Intereting Posts

If $A,B,$ and $C$ are sets, then $A\times(B-C)$ = $(A \times B)$ $-$ $(A \times C)$.
Proving mathematical induction with arbitrary base using (weak) induction
Find an infinite set of positive integers such that the sum of any two distinct elements has an even number of distinct prime factors
Proof related to Fibonacci sequence
If $a\in \mathrm{clo}(S)$, does it follow that there exists a sequence of points in $S$ that converges to $a$?
Summing the series $(-1)^k \frac{(2k)!!}{(2k+1)!!} a^{2k+1}$
How do we identify $\mathfrak{R}$-automorphisms of a group?
Inequality $\sqrt{1+x^2}+\sqrt{1+y^2}+\sqrt{1+z^2} \le \sqrt{2}(x+y+z)$
Computing the properties of the 3D-projection of an ellipse.
Proving a function is onto and one to one
Some inequalities for an entire function $f$
Rings of integers are noetherian (question about a specific proof)
finding right quotient of languages
Column Vectors orthogonal implies Row Vectors also orthogonal?
Smash product of pointed spaces is not associative

Just came across the following question:

Let $S=\{2,5,13\}$. Notice that $S$ satisfies the following property: for any $a,b \in S$ and $a \neq b$, $ab-1$ is a perfect square. Show that for any positive integer $d \not\in S$, $S \cup \{d\}$ does not satify the above property.

This question can be done by considering modulo 4.

- Proof: if $p$ is prime, and $0<k<p$ then $p$ divides $\binom pk$
- Find five positive integers whose reciprocals sum to $1$
- Algebraic independence over $\overline{\mathbb Q}$ and over $\mathbb Q$
- Generalizing Ramanujan's sum of cubes identity?
- $n$-Bit Strings Not Containing $010$
- Reference request: Clean proof of Fermat's last theorem for $n=3$.

Here comes my question:

What is the greatest value of $|A|$ if all elements in $A$ are different and for any $a,b \in A$ and $a \neq b$, $ab-1$ is a perfect square?

Remark: $A$ may not contain any of $\{2,5,13\}$, example: $\{17, 26, 85\}$

Edit: From the link here, there are infinite many 3-element sets statisfy the poperty. These sets are of the form $\{a, b, a+b+2r\}$ where $r^2 = ab-1$. Are we able to find a 4-element set that statisfies the property?

- Use division algorithm to prove for any odd integer n, $n^2 -1$ is a multiple of 8.
- Prove $(k,n+k)=1$ iff $(k,n)=1$
- When the $a^2+b+c+d,b^2+a+c+d,c^2+a+b+d,d^2+a+b+c$ are all perfect squares?
- What's the quickest way to solve $3^i \equiv 1 \mod 28$
- Prove by induction that $2^{2n} – 1$ is divisible by $3$ whenever n is a positive integer.
- How to do +, -, *, / with number in a base b?
- Prove that every number ending in a $3$ has a multiple which consists only of ones.
- If $n = 2^k - 1$ for $k \in \mathbb{N}$, then every entry in row $n$ of pascal's triangle is odd.
- Fibonacci identity: $f_{n-1}f_{n+1} - f_{n}^2 = (-1)^n$
- Multiples of an irrational number forming a dense subset

By the paper A. Dujella and C. Fuchs, Complete solution of a problem of Diophantus and Euler, J. London Math. Soc. 71 (2005), 33-52. (see Theorem 1b),

there does not exist 4-element set the considered property and with all elements greater than 1.

For results on sets contaning 1, see e.g. N. C. Bonciocat, M. Cipu, M. Mignotte, On D(-1)-quadruples, Publ. Mat. 56 (2012), 279-304. In particular, if {1,b,c,d} has considered property, then b>10^{13}.

I have been trying to find a 4-element set that satisfies the property. I ran a computer program to check all 4-element sets of integers <= 20,000 and found nothing, so I am guessing that there are no such sets, although of course I have nothing conclusive.

The program also listed all of the 3-element sets that it found, and I’ve noticed that each set, taken modulo 4, is either {1, 1, 1} or {1, 1, 2}. I can now prove this:

It is fairly easy to see that, if a*b-1 is square, then {a, b} modulo 4 is one of {1, 1}, {3, 3}, {1, 2}, and {2, 3}.

Next, looking through some notes on number theory that I printed from MIT OpenCourseware, there is a lemma which states that, because a*b is the sum of two squares (n^2 + 1^2), any prime factors of a*b which are congruent to 3 mod 4 must divide both squares, i.e., n and 1. Since no primes divide 1, this implies that a*b has no prime factors congruent to 3 mod 4, from which it follows that a and b are both not congruent to 3 mod 4.

Thus, {a, b} modulo 4 is one of {1, 1}, {1, 2}. From this, it is easy to see that any 3-element set satisfying the property is of one of the forms {1, 1, 1}, {1, 1, 2}, as desired.

@pipi: how did you prove the original problem “modulo 4”? Does your result imply that *any* set of the form {1, 1, 2} modulo 4 cannot be extended to a 4-element set? If so, by using your argument and then modifying it to work on {1, 1, 1}, we may be able to prove that no 4-element sets satisfying the property exist.

- Cardinality of algebraic extensions of an infinite field.
- Field of sets versus a field as an algebraic structure
- Continuity of a convex function
- When are two semidirect products isomorphic?
- One sided Chebyshev's inequality
- How fundamental is Euler's identity, really?
- Why is $PGL(2,4)$ isomorphic to $A_5$
- Ultrafilters – when did it start?
- Motivation for/history of Jacobi's triple product identity
- What condition need to be imposed on Havel-Hakimi theorem to check for connected graph?
- Eisenstein Criterion with a twist
- How to prove $\int_{\Omega}\frac{1}{\kappa}uv\ +\ \int_{\Omega}\kappa\nabla u\cdot\nabla v$ is an inner product in $H^1$
- Integral $\int_{0}^{\infty}x^{-x}dx$
- $ 2|f^{'}(0)| = \sup_{z, w \in D} |f(z)-f(w)|$
- Residue integral: $\int_{- \infty}^{+ \infty} \frac{e^{ax}}{1+e^x} dx$ with $0 \lt a \lt 1$.