1
$\begingroup$

This morning, it occurred to me that there's a straight forward argument for Catalan's conjecture for a specific prime and a power of $2$.

I doubt that it will generalize to the full Catalan's conjecture but I am planning to invest time this evening to see what I can come up with.

Is this interesting? Is Catalan's conjecture trivial for a specific prime and a power of $2$?

Has anyone attempted to simplify the argument made by Mihailescu?


Edit: Added an example of the method

Here is a very simple example of the method that I am talking about.

I will use it to argue that there is no solution to $2^a - 5^b = 1$ for any integer $a,b$

Apologies for any typos. Please let me know if you see any mistakes or if any point is not clear.

(1) Assume that such a solution exists.

(2) Then, there exists positive integers $a>0,b>0,x>0,c < 10$ such that:

$2^a = 10x + (c+1)$

$5^b = 10x + c$

(3) Clearly, $c = 5$ so that we have:

$2^a = 10x + 6 = 2(5x + 3)$

$5^b = 10x + 5 = 5(2x + 1)$

(4) We can see that $x$ must be odd otherwise:

$2^a \ne 2(5x+3)$

(5) Further, we can see that $x \equiv 2 \pmod 5$ otherwise:

$5^b \ne 5(2x+1)$

(6) $x \ne 2$ so $x \ge 7$ and there exists $x_2$ such that:

$5^b = 5(2(5x_2+2)+1) = 5(5(2x_2+1))$

(7) We see that $x_2 \equiv 2$ which means that $x = 5x_2 + 2$ but $x_2 \ne 2$ since that would make $x$ even.

(8) We can repeat this argument up until we reach an $x_i = 2$ but then we have the following situation:

$x = 5(5(\dots(5(2)+2)\dots)+2) = (5^{i-1} + 5^{i-2} + \dots + 1)(2)$ which is even.

(9) But this contradicts step #4 where $x$ is odd so we reject our assumption in step #1.

  • 0
    Plz tell us more2017-02-17
  • 0
    What details would be interesting? Would this method be interesting? It will work with $3^x - 2^x$ but doesn't help with a conclusion about any prime power and the power of $2$.2017-02-17
  • 1
    Show us what you did. Then we can judge its value !2017-02-17
  • 1
    @mick, added an example of my method.2017-02-18

0 Answers 0