Completely Multiplicative Function

In the study of integers, number theorists seek functions that respect the fundamental structure of multiplication. These tools, known as arithmetic functions, help reveal deep patterns, but their interaction with products is not always straightforward. This raises a crucial question: How can we classify functions based on their multiplicative behavior, and what does this classification tell us about the nature of numbers? This article addresses this question by focusing on a particularly "perfect" class of functions—the completely multiplicative functions.

We will first delve into the "Principles and Mechanisms," where we define completely multiplicative functions, contrast them with their more general multiplicative counterparts using a prime power test, and explore their algebraic properties under Dirichlet convolution. Subsequently, in "Applications and Interdisciplinary Connections," we will uncover the power of these functions as essential tools in analytic number theory, connecting discrete integer properties to continuous analysis and revealing surprising links to fields like abstract algebra and physics. Through this exploration, we will see how a simple definition blossoms into a concept of profound importance.

In our journey through the world of numbers, we often seek patterns, rules that simplify the chaos. Imagine you are a physicist studying particles. You would want to know how they behave when they are alone, and how they interact when they are brought together. In number theory, our "particles" are the integers, and the "fundamental particles," the indivisible building blocks, are the prime numbers. The ​​Fundamental Theorem of Arithmetic​​ is our guiding principle: every integer greater than 1 can be written as a unique product of prime numbers. This isn't just a curiosity; it's the bedrock upon which much of number theory is built.

An ​​arithmetic function​​ is simply a function that takes a positive integer and gives back a number, usually a complex number. We can think of it as a property or a measurement we assign to each integer. A natural question arises: if we know the value of a function on the primes, do we know its value for all numbers? If a function is to be considered "simple" or "fundamental," we might hope that it respects the multiplicative way numbers are built from primes. This very idea leads us to a crucial distinction.

It turns out there are two levels of "good behavior" when it comes to multiplication.

First, we have ​​multiplicative​​ functions. A function $f$ is multiplicative if $f(1)=1$ and whenever you take two numbers $m$ and $n$ that are coprime—meaning they share no common prime factors (like 8 and 15)—the function value of their product is just the product of their function values:

$f(mn) = f(m)f(n) \quad \text{whenever } \gcd(m,n)=1$

This is a beautiful property. It means if we break a number down into its pairwise coprime prime-power components, say $n = p_1^{k_1} p_2^{k_2} \cdots p_r^{k_r}$, we can understand the function's value on $n$ by understanding its value on each prime power piece, $p_i^{k_i}$. Specifically, $f(n) = f(p_1^{k_1}) f(p_2^{k_2}) \cdots f(p_r^{k_r})$. This is powerful: to understand a multiplicative function, we don't need to study it on all integers, just on prime powers!.

But there is an even stronger, more "perfect" form of behavior. A function is ​​completely multiplicative​​ if it doesn't care about the coprime condition at all. For any two positive integers $m$ and $n$, the property holds:

$f(mn) = f(m)f(n) \quad \text{for all } m,n \in \mathbb{N}$

This is the ultimate dream of multiplicative simplicity. Any completely multiplicative function is, by definition, also multiplicative. But as we will see, the reverse is certainly not true, and the distinction is not trivial—it reveals a deep structural truth about the world of numbers.

How can we tell these two types of functions apart? The secret lies in what we call the ​​prime power test​​. Let's look at what happens when we feed a function powers of a single prime, like $p^2, p^3$, and so on.

If a function $f$ is ​​completely multiplicative​​, then we can write $f(p^k) = f(p \cdot p^{k-1}) = f(p)f(p^{k-1})$. Repeating this, we find a rigid constraint:

$f(p^k) = (f(p))^k$

This is a staggering simplification! It means that a completely multiplicative function is entirely determined by its values on the prime numbers alone. If you tell me $f(2)$, $f(3)$, $f(5)$, and so on for all primes, I can tell you $f(n)$ for any integer $n$ just by using its prime factorization.

