = z3 + 3ε.
On the other hand, applying the conjecture to this triple, we have that for ε > 0,
rad(xnynzn)1 + ε > zn
in all but finitely many cases. Since we can choose ε to be any positive number, we can make it small enough so that 3 + 3ε < 4 (e.g. if ε = 0.1). Then if n ≥ 4, the two inequalities above directly contradict each other. Since the top one always holds and the bottom holds in all but finitely many cases, we conclude that there can be at most finitely many exceptions to Fermat's Last Theorem when n ≥ 4.
So the abc conjecture does not quite imply Fermat's Last Theorem, but it comes very close. If, in addition, we knew just a bit more about how the exceptional abc triples behaved, we could manually verify that there are no counterexamples to Fermat's Last Theorem for n ≥ 4. Interestingly, this argument does not say anything about the n = 3 case, that is, about the non-existence of solutions to x3 + y3 = z3. This special case, however, had already been proven by Euler in the mid-1700s.
Of course, the abc conjecture remains unproven, while Fermat's Last Theorem was finally proven by Andrew Wiles in 1995. This was done by entirely different means. Nevertheless, this serves as a relatively simple example of how the conjecture can prove results about Diophantine equations without invoking very difficult mathematics. Another example of a consequence is the following statement, sometimes called Pillai's conjecture:
Conjecture: Every natural number k occurs only finitely many times as the difference of two perfect powers.
For example, the special case k = 1 is the subject of Catalan's conjecture, and states that xp - yq = 1 has only one solution: 32 - 23 = 1. This was proven by Preda Mihăilescu in 2002 (again by very different means from those above and from Wiles' methods), but the general case remains unsolved. If we knew for a fact that the abc conjecture were true, we would be able to prove this result by a very similar argument to the one given above for Fermat's Last Theorem (the reader is encouraged to try this!). Note that Pillai's conjecture also implies that the original equation that motivated the abc conjecture, namely y2 = x3 + k, also has only finitely many solutions (for fixed k). This is the result David Masser and Joseph Oesterlé sought on their way to first formulating the statement.
These examples start to indicate how important the abc conjecture is to the study of Diophantine equations; if it were proven, it would resolve many different problems that are currently treated separately in a single stroke. Even reproving known results in a new and simple way would be greatly beneficial to the theory, since a set of tools that could prove abc would help to unify disparate parts of number theory. As a result, mathematicians will doubtlessly continue work toward solving the conjecture and probing the most fundamental structure of numbers.
Sources: http://projectwordsworth.com/the-paradox-of-the-proof/, Shinichi Mochizuki: Inter-Universal Teichmüller Theory I: Construction of Hodge Theaters, http://mathworld.wolfram.com/PillaisConjecture.html, https://rlbenedetto.people.amherst.edu/talks/abc\_intro14.pdf, Brian Conrad: The abc Conjecture, 12 sep 2013.
Thank you, I thoroughly enjoyed reading this short series of three short articles on the abc conjecture. You wrote them superbly.
ReplyDelete