Skip to main content
Have a personal or library account? Click to login
On the Arithmetic Properties of A Truncated Power-Residue Function and Related Diophantine Equations Cover

On the Arithmetic Properties of A Truncated Power-Residue Function and Related Diophantine Equations

Open Access
|Nov 2026
Read online

Full article

1. Introduction

Let n be a positive integer with canonical prime factorization

n=p1α1p2α2⋯pkαk,
where k ≥ 0, α1, α2, . . . , αk are positive integers, and p1, p2, . . . , pk are distinct primes. The Euler totient function ϕ(n) counts the number of positive integers up to n that are coprime to n and is given by the well-known formula:
ϕ(n)=p1α1−1(p1−1)p2α2−1(p2−1)⋯pkαk−1(pk−1).

For a fixed integer m ≥ 2, and for any integer n ≥ 1, we define the arithmetic function κm(n) by the product formula

Eq (1)
κm(n)=∏pm∤npvp(n)×∏pm‖npm−1,
where the first product runs over all primes p such that pm does not divide n, and the second product runs over all primes p such that pm exactly divides n. Here, vp(n) denotes the p-adic valuation of n. This function extracts from n its largest divisor that is free of m-th powers. In particular, κ2(n) coincides with the classical square-free kernel (or radical) of n, often denoted rad(n) = Πp|n p.

Many mathematicians have investigated Diophantine equations involving arithmetic functions, particularly those combining ϕ(n) with other number-theoretic functions. For example, Sándor and Bhattacharjee [8] studied certain equations and inequalities involving the arithmetical functions ϕ(n) and d(n). In [5], Ford, Luca, and Pomerance examined common values of the arithmetic functions ϕ and σ. The equation ϕ(n) = ϕ(n + 1) was investigated by Kinlaw, Kobayashi, and Pomerance [7], who established new results on this equation. More recently, Bellaouar, Togbé, and Jakimczuk [2] studied equations of the form d(n) = d(ϕ(n)) and established several new results on the distribution of solutions.

The function κm(n) belongs to a broader class of arithmetic functions defined through constraints on prime power exponents. Related functions have been studied in the context of power-free integers [1], k-free numbers [9], and divisor problems [6]. However, the specific properties of κm(n) as defined above, particularly its interaction with Euler’s totient function, have not been systematically investigated.

The structural characterization of κm(n) as the largest m-power-free divisor plays a central role in simplifying the Diophantine analysis.

The outline of our work is as follows. First, we provide equivalent definitions of κm(n) and establish its fundamental properties, in particular multiplicativity and the characterisation of the greatest divisor. Second, we completely solve the Diophantine equation ϕ(n) = κm(n), thus studying the divisibility relation ϕ(n) | κm(n) and classifying all solutions. Finally, we study the equation κm(n) = d for fixed d and derive the structural properties of the set of solutions.

Throughout this paper, we use standard notation from number theory: ℕ* denotes the set of positive integers, 𝒫 denotes the set of prime numbers, vp(n) denotes the p-adic valuation of n. The symbol pα|n means that pα | n but pα+1 ∤ n. In this work, we focus on a specific truncation function that captures the largest divisor of an integer free of m-th powers.

2. Preliminaries

We now establish equivalent formulations, which are often more convenient for calculations, and thus we deal with some Diophantine equations that essentially use the function κm.

Lemma 2.1.

Let n=∏i=1kpiαi be the canonical prime factorization of n, with αi ≥ 1 for all i ∈ {1, . . . , k}. Then the following identity holds:

Eq (2)
κm(n)=∏i=1kpimin{m−1,αi}.

Proof.

Let n=∏i=1kpiαi with αi ≥ 1. For each prime pi appearing in the factorization of n, we have vpi(n) = αi. Moreover, we observe the elementary equivalence:

pim‖n⇔m≤αi, pim∤n⇔αi<m.

