The Mean Value of A New Arithmetical Function
The Mean Value of A New Arithmetical Function
The Mean Value of A New Arithmetical Function
i
+ 1
_
,
where n = p
1
1
p
2
2
p
k
k
be the factorization of n into prime powers. The rst few values of
f(n) are f(1) = 1, f(2) =
1
2
, f(3) =
1
2
, f(4) =
1
3
, f(5) =
1
2
, f(6) =
1
2
, f(7) =
1
2
, f(8) =
1
4
,
f(9) =
1
3
, f(10) =
1
2
, f(11) =
1
2
, .
Generally, for any prime p and positive integer , we have f(p
) =
1
1 +
. About the
elementary properties of f(n), it seems that none had studied it before. This function is
interesting, because its value only depend on the power of primes. The main purpose of this
paper is using the elementary and the analytic methods to study the mean value properties of
f(n), and give two sharper asymptotic formulas for it. That is, we shall prove the following:
Theorem 1. For any real number x > 1, we have the asymptotic formula
nx
f(n) =
1
2
xlnlnx +c x +O
_
x
lnx
_
,
where c is a computable constant.
Theorem 2. For any real number x > 1, we also have the asymptotic formula
nx
_
f(n)
1
2
_
2
=
1
36
_
3
2
_
(3)
x lnlnx +d
x +O
_
x
1
3
_
,
where (s) is the Riemann zeta-function, and d is a computable constant.
80 Jin Zhang and Pei Zhang No. 1
2. Proof of the theorems
In this section, we shall using the elementary and the analytic methods to prove our
Theorems. First we give following two simple lemmas:
Lemma 1. Let A denotes the set of all square-full numbers. Then we have the asymptotic
formula
nx
nA
1 =
_
3
2
_
(3)
x
1
2
+
_
2
3
_
(2)
x
1
3
+O
_
x
1
6
_
,
where (s) is the Riemann zeta-function.
Lemma 2. Let B denotes the set of all cubic-full numbers. Then we have
nx
nB
1 = N x
1
3
+O
_
x
1
4
_
,
where N is a computable constant.
Proof. The proof of these two Lemmas can be found in reference [3].
Now we use these two simple Lemmas to complete the proof of our Theorems. In fact, for
any positive integer n > 1, we can write it as n = p
1
1
p
2
2
p
k
k
, then from the denition of
f(n), we have
nx
f(n) =
nx
nA
f(n) +
nx
nB
f(n),
where A denotes the set of all square-full numbers. That is, n > 1, and for any prime p, if p | n,
then p
2
| n. B denotes the set of all positive integers with n / A. Note that f(n) 1, from
the denition of A and Lemma 1 we have
nx
nA
f(n) = O
_
x
1
2
_
. (1)
nx
nB
f(n) =
npx
(n, p)=1
f(n) =
px
n
x
p
(n, p)=1
1
2
=
1
2
px
_
x
p
x
p
2
+O(1)
_
=
x
2
px
1
p
x
2
px
1
p
2
+O
_
_
1
2
px
1
_
_
. (2)
Note that
px
1
p
= ln lnx +c +O
_
1
lnx
_
( see Theorem 4.12 of reference [2] ),
px
1
p
2
=
p
1
p
2
p>x
1
p
2
= d +O
_
1
x
_
,
Vol. 4 The mean value of a new arithmetical function 81
where c and d are two computable constants.
And the Prime Theorem ( see Theorem 3.2 of reference [3]):
(x) =
px
1 =
x
lnx
+O
_
x
ln
2
x
_
.
So from (2) we have
nx
nB
f(n) =
x
2
_
lnlnx +c +O
_
1
lnx
__
x
2
_
d +O
_
1
x
__
+O
_
1
2
_
x
lnx
+O
_
x
ln
2
x
___
=
1
2
xlnlnx +
c
2
x
d
2
x +O
_
x
lnx
_
=
1
2
xlnlnx +x +O
_
x
lnx
_
, (3)
where is a computable constant.
Now combining (1) and (3) we may immediately get
nx
f(n) = 1 +
nx
nA
f(n) +
nx
nB
f(n)
= 1 +O
_
x
1
2
_
+
1
2
xlnlnx + x +O
_
x
lnx
_
=
1
2
xlnlnx + x +O
_
x
lnx
_
,
where is a computable constant.
This proves Theorem 1.
Now we complete the proof of Theorem 2. From the denition of f(n) and the properties
of square-full numbers, we have
nx
_
f(n)
1
2
_
2
=
1
4
+
nx
nA
_
f(n)
1
2
_
2
+
nx
n/ A
_
f(n)
1
2
_
2
=
1
4
+
nx
nA
_
f(n)
1
2
_
2
.
where A also denotes the set of all square-full numbers. Let C denotes the set of all cubic-full
82 Jin Zhang and Pei Zhang No. 1
numbers. Then from the properties of square-full numbers, Lemma 1 and Lemma 2 we have
nx
nA
_
f(n)
1
2
_
2
=
np
2
x
(n, p)=1, nA
_
f(n)
1
2
_
2
+
nx
nC
_
f(n)
1
2
_
2
=
p
2
x
n
x
p
2
(n, p)=1, nA
_
1
3
1
2
_
2
+O
_
_
_
_
nx
nC
1
_
_
_
_
=
p
2
x
_
_
_
_
_
n
x
p
2
nA
1
36
n
x
p
4
nA
1
36
_
_
_
_
_
+O
_
x
1
3
_
=
1
36
p
2
x
_
c
x
1
2
p
c
x
1
2
p
2
_
+O
_
x
1
3
_
=
c
36
x
_
1
p
1
p
2
_
+O
_
x
1
3
_
=
1
36
c
x lnlnx +d
x +O
_
x
1
3
_
.
where c =
_
3
2
_
(3)
, d is a computable constant.
So we have the asymptotic formula
nx
_
f(n)
1
2
_
2
=
1
36
_
3
2
_
(3)
x lnlnx +d
x +O
_
x
1
3
_
.
This completes the proof of Theorem 2.
References
[1] F.Smarandache, Only problems, Not solutions, Xiquan Publishing House, Chicago,
1993.
[2] Tom M. Apostol, Introduction to Analytical Number Theory, Spring-Verlag, New York,
1976.
[3] Aleksandar Ivic, The Riemann Zeta-Function, Dover Publications, New York, 2003.
[4] Zhefeng Xu, On the k-full Number Sequences, Smarandache Notions Journal, 14(2004),
159-163.
[5] Jinping Ma, The Smarandache multiplicative function, Scientia Magna, 1(2005), No.1,
125-128.