I've managed to reduce down the problem to showing that the solutions of \(a^2-a+1 = 3^k\) in integers satisfy \(a \le 2\), but can't proceed anywhere afterwards.

Note that since \(a^3+1 = (a+1)(a^2-a+1)\), the prime number we're looking for cannot divide \(a+1\) and hence must divide \(a^2-a+1\). Thus it suffices to find a prime number \(p\) that divides \(a^2-a+1\) but not \(a+1\).

However, \(\gcd(a^2-a+1, a+1) = \gcd(3, a+1)\), thus it suffices to find some prime factor \(p\) of \(a^2-a+1\) that is not \(3\). If this is found, \(p\) cannot divide \(a+1\), since if it does then \(p\) divides their GCD, but the GCD is either \(1\) or \(3\), impossible to be divisible by \(p\) whatever it is.

Also note that \(a^2-a+1 > 1\) for \(a > 2\), so it suffices to show that \(a^2-a+1\) is not in the form \(3^k\) for \(a > 2\), or in other words the only solutions of \(a^2-a+1 = 3^k\) in integers satisfy \(a \le 2\).
–
Ivan Koswara
·
2 years, 3 months ago

## Comments

Sort by:

TopNewestI've managed to reduce down the problem to showing that the solutions of \(a^2-a+1 = 3^k\) in integers satisfy \(a \le 2\), but can't proceed anywhere afterwards.

Note that since \(a^3+1 = (a+1)(a^2-a+1)\), the prime number we're looking for cannot divide \(a+1\) and hence must divide \(a^2-a+1\). Thus it suffices to find a prime number \(p\) that divides \(a^2-a+1\) but not \(a+1\).

However, \(\gcd(a^2-a+1, a+1) = \gcd(3, a+1)\), thus it suffices to find some prime factor \(p\) of \(a^2-a+1\) that is not \(3\). If this is found, \(p\) cannot divide \(a+1\), since if it does then \(p\) divides their GCD, but the GCD is either \(1\) or \(3\), impossible to be divisible by \(p\) whatever it is.

Also note that \(a^2-a+1 > 1\) for \(a > 2\), so it suffices to show that \(a^2-a+1\) is not in the form \(3^k\) for \(a > 2\), or in other words the only solutions of \(a^2-a+1 = 3^k\) in integers satisfy \(a \le 2\). – Ivan Koswara · 2 years, 3 months ago

Log in to reply

– Joel Tan · 2 years, 3 months ago

Another hint: Use the quadratic formula.Log in to reply

Hopefully that is a sufficient hint (and you shouldn't look down further).

If not,

Hint:If \( \gcd (3, a+1) = 3 \), what does that tell us about \(a\)? – Calvin Lin Staff · 2 years, 3 months agoLog in to reply

Yes.Corrected it. – Lawrence Bush · 2 years, 3 months ago

Log in to reply

I've updated it to "any integer \( a > 2 \)" for clarity.

This is an interesting problem. – Calvin Lin Staff · 2 years, 3 months ago

Log in to reply