The phenomena of eventual counterexamples - MathOverflow

The phenomena of eventual counterexamples

2010-02-16T13:00:56Z

Define an "eventual counterexample" to be

$P(a) = T $ for $a < n$
$P(n) = F$
$n$ is sufficiently large for $P(n) = T\ \ \forall n \in \mathbb{N}$ to be a 'reasonable' conjecture to make.

where 'reasonable' is open to interpretation, and similar statements for rational, real, or more abstractly ordered sets for $n$ to belong to are acceptable answers.

What are some examples of eventual counterexamples, famous or otherwise, and do different eventual counterexamples share any common features? Could we build an 'early warning system' set of heuristics for seemingly plausible theorems?

edit: The Polya conjecture is a good example of what I was trying to get at, but answers are not restricted to number theory or any one area.

Answer by Steve Huntsman for The phenomena of eventual counterexamples

2010-02-16T14:05:48Z

The Mertens conjecture.

Answer by Steve D for The phenomena of eventual counterexamples

2010-02-16T14:21:30Z

Strong Law of Small Numbers by Guy.

Steve

Answer by Steve Huntsman for The phenomena of eventual counterexamples

2010-02-16T14:28:06Z

The Pólya conjecture.

Answer by Steve Huntsman for The phenomena of eventual counterexamples

2010-02-16T14:35:02Z

From the Wikipedia category of disproved conjectures:

Borsuk's conjecture
The Chinese hypothesis
Euler's sum of powers conjecture

Answer by Willie Wong for The phenomena of eventual counterexamples

2010-02-16T19:03:56Z

The De Giorgi conjecture is true for dimensions $\leq 8$. I guess this doesn't really count because De Giorgi himself only conjectured it for those dimensions based on the fact that Bernstein Theorem of minimal graphs is only true in dimensions $\leq 8$...

(To stay within the realm of geometry, if someone finds a counterexample to the positive mass theorem in high dimensions, that would be an example too.)

Answer by Gerry Myerson for The phenomena of eventual counterexamples

2010-02-16T23:05:18Z

It was once conjectured that factors of $x^n-1$ over the rationals had no coefficient exceeding 1 in absolute value. The first counterexample comes at $n=105$.

Answer by Gerry Myerson for The phenomena of eventual counterexamples

2010-02-17T01:06:33Z

I'm trying to reconstruct an example I saw somewhere some years back. It goes something like this: $\gcd(n^5-5,(n+1)^5-5)=1$ is true for $n=1,2,\dots,1435389$ but fails for $n=1435390$ (when the gcd is 1968751).

Answer by Gerhard Paseman for The phenomena of eventual counterexamples

2010-02-17T08:16:56Z

In answering another MathOverflow question on Graham's number, I quoted from Harvey Friedman's Enormous Numbers in Real Life. Perhaps eventual counterexamples bear some relation to proof strength in certain systems of logic? Anyway, that example there could be rephrased to fit the current question.

Suppose I look at strings on three symbols, and given a word w of length n I look at subwords of the form (forgive the AWK notation) spc[i] = substr(w,i,i+1), i.e. those substrings starting at the ith character going for length i+1 characters. So spc[1] gets the first two characters of w, spc[2] == w[2]w[3]w[4], and so on.

I manage to find, for every n that I can compute, a string w_n that I use for w above such that for 0 < i < j < = n/2, spc[i] is not a subsequence of spc[j]. Others find such examples for even larger values of n. It would be reasonable for me to believe I could find arbitrarily long strings with this property.

Enter Harvey Friedman:

"THEOREM 8.1. Let k >= 1. There is a longest finite sequence x1,...,xn from {1,...,k} such that for no i < j <= n/2 is xi,...,x2i a subsequence of xj,...,x2j.

For k >= 1, let n(k) be the length of this longest finite sequence.

Paul Sally runs a program for gifted high school students at the University of Chicago.