Consequently, the set {p1, . . . , pk} admits the disjoint partition:

{p1, …, pk}={pi:αi<m}⊔{pi:αi≥m}.

Now, let us consider the product ∏i=1kpimin{m−1,αi} . We distinguish two cases:

  • i) If αi < m, then αi ≤ m − 1, whence min{m − 1, αi} = αi = vpi(n). In this situation, pim∤n by the above equivalence.

  • ii) If αi ≥ m, then m − 1 < αi, whence min{m − 1, αi} = m − 1. In this situation, pim||n .

Therefore, we may rewrite

∏i=1kpimin{m−1,αi}=(∏i:αi<mpiαi)×(∏i:αi≥mpim−1)                                                                    =(∏pm∤npvp(n))×(∏pm||npm−1),
which establishes the desired identity.□

The following lemma provides a structural interpretation of κm(n) in terms of divisor theory.

Lemma 2.2.

Let n ∈ ℕ* be a positive integer with canonical factorization n=∏i=1kpiαi . The function κm(n) defined by

κm(n)=∏i=1kpimin{m−1,αi}
coincides with the largest divisor of n that is free of m-th powers. More precisely
Eq (3)
κm(n)=max {d∈ℕ*:d|n and ∀p∈P,pm∤d}.

Proof.

Denote by

En={d∈ℕ*:d|n  and  ∀p∈P,pm∤d}
the set of divisors of n that are m-power-free. We will prove that κm(n) = max(En).

By Lemma 2.1, the expression (2) is equivalent to the partitioned product (1). However, for the maximality argument, the form ∏i=1kpiβi is more convenient.

  • i) Verification that κm(n) ∈ En. For each i ∈ {1, . . . , k}, we have by definition βi ≤ αi. This implies that κm(n) divides n. Furthermore, again by definition, βi ≤ m − 1 < m. Thus, for every prime p, the valuation vp(κm(n)) is strictly less than m. Consequently, pm ∤ κm(n) for all primes p, which means that κm(n) is m-power-free. Hence κm(n) ∈ En.

  • ii) Proof of maximality. Let d ∈ En be an arbitrary divisor of n that is m-power-free. Write d=∏i=1kpiγi for its prime factorization (where some γi may be zero). The condition d | n implies

Eq (4)
0≤γi≤αi, ∀i∈{1, …, k}.

The m-power-free condition implies that for each i, vpi(d) < m, i.e.

Eq (5)
γi≤m−1, ∀i∈{1, …, k}.

Combining inequalities (4) and (5), we obtain for each index i:

γi≤ {αi, m−1}=βi.

This implies that d divides κm(n) and consequently d ≤ κm(n). Since κm(n) belongs to En and dominates every element of En, we conclude that κm(n) = max(En), as claimed. □

Remark 2.3.

The characterization (3) reveals that κm(n) may be viewed as a natural generalization of the square-free kernel (or radical) of n, which corresponds to the special case m = 2. Indeed, when m = 2, we recover κ2(n) = Πp|n p, the classical radical function.

Corollary 2.4.

For any prime power pα with α ≥ 1, we have:

Eq (6)
κm(pα)={pαif α<m,pm−1if α≥m.

Proof.

Formula (6) is a direct specialization of (2) to the case of a single prime factor.□

Proposition 2.5.

The arithmetic function κm is multiplicative. That is, κm(1) = 1 and for any coprime positive integers a and b (i.e., gcd(a, b) = 1), we have:

κm(ab)=κm(a)κm(b).

However, κm is not completely multiplicative in general.

Proof.

We first verify the normalization condition. For n = 1, the prime factorization is empty, and the empty product in (2) yields κm(1) = 1.

Now let a, b ∈ ℕ* with gcd(a, b) = 1. Write their canonical prime factorizations as:

a=∏i=1rpiαi, b=∏j=1sqjβj,
where {p1, . . . , pr} ∩ {q1, . . . , qs} = ∅ since a and b are coprime. Then the factorization of the product is:
ab=∏i=1rpiαi×∏j=1sqjβj.