However, if a function is merely ​​multiplicative​​, there is no such constraint. The values $f(p)$, $f(p^2)$, $f(p^3)$, ... for a given prime $p$ can be completely unrelated to each other. They can be assigned arbitrarily, and the function's value on other integers will be built from these prime-power values. This freedom is the defining difference between the two classes.

Let's see these principles in action by visiting a small zoo of famous arithmetic functions.

​​The Straight Arrows (Completely Multiplicative):​​

• ​​Power Functions:​​ The function $f(n) = n^s$ for some fixed complex number $s$ is completely multiplicative. This is clear from the laws of exponents: $(mn)^s = m^s n^s$ for all $m,n$.
• ​​The Liouville Function $\lambda(n)$:​​ This fascinating function is defined as $\lambda(n) = (-1)^{\Omega(n)}$, where $\Omega(n)$ is the total number of prime factors of $n$ counted with multiplicity (e.g., $\Omega(12) = \Omega(2^2 \cdot 3) = 2+1=3$). Since the total number of prime factors in a product is the sum of the totals for each factor, $\Omega(mn) = \Omega(m)+\Omega(n)$, it follows directly that $\lambda(mn) = \lambda(m)\lambda(n)$ for all $m,n$.

​​The More Nuanced Characters (Multiplicative Only):​​

• ​​Euler's Totient Function $\varphi(n)$:​​ This function counts the number of positive integers up to $n$ that are relatively prime to $n$. It is famously multiplicative, a cornerstone of number theory. But is it completely multiplicative? Let's use the prime power test. We find $\varphi(2)=1$. If it were completely multiplicative, we would expect $\varphi(4) = \varphi(2^2) = (\varphi(2))^2 = 1^2 = 1$. But a quick count shows that the numbers coprime to 4 are {1, 3}, so $\varphi(4)=2$. The property fails! The values for prime powers, $\varphi(p^k) = p^k - p^{k-1}$, do not follow the simple $(f(p))^k$ rule.
• ​​The Divisor Function $\tau(n)$:​​ This function counts the number of positive divisors of $n$. It is also multiplicative. Let's test it. The divisors of 2 are {1, 2}, so $\tau(2)=2$. If it were completely multiplicative, we'd need $\tau(4) = \tau(2)^2 = 4$. But the divisors of 4 are {1, 2, 4}, so $\tau(4)=3$. Again, a failure! The prime power test reveals its true nature.
• ​​The Möbius Function $\mu(n)$:​​ This is another central function in number theory. It is multiplicative, but a quick check shows $\mu(4) = 0$ while $\mu(2)\mu(2) = (-1)(-1) = 1$. It is not completely multiplicative. Similarly, the ​​square-free indicator function​​ $\mu(n)^2$, which is 1 if $n$ has no repeated prime factors and 0 otherwise, is also multiplicative but not completely multiplicative.

Now for a more subtle idea. What happens when we combine functions? A particularly fruitful way to do this is with the ​​Dirichlet convolution​​, denoted by a star *. The convolution of two functions $f$ and $g$ is a new function $(f*g)$ defined as:

$(f*g)(n) = \sum_{d|n} f(d) g(n/d)$

This operation might seem strange at first, but it appears naturally all over number theory. It's a way of "blending" the functions' values over all the divisors of a number. A beautiful and profound fact is that the set of multiplicative functions is closed under this operation. If you convolve two multiplicative functions, you get another multiplicative function! The proof of this relies directly on unique factorization to cleverly split the sum over divisors.

This leads to a wonderful question: is the same true for the "purer" class of completely multiplicative functions? Is the convolution of two completely multiplicative functions also completely multiplicative?

The answer is a surprising and illuminating ​​no​​.