He asked them to find n(1), n(2), n(3). They all got n(1) = 3. One got n(2) = 11. Nobody reported much on n(3). I then started to ask several mathematicians to give an estimate on n(3), some of them very famous. I got guesses like this: 60, 100, 150, 200, 300. They were not in combinatorics. Recently I asked Lovasz, telling him about these five guesses. He guessed 20,000.

THEOREM 8.2. n(3) > A(7,184). Lovasz wins, as his guess is closer to A(7,184) than the other guesses.

Recall the discussion about A(5,5) being incomprehensibly large. With the help of computer investigations (with R. Dougherty), I got:

THEOREM 8.3. n(3) > A(7198, 158386).

A good upper bound for n(3) is work in progress. Crude result: A(n,n) where n = A(5,5). "

Here A(n,n) is defined earlier in Friedman's paper as an Ackermann-like sequence. I suspect n(3) squishes Graham's number quite unlike a galactic black hole absorbing a prion or even a quark.

EDIT: I have been corrected; in the squishing hierarchy, n(4) squishes Graham's number, which squishes n(3). Again, unlike any physical realization I can imagine. END EDIT

The moral here is: "Don't jump to conclusions without a sufficiently strong proof system as back up".

Gerhard "Ask Me About System Design" Paseman, 2010.02.17

Answer by Gerald Edgar for The phenomena of eventual counterexamples

2010-02-17T12:45:57Z

In this thread search down for the answer by sigfpe .

Answer by Pandelis Dodos for The phenomena of eventual counterexamples

2010-02-17T14:31:48Z

The Busemann-Petty Problem.

Answer by Andrew Mullhaupt for The phenomena of eventual counterexamples

2010-02-18T21:47:31Z

R. M. Grassl and A. P. Mullhaupt, "Hook and Shifted Hook Numbers", Discrete Mathematics, Volume 79, Number 2, January (1990) pp. 153-167

"An infinite number of counter examples is provided for the conjecture that a shifted tableau shape is uniquely determined by its multiset of shifted hook numbers. Nevertheless, the previous conjecture of the first author that there was only one example of nonuniqueness is discussed and it is shown that it is «almost» true, based on computer search."

There were about five million examples before the counterexample, and approximately 1 mole of examples before the next counterexample is thought to occur.

Answer by Kevin O'Bryant for The phenomena of eventual counterexamples

2010-06-09T02:35:30Z

Freeman Dyson observed in my presence that the sequence with initial condition $a_0=3,a_1=0,a_2=2$, and recurrence $a_{n+3}=a_{n+1}+a_{n}$ almost has the property that $n\mid a_n$ if and only if $n$ is prime or 1, except that it doesn't.

He challenged us (grad students) to explain this ``near-phenomenon'', as it seems too close to being too good to be true to be coincidence. I've never seen an explanation.

Since this is Math Overflow, I'll give the spoiler, the first counterexample is $n=271441$.

Answer by S. Carnahan for The phenomena of eventual counterexamples

2010-06-09T04:32:12Z

D. H. Lehmer showed that the first prime value of the Ramanujan tau-function, defined by $$\sum_{n=1}^\infty \tau(n) q^n = q \prod_{n=1}^\infty (1-q^n)^{24} = q - 24q^2 + 252q^3 - 1472q^4 + \dots,$$ occurs at the 63001st term. This is slightly less surprising when one knows that prime values can only occur for odd square inputs.

Answer by Sune Jakobsen for The phenomena of eventual counterexamples

2010-06-12T16:22:49Z

Shapiro inequality: Let $x_1,x_2\dots x_n,x_{n+1},x_{n+2}$ be positive real numbers with $x_{n+1}=x_1$ and $x_{n+2}=x_2$. Now the inequality $\sum_{i=1}^{n} \frac{x_i}{x_{i+1}+x_{i+2}} \geq \frac{n}{2}$ must be true if $n<14$ or if $n\leq 23$ and $n$ is odd. So $n=14$ is the first $n$ where a counterexample can be found. I know that 14 is not that large a number, but remember that for each n we have a problem with a lot of freedom.

