logo

Arithmetic Functions' Multiplicative Properties 📂Number Theory

Arithmetic Functions' Multiplicative Properties

Definition 1

  1. For nN\forall n \in \mathbb{N}, if an arithmetic function ff that is not f(n)=0f(n) = 0 satisfies the following, it is called a multiplicative function. f(mn)=f(m)f(n),gcd(m,n)=1 f(mn) = f(m) f(n) \qquad,\gcd(m,n)=1
  2. If a multiplicative function satisfies the following condition, it is called a completely multiplicative function. f(mn)=f(m)f(n),m,nN f(mn) = f(m) f(n) \qquad,m,n \in \mathbb{N}

Basic Properties

  • [1]: If ff is multiplicative, then f(1)=1f(1) = 1.
  • [2]: That ff is a multiplicative function is equivalent to for all primes p1,,prp_{1} , \cdots , p_{r} and all a1,,arNa_{1} , \cdots, a_{r} \in \mathbb{N}, f(p1a1prar)=f(p1a1)f(prar)f \left( p_{1}^{a_{1}} \cdots p_{r}^{a_{r}} \right) = f \left( p_{1}^{a_{1}} \right) \cdots f \left( p_{r}^{a_{r}} \right).
  • [3]: If ff is multiplicative, for primes pp that satisfy pnp \mid n, dnμ(d)f(d)=pn(1f(p)) \sum_{d \mid n} \mu (d) f(d) = \prod_{p \mid n} \left( 1 - f(p) \right)
  • [4]: If ff is multiplicative, then ff being a completely multiplicative function is equivalent to for all primes pp and all aNa \in \mathbb{N}, f(pa)=[f(p)]af \left( p^{a} \right) = \left[ f(p) \right]^{a}.
  • [5]: If ff is multiplicative, then ff being a completely multiplicative function is equivalent to the inverse of ff with respect to Dirichlet convolution f1f^{-1} being represented as follows. f1(n)=μ(n)f(n) f^{-1} (n) = \mu (n) f (n)
  • [6]: If ff is completely multiplicative, then F(n):=dnf(d)F(n) := \sum_{d \mid n} f(d) is multiplicative.

Description

Naturally, being completely multiplicative implies being multiplicative.

An arithmetic function being multiplicative is akin to the ‘independent’ condition spoken of across mathematics. Since function values can be thought of in smaller fragments, it naturally harbors rich mathematical properties.

Theorem [1] is derived very simply but implies f(1)0f(1) \ne 0, which becomes a critical condition for the set of multiplicative functions to form a group.

Proof

[1]

Since f(1)=f(11)=f(1)f(1)f(1) = f( 1 \cdot 1) = f(1) f(1), f(1)=1f(1) = 1 must hold.

[2]

()(\Rightarrow)

Since ff is multiplicative and always gcd(pa,qb)=1\gcd \left( p^{a} , q^{b} \right) = 1 for different primes pqp \ne q, f(p1a1prar)=f(p1a1)f(p2a2prar)=f(p1a1)f(p2a2)f(p3a3prar)=f(p1a1)f(prar) \begin{align*} f \left( p_{1}^{a_{1}} \cdots p_{r}^{a_{r}} \right) =& f \left( p_{1}^{a_{1}} \right) f \left( p_{2}^{a_{2}} \cdots p_{r}^{a_{r}} \right) \\ =& f \left( p_{1}^{a_{1}} \right) f \left( p_{2}^{a_{2}} \right) f \left( p_{3}^{a_{3}} \cdots p_{r}^{a_{r}} \right) \\ =& f \left( p_{1}^{a_{1}} \right) \cdots f \left( p_{r}^{a_{r}} \right) \end{align*}


()(\Leftarrow)