Applying the definition given by (2) to ab, we obtain

κm(ab)=∏i=1rpimin {m−1,αi}×∏j=1sqjmin{m−1,βj}=κm(a)κm(b),
which establishes the multiplicativity of κm. Multiplicativity follows immediately from the definition (1), as the condition pm|n and the valuation vp(n) depend only on the local behavior at each prime.

To see that κm is not completely multiplicative, let us give a counterexample. Fix m ≥ 3 and let p be any prime. Take a = b = pm−1. Then, we get κm(a) = κm(pm−1) = pm−1, κm(b) = pm−1, so κm(a)κm(b) = p2(m−1). On the other hand, ab = p2(m−1) with 2(m − 1) ≥ m (since m ≥ 3). Hence, one can see that κm(ab) = κm(p2(m−1)) = pm−1 ≠ p2(m−1) = κm(a)κm(b). Since pm−1 ≠ p2(m−1) for m ≥ 3, we conclude that κm(ab) ≠ κm(a)κm(b) when a and b are not coprime. Therefore, κm is multiplicative but not completely multiplicative.□

Remark 2.6.

The multiplicativity of κm is a direct consequence of its definition via local factors at each prime, as seen in (1). This property is essential for deriving Dirichlet series representations and mean value estimates, which remains a direction for future work.

Proposition 2.7.

Let 2 ≤ m < r be fixed integers. The equation

κm(n)=κr(n)
holds if and only if n is m-power-free (i.e., ∀p ∈ 𝒫, pm ∤ n). In this case, κm(n) = κr(n) = n.

Proof.

We begin by assuming that n is m-power-free. Then for every prime p, we have vp(n) ≤ m − 1 < r − 1. By Lemma 2.1 we have

κm(n)=∏ppmin{m−1,vp(n)}=∏ppvp(n)=n,κr(n)=∏ppmin{r−1,vp(n)}=∏ppvp(n)=n.

Therefore, κm(n) = κr(n) = n.

Now suppose that κm(n) = κr(n). Let n=∏i=1kpiαi be the canonical factorization. By Lemma 2.2, κm(n) is the largest m-power-free divisor of n, and κr(n) is the largest r-power-free divisor of n. Since m < r, every m-power-free divisor is also r-power-free. Therefore, one can see that κm(n) | κr(n).

The equality κm(n) = κr(n) implies that the largest r-power-free divisor of n is actually m-power-free. We prove this forces n to be m-power-free. Suppose, for contradiction, that n is not m-power-free. Then there exists a prime p such that vp(n) = α ≥ m. We have:

vp(κm(n))=min{m−1,α}=m−1,vp(κr(n))= min{r−1,α}.
  • – If m ≤ α ≤ r − 1, then vp(κr(n)) = α ≥ m > m − 1 = vp(κm(n)), contradiction.

  • – If α ≥ r, then vp(κr(n)) = r − 1 > m − 1 = vp(κm(n)) (since r > m), contradiction.

Therefore, no such prime exists, and n must be m-power-free.□

3. Statements and proofs of the main results

In this section, we will prove three main results related to Diophantine equations involving κm(n). Here is the first theorem.

Theorem 3.1.

Let m ≥ 2 be a fixed integer. Then the equation

Eq (7)
ϕ(n)=κm(n)
has exactly m + 1 solutions, namely n = 1, n = 2m, and n = 2α · 3m for α = 1, 2, . . . , m − 1, where ϕ denotes Euler’s totient function.

Proof.

Recall that Euler’s totient function ϕ(n) is defined for any integer n ≥ 1 as the number of positive integers not exceeding n that are relatively prime to n. If n=∏i=1kpiαi denotes the canonical prime factorization of n, then ϕ(n) admits the explicit product formula

Eq (8)
ϕ(n)=∏i=1kpiαi−1(pi−1)=p1α1−1(p1−1)⋯pkαk−1(pk−1).