Let's take the simplest completely multiplicative function imaginable: the constant-one function, $\mathbf{1}(n) = 1$ for all $n$. Let's convolve it with itself. $(\mathbf{1} * \mathbf{1})(n) = \sum_{d|n} \mathbf{1}(d) \mathbf{1}(n/d) = \sum_{d|n} 1 \cdot 1 = \sum_{d|n} 1$ This sum simply counts how many divisors $n$ has. We have just rediscovered the divisor function, $\tau(n)$! So, $\mathbf{1} * \mathbf{1} = \tau$. And we just saw in our gallery that $\tau(n)$ is a classic example of a function that is not completely multiplicative.

This discovery is more than just a curiosity; it points to a deep structural feature of arithmetic functions. In abstract algebra, a ​​group​​ is a set with an operation that satisfies certain closure, identity, and inverse properties. The set of all invertible arithmetic functions (those with $f(1) \neq 0$) forms a group under Dirichlet convolution.

Within this vast universe of functions, the set of ​​multiplicative functions​​ forms a beautiful, self-contained sub-universe. It is a ​​subgroup​​. It contains the identity element, the convolution of any two multiplicative functions is still multiplicative, and the inverse of any multiplicative function is also multiplicative.

But the set of ​​completely multiplicative functions​​ does not form a subgroup. As we just witnessed, it is not closed under the group operation—the convolution of two completely multiplicative functions can produce a result that is merely multiplicative. This property of complete multiplicativity is fragile; it can be broken by the very operations that define the algebraic landscape.

This distinction, which began as a simple definition about how functions interact with products, has blossomed into a fundamental insight. It is a "broken symmetry" at the heart of number theory, reminding us that even in the most abstract of worlds, some properties are more robust than others, and understanding which ones break—and how—is often the key to deeper discovery.

Having grasped the foundational principle of complete multiplicativity—that a function's behavior is dictated entirely by its dialogue with the prime numbers—we might wonder: What is this concept for? Is it merely a classificatory scheme, a label we affix to certain functions in the vast zoo of number theory? The answer, you will be delighted to find, is a resounding no. Completely multiplicative functions are not static exhibits; they are active, powerful tools. They are the finely tuned instruments through which we can listen to the symphony of the integers, uncovering harmonies and structures that would otherwise remain hidden.

Their power stems from a single, elegant fact we have already encountered: a completely multiplicative function is uniquely determined by its values on the prime numbers. This is not just a definition; it is a license to build the infinite from the finite. If you tell me how such a function $f$ behaves on the primes $2, 3, 5, \dots$, I can tell you its value on any integer, no matter how colossal. This "atomic principle" has profound consequences. For instance, if a completely multiplicative function is defined to be zero on all primes greater than 3, its associated infinite Dirichlet series $\sum f(n)/n^s$ miraculously collapses into the product of just two simple geometric series, a calculation that would be impossible otherwise. This is the first hint of their magic: they transform the intractable multiplicative structure of all integers into the more manageable, additive structure of their prime exponents.

This magic is most powerfully expressed when we move from the discrete world of integers to the continuous landscape of complex analysis. The bridge between these worlds is the Dirichlet series, $L(s, f) = \sum_{n=1}^\infty f(n) n^{-s}$. For a completely multiplicative function, this infinite sum blossoms into an infinite product over the primes, known as an Euler product:

This formula is a Rosetta Stone. It tells us that the global behavior of the function, encoded in the infinite sum over all integers, is equivalent to its local behavior at each prime, encoded in the infinite product.

Nature, it seems, has its favorite completely multiplicative functions—a cast of characters that appear time and again in the drama of numbers. We have already met some of the simplest ones: the constant function $\mathbf{1}(n) = 1$, whose Dirichlet series is the famed Riemann zeta function $\zeta(s)$; the identity function $\varepsilon(n)$ (1 at $n=1$, 0 otherwise), which acts as a neutral element; and the power functions $N^k(n) = n^k$, which serve to "twist" or shift other series.

