Loading [MathJax]/jax/element/mml/optable/BasicLatin.js
Research article

Some results on p-adic valuations of Stirling numbers of the second kind

  • Received: 25 December 2019 Accepted: 16 April 2020 Published: 30 April 2020
  • MSC : 11B73, 11A07

  • Let n and k be nonnegative integers. The Stirling number of the second kind, denoted by S(n,k), is defined as the number of ways to partition a set of n elements into exactly k nonempty subsets and we have S(n,k)=1k!ki=0(1)i(ki)(ki)n. Let p be a prime and vp(n) stand for the p-adic valuation of n, i.e., vp(n) is the biggest nonnegative integer r with pr dividing n. Divisibility properties of Stirling numbers of the second kind have been studied from a number of different perspectives. In this paper, we present a formula to calculate the exact value of p-adic valuation of S(n,nk), where nk+1 and 1k7. From this, for any odd prime p, we prove that vp((nk)!S(n,nk))<n if nk+1 and 0k7. It confirms partially Clarke's conjecture proposed in 1995. We also give some results on vp(S(apn,apnk)), where a and n are positive integers with (a,p)=1 and 1k7.

    Citation: Yulu Feng, Min Qiu. Some results on p-adic valuations of Stirling numbers of the second kind[J]. AIMS Mathematics, 2020, 5(5): 4168-4196. doi: 10.3934/math.2020267

    Related Papers:

    [1] Taekyun Kim, Dae San Kim, Jin-Woo Park . Degenerate r-truncated Stirling numbers. AIMS Mathematics, 2023, 8(11): 25957-25965. doi: 10.3934/math.20231322
    [2] Taekyun Kim, Dae San Kim, Hye Kyung Kim . Some identities involving degenerate Stirling numbers arising from normal ordering. AIMS Mathematics, 2022, 7(9): 17357-17368. doi: 10.3934/math.2022956
    [3] Waseem Ahmad Khan, Kottakkaran Sooppy Nisar, Dumitru Baleanu . A note on (p, q)-analogue type of Fubini numbers and polynomials. AIMS Mathematics, 2020, 5(3): 2743-2757. doi: 10.3934/math.2020177
    [4] Dae San Kim, Hye Kyung Kim, Taekyun Kim . New approach to λ-Stirling numbers. AIMS Mathematics, 2023, 8(12): 28322-28333. doi: 10.3934/math.20231449
    [5] Aimin Xu . Some identities connecting Stirling numbers, central factorial numbers and higher-order Bernoulli polynomials. AIMS Mathematics, 2025, 10(2): 3197-3206. doi: 10.3934/math.2025148
    [6] Feng Qi, Da-Wei Niu, Bai-Ni Guo . Simplifying coefficients in differential equations associated with higher order Bernoulli numbers of the second kind. AIMS Mathematics, 2019, 4(2): 170-175. doi: 10.3934/math.2019.2.170
    [7] Hye Kyung Kim . Note on r-central Lah numbers and r-central Lah-Bell numbers. AIMS Mathematics, 2022, 7(2): 2929-2939. doi: 10.3934/math.2022161
    [8] Letelier Castilla, William Ramírez, Clemente Cesarano, Shahid Ahmad Wani, Maria-Fernanda Heredia-Moyano . A new class of generalized Apostol–type Frobenius–Euler polynomials. AIMS Mathematics, 2025, 10(2): 3623-3641. doi: 10.3934/math.2025167
    [9] Taekyun Kim, Dae San Kim, Hyunseok Lee, Lee-Chae Jang . A note on degenerate derangement polynomials and numbers. AIMS Mathematics, 2021, 6(6): 6469-6481. doi: 10.3934/math.2021380
    [10] Dojin Kim, Patcharee Wongsason, Jongkyum Kwon . Type 2 degenerate modified poly-Bernoulli polynomials arising from the degenerate poly-exponential functions. AIMS Mathematics, 2022, 7(6): 9716-9730. doi: 10.3934/math.2022541
  • Let n and k be nonnegative integers. The Stirling number of the second kind, denoted by S(n,k), is defined as the number of ways to partition a set of n elements into exactly k nonempty subsets and we have S(n,k)=1k!ki=0(1)i(ki)(ki)n. Let p be a prime and vp(n) stand for the p-adic valuation of n, i.e., vp(n) is the biggest nonnegative integer r with pr dividing n. Divisibility properties of Stirling numbers of the second kind have been studied from a number of different perspectives. In this paper, we present a formula to calculate the exact value of p-adic valuation of S(n,nk), where nk+1 and 1k7. From this, for any odd prime p, we prove that vp((nk)!S(n,nk))<n if nk+1 and 0k7. It confirms partially Clarke's conjecture proposed in 1995. We also give some results on vp(S(apn,apnk)), where a and n are positive integers with (a,p)=1 and 1k7.


    Let n and k be nonnegative integers. Let (x)k and xk stand for the falling factorial and the rising factorial, which are defined by (x)k:=x(x1)(x2)(xk+1) if k1 and (x)0:=1, and xk:=x(x+1)(x+k1) if k1 and x0:=1, respectively. The Stirling number of the first kind, denoted by s(n,k), counts the number of permutations of n elements with k disjoint cycles. One can also characterize s(n,k) by

    (x)n=nk=0(1)nks(n,k)xk.

    The Stirling number of the second kind S(n,k) is defined as the number of ways to partition a set of n elements into exactly k nonempty subsets, and we have

    S(n,k)=1k!ki=1(1)ki(ki)in, (1.1)

    where (ki) represents the binomial coefficient, which is defined by

    (ki)=(k)ii!=k!i!(ki)!  (ki).

    One can also characterize the Stirling number of the second kind by

    xn=k=0S(n,k)(x)k.

