Notes on Number Theory and Discrete Mathematics

ISSN 1310–5132
Vol. XX, XXXX, No. X, XX–XX

Embedding the unitary divisor meet semilattice

in a lattice
Pentti Haukkanen1
1
School of Information Sciences,
FI-33014 University of Tampere, Finland
e-mail: [email protected]

Abstract: A positive divisor d of a positive integer n is said to be a unitary divisor of n if

(d, n/d) = 1. The set of positive integers is a meet semilattice under the unitary divisibility
relation but not a lattice since the least common unitary multiple (lcum) does not always exist.
This meet semilattice can be embedded to a lattice; two such constructions have hitherto been
presented in the literature. Neither of them is distributive nor locally finite. In this paper we
embed this meet semilattice to a locally finite distributive lattice. As applications we consider
semimultiplicative type functions, meet and join type matrices and the Möbius function of this
lattice.
Keywords: unitary divisor, meet semilattice, distributive lattice, semimultiplicative function,
meet matrix, Möbius function.
AMS Classification: 06A12, 06D99, 11A25, 11A51, 11C20, 15B36.

1 Introduction
It is well known that the set Z+ of positive integers is a poset under the usual divisibility relation.
It is likewise well known that the gcd and the lcm operations serve as the meet and the join on
this poset. Thus Z+ is a lattice under the usual divisibility relation, known as the divisor lattice.
This lattice is distributive.
A divisor d ∈ Z+ of n ∈ Z+ is said to be a unitary divisor of n and is denoted by d∥n if
(d, n/d) = 1. For example, the unitary divisors of 200 (= 23 52 ) are 1, 8, 25, 200. The unitary
divisors of a prime power pa are 1 and pa . A general formula for the unitary divisors of n =
∏p∈P pvp (n) can be written as
∏ pup (n) ,
p∈P

where up (n) runs over the (one or two) values 0 and vp (n) for all primes p. The number of
the unitary divisors of n is 2ω(n) , where ω(n) is the number of distinct prime divisors of n with
ω(1) = 0.

1
 The concept of a unitary divisor (or a “block factor”) originates from Vaidyanathaswamy [22]
and was further studied for example by Cohen [4] and Tóth [21]. General accounts can be found,
e.g., in [14, 18].
We denote the greatest common unitary divisor (gcud) of m and n as (m, n)⊕⊕ . The gcud of
m and n exists for all m, n ∈ Z+ and

(m, n)⊕⊕ = ∏ pρ(vp (m),vp (n)) ,

p∈P

where ρ(vp (m), vp (n)) = vp (m) if vp (m) = vp (n), and ρ(vp (m), vp (n)) = 0 if vp (m) ≠ vp (n).
We denote the least common unitary multiple (lcum) of m and n as [m, n]⊕⊕ . The lcum of m and
n exists if and only if vp (m) = vp (n), vp (n) = 0 or vp (m) = 0 for each prime p. For example,
the lcum of 2 and 4 does not exist and the lcum of 12 = 22 3 and 20 = 22 5 exists and is equal to
60 = 22 ⋅ 3 ⋅ 5. Hansen and Swanson [7] overcame the difficulty of the nonexistence of the lcum
by defining
mn
[m, n]∗ = .
(m, n)⊕⊕
It is easy to see that [m, n]∗ exists for all m, n ∈ Z+ and [m, n]∗ = [m, n]⊕⊕ when [m, n]⊕⊕
exists. Naturally, [m, n]∗ ≠ [m, n]⊕⊕ when [m, n]⊕⊕ does not exist. For example, [2, 4]∗ = 8 but
[2, 4]⊕⊕ does not exist. It is not reasonable to say that 8 is the lcum of 2 and 4, since 2, 4 ∦ 8.
It is easy to see that the unitary divisibility relation is a partial ordering on Z+ . The gcud
operation serves as the meet on this poset. Thus Z+ is a meet semilattice under the unitary
divisibility relation. Unfortunately, however, it is not a lattice, since the lcum does not always
exist. Korkee [12] embedded the unitary divisor meet semilattice (Z+ , ∥) in a lattice by adding an
element, denoted as ∞, so that each n ∈ Z+ is a unitary divisor of ∞. Then

[m, n]⊕⊕ if [m, n]⊕⊕ exists,

m∨n={
∞ otherwise.