This classical result is fundamental in elementary number theory (see, for instance, Apostol [1] or Hardy and Wright [6]).

According to definition (2) in Lemma 2.1, we have

Eq (9)
κm(n)=∏i=1kpimin{m−1,αi}.

It is clear that n = 1 is a solution to equation (7) since ϕ(1) = 1 = κm(1). Let us now assume that n > 1 and consider its prime factorisation n=∏i=1kpiαi with αi ≥ 1 for all i. We will distinguish between four cases.

  • Case (i): If there exists an index i such that αi > m, then αi − 1 ≥ m while max1≤j≤k{min{m − 1, αj}} ≤ m − 1. In this case, we have the valuation inequality:

    vpi(ϕ(n))≥αi−1+∑j≠ivpi(pj−1)≥αi−1≥m,
    whereas
    vpi(κm(n))=m−1.

    Thus equality is impossible. By comparing (8) and (9), we see that ϕ(n) ≠ κm(n). Thus, there is no solution in this case.

  • Case (ii): If α1 = α2 = · · · = αk = m, so according to (9) we get κm(n)=p1m−1p2m−1⋯pkm−1 , and according to (8) we have ϕ(n)=p1m−1(p1−1)p2m−1(p2−1)⋯pkm−1(pk−1) . Therefore, the equality ϕ(n) = κm(n) implies (p1 − 1)(p2 − 1) · · · (pk − 1) = 1, which is only possible if k = 1 and p1 = 2. Therefore, the only solution in this case is n = 2m.

  • Case (iii): If max1≤i≤k{αi} < m, for all i such that αi ≤ m − 1, then for all i, we have min{m − 1, αi} = αi. Thus, one obtains κm(n)=∏ki=1piαi=n , whereas ϕ(n) < n for all n > 1. Therefore, ϕ(n) ≠ κm(n) for all n in this case and equation (7) has no solution.

  • Case (iv): We assume that among the exponents of the prime factorization of n, at least one is equal to m and at least one is strictly less than m. To understand this delicate situation well, we will raise four basic partial assumptions.

    • a) Let n=p1α1p2α2⋯pkαk with p1 < p2 < · · · < pk. We claim that p1 = 2. Let us reason by contradiction. Suppose that all prime factors of n are odd (i.e., pi ≥ 3 for all i). Then:

      • – ϕ(n)=∏i=1kpiαi−1(pi−1) is even, since each factor pi − 1 is even.

      • – κm(n)=∏i=1kpimin{m−1,αi} is odd, as it is a product of odd primes.

        This contradicts ϕ(n) = κm(n). Therefore, we must have p1 = 2.

    • b) We now determine the possible value of α1. There are two possibilities:

      • – If α1 = m and all other exponents also equal m, then we are in Case (ii), which yields the solution n = 2m. However, Case (iv) requires at least one exponent to be strictly less than m. If α1 = m but some αk < m for k ≥ 2, then comparing pk-adic valuations vpk(ϕ(n)) = αk − 1, vpk(κm(n)) = min{m − 1, αk} = αk (since αk < m). This gives αk − 1 = αk, a contradiction. Hence α1 ≠ m.

      • – Let α1 < m. This is consistent with the assumptions in Case (iv). Under this assumption, we compute the 2-adic valuations of both sides of the equation ϕ(n) = κm(n): v2(ϕ(n)) = α1 − 1, v2(κm(n)) = min{m − 1, α1} = α1 (since α1 < m). Consequently, κm(n) contains exactly one more factor of 2 than ϕ(n) from the power of 2 alone. This deficit must be compensated by the factors (pi − 1) arising from the odd prime divisors pi of n, each of which contributes at least one factor of 2 to ϕ(n). This observation will be crucial in restricting the number of odd prime factors of n. Therefore, in Case (iv), we must have α1 < m.

    • c) Since α1 < m, there must exist at least one index k ≥ 2 such that αk = m (by the hypothesis of Case (iv)). Without loss of generality, after relabeling, we may assume αk = m.