If we set m=p1a1prarm = p_{1}^{a_{1}} \cdots p_{r}^{a_{r}}, n=q1b1qsbsn = q_{1}^{b_{1}} \cdots q_{s}^{b_{s}}, and gcd(m,n)=1\gcd(m,n) =1, then f(mn)=f(p1a1prarq1b1qsbs)=f(p1a1)f(prar)f(q1b1)f(qsbs)=f(p1a1prar)f(q1b1qsbs)=f(m)f(n) \begin{align*} f(mn) =& f \left( p_{1}^{a_{1}} \cdots p_{r}^{a_{r}} q_{1}^{b_{1}} \cdots q_{s}^{b_{s}} \right) \\ =& f \left( p_{1}^{a_{1}} \right) \cdots f \left( p_{r}^{a_{r}} \right) f \left( q_{1}^{b_{1}} \right) \cdots f \left( q_{s}^{b_{s}} \right) \\ =& f \left( p_{1}^{a_{1}} \cdots p_{r}^{a_{r}} \right) f \left( q_{1}^{b_{1}} \cdots q_{s}^{b_{s}} \right) \\ =& f(m)f(n) \end{align*}

[3]

Strategy: The properties of the Möbius function must be used.

Definition of the Möbius function: Given a prime p1,,pkp_{1} , \cdots , p_{k}, let a natural number nn be represented as n=p1a1pkakn = p_{1}^{a_{1}} \cdots p_{k}^{a_{k}}. The arithmetic function μ\mu defined as follows is called the Möbius function. μ(n):={1,n=1(1)k,a1==ak=10,otherwise \mu (n) := \begin{cases} 1 &, n=1 \\ (-1)^{k} &, a_{1} = \cdots = a_{k} = 1 \\ 0 & , \text{otherwise} \end{cases} Multiplicativity: For all m,nNm, n \in \mathbb{N} that satisfy gcd(m,n)=1\gcd (m,n) = 1, μ(mn)=μ(m)μ(n)\mu (mn) = \mu (m) \mu (n)


g(n):=dnμ(d)f(d) g(n) := \sum_{d \mid n} \mu (d) f(d) And let’s say mm satisfies g(m,n)=1g(m,n) = 1. Then, for d=abd = ab that satisfies ama \mid m and bnb \mid n, gcd(a,b)=1\gcd(a,b) = 1 also holds, so we can consider sigma separately. Also, since μ\mu is multiplicative and ff is multiplicative by the premise, g(mn)=abmnμ(ab)f(ab)=ambnμ(ab)f(ab)=ambnμ(a)μ(b)f(a)f(b)=amμ(a)f(a)bnμ(b)f(b)=g(m)g(n) \begin{align*} g(mn) =& \sum_{ab \mid mn} \mu (ab) f(ab) \\ =& \sum_{a \mid m \\ b \mid n} \mu (ab) f(ab) \\ =& \sum_{a \mid m \\ b \mid n} \mu (a) \mu (b) f(a) f(b) \\ =& \sum_{a \mid m} \mu (a) f(a) \sum_{b \mid n} \mu (b) f(b) \\ =& g(m)g(n) \end{align*}

Therefore, gg is also multiplicative. Then, according to Theorem [2], g(p1a1prar)=g(p1a1)g(prar)g \left( p_{1}^{a_{1}} \cdots p_{r}^{a_{r}} \right) = g \left( p_{1}^{a_{1}} \right) \cdots g \left( p_{r}^{a_{r}} \right), so we only need to calculate g(pa)g \left( p^{a} \right) for a fixed prime pp. According to Theorem [1], f(1)=1f(1)=1 and μ(1)=1\mu (1) = 1, and for higher orders k2k \ge 2, all μ(pk)=0\mu (p^{k}) = 0, g(pa)=dpaμ(d)f(d)=μ(1)f(1)+μ(p)f(p)=111f(p)=1f(p) \begin{align*} g \left( p^{a} \right) =& \sum_{d \mid p^{a}} \mu (d) f(d) \\ =& \mu (1) f(1) + \mu (p) f(p) \\ =& 1 \cdot 1 - 1 \cdot f(p) \\ =& 1 - f(p) \end{align*} Summarizing for natural numbers n=p1a1prarn = p_{1}^{a_{1}} \cdots p_{r}^{a_{r}} results in g(n)=png(pa)=pn(1f(p)) g(n) = \prod_{p \mid n} g \left( p^{a} \right) = \prod_{p \mid n} \left( 1 - f(p) \right)