Thus (Z+ ∪ {∞}, ∥) is a lattice. For example, the join of 2 and 3 is 6, and the join of 2 and 4 is
∞. This approach may be considered the one-point compactification of the discrete topological
space Z+ , [6].
The authors in [9] propose another way to embed the unitary divisor meet semilattice (Z+ , ∥)
in a lattice. They apply the one-point compactification with respect to each set of prime powers
{pa ∶ a = 0, 1, 2, . . .} and then apply the fundamental theorem of arithmetic. This lattice may be
considered a refinement of the lattice adopted by Korkee [12]. This extension goes as follows.
Let p be a prime and denote Up = {pa ∶ a = 0, 1, 2, . . .}. Let (Up , τ ) denote the corresponding
discrete topological space. Let (Up∗ , τ ∗ ) be the one-point compactification [6] of (Up , τ ). Denote
the inserted point (often called the point of infinity) by p∞ . Thus Up∗ = {pa ∶ a = 0, 1, 2, . . . or a =
∞} = {1, p, p2 , . . . , p∞ }, where 1∥pa ∥p∞ for all a = 0, 1, 2, . . . , ∞ and pa ∦ pb for all 0 < a, b < ∞
with a ≠ b. Then (Up∗ , ∥) is a lattice. In this lattice, for example, 1∨p = p, p∨p2 = p∞ , p2 ∨p2 = p2 ,
p2 ∨ p∞ = p∞ for all p ∈ P.
The lattice (Z∗+ , ∥) is the direct product

Z∗+ = ∏ Up∗
p∈P

of the lattices (Up∗ , ∥). In [9], Z∗+ is defined topologically; we do not present these details here.
As an illustration of the lattice (Up∗ , ∥) = ({1, p, p2 , . . . , p∞ }, ∥) its sublattice ({1, p, p2 , p3 , p∞ }, ∥)
is shown in Figure 1.

2
 p∞

p p2 p3

Figure 1: The sublattice ({1, p, p2 , p3 , p∞ }, ∥) of the lattice (Up∗ , ∥)

It should be noted that neither of the above extensions (Z+ ∪ {∞}, ∥) and (Z∗+ , ∥) of (Z+ , ∥)
to a lattice is distributive nor locally finite. The main purpose of this paper is to emped (Z+ , ∥)
to a locally finite distributive lattice, see Section 2. As an application of distributivity, in Section
3 we consider semimultiplicative type functions, which we in turn utilize in meet and join type
matrices in Section 4. Locally finity is needed to consider the Möbius function of the lattice, see
Section 5.

2 Embedding to a locally finite distributive lattice

Consider first the meet semilattice ({1, p, p2 , p3 }, ∥), where p is a prime number. Its Hasse dia-
gram is given in Figure 2.
p p2 p3

Figure 2: ({1, p, p2 , p3 }, ∥)

We extend this to a distributive lattice as is shown in Figure 3. We denote this distributive

extension as ({1, p, p2 , p3 }d , ∥), where d stands for distributive.

p ∨ p2 ∨ p3
p ∨ p3
p ∨ p2 p2 ∨ p3
p p2 p3

Figure 3: The distributive lattice ({1, p, p2 , p3 }d , ∥)

Let P(S) denote the power set of S. It is easy to see that

({1, p, p2 , p3 }d , ∥) ≃ (P({1, 2, 3}), ⊆).

We proceed in a similar manner to obtain

({1, p, p2 , . . . , pn }d , ∥) ≃ (P({1, 2, . . . , n}), ⊆).

Further, define ({1, p, p2 , . . .}d , ∥) by

({1, p, p2 , . . .}d , ∥) ≃ (P ′ (Z+ ), ⊆), (1)

where P ′ (Z+ ) is the set of all finite subsets of Z+ .

3
 Each element np ∈ {1, p, p2 , . . .}d (n ≠ 1) is of the form

np = pi1 ∨ ⋯ ∨ pik , k ≥ 1, (2)

where i1 , . . . , ik ∈ Z+ are distinct. In other words, for each element np ∈ {1, p, p2 , . . .}d (n ≠ 1)
there is a unique finite set Inp ⊆ Z+ such that

np = ⋁ pi . (3)
i∈Inp

In addition, we set Inp = ∅ for np = 1.