    Given a prime p and a nonzero integer m, there exist unique integers a and r, with pa and r0, such that m=apr. The number r is called the p-adic valuation of m, denoted by r=vp(m). If x=m1m2, where m1 and m2 are integers and m20, then we define vp(x):=vp(m1)vp(m2). Let dp(m) denote the base p digital sum of m. Many authors studied the divisibility properties of Stirling numbers of the second kind. For each given k, the sequence {S(n,k),nk} is known to be periodic modulo prime powers. Carlitz [1] and Kwong [2] have studied the length of this period. Chan and Manna [3] characterized S(n,k) modulo prime powers in terms of binomial coefficients when k is a multiple of prime powers. Let p be a prime. For any positive integer n and k, define

    Tp(n,k):=ki=0pi(1)ki(ki)in.

    By formula (1.1) we can see that k!S(n,k)Tp(n,k) is divisible at least by pn. In this sense, Tp(n,k) is also known as Stirling-like numbers [4]. In 1990, Davis [5] gave a method for calculating v2(T2(n,5)) and v2(T2(n,6)). Then Clarke [4] generalized this result by applying Hensel's Lemma on the p-adic integers. He also conjectured that vp(k!S(n,k))=vp(Tp(n,k)) if nk. We note that Hong, Zhao and Zhao [6,7,8] have presented some important results on the divisibility properties of Stirling numbers of the second kind. In fact, they proved several conjectures proposed by Amdeberhan, Manna and Moll [9] and by Lengyel [10]. We refer the readers to [11,12,13,14,15,16] for other results on the divisibility properties of Stirling numbers of the both kinds.

    Let n and k be positive integers. For any positive integer r, let Sr(n,k) denote the r-associated Stirling number of the second kind, which is defined as the number of ways to partition a set of n elements into k nonempty subsets such that each of the k subsets has at least r elements (see, for example, [17]). Define Sr(0,0)=1,Sr(n,0)=Sr(0,k)=0 and Sr(n,k)=0 if n<rk. Then S1(n,k)=S(n,k) and for r2 and nrk one has

    Sr(n+1,k)=kSr(n,k)+(nr1)Sr(nr+1,k1).

    Note that

    S2(n+1,k)=kS2(n,k)+nS2(n1,k1),

    and the first values of S2(n,k) can be listed as follows (see [17]):

    We can now state the first main result of this paper as follows.

    Theorem 1.1. Let p be a prime number. For any positive integers n and k with nk+1, we have

    vp(S(n,nk))=vp((nk+1))+tp(n,k)=ki=0vp(ni)vp((k+1)!)+tp(n,k),

    where tp(n,k):=0 if k=1 and tp(n,k):=vp(fk(n))vp(k+2k1) if k2 with

    fk(n):=k+2k1+2ki=k+2i+12ki(nk1)ik1S2(i,ik).

    In particular, we have

    tp(n,k)={vp(3n5)vp(4),if k=2,vp(n25n+6)vp(2),if k=3,vp(15n3150n2+485n502)vp(48),if k=4,vp(3n450n3+305n2802n+760)vp(16),if k=5,vp(63n51575n4+15435n373801n2+171150n152696)vp(576),if k=6

    and

    tp(n,7)=vp(9n6315n5+4515n433817n3+139020n2295748n+252336)vp(144).

    From Theorem 1.1, we can deduce the following result.

    Theorem 1.2. Let p be a prime and n be a positive integer.

    (i). If n2, then

    vp(S(n,n1))=vp(n)+vp(n1)vp(2).

    (ii). If n3, then

    vp(S(n,n2))=vp(n)+vp(n1)+vp(n2)+vp(3n5)vp(4)vp(3!).

    (iii). If n4, then

    vp(S(n,n3))=vp(n)+vp(n1)+2vp(n2)+2vp(n3)vp(2)vp(4!).

    (iv). If n5, then

    vp(S(n,n4))=4i=0vp(ni)+vp(15n3150n2+485n502)vp(48)vp(5!).

    (v). If n6, then

    vp(S(n,n5))=5i=0vp(ni)+vp(n4)+vp(n5)+vp(3n223n+38)vp(16)vp(6!).

    (vi). If n7, then

    vp(S(n,n6))=6i=0vp(ni)+vp(63n51575n4+15435n373801n2+171150n152696)     vp(576)vp(7!).

    (vii). If n8, then

    vp(S(n,n7))=7i=0vp(ni)+vp(n6)+vp(n7)     +vp(9n4198n3+1563n25182n+6008)vp(144)vp(8!).

    In 1995, Clarke proposed the following conjecture:

    Conjecture 1.3. [4] Let n and k be nonnegative integers. If nk+1 and p is an odd prime, then

    vp((nk)!S(n,nk))<n.

    Evidently, Conjecture 1.3 implies that vp((nk)!S(n,nk))=vp(T(n,nk)) holds if nk+1 and p is an odd prime. For the cases that 1k4, Clarke [4] checked the truth of above conjecture numerically. By using Theorem 1.2, we can derive the following result which proves Conjecture 1.3 for the cases 0k7 and is the third main result of this paper.

    Theorem 1.4. Let p be an odd prime. Let k be an integer such that 0k7. For any positive integer n with nk+1, one has

    vp((nk)!S(n,nk))<n. (1.2)

    This paper is organized as follows. First of all, in the next section, we reveal some preliminaries. Subsequently, we prove Theorems 1.1 and 1.2 in Section 3. In Section 4, we give the proof of Theorem 1.4. Finally, we present some interesting consequences of Theorem 1.2 in Section 5.

    In this section, we present several auxiliary lemmas that are needed.

    As introduced before, S2(n,k) represents the 2-associated Stirling number of the second kind. The exact values of Stirling number of the second kind can be computed by the following result.

    Lemma 2.1. [17] For any positive integers n and k with nk, we have

    S(n,nk)=2ki=k+1(ni)S2(i,ik).

    Lemma 2.2. Let p be an odd prime and m be a positive integer. For positive real number x, we define the function Gm,p as follows:

    Gm,p(x):=xm(p1)px(p2).

    Then Gm,p is strictly monotonic decreasing on the interval [2m,+) if p=3, and on the interval [m,+) if p5.

    Proof. Since p3 and

    ddxGm,p(x)=xm(p1)1px(p2)(m(p1)x(p2)lnp),

    one gets that ddxGm,p(x)<0 if mx2 with p=3 or mx with p5.

    For any given real number y, let y stand for the largest integer no more than y. We have the following result.