[4]

()(\Rightarrow)

Since ff is completely multiplicative, f(pa)=f(p)(pa1)=[f(p)]2(pa2)==[f(p)]a f \left( p^{a} \right) = f(p) \left( p^{a-1} \right) = \left[ f(p) \right]^{2} \left( p^{a-2} \right) = \cdots = \left[ f \left( p \right) \right]^{a}


()(\Leftarrow)

Let’s consider that for a prime p1,,psp_{1} , \cdots , p_{s}, an integer a1,,as,b1,,bsN0a_{1} , \cdots, a_{s} , b_{1} , \cdots , b_{s} \in \mathbb{N}_{0}, and a natural number c1,,ctNc_{1}, \cdots , c_{t} \in \mathbb{N}, any mm and nn are represented as follows. m=p1a1ptatn=p1b1ptbtmn=p1c1ptct m = p_{1}^{a_{1}} \cdots p_{t}^{a_{t}} \\ n = p_{1}^{b_{1}} \cdots p_{t}^{b_{t}} \\ mn = p_{1}^{c_{1}} \cdots p_{t}^{c_{t}} If mm and nn do not have prime pip_{i} as a divisor, then ai=0a_{i} = 0 and bi=0b_{i} = 0, respectively. Since ff is multiplicative, f(mn)=f(p1c1ptct)=f(p1c1)f(ptct)=[f(p1)]c1[f(pt)]ct=[f(p1)]a1[f(p1)]b1[f(pt)]at[f(pt)]bt=[f(p1)]a1[f(pt)]at[f(p1)]b1[f(pt)]bt=[f(p1a1)][f(ptat)][f(p1b1)][f(ptbt)]=f(p1a1ptat)f(p1b1ptbt)=f(m)f(n) \begin{align*} f(mn) =& f \left( p_{1}^{c_{1}} \cdots p_{t}^{c_{t}} \right) \\ =& f \left( p_{1}^{c_{1}}) \cdots f( p_{t}^{c_{t}} \right) \\ =& \left[ f ( p_{1}) \right]^{c_{1}} \cdots \left[ f ( p_{t}) \right]^{c_{t}} \\ =& \left[ f ( p_{1}) \right]^{a_{1}} \left[ f ( p_{1}) \right]^{b_{1}} \cdots \left[ f ( p_{t}) \right]^{a_{t}} \left[ f ( p_{t}) \right]^{b_{t}} \\ =& \left[ f ( p_{1}) \right]^{a_{1}} \cdots \left[ f ( p_{t}) \right]^{a_{t}} \left[ f ( p_{1}) \right]^{b_{1}} \cdots \left[ f ( p_{t}) \right]^{b_{t}} \\ =& \left[ f ( p_{1}^{a_{1}}) \right] \cdots \left[ f ( p_{t}^{a_{t}}) \right] \left[ f ( p_{1}^{b_{1}}) \right] \cdots \left[ f ( p_{t}^{b_{t}}) \right] \\ =& f \left( p_{1}^{a_{1}} \cdots p_{t}^{a_{t}} \right) f \left( p_{1}^{b_{1}} \cdots p_{t}^{b_{t}} \right) \\ =& f(m)f(n) \end{align*}

[5]

()(\Rightarrow)

Let’s say g(n):=μ(n)f(n)g(n) := \mu (n) f(n).