      The equation ϕ(n) = κm(n) becomes

      2a1−1·p2a2−1(p2−1)…pkm−1(pk−1)=2a1·p2min{m−1,a2}…pk−1min{m−1,ak−1}·pkm−1.

      Canceling 2α1−1 from both sides:

      Eq (10)
      p2α2−1(p2−1)⋯pkm−1(pk−1)                                                =2⋅p2min{m−1,α2}⋯pk−1min{m−1,αk−1}⋅pkm−1.

      We perform a 2-adic evaluation analysis where we see that each factor (pi − 1) for i ≥ 2 is even (since pi is odd). The right-hand side of (10) contains exactly one factor of 2. Therefore, there can be only one factor of the form (pi − 1) on the left-hand side. This implies k = 2, meaning n has exactly two distinct prime factors n = 2α · pm, where α = α1 < m.

    • d) Substituting n = 2αpm into ϕ(n) = κm(n):

      ϕ(2αpm)=2α−1⋅pm−1(p−1),       κm(2αpm)=2min{m−1,α}⋅pm−1.

    The equality becomes 2α−1 · pm−1(p − 1) = 2min{m−1,α} · pm−1. Canceling pm−1 and using α < m (so min{m − 1, α} = α):

    2α−1(p−1)=2α⇒p−1=2⇒p=3.

    If we summarize the study conducted in Case (iv), we find that the solutions in this case are exactly as follows:

    n=2α⋅3m    with     α∈{1,2, …, m−1}.

    This yields exactly m − 1 distinct solutions. By combining the above four cases, we immediately obtain all m+1 solutions to the equation ϕ(n) = κm(n), namely n = 1, n = 2m, and n = 2α ·3m for α = 1, 2, . . . , m−1. This completes the proof of the theorem.□

The second result involves the divisibility of κm(n) by ϕ(n).

Theorem 3.2.

Let m ≥ 2 be a fixed integer. The divisibility ϕ(n) | κm(n) holds if and only if n is of one of the following conditions:

  • i) n = 1;

  • ii) n = 2α with 1 ≤ α ≤ m;

  • iii) n = 2α · 3β with 1 ≤ α ≤ m − 1 and 1 ≤ β ≤ m.

Consequently, the equation ϕ(n) | κm(n) has exactly m2 + 1 positive integer solutions.

Proof.

Recall that for n=∏i=1kpiαi ,

ϕ(n)=∏i=1kpiαi−1(pi−1),        κm(n)=∏i=1kpimin{m−1,αi}

The condition ϕ(n) | κm(n) is equivalent to, for every prime p, we have vp(ϕ(n)) ≤ vp(κm(n)).

  • i) If n = 1, then we have ϕ(1) = 1 and κm(1) = 1. Clearly, 1 | 1, so n = 1 is a solution.

  • ii) If n = pα, then one can see that ϕ(pα) = pα−1(p − 1), κm(pα) = pmin{m−1,α}. If pα−1(p − 1) | pmin{m−1,α}, we need to consider two conditions:

    • – Condition A: α − 1 ≤ min{m − 1, α}.

    • – Condition B: (p − 1) | pmin{m−1,α}−(α−1).

    According to Condition B, if p > 2, then p − 1 is even and p − 1 ≥ 2. But pk is odd, for all k ≥ 0. Therefore, (p − 1) ∤ pk, for any k. We deduce that p = 2.

    For p = 2, Condition A becomes α − 1 ≤ min{m − 1, α}.

    • – If α ≤ m − 1, then min{m − 1, α} = α, so α − 1 ≤ α. This is always true.

    • – If α ≥ m, then min{m − 1, α} = m − 1. So, we get α − 1 ≤ m − 1, which gives α ≤ m.

    Therefore, we must have 1 ≤ α ≤ m. So, we see that all the solutions must be of the form n = 2α with 1 ≤ α ≤ m.