For mp , np ∈ {1, p, p2 , . . .}d we have

mp ∥np ⇐⇒ Imp ⊆ Inp . (4)

The meet of mp , np ∈ {1, p, p2 , . . .}d is given as

mp ∧ np = ⋁ pi (5)
i∈Imp ∩Inp

and their join is

mp ∨ np = ⋁ pi . (6)
i∈Imp ∪Inp

The expression of np is the direct join; in symbols,

np = pi 1 / ⋯ / pik = pi . (7)
i∈Inp

This means that np is the join of A = {pi1 , . . . , pik } and the set A is independent (that is, for each
pij ∈ A, pij ∧ ⋁(A ∖ {pij }) = 1 (the least element of the lattice). For direct join, see [5].
Finally, we define (Zd+ , ∥) as a “restricted” direct product (or a “discrete” direct product [20])
by
∗
(Zd+ , ∥) = ∏ ({1, p, p2 , . . .}d , ∥). (8)
p∈P

This means that each element n ∈ Zd+ is of the form

n = (n2 , n3 , . . . , np , . . .), (9)

where np ∈ {1, p, p2 , . . .}d for all primes p ∈ P and np ≠ 1 for at most finitely many p ∈ P. The
word “restricted” stands for “np ≠ 1 for at most finitely many p ∈ P”. This is denoted in (8) by the
asterisk in ∏∗ .
In other words, we have
∗
(Zd+ , ∥) ≃ ⨉ (Ai , ⊆), (10)
i∈Z+

where Ai = P ′ (Z+ ) for all i ∈ Z+ . This is again a “restricted” direct product, denoted by ⨉∗ ,
which means that for each (A1 , A2 , . . .) in this restricted direct product, Ai ∈ Ai is nonempty for
at most finitely many i ∈ Z+ .
For m, n ∈ Zd+ we have
m∥n ⇐⇒ mp ∥np ∀p ∈ P. (11)
The meet and join of m, n ∈ Zd+ are given as

m ∧ n = (m2 ∧ n2 , m3 ∧ n3 , . . . , mp ∧ np , . . .) (12)

4
and
m ∨ n = (m2 ∨ n2 , m3 ∨ n3 , . . . , mp ∨ np , . . .) (13)
For n ∈ Zd+ let Pn denote the set of primes p for which np ≠ 1. It is easy to see that if m∥n,
then Pm ⊆ Pn , and if Pm ∩ Pn = ∅, then m ∧ n = 1. The converse results do not hold.
For m, n ∈ Z+ we have m ∧ n = (m, n)⊕⊕ , and for m, n ∈ Z+ possessing the lcum we have
m ∨ n = [m, n]⊕⊕ . We here mean that for n ∈ Z+ ,

n = ∏ pvp (n) = (2v2 (n) , 3v3 (n) , . . . , pvp (n) , . . .) = (pvp (n) ∶ p ∈ P). (14)
p∈P

Theorem 2.1. The pair (Zd+ , ∥) is a locally finite distributive lattice.

Proof. It is easy to see that (P ′ (Z+ ), ⊆) is a distributive lattice; thus ({1, p, p2 , . . .}d , ∥) is a
distributive lattice. Then (Zd+ , ∥) is distributive lattice as a restricted direct product of distributive
lattices.
Since (P ′ (Z+ ), ⊆) concerns finite sets, it is locally finite. We show that (Zd+ , ∥) is locally
finite. Let m, n ∈ Zd+ with m∥n, and consider the interval [m, n] ⊆ Zd+ . Then

m = (m2 , m3 , . . . , mp , . . .), n = (n2 , n3 , . . . , np , . . .),

where mp ∥np for all primes p ∈ P and mp , np ≠ 1 for at most finitely many p ∈ P. Since
({1, p, p2 , . . .}d , ∥) is locally finite, the number of elements in each interval [mp , np ] is finite,
and since mp , np ≠ 1 for at most finitely many p ∈ P, the number of elements in [mp , np ] is greater
than 1 for at most finitely many p ∈ P. Thus the number of elements in the interval [m, n] ⊆ Zd+ is
finite. This completes the proof.

Recall that (Zd+ , ∥) is presented as the restricted direct product of the lattice of all finite subsets
of Z+ . We could also consider the unrestricted [3] (or complete [20]) direct product of the lattice
of all subsets of Z+ as a distributive extension of the meet semilattice (Z+ , ∥). This extension is
even a Boolean lattice but not locally finite.
It should be noted that the unitary divisibility in Zd+ is the divisibility relation with respect to
multiplication m⊙n in Zd+ defined by m⊙n = m∨n. Now, (Zd+ , ⊙) is a commutative semigroup of
idempotents with identity, and we have m∥n if and only if there exists k ∈ Zd+ such that n = m ⊙ k.
In particular, for m, n ∈ Z+ , we have m∥n if and only if there exists k ∈ Z+ such that n = m ⊙ k.