Answer by Richard Stanley for The phenomena of eventual counterexamples

2010-06-12T16:32:42Z

The least positive integer for which the equality $$ \left\lceil \frac{2}{2^{1/n}-1}\right\rceil = \left\lfloor
\frac{2n}{\log 2} \right\rfloor. $$ fails is $n=777,451,915,729,368$. See http://oeis.org/A129935.

Another example that I like is the number $f(n)$ of inequivalent differentiable structures on $\mathbb{R}^n$. We have $f(n)=1$ if $n\neq 4$, while $f(4)=c$, the cardinality of the continuum.

Answer by Torsten Ekedahl for The phenomena of eventual counterexamples

2010-06-12T18:52:19Z

One of my favourite examples in this context is the following: Define a sequence $(s_n)$ by $s_1=8$, $s_2=55$ and for $n\geq3$ $s_n$ the smallest integer such that $s_n/s_{n-1}>s_{n-1}/s_{n-2}$ so that $s_3=379$ as $379/55>55/8$. Then we have $s_n=6s_{n-1}+7s_{n-2}-5s_{n-3}-6s_{n-4}$ for $5\leq n\leq11056$ but not for $n=11057$ (I have lost track of the name of the person to whom this is due, but it is, nowadays, easily verified on a computer).

Answer by Yiftach Barnea for The phenomena of eventual counterexamples

2010-06-12T23:32:54Z

A famous example is the isomorphism problem for integral group rings: suppose $G$ and $H$ are two finite groups of order $n$ such that $\mathbb{Z}G \cong \mathbb{Z}H$ does it mean that $G \cong H$? It was proved to be true for many cases and for many $n$'s and I think it was believed to be true in all cases. Nonetheless, eventually a counter example was found, see http://www.jstor.org/pss/3062112.

Answer by Gerry Myerson for The phenomena of eventual counterexamples

2010-06-15T02:05:11Z

Smallest counterexample to "There is no positive integer $n$ such that the concatenation of (the decimal representation of) $n$ with itself is a square" is $n=13223140496$, according to http://www.research.att.com/~njas/sequences/A102567; $1322314049613223140496 = 36363636364^2$.

Answer by Joel David Hamkins for The phenomena of eventual counterexamples

2010-06-15T03:21:10Z

The essence of the phenomenon of eventual counterexamples is that a certain pattern that holds among small numbers, turns out not to be universal. In the very best examples, such as the examples provided in the other answers, which I have enjoyed very much, what we have is an easily described property $P(n)$, whose first failing instance is very large in comparison. Indeed, the quality of answer might be measured by the difference between the size of the description of the property and the size of the first failing instance of it. When an easily described property holds for a very long time and then suddenly fails at some very large number, we are surprised. Therefore, to my mind the phenomenon of eventual counterexamples is intimately wrapped up with the possibility of providing very short descriptions of enormous numbers.

Surely we are all able easily to provide short descriptions of some very large numbers, such as $2^{100}$ or $2^{2^{100!}}$. In order to go beyond exponentiation and factorials, we might make use of other easily described functions exhibiting even more enormous growth. The Ackermann function, for example, defined by a simple one-line recursion, has diagonal values 1, 3, 7, 61, $2^{2^{2^{65536}}}$, with the next value $A(5)$ mind-bogglingly huge.

All such examples, short descriptions of large numbers, can be systematically transformed into instances of eventual counterexamples. For if $d$ is a short description of an enormous number $N$, then the property $P(k)=$"$k$ does not exhibit $d$" is easily described and holds for all values $k$ below $N$, but not of $N$ itself. Thus, it does very well by the quality measure I mentioned above.

So to my mind, the real issue is: what are the largest numbers that you can describe by a very short description?