    Lemma 2.3. Let p be an odd prime. Let n and k be positive integers such that kmin. We have

    \begin{align*} \sum\limits_{i = 0}^kv_p(n-i)\le \min\Big\{1,\Big\lfloor\frac{k}{3}\Big\rfloor\Big\} +\max\Big\{1,\Big\lfloor\frac{k}{3}\Big\rfloor\Big\}\log_p n. \end{align*}

    Proof. If \sum_{i = 0}^kv_p(n-i) = 0 , then Lemma 2.3 clearly holds. In the following we let

    \sum\limits_{i = 0}^kv_p(n-i)\ge 1.

    Then at least one element in the set \{v_p(n), ..., v_p(n-k)\} is nonzero. We now divide rest of the proof into the following cases.

    CASE 1. k\in \{1, 2\} . In this case, one can easily see that only one of \{v_p(n), ..., v_p(n-k)\} is nonzero. This infers that

    \sum\limits_{i = 0}^k v_p(n-i) = \max\{v_p(n),...,v_p(n-k)\}\le \log_p n.

    CASE 2. k\in \{3, 4, 5\} . We need to show that

    \begin{align} \sum\limits_{i = 0}^kv_p(n-i)\le 1+\log_p n. \end{align} (2.1)

    Since 3\le k\le 5 , one gets that no more than two terms of \{v_p(n), ..., v_p(n-k)\} can be nonzero. If there is only one term being nonzero, then we must have

    \sum\limits_{i = 0}^k v_p(n-i) = \max\{v_p(n),...,v_p(n-k)\}\le \log_p n.

    So (2.1) holds. Otherwise, we can assume that v_p(n-i)\ge 1 and v_p(n-j)\ge 1 with 0\le i < j\le k . Then one can deduce that v_p(j-i)\ge 1 and so p\in\{3, 5\} . But 1\le j-i\le k\le 5 . Hence we must have v_p(j-i) = 1 . By using the isosceles triangle principle (see, for example, [18]), we derive that one of \{v_p(n-i), v_p(n-j)\} must be 1. Therefore

    \sum\limits_{i = 0}^k v_p(n-i) = 1+\max\{v_p(n),...,v_p(n-k)\}\le 1+\log_p n.

    Thus (2.1) is proved.

    CASE 3. k\in \{6, 7\} . We need to show that

    \begin{align} \sum\limits_{i = 0}^kv_p(n-i)\le 1+2\log_p n. \end{align} (2.2)

    Since k = 6 or k = 7 , one gets that no more than three terms of \{v_p(n), ..., v_p(n-k)\} can be nonzero. If there are at most two terms being nonzero, then

    \sum\limits_{i = 0}^k v_p(n-i)\le 2\log_p n.

    So (2.2) holds.

    If three terms of \{v_p(n), ..., v_p(n-k)\} are nonzero, then we must have p = 3 . Suppose that v_3(n-i_1)\ge 1 , v_3(n-i_2)\ge 1 and v_3(n-i_3)\ge 1 with 0\le i_1 < i_2 < i_3\le k\le 7 . Then one can derive that v_3(i_3-i_2) = v_3(i_2-i_1) = 1 . From this, we can obtain that one of \{v_3(n-i_1), v_3(n-i_2), v_3(n-i_3)\} must be 1. Hence one arrives at

    \sum\limits_{i = 0}^k v_p(n-i)\le 1+2\log_p n

    as (2.2) desired.

    This completes the proof of Lemma 2.3.

    This section is dedicated to the proofs of Theorems 1.1 and 1.2. We begin with the proof of Theorem 1.1.

    Proof of Theorem 1.1. First of all, Theorem 1.1 is clearly true when k = 1 , since

    S(n, n-1) = \binom{n}{2}.

    So in what follows, we let n\ge k+1\ge 3 . From Lemma 2.1, one knows that

    \begin{align} S(n, n-k) & = \binom{n}{k+1}S_2(k+1, 1)+\sum\limits_{i = k+2}^{2k}\binom{n}{i}S_2(i, i-k)\\ & = \binom{n}{k+1}+\sum\limits_{i = k+2}^{2k}\binom{n}{i}S_2(i, i-k). \end{align} (3.1)

    Let i be a positive integer with k+2\le i\le 2k . It is easy to see that

    \begin{align*} \binom{n}{i}& = \frac{(n)_i}{i!} = \frac{(n)_{k+1}\cdot(n-k-1)_{i-k-1}}{(k+1)!\cdot\langle k+2\rangle_{i-k-1}} = \binom{n}{k+1}\frac{(n-k-1)_{i-k-1}}{\langle k+2\rangle_{i-k-1}}. \end{align*}

    Together with (3.1) give us that

    \begin{align} &v_p(S(n, n-k))\\ & = v_p\Big(\binom{n}{k+1}+\sum\limits_{i = k+2}^{2k}\binom{n}{i}S_2(i, i-k)\Big)\\ & = v_p\Big(\binom{n}{k+1}\Big(1+\sum\limits_{i = k+2}^{2k} \frac{(n-k-1)_{i-k-1}}{\langle k+2\rangle_{i-k-1}}S_2(i, i-k)\Big)\Big)\\ & = v_p\Big(\binom{n}{k+1}\Big)+v_p\Big(1+ \sum\limits_{i = k+2}^{2k}\frac{(n-k-1)_{i-k-1}}{\langle k+2\rangle_{i-k-1}}S_2(i, i-k)\Big). \end{align} (3.2)

    Note that S_2(i, i-k) is an integer. Let

    f_k(n): = \langle k+2\rangle_{k-1} +\sum\limits_{i = k+2}^{2k}\langle i+1\rangle_{2k-i}\cdot(n-k-1)_{i-k-1}S_2(i, i-k)

    and

    t_p(n, k): = v_p(f_k(n))-v_p(\langle k+2\rangle_{k-1}).

    Then f_k(n) is a polynomial with integer coefficients and \deg f_k = k-1 , and by (3.2) one derives that

    v_p(S(n, n-k)) = v_p\Big(\binom{n}{k+1}\Big)+t_p(n, k).

