3
$\begingroup$

Let $\delta>0$. I am interested in obtaining a bound for the sum $\sum_{1 \leq x_1, ..., x_n \leq N} \operatorname{lcm}(x_1, ..., x_n)^{- \delta}$ where lcm denotes the lowest common multiple of the numbers. I would appreciate any comments and suggestions! Thank you very much.

$\endgroup$

2 Answers 2

8
$\begingroup$

For $\delta>1$ the sum is bounded, while for $\delta=1$ it grows by a power of $\log N$.

So let me focus on $0<\delta<1$ and $N\geq 2$. For a simple lower bound, we have $$ \sum_{1 \leq x_1, ..., x_n \leq N} \operatorname{lcm}(x_1, ..., x_n)^{- \delta}\geq \left(\sum_{1 \leq x \leq N} x^{- \delta}\right)^n\gg_{n,\delta}N^{n(1-\delta)}. $$ For a simple upper bound, we have $$ \sum_{1 \leq x_1, ..., x_n \leq N} \operatorname{lcm}(x_1, ..., x_n)^{- \delta}\leq \sum_{1\leq m\leq N^n}\frac{\tau(m)^n}{m^\delta}\ll_{n,\delta}N^{n(1-\delta)}(\log N)^{2^n-1}.$$ So, up to a logarithmic factor, the sum grows like $N^{n(1-\delta)}$.

$\endgroup$
0
$\begingroup$

Here is a start on the problem. For each value k of lcm in your sum, you find how many tuples M_k give that lcm, and then you sum M_k(k^{-delta}) over enough values of k, which includes values up to N and some smattering up to near N^n.

Let's say k is a prime power, p^e , so k is less than N. Then the number of tuples is seen to be (e+1)^n -e^n. So already we have about N/log N values of k solved.

Now let's say k is the proper product of two prime powers. We get an upper bound of ((e+1)^n - e^n)((f+1)^n - f^n), where I bet you can guess what f is. For a more precise enumeration, you have to toss out numbers of the form p^cq^d that are larger than N, or decide that they are small enough to include so that you have an upper bound.

This extends to k being a product of more than one prime factor, but one has to watch for counting things outside the sum. However, as k gets large, k^{-delta} may get small enough that it won't hurt to include additional terms. The major contribution will come from those k at most N anyway, and provide an informative lower bound.

Gerhard "That's It, More Or Less" Paseman, 2017.12.19.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.