This question can be made precise by requiring the description to be expressible in a particular formal language. Once the language is rich enough, however, this problem will certainly wade into interesting foundational waters, for the question of whether a given description actually succeeds in describing a number---for example, "the length of the shortest proof of a contradiction in ZFC"---may be independent of our basic axioms, even if it is enormous.

Answer by AL for The phenomena of eventual counterexamples

2010-06-15T08:38:21Z

In reference to the Prime Number Theorem (then Conjecture) both Gauss and Riemann further conjectured that $\pi(n) < Li(n)$ (where $\pi(n)$ is the number of primes from $1$ to $n$ and $Li(n)$ is the logarithmic integral, $\int_2^n \frac{1}{ln(t)}dt$).

Although it has been proven that this does not hold (Littlewood), that there exists some $n$ such that $\pi(n) \geq Li(n)$, the first $n$ where this takes place is so huge no-one has worked it out yet (allegedly). The number is known as Skewes' Number. It is known to be between $10^{14}$ and $1.39822\times 10^{316}$, and strongly believed to be about $1.397162914\times 10^{316}$. (References at the foregoing link.)

Answer by Fedor Petrov for The phenomena of eventual counterexamples

2010-10-10T06:43:32Z

It was a conjecture that number of three-dimensional Young diagram of volume $n$ is counted by the generating function $\prod(1-x^n)^{-n(n+1)/2}$, as analogous facts are true for usual Young diagrams (Euler) and two-dimensional (Macmahon?) It is so for first few coefficients, but fails in general.

Answer by Fedor Petrov for The phenomena of eventual counterexamples

2010-10-10T16:16:32Z

Let $a_1=1$, $a_{n+1}=(1+a_1^2+a_2^2+\dots+a_n^2)/n$. Are all terms integer? No, the first non-integer is $a_{44}$. I do not know neither reference (my source is private communication by Dmitry Rostovsky, and he does not remember where is it from), nor deep reason (if they exist) why first 43 terms are integer.

Answer by Jonathan Kiehlmann for The phenomena of eventual counterexamples

2010-11-13T14:42:40Z

I'm surprised no one has mentioned Graeco-Latin Squares http://en.wikipedia.org/wiki/Graeco-Latin_square

Euler showed these exist for $n$ odd, or any multiple of 4. As none exist for $n=2$ or $6$, he conjectured that none exist for any $n\equiv 2 (mod 4)$.

As it happens, such exist for any $n\geq 3$ except $6$. This is quite a famous example, if small.

Answer by Gerry Myerson for The phenomena of eventual counterexamples

2011-01-18T05:29:27Z

The numbers 12, 121, 1211, 12111, 121111, etc., are all composite - until you get to the one with 138 digits, that's a prime. Saw this in a talk Lenny Jones gave at the New Orleans meeting earlier this month.

Answer by Willie Wong for The phenomena of eventual counterexamples

2011-04-21T00:02:57Z

The Borwein Integrals are integrals of products of the sinc function. They exhibit certain "apparent patterns" which, while eventually breaking down, are actually indicative of something larger at work. (The example given on the Wikipedia page is a good one.)

Answer by John Sidles for The phenomena of eventual counterexamples

2011-05-01T16:13:29Z

Hmmm ... as yet, no examples have been given from geometry or dynamics. So here's one.

Supposing that we interpret $P(a)=T$ for $a<n$ to mean "geometric objects have property $P$ for most objects that arise naturally", and let $P$ be the ergodic property, then the Kolmogorov–Arnold–Moser theorem suggests itself as providing the "eventual counterexample."

Domokos Szasz' article "Boltzmann's Ergodic Hypothesis, a Conjecture for Centuries?" (1994) provides an historical overview of the long slow process by which dynamical conjectures that for centuries were widely believed, were eventually proved to be wrong.

Another (related) answer:

In Conway's LIFE game, if the starting patterns are arranged in lexical order, the first self-replicating life-form (known at present) is Andrew J. Wade's Gemini.