    In the following, we deal with t_p(n, k) for 2\le k\le 7 . We divide this into six cases.

    CASE 1. k = 2 . Since S_2(4, 2) = 3 , by the definition of f_2(n) we get that

    f_2(n) = 4+3(n-3) = 3n-5,

    and so

    t_p(n, 2) = v_p(f_2(n))-v_p(4) = v_p(3n-5)-v_p(4).

    CASE 2. k = 3 . From S_2(5, 2) = 10 and S_2(6, 3) = 15 one deduces that

    f_3(n) = 5\cdot6+6\cdot10(n-4)+15(n-4)(n-5) = 15(n^2-5n+6)

    and

    \begin{align*} t_p(n, 3)& = v_p(f_3(n))-v_p(5\cdot6) = v_p(n^2-5n+6)-v_p(2). \end{align*}

    CASE 3. k = 4 . By S_2(6, 2) = 25 , S_2(7, 3) = 105 and S_2(8, 4) = 105 , we derive that

    \begin{align*} f_4(n)& = 6\cdot7\cdot8+7\cdot8\cdot25(n-5)+8\cdot105(n-5)(n-6)+105(n-5)(n-6)(n-7)\\ & = 7(15n^3-150n^2+485n-502) \end{align*}

    and

    \begin{align*} t_p(n, 4)& = v_p(f_4(n))-v_p(6\cdot7\cdot8)\\ & = v_p(7)+v_p(15n^3-150n^2+485n-502)-v_p(6\cdot7\cdot8)\\ & = v_p(15n^3-150n^2+485n-502)-v_p(48). \end{align*}

    CASE 4. k = 5 . It follows from S_2(7, 2) = 56 , S_2(8, 3) = 490 , S_2(9, 4) = 1260 and S_2(10, 5) = 945 that

    \begin{align*} f_5(n)& = 7\cdot8\cdot9\cdot10+8\cdot9\cdot10\cdot56(n-6)+9\cdot10\cdot490(n-6)(n-7)\\ &\ \ \ \ \ +10\cdot1260(n-6)(n-7)(n-8)+945(n-6)(n-7)(n-8)(n-9)\\ & = 315(3n^4-50n^3+305n^2-802n+760) \end{align*}

    and

    \begin{align*} t_p(n, 5)& = v_p(f_5(n))-v_p(7\cdot8\cdot9\cdot10)\\ & = v_p(315)+v_p(3n^4-50n^3+305n^2-802n+760)-v_p(7\cdot8\cdot9\cdot10)\\ & = v_p(3n^4-50n^3+305n^2-802n+760)-v_p(16). \end{align*}

    CASE 5. k = 6 . Since S_2(8, 2) = 119 , S_2(9, 3) = 1918 , S_2(10, 4) = 9450 , S_2(11, 5) = 17325 and S_2(12, 6) = 10395 , one deduces that

    \begin{align*} f_6(n)& = \langle8\rangle_5+\langle9\rangle_4\cdot119(n-7) +\langle10\rangle_3\cdot1918(n-7)(n-8)\\ &\ \ \ \ \ +\langle11\rangle_2\cdot9450(n-7)(n-8)(n-9)+12\cdot17325(n-7)(n-8)(n-9)(n-10)\\ &\ \ \ \ \ +10395(n-7)(n-8)(n-9)(n-10)(n-11)\\ & = 165(63n^5-1575n^4+15435n^3-73801n^2+171150n-152696) \end{align*}

    and

    \begin{align*} t_p(n, 6) & = v_p(f_6(n))-v_p(\langle8\rangle_5)\\ & = v_p(63n^5-1575n^4+15435n^3-73801n^2+171150n-152696)-v_p(576). \end{align*}

    CASE 6. k = 7 . By S_2(9, 2) = 246 , S_2(10, 3) = 6825 , S_2(11, 4) = 56980 , S_2(12, 5) = 190575 , S_2(13, 6) = 270270 and S_2(14, 7) = 135135 , we obtain that

    \begin{align*} f_7(n)& = \langle9\rangle_6+\langle10\rangle_5\cdot246(n-8) +\langle11\rangle_4\cdot6825(n-8)(n-9)\\ &\ \ \ \ \ +\langle12\rangle_3\cdot56980(n-8)(n-9)(n-10)\\ &\ \ \ \ \ +\langle13\rangle_2\cdot190575(n-8)(n-9)(n-10)(n-11)\\ &\ \ \ \ \ +14\cdot270270(n-8)(n-9)(n-10)(n-11)(n-12)\\ &\ \ \ \ \ +135135(n-8)(n-9)(n-10)(n-11)(n-12)(n-13)\\ & = 15015 (9 n^6-315 n^5 + 4515 n^4 - 33817 n^3 + 139020 n^2- 295748 n+ 252336) \end{align*}

    and

    \begin{align*} t_p(n, 7) & = v_p(f_7(n))-v_p(\langle9\rangle_6)\\ & = v_p(9 n^6-315 n^5 + 4515 n^4 - 33817 n^3 + 139020 n^2- 295748 n+ 252336)-v_p(144). \end{align*}

    This concludes the proof of Theorem 1.1.

    From the proof of Theorem 1.1, one may see that for any positive integers n and k with n\ge k+1 , f_k(n) is a polynomial with integer coefficients and \deg f_k = k-1 . We also note that the expression of f_k(n) depends on the values of the 2-associated Stirling number of the second kind S_2(n, k) . However, the computation of the exact value of S_2(n, k) becomes complicated when k\ge 8 . Therefore we only present the formulas for f_k(n) and t_p(n, k) for the case 1\le k\le 7 .

    Subsequently, we give the proof of Theorem 1.2.

    Proof of Theorem 1.2. (ⅰ). Since n\ge 2 , from Theorem 1.1 one knows that

    v_p(S(n, n-1)) = v_p(n)+v_p(n-1)-v_p(2).

    So part (ⅰ) of Theorem 1.2 is proved.