  • iii) If n has at least two distinct prime factors, then we write n=p1α1⋯pkαk with k ≥ 2 and p1 < p2 < · · · < pk. Thus, we have ϕ(n)=∏i=1kpiαi−1(pi−1) .

    In order for us to continue, we have this key observation: For each i, the factor pi − 1 must divide κm(n). Since κm(n)=∏j=1kpjmin{m−1,αj} , any prime divisor q of pi − 1 must divide κm(n), which means q ∈ {p1, p2, . . . , pk}.

Now consider the smallest prime factor p1 of n. All prime divisors of p1 − 1 are strictly less than p1 (since p1 − 1 < p1). But these divisors must belong to {p1, p2, . . . , pk}, all of which are ≥ p1. Therefore, the only possibility is p1 − 1 = 1, i.e. p1 = 2.

  • a) If n = 2α · pβ, with p an odd prime, then we have

    ϕ(n)=2α−1⋅pβ−1(p−1),       κm(n)=2min{m−1,α}⋅pmin{m−1,β}.

    Since (p − 1) | 2min{m−1,α} · pmin{m−1,β} and gcd(p − 1, p) = 1 (as p is odd), we must have (p − 1) | 2min{m−1,α}. Thus, p − 1 is a power of 2. Furthermore, we have v2(ϕ(n)) = (α − 1) + v2(p − 1), v2(κm(n)) = min{m − 1, α}. Therefore, we obtain α − 1 + v2(p − 1) ≤ min{m − 1, α}.

    This results in the following two sub-cases.

    • – If α ≤ m−1, then min{m−1, α} = α. So, one can see that α−1+v2(p−1) ≤ α ⇒ v2(p − 1) ≤ 1. Thus, p − 1 ∈ {1, 2}, giving p ∈ {2, 3}. Since p is odd, we must have p = 3.

    • – If α ≥ m, then min{m − 1, α} = m − 1. So, we get α − 1 + v2(p − 1) ≤ m − 1 ⇒ v2(p − 1) ≤ m − α. Since α ≥ m, we have m − α ≤ 0. Hence, v2(p − 1) = 0, which gives p = 2. This contradicts the fact that p is odd. In conclusion, we must have α ≤ m − 1 and p = 3.

      To obtain a condition on 3-adic valuations (for p = 3), we have v3(ϕ(n)) = β − 1, v3(κm(n)) = min{m − 1, β}. Thus, we see that β − 1 ≤ min{m − 1, β}.

    • – If β ≤ m − 1, then β − 1 ≤ β, which is always true.

    • – If β ≥ m, then β − 1 ≤ m − 1, so β ≤ m. Therefore, 1 ≤ β ≤ m.

    So, in this case the solutions are integers of the form n = 2α · 3β, with 1 ≤ α ≤ m − 1 and 1 ≤ β ≤ m.

  • b) Suppose n = 2α · 3β · qγ · · · with q ≥ 5. Let q ≥ 5 be the smallest prime divisor of n greater than 3. Then every prime divisor r of q − 1 satisfies r < q.

Since r | n, the minimality of q implies r ∈ {2, 3}. Hence q − 1 = 2a3b with a ≥ 1, since q is odd.

The valuation argument then gives

v2(ϕ(n))≥(α−1)+1+a=α+a,
whereas
v2(κm(n))=min{m−1,α}.

The divisibility condition would therefore imply

α+a≤min{m−1,α},
which is impossible because a ≥ 1.

We conclude from the study that it is impossible to have three or more distinct prime factors. The total number of solutions for the equation stated in the theorem is equal to 1 + m + (m − 1)m = 1 + m + m2 − m = m2 + 1. This completes the proof of the theorem. □

The last result involves the Diophantine equation κm(n) = d.

Theorem 3.3.

Let m ≥ 2 be a fixed integer and let d ∈ ℕ*. The equation

κm(n)=d
admits a positive integer solution n if and only if d is m-power-free. In this case, the set of all solutions is
Sd={n=d⋅∏p|dvp(d)=m−1pγp:γp≥0}.