The Gemini life-form can be viewed as the first (known) counter-example to the hypothesis "life-forms are not self-replicating". The lexical index of Gemini (as computed from its bounding-box) is $2^{4217807\times4220191}$ ... obviously too large to find by a blind search.

It seems to be generically true of life-forms (both biological-type and Conway-type)—and perhaps formal proofs too?—that special properties are emergent at very large lexical order-number of starting structures.

Answer by Sándor Kovács for The phenomena of eventual counterexamples

2011-05-01T17:35:54Z

Let $S_m$ denote the symmetric group on $n$ letters and let $P(m)$ denote the size of the outer automorphism group of $S_m$, i.e., the size of the quotient $\mathrm{Aut}/\mathrm{Inn}$ where $\mathrm{Inn}$ is the group of inner automorphisms (the ones induced by conjugation by an element of the group). Then $$\begin{cases} P(m)=1 &\text{ if } m\neq 6 \\ P(m)=2 &\text{ if } m= 6. \end{cases}$$ Of course, the "counter example" is not for a particularly large value, but only for a single one.

Answer by pageman for The phenomena of eventual counterexamples

2011-05-01T19:28:34Z

the Weaire–Phelan structure was found to be

a better solution of the "Kelvin problem" than the previous best-known solution, the Kelvin structure.

Answer by Tilman for The phenomena of eventual counterexamples

2011-05-01T21:13:21Z

Any finite loop space has the rational cohomology of a Lie group -- up to rank 65. From then on, there are counterexamples in every dimension. The smallest known dimension of a counterexampe is 1250, but whatever the actual smallest dimension is, counterexamples will occur in every dimension after that.

Answer by Tarun Chitra for The phenomena of eventual counterexamples

2011-05-02T05:59:03Z

This is a bit tongue-in-cheek, but what about Special Relativity? In this case let property $P(x), x\in \mathbb{R}$ be the property that a given velocity $x$ is attainable. After all, Galilean Transforms allow one to change to a frame moving at an arbitrary velocity. Only Einstein's interpretation of the discoveries of Lorenz and Poincaré allowed for us to realize that property $P$ is only true if $x \in [-3 \times 10^8, 3 \times 10^8]$

Answer by Tilman for The phenomena of eventual counterexamples

2011-05-02T06:16:56Z

Here's another one, maybe mostly of historical interest. Fermat once conjectured that all numbers of the form $$ p=2^{2^n}+1 $$ are prime, which he had the means to verify up to $n=4$. It took more than 100 years until Euler showed that this fails at $n=5$. Today we still don't know if there are any other Fermat primes, so quite possibly Fermat's conjecture fails in the worst possible way.

Answer by Seva for The phenomena of eventual counterexamples

2011-05-20T06:37:18Z

How about this paper?

Answer by Granger for The phenomena of eventual counterexamples

2011-05-20T10:19:13Z

Robert Baillie has a paper on arxiv today (http://arxiv.org/abs/1105.3943) which shows how in principle one can construct examples of formulae which hold for $N=0,1,2,\ldots,k$, for arbirtrarily large $k$, then fail for all larger $N$.

His largest example holds with $k\approx \exp(\exp(\exp(\exp(\exp(\exp(e))))))$.

Answer by Terry Tao for The phenomena of eventual counterexamples

2012-03-30T06:04:05Z

The first counterexample to the second Hardy-Littlewood conjecture is expected to occur somewhere between $10^{174}$ and $10^{1199}$ (at least, according to the references from the Wikipedia page), though it has not yet been definitively established that such a counterexample exists.

Answer by Dan Glasscock for The phenomena of eventual counterexamples

2012-07-08T10:27:23Z

This came up a few years ago from an error I noticed in the OEIS database. For all $0 \leq n \leq 58$, the numerator of $\sum_{k=0}^n \frac{2^{k+1}-1}{k+1}$ is equal to the numerator of $\sum_{k=0}^n \binom{n}{k}/(k+1)^2$. This fails first at $n=59$ and then at $n=1519, 7814, \ldots$. See A134652.