    (ⅱ). Let n\ge 3 . By Theorem 1.1 we derive that

    \begin{align*} v_p(S(n, n-2)) & = v_p(n)+v_p(n-1)+v_p(n-2)+t_p(n, 2)-v_p(3!)\\ & = v_p(n)+v_p(n-1)+v_p(n-2)+v_p(3n-5)-v_p(4)-v_p(3!) \end{align*}

    as (ⅱ) desired. Hence part (ⅱ) of Theorem 1.2 is proved.

    (ⅲ). Since n\ge 4 and n^2-5n+6 = (n-2)(n-3) , it follows from Theorem 1.1 that

    \begin{align*} v_p(S(n, n-3))& = \sum\limits_{i = 0}^{3}v_p(n-i)-v_p(4!)+t_p(n, 3)\notag\\ & = \sum\limits_{i = 0}^{3}v_p(n-i)-v_p(4!)+v_p(n^2-5n+6)-v_p(2)\notag\\ & = v_p(n)+v_p(n-1)+2v_p(n-2)+2v_p(n-3)-v_p(2)-v_p(4!). \end{align*}

    This completes the proof of part (ⅲ).

    (ⅳ). Let n\ge 5 . By Theorem 1.1 one obtains that

    \begin{align*} v_p(S(n, n-4))& = \sum\limits_{i = 0}^{4}v_p(n-i)-v_p(5!)+t_p(n, 4)\\ & = \sum\limits_{i = 0}^{4}v_p(n-i)+v_p(15n^3-150n^2+485n-502)-v_p(48)-v_p(5!) \end{align*}

    as part (ⅳ) asserted.

    (ⅴ). Since n\ge 6 and 3n^4-50n^3+305n^2-802n+760 = (n-4)(n-5)(3n^2-23n+38) , it follows from Theorem 1.1 that

    \begin{align} v_p(S(n, n-5))& = \sum\limits_{i = 0}^{5}v_p(n-i)-v_p(6!)+t_p(n, 5) \end{align} (3.3)

    and

    \begin{align} t_p(n, 5) & = v_p(3n^4-50n^3+305n^2-802n+760)-v_p(16)\\ & = v_p(3n^2-23n+38)+v_p(n-4)+v_p(n-5)-v_p(16). \end{align} (3.4)

    Then by using (3.3) and (3.4), we deduce that

    \begin{align*} v_p(S(n, n-5)) & = \sum\limits_{i = 0}^{5}v_p(n-i)+v_p(n-4)+v_p(n-5)+v_p(3n^2-23n+38)-v_p(16)-v_p(6!). \end{align*}

    So part (ⅴ) is proved.

    (ⅵ). Let n\ge 7 . By Theorem 1.1 we deduce that

    \begin{align*} &v_p(S(n, n-6))\\ & = \sum\limits_{i = 0}^{6}v_p(n-i)-v_p(7!)+t_p(n, 6)\\ & = \sum\limits_{i = 0}^{6}v_p(n-i) + v_p(63n^5-1575n^4+15435n^3-73801n^2+171150n-152696)-v_p(576)-v_p(7!). \end{align*}

    Hence the proof of part (ⅵ) is finished.

    (ⅶ). Since n\ge 8 and

    \begin{align} &9 n^6-315 n^5 + 4515 n^4 - 33817 n^3 + 139020 n^2- 295748 n+ 252336\\ & = (n-6)(n-7)(9n^4-198n^3+1563n^2-5182n+6008), \end{align} (3.5)

    it follows from Theorem 1.1 and (3.5) that

    \begin{align*} &v_p(S(n, n-7))\notag\\ & = \sum\limits_{i = 0}^{7}v_p(n-i)-v_p(8!)+t_p(n, 7)\\ & = \sum\limits_{i = 0}^{7}v_p(n-i)-v_p(8!)-v_p(144)\\ &\ \ \ \ \ +v_p(9 n^6-315 n^5 + 4515 n^4 - 33817 n^3 + 139020 n^2- 295748 n+ 252336)\\ & = \sum\limits_{i = 0}^{7}v_p(n-i)+v_p(n-6)+v_p(n-7)\\ &\ \ \ \ \ +v_p(9n^4-198n^3+1563n^2-5182n+6008)-v_p(144)-v_p(8!). \end{align*}

    Therefore part (ⅶ) is proved. This completes the proof of Theorem 1.2.

    In this section, we present the proof of Theorem 1.4.

    Proof of Theorem 1.4. Let p be an odd prime. First, since S(n, n) = 1 and

    v_p(n!) = \frac{n-d_p(n)}{p-1} \lt n,

    (1.2) is clearly true when k = 0 . Now let 1\le k\le 7 . Consider the following cases.

    CASE 1. k = 1 . Since p\ge 3 , from Theorem 1.2 one knows that

    v_p(S(n, n-1)) = v_p(n)+v_p(n-1).

    It then follows from Lemma 2.3 that

    \begin{align} &v_p((n-1)!S(n, n-1))-n\\ & = \frac{n-1-d_p(n-1)}{p-1}+v_p(n)+v_p(n-1)-n\\ &\le \frac{1}{p-1}\Big(n-2+(p-1)(\log_p n-n)\Big)\\ & = \frac{1}{p-1}\Big(n(2-p)+(p-1)\log_p n-2\Big)\\ & = \frac{1}{p-1}\Big(\log_p\frac{n^{p-1}}{p^{n(p-2)}}-2\Big). \end{align} (4.1)

    Note that n\ge 2 . Hence we derive from Lemma 2.2 that

    \begin{align} \log_p\frac{n^{p-1}}{p^{n(p-2)}}-2 & = \log_pG_{1, p}(n)-2\\ &\le \log_pG_{1, p}(2)-2\\ & = \log_p\frac{2^{p-1}}{p^{2(p-2)}}-2\\ & = (p-1)\log_p\frac{2}{p^{2}} \lt 0. \end{align} (4.2)

    By (4.1) and (4.2), one then obtains that v_p((n-1)!S(n, n-1)) < n as desired.

    CASE 2. k = 2 . Note that n\ge 3 and p\ge 3 . It follows from Theorem 1.2 that

    v_p(S(n, n-2)) = \sum\limits_{i = 0}^2v_p(n-i)+v_p(3n-5)-v_p(3!).