3 An analog of semimultiplicative functions

As an application of distributivity we introduce an analog of semimultiplicative functions. Dis-
tributivity is utilized in the proof of Theorem 3.1.
An arithmetical function f ∶ Z+ → C is said to be semimultiplicative if f ((m, n))f ([m, n]) =
f (m)f (n) for all m, n ∈ Z+ , where (m, n) and [m, n] stand for the gcd and lcm of m, n ∈ Z+ .
For information on semimultiplicative arithmetical functions we refer to [8, 15, 17, 18]
We propose an analog of this concept as follows. Let (m, n)d and [m, n]d denote the meet
and join in (Zd+ , ∥). We say that a function f ∶ Zd+ → C is d-unitarily semimultiplicative if

f ((m, n)d )f ([m, n]d ) = f (m)f (n), ∀m, n ∈ Zd+ . (15)

5
Theorem 3.1. Let f be a d-unitarily semimultiplicative function with f (1) ≠ 0. Then

f (pi1 ∨ ⋯ ∨ pik ) = f (1)1−k f (pi1 )⋯f (pik ) (16)

for all primes p and distinct positive integers i1 , . . . , ik (k ≥ 1), and

f (m ∨ n) = f (1)−1 f (m)f (n) (17)

for all m, n ∈ Zd+ with Pm ∩ Pn = ∅. (Thus f is completely determined by its values at prime
powers pa , where p ∈ P and a = 0, 1, 2, . . .)

Proof. We proceed by induction on k to prove (16). It is trivial for k = 1. Assume that (16)
holds for some k ≥ 1. We show that it holds for k + 1. Now,

f (pi1 ∨ ⋯ ∨ pik )f (pik+1 ) = f ((pi1 ∨ ⋯ ∨ pik ) ∧ pik+1 )f ((pi1 ∨ ⋯ ∨ pik ) ∨ pik+1 ).

By distributivity,

(pi1 ∨ ⋯ ∨ pik ) ∧ pik+1 = (pi1 ∧ pik+1 ) ∨ ⋯ ∨ (pik ∧ pik+1 ) = 1 ∨ ⋯ ∨ 1 = 1.

Thus

f (pi1 ∨ ⋯ ∨ pik ∨ pik+1 ) = f (1)−1 f (pi1 ∨ ⋯ ∨ pik )f (pik+1 )

= f (1)−1 f (1)1−k f (pi1 )⋯f (pik )f (pik+1 )
= f (1)1−(k+1) f (pi1 )⋯f (pik )f (pik+1 ).

This proves (16).

If Pm ∩ Pn = ∅, then m ∧ n = (m, n)d = 1. We thus obtain (17). This completes the proof.

4 Matrices associated with gcud and lcum

In this section we apply the new analog of semimultiplicative functions to a certain meet and join
type matrix. For information on these type matrices, see, e.g., [2, 10, 11, 13].
Let S = {x1 , x2 , . . . , xn } be a set of distinct positive integers with x1 < x2 < ⋅ ⋅ ⋅ < xn , and let
f be an arithmetical function such that f (x) ≠ 0 for all x ∈ Z+ . Consider the n × n matrix (S × )f
having
f 2 ((xi , xj )⊕⊕ )
(18)
f (xi )(xj )
as its ij entry. We adopt the notation

Bf∗ (xi ) = ∑ gf (d)

d∥xi
d∦xt
t<i

for all i = 1, 2, . . . , n, where

x
gf (x) = ∑ f 2 ( )µ∗ (d).
d∥x d
The function µ∗ is the unitary analog of the number-theoretic Möbius function [4].

6
 Let S = {d1 , d2 , . . . , dm } be the minimal gcud-closed set containing S. Define the n × m
matrix C = (cij ) by √
⎧
⎪
⎪
⎪ Bf∗ (dj )
⎪
⎪ if dj ∥ xi
cij = ⎨ f (xi )
⎪
⎪
⎪
⎩0 otherwise.
⎪
⎪
It is shown in [9] that
(S × )f = CC T . (19)
It is also shown in [9] that if S is gcud-closed, then

×
n Bf∗ (xk )
det(S )f = ∏ . (20)
k=1 f 2 (xk )

If f is a d-unitarily semimultiplicative function on Zd+ with f (n) ≠ 0 for all n ∈ Zd+ , then
f 2 ((xi , xj )⊕⊕ ) f ((xi , xj )d )
= , (21)
f (xi )f (xj ) f ([xi , xj ]d )
f ((xi ,xj )d )
that is, the ij element of (S × )f is equal to f ([xi ,xj ]d ) .

