# Remainder of summation of $g(n)$ when divided by $8$

I was given the following problem:

Let $g(n)$ be the smallest prime that divides $n^4+1$. What is the remainder of $g(1)+g(2)+...+g(2015)$ when divided by $8$?

My attempt so far:

Instead of summing up then divide $8$, i tried to find the remainder of each individual $g(n)$ when divided by $8$, then sum them up, and reduce it with modular arithmetic. So clearly for $n$ odd, $g(n)$ leaves a remainder of $2$ when divided by $8$. But for $n$ even I have no clue how to proceed.

If you can help solving this, I will greatly appreciate it.

You're on the right track. As you noted, for odd $n$, we have that $n^4 + 1$ is even, and thus its smallest prime factor is $2$; i.e., $g(n) = 2$ for odd $n$.
On the other hand, for even $n$, we have that $n^4 + 1$ is odd, and thus its smallest prime factor is odd. Furthermore, if a prime $p$ divides $n^4 + 1$, it must be the case that $-1$ has a 4th root modulo $p$; for an odd prime $p$, this is the same as saying $1$ has a primitive 8th root. Because multiplicative groups modulo primes are cyclic (as multiplicative groups in any finite field are cyclic), this will happen if and only if $p - 1$ is divisible by $8$; thus $g(n)$ for even $n$ will be 1 modulo 8.
Combining these and the fact that in the range from 1 to 2015, there are 1008 odd and 1007 even values, our final answer is $1008 * 2 + 1007 * 1 = 7 \pmod 8$.
• You actually don't need the full fact that multiplicative groups modulo primes are cyclic to pull this off; just Fermat's Little Theorem (that every element of the multiplicative group modulo $p$ has order dividing $p - 1$) suffices to carry this argument through (since we just need the entailment from "There exists a primitive 8th root modulo $p$" to "$p - 1$ is divisible by 8", and not the converse). Anyway, if you have any further questions, let me know and I'll try to explain more. Dec 13, 2016 at 9:16
• Alright. I presume you are perfectly comfortable with the first line (that if $n$ is odd, then so must be $g(n)$). As for the rest: are you familiar with Fermat's Little Theorem? Dec 13, 2016 at 12:18