    Thus one deduces that

    \begin{align} &v_p((n-2)!S(n, n-2))-n\\ & = \frac{n-2-d_p(n-2)}{p-1}+\sum\limits_{i = 0}^2v_p(n-i)+v_p(3n-5)-v_p(3!)-n\\ &\le \frac{1}{p-1}\Big(n-3+(p-1)\Big(\sum\limits_{i = 0}^2v_p(n-i) +v_p(3n-5)-v_p(3)-n\Big)\Big): = \frac{D_{2,p}(n)}{p-1} \end{align} (4.3)

    with

    \begin{align} D_{2,p}(n) = (p-1)\Big(\sum\limits_{i = 0}^2v_p(n-i)+v_p(3n-5)-v_p(3)\Big)-n(p-2)-3. \end{align} (4.4)

    If v_p(3n-5) = 0 , then by Lemma 2.3 we obtain that

    \begin{align} \sum\limits_{i = 0}^2v_p(n-i)+v_p(3n-5)-v_p(3)\le \log_p n. \end{align} (4.5)

    If v_p(3n-5)\ge 1 , then we must have p\ge 5 and v_p(n-1) = v_p(n-2) = 0 . Also note that v_p(n) = v_p(3n) = v_p(3n-5+5) and v_p(3n-5) = v_p(\frac{3n-5}{3})\le \log_p n . This implies that \min\{v_p(n), v_p(3n-5)\}\le 1 , and so

    \begin{align} &\sum\limits_{i = 0}^2v_p(n-i)+v_p(3n-5)-v_p(3)\\ & = v_p(n)+v_p(3n-5)\\ &\le 1+\max\{v_p(n),v_p(3n-5)\}\\ &\le 1+\log_p n. \end{align} (4.6)

    Since p\ge 3 , we derive from Lemma 2.2 together with (4.4) to (4.6) that

    \begin{align} D_{2, p}(n) &\le(p-1)(1+\log_p n)-n(p-2)-3\\ & = \log_p \frac{n^{p-1}}{p^{n(p-2)}}+p-4\\ & = \log_p G_{1, p}(n)+p-4\\ &\le \log_p G_{1, p}(3)+p-4\\ & = (p-1)\log_p \frac{3}{p^2} \lt 0. \end{align} (4.7)

    So (4.3) and (4.7) imply that v_p((n-2)!S(n, n-2)) < n .

    CASE 3. k = 3 . Since p\ge 3 , by Theorem 1.2, we get that

    \begin{align*} v_p(S(n, n-3)) & = v_p(n)+v_p(n-1)+2v_p(n-2)+2v_p(n-3)-v_p(2)-v_p(4!)\\ & = \sum\limits_{i = 0}^3v_p(n-i)+v_p(n-2)+v_p(n-3)-v_p(3). \end{align*}

    It follows that

    \begin{align} &v_p((n-3)!S(n, n-3))-n\\ & = \frac{n-3-d_p(n-3)}{p-1}+v_p(S(n, n-3))-n\\ &\le \frac{1}{p-1}\Big(n-4+(p-1)\big(v_p(S(n, n-3))-n\big)\Big): = \frac{D_{3, p}(n)}{p-1} \end{align} (4.8)

    with

    \begin{align} D_{3, p}(n)& = n-4+(p-1)\big(v_p(S(n, n-3))-n\big)\\ & = (p-1)\Big(\sum\limits_{i = 0}^3v_p(n-i)+v_p(n-2)+v_p(n-3)-v_p(3)\Big)-n(p-2)-4. \end{align} (4.9)

    Note that v_p(n-2)+v_p(n-3)\le \log_p n . Hence, by Lemma 2.3 one deduces that

    \sum\limits_{i = 0}^3v_p(n-i)+v_p(n-2)+v_p(n-3)-v_p(3)\le 1+2\log_p n.

    It then follows from (4.9) that

    \begin{align} D_{3, p}(n) &\le (p-1)(1+2\log_p n)-n(p-2)-4\\ & = \log_p \frac{n^{2(p-1)}}{p^{n(p-2)}}+p-5\\ & = \log_p G_{2, p}(n)+p-5\\ &\le \log_p G_{2, p}(4)+p-5\\ & = (p-1)\log_p \frac{4^2}{p^{3}} \lt 0 \end{align} (4.10)

    since p\ge 3 and n\ge 4 . So (4.8) and (4.10) give us that v_p((n-3)!S(n, n-3)) < n .

    CASE 4. k = 4 . Note that n\ge 5 and p\ge 3 . From Theorem 1.2, we derive that

    \begin{align*} v_p(S(n, n-4)) & = \sum\limits_{i = 0}^4v_p(n-i)+v_p(15n^3-150n^2+485n-502)-v_p(48)-v_p(5!)\\ & = \sum\limits_{i = 0}^4v_p(n-i)+v_p(15n^3-150n^2+485n-502)-2v_p(3)-v_p(5) \end{align*}

    and

    \begin{align} &v_p((n-4)!S(n, n-4))-n\\ & = \frac{n-4-d_p(n-4)}{p-1}+v_p(S(n, n-4))-n\\ &\le \frac{1}{p-1}\Big(n-5+(p-1)\big(v_p(S(n, n-4))-n\big)\Big) : = \frac{D_{4, p}(n)}{p-1} \end{align} (4.11)

    with

    \begin{align*} D_{4, p}(n)& = (p-1)v_p(S(n, n-4))-n(p-2)-5\notag\\ & = (p-1)\Big(\sum\limits_{i = 0}^4v_p(n-i)+v_p(15n^3-150n^2+485n-502)-2v_p(3)-v_p(5)\Big)-n(p-2)-5. \end{align*}

    First of all, it is easy to check that D_{4, p}(n) < 0 for 5\le n\le 9 . In what follows, let n\ge 10 .