5 Möbius function
Let (P, ⪯) be a locally finite poset. The incidence algebra I(P, ⪯) of (P, ⪯) is the set of functions
f ∶ P × P → C, which satisfy f (x, y) = 0 unless x ⪯ y. Addition and scalar multiplication are
defined pointwise, and the multiplication or convolution is defined as

(f ⋆ g)(x, y) = ∑ f (x, z)g(z, y).

x⪯z⪯y

The function δ ∈ I(P, ⪯) is defined by δ(x, y) = 1 if x = y ∈ P and δ(x, y) = 0 otherwise. The

function δ serves as the identity under the convolution. The function ζ ∈ I(P, ⪯) is defined by
ζ(x, y) = 1 if x ⪯ y. The Möbius function µ of (P, ⪯) is the inverse of the function ζ under the
convolution. This means that

(ζ ⋆ µ)(x, y) = ∑ µ(z, y) = δ(x, y).

x⪯z⪯y

Material on incidence algebras and Möbius functions can be found in [1, 14, 16, 19].
Consider the lattices ({1, p, p2 , . . .}d , ∥) and (Zd+ , ∥). According to Theorem 2.1 they are lo-
cally finite, and since they are lattices, they are posets. Therefore we can consider their Möbius
functions for which we present expressions in the following theorem.
Theorem 5.1. The Möbius function of the lattice ({1, p, p2 , . . .}d , ∥) is given as

µp (mp , np ) = (−1)∣Inp ∣−∣Imp ∣ for mp ∥np . (22)

The Möbius function of the lattice (Zd+ , ∥) is given as

µZd+ (m, n) = ∏ µp (mp , np ), (23)

p∈Pm ∪Pn

where Pk is the set of primes p for which kp ≠ 1 (as is given in Section 2).

7
 Proof. We prove (22) by showing that the function µp (mp , np ) defined in (22) satisfies
∑ µp (dp , np ) = δ(mp , np ),
mp ∥dp ∥np

for mp , np ∈ {1, p, p2 , . . .}d , where δ(mp , np ) is the delta-function of ({1, p, p2 , . . .}d , ∥). In fact,
∑ µp (dp , np ) = ∑ µp (⋁ pi , ⋁ pi ) = ∑ (−1)∣Inp ∣−∣I∣ .
mp ∥dp ∥np Imp ⊆I⊆Inp i∈I i∈Inp Imp ⊆I⊆Inp

Let ∣Imp ∣ = r and ∣Inp ∣ = s. Since Imp ⊆ Inp , we have ∣Inp ∖ Imp ∣ = s − r. Thus we can write
s−r
s−r
∑ µp (dp , np ) = ∑ (−1)s−r−∣I∣ = ∑ ( )(−1)s−r−∣I∣ = (1 + (−1))s−r .
mp ∥dp ∥np I⊆(Inp ∖Imp ) k=0 k

This is equal to 1 if s = r and 0 if s ≠ r. Here s = r if and only if mp = np . Thus

∑ µp (dp , np ) = δ(mp , np ),
mp ∥dp ∥np

This shows that the function µp defined in (22) is the Möbius function of ({1, p, p2 , . . .}d , ∥).
We next prove (23). Assume first that m ∦ n. Then
µZd+ (m, n) = 0.
On the other hand, mp0 ∦ np0 for some prime p0 . Then µp (mp0 , np0 ) = 0. Now, p0 ∈ Pm ∪ Pn
(since otherwise mp0 = np0 = 1 and µp (mp0 , np0 ) = 1), and therefore
∏ µp (mp , np ) = 0.
p∈Pm ∪Pn

Thus, in this case,

µZd+ (m, n) = ∏ µp (mp , np ).
p∈Pm ∪Pn

Assume second that m ∥ n. Denote

S(m, n) = ∑ ∏ µp (dp , np ).
m∥d∥n p∈Pd ∪Pn

We show that S(m, n) = δ(m, n). Now, in the above sum, Pm ⊆ Pd ⊆ Pn . Let Pn = {p1 , p2 , . . . , pu }.
Since µp (dp , np ) = 1 for p ∉ {p1 , p2 , . . . , pu }, we have
S(m, n) = ∑ ⋯ ∑ µp1 (dp1 , np1 )⋯µpu (dpu , npu ).
mp1 ∥dp1 ∥np1 mpu ∥dpu ∥npu