Since ff is completely multiplicative and dnμ(d)=I(n)\displaystyle \sum_{d \mid n } \mu (d) = I(n), $$ \begin{align*} (gf)(n) =& \sum_{d \mid n} \mu (d) f (d) f \left( {{ n } \over { d }} \right) \\ =& f(n) \sum_{d \mid n} \mu (d) \\ =& f(n) I(n) \end{align} $$ In Theorem [1], since $f(1) = 1$ and the identity function II is I(n)=0I(n) = 0 in n>1n > 1, f(n)I(n)=I(n)f(n) I(n) = I(n). Summarizing, (g f)(n)=I(n)(g \ast\ f) (n) = I (n), so g=f1g = f^{-1}.


()(\Leftarrow)

For any prime pp and natural number aNa \in \mathbb{N}, let’s set n=pan = p^{a}.

If f1(n):=μ(n)f(n)f^{-1} (n) := \mu (n) f (n) then dnμ(d)f(d)f(nd)=dnf1(d)f(nd)=(f1 f)(n)=I(n)=0 \begin{align*} \sum_{d \mid n} \mu (d) f (d) f \left( {{ n } \over { d }} \right) =& \sum_{d \mid n} f^{-1} (d) f \left( {{ n } \over { d }} \right) \\ =& (f^{-1} \ast\ f ) (n) \\ =& I(n) \\ =& 0 \end{align*}

Definition of the Möbius function: Given a prime p1,,pkp_{1} , \cdots , p_{k}, let a natural number nn be represented as n=p1a1pkakn = p_{1}^{a_{1}} \cdots p_{k}^{a_{k}}. The arithmetic function μ\mu defined as follows is called the Möbius function. μ(n):={1,n=1(1)k,a1==ak=10,otherwise \mu (n) := \begin{cases} 1 &, n=1 \\ (-1)^{k} &, a_{1} = \cdots = a_{k} = 1 \\ 0 & , \text{otherwise} \end{cases}

According to the definition of the Möbius function, if k2k \ge 2 then μ(pk)=0\mu (p^{k} ) = 0, μ(1)f(1)f(pa)+μ(p)f(p)f(pa1)+0++0=0 \mu (1) f(1) f \left( p^{a} \right) + \mu (p) f(p) f \left( p^{a-1} \right) + 0 + \cdots + 0 = 0 Since the premise is that ff is a multiplicative function, according to Theorem [1], f(1)=1f(1) = 1. Meanwhile, the Möbius function is μ(1)=1\mu (1) = 1 and μ(p)=1\mu (p) = -1, therefore, 11f(pa)+(1)f(p)f(pa1)=0 1 \cdot 1 \cdot f \left( p^{a} \right) + (-1) \cdot f(p) f \left( p^{a-1} \right) = 0 Summarizing, f(pa)=f(p)f(pa1) f \left( p^{a} \right) = f(p) f \left( p^{a-1} \right) When recursively resolved, f(pa)=f(p)af \left( p^{a} \right) = f(p)^{a}, therefore, by Theorem [4], ff is completely multiplicative.

[6]

If we assume gcd(m,n)=1\gcd (m,n) = 1, we can distinguish the divisors of mm from those of nn. For convenience, let’s say a1,,aMa_{1} , \cdots, a_{M} are the divisors of mm, and b1,,bNb_{1} , \cdots , b_{N} are the divisors of nn. F(mn)=dmnf(d)=f(1)+f(a1)++f(aM)+f(b1)++f(bN)+f(a1bN)++f(aMb1)++f(a1a2bN)++f(mn)=amf(a)bnf(b)=F(m)F(n) \begin{align*} F(mn) =& \sum_{d \mid mn} f(d) \\ =& f(1) + f(a_{1}) + \cdots + f (a_{M}) + f(b_{1}) + \cdots + f(b_{N}) \\ & + f(a_{1}b_{N}) + \cdots + f(a_{M} b_{1}) + \cdots + f(a_{1} a_{2} b_{N}) + \cdots + f(mn) \\ =& \sum_{a \mid m} f(a) \sum_{b \mid n} f(b) \\ =& F(m) F(n) \end{align*}


  1. Apostol. (1976). Introduction to Analytic Number Theory: p33. ↩︎