    If v_p(15n^3-150n^2+485n-502) = 0 , then by Lemma 2.3 we obtain that

    \begin{align} \sum\limits_{i = 0}^4v_p(n-i)+v_p(15n^3-150n^2+485n-502)-2v_p(3)-v_p(5)\le1+ \log_p n. \end{align} (4.12)

    Now let v_p(15n^3-150n^2+485n-502)\ge 1 . Note that

    \begin{align*} 15n^3-150n^2+485n-502 & = 15(n-1)^3-105(n-1)^2+230(n-1)-152\\ & = 15(n-2)^3-60(n-2)^2+65(n-2)-12\\ & = 15(n-3)^3-15(n-3)^2-10(n-3)+8\\ & = 15(n-4)^3+30(n-4)^2+5(n-4)-2 \end{align*}

    and 502 = 2\cdot 251 , 152 = 2^3\cdot 19 , 12 = 2^2\cdot 3 . Then we deduce that p\neq5 , v_p(n-3) = v_p(n-4) = 0 and

    \begin{align*} v_p(15n^3-150n^2+485n-502) &\le\left\{\begin{array}{ll} 1+\log_p n^3,& {\rm if}\ p = 3,\\ \log_p n^3, & {\rm if}\ p\ge 7 \end{array}\right. \end{align*}

    since v_p(15n^3-150n^2+485n-502) = v_p(\frac{15n^3-150n^2+485n-502}{15})+v_p(15)\le \log_p n^3+v_p(15) . If p = 3 , then one gets that v_3(n) = v_3(n-1) = 0 and v_3(n-2)\ge1 . Also note that \min\{v_3(n-2), v_3(15n^3-150n^2+485n-502)\}\le 1 . It follows that

    \begin{align} &\sum\limits_{i = 0}^4v_3(n-i)+v_3(15n^3-150n^2+485n-502)-2v_3(3)-v_3(5)\\ & = v_3(n-2)+v_3(15n^3-150n^2+485n-502)-2\\ &\le\log_3 n^3. \end{align} (4.13)

    If p\ge 7 , then one has v_p(n-2) = 0 and only one of \{v_p(n), v_p(n-1)\} can be nonzero. Without loss of generality, assume that v_p(n) = 0 and v_p(n-1)\ge 1 . Then it follows that \min\{v_p(n-1), v_p(15n^3-150n^2+485n-502)\}\le 1 . This gives us that

    \begin{align} &\sum\limits_{i = 0}^4v_p(n-i)+v_p(15n^3-150n^2+485n-502)-2v_p(3)-v_p(5)\\ & = v_p(n)+v_p(n-1)+v_p(15n^3-150n^2+485n-502)\\ &\le1+\log_p n^3. \end{align} (4.14)

    Since p\ge 3 and n\ge 10 , by (4.12) to (4.14) together with Lemma 2.2 one derives that

    \begin{align} D_{4, p}(n) &\le(p-1)(1+\log_p n^3)-n(p-2)-5\\ & = \log_p \frac{n^{3(p-1)}}{p^{n(p-2)}}+p-6\\ & = \log_p G_{3, p}(n)+p-6\\ &\le \log_p G_{3, p}(10)+p-6\\ & = \log_p \frac{10^{3(p-1)}}{p^{9p-14}} \lt 0. \end{align} (4.15)

    Thus (4.11) and (4.15) tell us that v_p((n-4)!S(n, n-4)) < n .

    CASE 5. k = 5 . Since p\ge 3 , from Theorem 1.2, we obtain that

    \begin{align*} v_p(S(n, n-5)) & = \sum\limits_{i = 0}^5v_p(n-i)+v_p(n-4)+v_p(n-5)+v_p(3n^2-23n+38)-v_p(16)-v_p(6!)\\ & = \sum\limits_{i = 0}^5v_p(n-i)+v_p(n-4)+v_p(n-5)+v_p(3n^2-23n+38)-2v_p(3)-v_p(5) \end{align*}

    and

    \begin{align} &v_p((n-5)!S(n, n-5))-n\\ & = \frac{n-5-d_p(n-5)}{p-1}+v_p(S(n, n-5))-n\\ &\le \frac{1}{p-1}\Big(n-6+(p-1)\big(v_p(S(n, n-5))-n\big)\Big) : = \frac{D_{5, p}(n)}{p-1} \end{align} (4.16)

    with

    \begin{align} D_{5, p}(n)& = (p-1)v_p(S(n, n-5))-n(p-2)-6\\ & = (p-1)\Big(\sum\limits_{i = 0}^5v_p(n-i)+v_p(n-4)+v_p(n-5)\\ &\ \ \ \ \ +v_p(3n^2-23n+38)-2v_p(3)-v_p(5)\Big)-n(p-2)-6. \end{align} (4.17)

    In what follows, we show that D_{5, p}(n) < 0 . If v_p(3n^2-23n+38) = 0 , then by Lemma 2.3 and v_p(n-4)+v_p(n-5)\le\log_p n we deduce that

    \begin{align} \sum\limits_{i = 0}^5v_p(n-i)+v_p(n-4)+v_p(n-5)+v_p(3n^2-23n+38)-2v_p(3)-v_p(5) &\le1+ 2\log_p n. \end{align} (4.18)

    Now let v_p(3n^2-23n+38)\ge 1 . Since 38 = 2\cdot 19 , 18 = 2\cdot 3^2 and

    \begin{align*} &3n^2-23n+38\\ & = 3(n-1)^2-17(n-1)+18 = 3(n-2)^2-11(n-2)+4\\ & = 3(n-3)^2-5(n-3)-4 = 3(n-4)^2+(n-4)-6 = 3(n-5)^2+7(n-5)-2, \end{align*}

    one gets that v_p(n-2) = v_p(n-3) = v_p(n-5) = 0 and one of v_p(n) and v_p(n-1) must be zero. If p = 3 , then we have v_3(n) = 0 and

    \begin{align*} &\left\{\begin{array}{ll} v_3(n-1)\le \log_3 n\ {\rm and}\ v_3(3n^2-23n+38)\le 1+\log_3 n^2,& {\rm if}\ v_3(n-4) = 1,\\ v_3(n-1) = v_3(3n^2-23n+38) = 1, & {\rm if}\ v_3(n-4)\ge2. \end{array}\right. \end{align*}

