On the sum of error terms associated to the sum-of-divisors function \(\sigma_a(n)\) for \(\frac{1}{2}<a\leq 1\) (Q6862929)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 8155743
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | On the sum of error terms associated to the sum-of-divisors function \(\sigma_a(n)\) for \(\frac{1}{2}<a\leq 1\) |
scientific article; zbMATH DE number 8155743 |
Statements
On the sum of error terms associated to the sum-of-divisors function \(\sigma_a(n)\) for \(\frac{1}{2}<a\leq 1\) (English)
0 references
6 February 2026
0 references
Let \(\sigma_a(n)\) be the classical arithmetic function defined by \N\[\N\sigma_a(n)=\sum_{d|n}d^a.\N\]\NThe well-known average behaviour of \(\sigma_a(n)\), for \(0<a<1\), is such that \N\[\N\sum_{n\leq x}\sigma_a(n)=\frac{\zeta(a+1)}{a+1}x^{a+1}+\zeta(1-a)x+E_a(x),\N\]\Nwhere \(E_a(x)=O(x^{(1+a)/2})\). For \(a=1\), this becomes \N\[\N\sum_{n\leq x}\sigma_1(n)=\frac{\zeta(2)}{2}x^{2}+E_1(x),\N\]\Nwhere \(E_1(x)=O(x\log x)\).\N\NThe paper under review investigates the average of \(E_a(n)\) for \(1/2< a \leq 1\). The main result states the following: For \(\delta_1(a)=1\) if a \(a=1\), or \(0\) otherwise, and for a exponent pair \((K_a,\Lambda_a)\) such that \N\[\N\Lambda_a-\frac{K_a}{2}+\frac{a}{2}-\frac{5}{4}=0,\N\]\Nwith \(K_1>0\) in the case \(a=1\), then \N\[\N\sum_{n\leq x}E_a(n)=\frac{\zeta(a+1)}{2(a+1)}x^{a+1}+\delta_1(a)\frac{\zeta(1-a)}{2}x^2+O(x^{a/2+3/4+K_a/2}\log x).\N\]\NBy applying the theory of exponent pairs, explicit choices of good exponents are given. In fact, for a certain choice of parameters, the \(O\)-term above \(O(x^{a/2+3/4+K_a/2}\log x)\), for explicit \(c\) nearly equal to \(0.28^+\), can be expressed as \(O(x^{1+(2a-1)c}\log x)\) in the case \(1/2<a<1\), and in the case \(a=1\), can be expressed as \(O(x^{1+c+\epsilon})\).\N\NThese results demonstrate a non-trivial cancellation among the terms \(E_a(n)\), improving upon the standard bound \(O(x^{1+(1+a)/2})\) for their sum.
0 references
generalized divisor problem
0 references
number theoretic error term
0 references
Dirichlet series of error term
0 references
Perron's formula
0 references
exponential sums
0 references
exponent pairs
0 references
sum of Chowla and Walum
0 references
0 references