Equivalently, n verifies one of the following conditions:

  • – For every prime p | d with vp(d) < m − 1, we have vp(n) = vp(d);

  • – For every prime p | d with vp(d) = m − 1, we get vp(n) ≥ m − 1;

  • – For every prime p ∤ d, we obtain vp(n) = 0.

Proof.

By Lemma 2.2, κm(n) is the largest m-power-free divisor of n. In particular, κm(n) itself is always m-power-free. Therefore, if κm(n) = d, then d must be m-power-free. This establishes the necessary condition on d.

Conversely, assume d is m-power-free. We characterize all n such that κm(n) = d. Let n = Πp pαp and d = Πp pδp be their prime factorizations.

The condition κm(n) = d is equivalent to min{m − 1, αp} = δp for all primes p. Therefore, we can observe the following cases:

  • – If p ∤ d (i.e., δp = 0), then min{m − 1, αp} = 0, which forces αp = 0. Thus, we see that p ∤ n.

  • – If p | d with δp < m − 1, then min{m − 1, αp} = δp < m − 1, which implies that αp = δp. Thus, we obtain vp(n) = vp(d).

  • – If p | d with δp = m − 1, then min{m − 1, αp} = m − 1, which is satisfied for any αp ≥ m − 1. Thus, we deduce that vp(n) ≥ m − 1 = vp(d).

Equivalently, we can write vp(n) = vp(d) + γp, where γp ≥ 0 is arbitrary. Combining all cases, n must be of the form

n=(∏p|dvp(d)<m−1pvp(d)).(∏p|dvp(d)=m−1pγp)

This is equivalent to

n=d⋅∏p|dvp(d)=m−1pγp,γp≥0.

This completes the proof of the theorem. □

Remark 3.4.

The results established in this paper resonate with the celebrated conjecture of Carmichael [3], which asserts that the equation ϕ(x) = n cannot have a unique solution for any positive integer n. While Carmichael’s conjecture remains open despite significant progress by K. Ford [4], our investigation reveals a complementary perspective through the lens of the truncated power-residue function κm(n).

Specifically, Theorem 3.1 demonstrates that the equation ϕ(n) = κm(n) admits exactly m + 1 solutions for each fixed m ≥ 2, namely n ∈ {1, 2m} ⋓ {2α · 3m : 1 ≤ α ≤ m − 1}. This explicit classification shows that when the codomain is restricted to values of the form κm(n), the preimage under ϕ is never a singleton, a phenomenon consistent with Carmichael’s assertion for the classical equation.

Moreover, the structural characterization of κm(n) as the largest m-power-free divisor of n (Lemma 2.2) provides a novel tool for analyzing the multiplicity structure of ϕ-preimages. An interesting open question, and a possible direction for future research, is to investigate whether for any m-power-free integer d in the range of κm, the equation ϕ(x) = d admits at least two solutions whenever it admits one. This suggests a potential refinement of Carmichael’s conjecture restricted to m-power-free codomains, which could be more tractable than the general case.

Future work could investigate whether the techniques developed here, in particular the p-adic valuation analysis and the maximal divisor characterization, extend to broader classes of arithmetic functions and yield new insights into the long-standing problem of solution multiplicity for ϕ(x) = n.

DOI: https://doi.org/10.2478/amsil-2026-0014 | Journal eISSN: 2391-4238 (formerly 0860-2107) | Journal ISSN: 0860-2107
Language: English
Submitted on: Jun 5, 2026
Accepted on: Oct 20, 2026
Published on: Nov 13, 2026
Published by: University of Silesia in Katowice, Institute of Mathematics
In partnership with: Paradigm Publishing Services

© 2026 Mihoub Bouderbala, Alain Togbé, published by University of Silesia in Katowice, Institute of Mathematics
This work is licensed under the Creative Commons Attribution 4.0 License.