Thus
u u
S(m, n) = ∏ ∑ µpi (dpi , npi ) = ∏ δpi (mpi , npi ) = ∏ δp (mp , np ).
i=1 mpi ∥dpi ∥npi i=1 p∈Pd ∪Pn

If p ∉ Pd ∪ Pn , then mp = np = 1 and therefore δp (mp , np ) = 1. Thus

S(m, n) = ∏ δp (mp , np ).
p∈P

By (9),
∏ δp (mp , np ) = δ(m, n).
p∈P

This completes the proof.

8
Remark 5.1. The unitary analog µ∗ of the number-theoretic Möbius function is given as

µ∗ (n) = (−1)ω(n) , n ∈ Z+ ,

where ω(n) is the number of distinct prime factors of n with ω(1) = 0, see [4]. It is easy to see
that µp (1, pi ) = −1 = µ∗ (pi ) for all prime powers pi , and further, µZd+ (1, n) = µ∗ (n) for all n ∈ Z+ .

References
[1] Aigner, M., Combinatorial Theory, Springer, 1979.

[2] Altinisik, E., B. E. Sagan & N. Tuglu, GCD matrices, posets, and nonintersecting paths,
Linear Multilinear Algebra 53 (2005) 75–84.

[3] Birkhoff, G., Lattice theory, Corrected reprint of the 1967 third edition, American Mathe-
matical Society Colloquium Publications, 25. American Mathematical Society, 1979.

[4] Cohen, E., Arithmetical functions associated with the unitary divisors of an integer, Math.
Z. 74 (1960) 66–80.

[5] Crawley, P. & R. P. Dilworth, Algebraic Theory of Lattices, Prentice-Hall, 1973.

[6] Dugundji, J., Topology, Allyn and Bacon, 1966.

[7] Hansen, R. T. & L. G. Swanson, Unitary divisors, Math. Mag. 52 (1979) 217–222.

[8] Haukkanen, P., Extensions of the class of multiplicative functions. East-West J. Math. 14
(2012), no. 2, 101–113.

[9] Haukkanen, P., P. Ilmonen, A. Nalli, Ayse & J. Sillanpää, On unitary analogs of GCD recip-
rocal LCM matrices. Linear Multilinear Algebra 58, No. 5–6 (2010) 599–616.

[10] Hong, S. & Q. Sun, Determinants of matrices associated with incidence functions on posets.
Czechoslovak Math. J. 54 (129) (2004), no. 2, 431–443.

[11] Ilmonen, P., P. Haukkanen & J. Merikoski, On eigenvalues of meet and join matrices asso-
ciated with incidence functions, Linear Algebra Appl. 429 (2008) 859–874.

[12] Korkee, I., On meet and join matrices associated with incidence functions, Ph. D. Thesis,
Acta Universitatis Tamperensis 1149, Tampere Univ. Press, 2006.

[13] Mattila, M., On the eigenvalues of combined meet and join matrices, Linear Algebra Appl.
466 (2015) 1–20.

[14] McCarthy, P. J., Introduction to Arithmetical Functions, Springer, 1986.

[15] Rearick, D., Semi-multiplicative functions. Duke Math. J. 33 (1966) 4–53.

[16] Sándor J. & B. Crstici, Handbook of Number Theory II, Kluwer Academic, 2004.

[17] Selberg, A., Remarks on multiplicative functions. In: Number Theory Day. Proc. Conf.,
Rockefeller Univ., New York, 1976, pp. 232–241. Springer, Berlin, 1977.

9
[18] Sivaramakrishnan, R., Classical Theory of Arithmetic Functions, Marcel Dekker, Inc., 1989.

[19] Stanley, R. P., Enumerative Combinatorics, Vol. 1, Wadsworth and Brooks/Cole, 1986.

[20] Szasz, G., Introduction to Lattice Theory, Third Ed., Academic Press, Budapest, 1963.

[21] Tóth, L., The unitary analogue of Pillai’s arithmetical function. Collect. Math. 40 (1989),
no. 1, 19–30.

[22] Vaidyanathaswamy, R., The theory of multiplicative arithmetic functions, Trans. Amer.
Math. Soc. 33 (1931) 579–662.

10