But the cast includes more subtle and fascinating players. Consider the ​​Liouville function​​, $\lambda(n) = (-1)^{\Omega(n)}$, where $\Omega(n)$ is the total number of prime factors of $n$ (counted with multiplicity). This function acts as a parity checker for the prime constituents of a number. Is the sequence of values $+1, -1, -1, +1, -1, +1, \dots$ random? One might think so. But its Euler product reveals a breathtaking secret:

This seemingly erratic sequence is governed by an elegant relationship between the Riemann zeta function and itself! This identity is a gateway to deep analytic number theory. It allows us to study the convergence properties of the Liouville series by analyzing the poles and zeros of the zeta function, revealing a rightmost pole at $s=1/2$ and thus an abscissa of convergence of $1/2$. The statistical properties of integers are encoded in the analytic properties of this complex function.

Perhaps the most important players are the ​​Dirichlet characters​​, $\chi(n)$. These functions are not only completely multiplicative but also periodic. One can think of them as the fundamental harmonics of the integers, the number-theoretic equivalent of sine and cosine waves in Fourier analysis. Just as any sound can be decomposed into pure tones, any arithmetic progression can be isolated using these characters. This is achieved through their remarkable "orthogonality relations", which allow them to act as perfect filters. By summing over all characters $\chi$ modulo some integer $q$, we can construct a function that is non-zero only for integers in a specific residue class modulo $q$. This is the key that unlocks the secrets of prime numbers in arithmetic progressions, a monumental achievement of Dirichlet.

The structures woven by completely multiplicative functions are so fundamental that their patterns resonate far beyond the confines of number theory. We find their echoes in abstract algebra, where the set of all arithmetic functions forms a rich algebraic structure known as a ring under addition and Dirichlet convolution. Within this ring, the complete multiplicativity of functions like the Liouville function $\lambda(n)$ allows for elegant computations that would otherwise be messy, such as finding a closed form for the threefold convolution $(\lambda * \lambda * \lambda)(n)$.

Even more surprisingly, we find a connection to signal processing and physics. The Laplace transform is a standard tool for engineers to move from a time-domain signal to a frequency-domain representation. What could this possibly have to do with prime numbers? Yet, if we take the function $F(s) = \zeta(2s)/\zeta(s)$, which we know to be the Dirichlet series of the Liouville function, and ask for its inverse Laplace transform, the answer is astonishing. The result is a "signal" composed of an infinite series of sharp spikes—Dirac delta functions—located at times $t = \ln n$, with the polarity of each spike determined by the Liouville function $\lambda(n)$:

It is as if the integers are broadcasting a signal through time, with pulses occurring at logarithmic intervals, and the Liouville function is flipping the phase of the transmission at each step. This beautiful correspondence underscores a deep unity in the mathematical sciences.

One might think that such a simple concept belongs to the classical age of mathematics. Yet, completely multiplicative functions are at the heart of some of the most exciting modern developments in analytic number theory. A central question is: when does a sum like $\sum_{n \le x} f(n)$ get large? When does it exhibit cancellation, hovering near zero?

The modern "pretentious" framework, pioneered by Andrew Granville and Kannan Soundararajan, provides a powerful and intuitive answer. The core idea is that a multiplicative function $f$ can only have a large average value if it "pretends" to be a very simple function, like $g(n) = n^{it}$ for some real number $t$. If the values of $f(p)$ at primes do not consistently align with the values of any such simple function, its partial sums will behave like a random walk, exhibiting significant cancellation.

This notion of "pretending" is made rigorous with a "pretentious distance," which measures the alignment between two functions $f$ and $g$ on the primes:

A small distance means $f$ mimics $g$ well. This simple but profound idea has led to new, simpler proofs of classical theorems and has become a central tool for understanding the distribution of number-theoretic sequences. It tells us that the "music" of a multiplicative function is either simple and resonant (if it pretends to be a basic tone) or complex and self-cancelling (if it does not).

From a simple definition, we have journeyed through the heart of analytic number theory, glimpsed its reflection in physics and algebra, and arrived at the edge of modern research. The completely multiplicative functions, these pure tones of the integers, are indeed far more than a mere curiosity. They are a fundamental concept, a key that continues to unlock the deepest secrets of the world of numbers.