    Hence one deduces that

    \begin{align} &\sum\limits_{i = 0}^5v_3(n-i)+v_3(n-4)+v_3(n-5)+v_3(3n^2-23n+38)-2v_3(3)-v_3(5)\\ & = v_3(n-1)+2v_3(n-4)+v_3(3n^2-23n+38)-2\le1+\log_3 n^3. \end{align} (4.19)

    If p\ge 5 , then one derives that v_p(n-1) = v_p(n-4) = 0 . Thus

    \begin{align} &\sum\limits_{i = 0}^5v_p(n-i)+v_p(n-4)+v_p(n-5)+v_p(3n^2-23n+38)-2v_p(3)-v_p(5)\\ & = v_p(n)+v_p(3n^2-23n+38)-v_p(5)\le\log_p n^3. \end{align} (4.20)

    Hence (4.17) to (4.20) imply that

    \begin{align} D_{5, p}(n) &\le (p-1)(1+\log_p n^3)-n(p-2)-6\\ & = \log_p \frac{n^{3(p-1)}}{p^{n(p-2)}}+p-7\\ & = \log_p G_{3,p}(n)+p-7\\ &\le\log_p G_{3,p}(6)+p-7\\ & = (p-1)\log_p \frac{6^3}{p^5} \lt 0. \end{align} (4.21)

    Thus (4.16) and (4.21) give us that v_p((n-5)!S(n, n-5)) < n .

    CASE 6. k = 6 . Note that n\ge 7 and p\ge3 . By Theorem 1.2, we deduce that

    \begin{align*} v_p(S(n, n-6)) & = \sum\limits_{i = 0}^6v_p(n-i)-v_p(576)-v_p(7!)\\ &\ \ \ \ \ + v_p(63n^5-1575n^4+15435n^3-73801n^2+171150n-152696)\\ & = \sum\limits_{i = 0}^6v_p(n-i)-4v_p(3)-v_p(5)-v_p(7)\\ &\ \ \ \ \ + v_p(63n^5-1575n^4+15435n^3-73801n^2+171150n-152696) \end{align*}

    and

    \begin{align} &v_p((n-6)!S(n, n-6))-n\\ & = \frac{n-6-d_p(n-6)}{p-1}+v_p(S(n, n-6))-n\\ &\le \frac{1}{p-1}\Big(n-7+(p-1)\big(v_p(S(n, n-6))-n\big)\Big) : = \frac{D_{6, p}(n)}{p-1} \end{align} (4.22)

    with

    \begin{align} D_{6, p}(n)& = (p-1)v_p(S(n, n-6))-n(p-2)-7\\ & = (p-1)\Big(\sum\limits_{i = 0}^6v_p(n-i)+v_p(E_{6,n})-4v_p(3)-v_p(5)-v_p(7)\Big)-n(p-2)-7 \end{align} (4.23)

    and

    E_{6, n}: = 63n^5-1575n^4+15435n^3-73801n^2+171150n-152696.

    If v_p(E_{6, n}) = 0 , then by Lemma 2.3 we derive that

    \begin{align} \sum\limits_{i = 0}^6v_p(n-i)-4v_p(3)-v_p(5)-v_p(7)+v_p(E_{6,n})\le1+ 2\log_p n. \end{align} (4.24)

    From (4.23) and (4.24) one obtains that

    \begin{align} D_{6, p}(n) &\le(p-1)(1+2\log_p n)-n(p-2)-7\\ & = \log_p\frac{n^{2(p-1)}}{p^{n(p-2)}}+p-8\\ & = \log_pG_{2,p}(n)+p-8\\ &\le \log_pG_{2,p}(7)+p-8\\ & = (p-1)\log_p\frac{7^2}{p^6} \lt 0. \end{align} (4.25)

    In what follows, we let v_p(E_{6, n})\ge 1 . Since

    \begin{align*} E_{6, n} & = 63n^5-1575 n^4+15435 n^3-73801 n^2+171150 n-152696\\ & = 63(n-1)^5 - 1260 (n-1)^4 + 9765 (n-1)^3 - 36316 (n-1)^2 + 63868(n-1) - 41424\\ & = 63 (n - 2)^5 - 945 (n - 2)^4 + 5355 (n - 2)^3 - 13951 (n - 2)^2 + 15806 (n - 2) - 5304\\ & = 63 (n - 3)^5 - 630 (n - 3)^4 + 2205 (n - 3)^3 - 2926 (n - 3)^2 + 504 (n - 3) + 1024\\ & = 63 (n - 4)^5 - 315 (n - 4)^4 + 315 (n - 4)^3 + 539 (n - 4)^2 - 938 (n - 4) + 240\\ & = 63 (n - 5)^5 - 315 (n - 5)^3 + 224 (n - 5)^2 + 140 (n - 5) - 96\\ & = 63 (n - 6)^5 + 315 (n - 6)^4 + 315 (n - 6)^3 - 91 (n - 6)^2 - 42 (n - 6) + 16 \end{align*}

    and 152696 = 2^3\cdot19087 , 41424 = 2^4\cdot3\cdot 863 , 5304 = 2^3\cdot3\cdot13\cdot17 , 1024 = 2^{10} , 240 = 2^4\cdot3\cdot5 , 96 = 2^5\cdot3 and 16 = 2^4 , one can deduce that v_p(n-3) = v_p(n-6) = 0 . If p = 3 , then only two of \{v_3(n), v_3(n-1), v_3(n-2), v_3(n-4), v_3(n-5)\} can be nonzero. Suppose that v_3(n-i)\ge 1 and v_3(n-j)\ge 1 for 0\le i < j\le 6 . Then either 1\le v_3(E_{6, n})\le \log_3(n-5)^5+v_3(63) with v_3(n-i) = v_3(n-j) = 1 or v_3(E_{6, n}) = 1 with \max\{v_3(n-i), v_3(n-j)\}\ge 2 , this implies that

    \begin{align*} &\sum\limits_{i = 0}^6v_3(n-i)+v_3(E_{6,n})-4v_3(3)-v_3(5)-v_3(7)\\ & = v_3(n-i)+v_3(n-j)+v_3(E_{6,n})-4\\ &\le 2+\log_3(n-5)^5+v_3(63)-4\\ &\le\log_3(n-5)^5. \end{align*}

    It is easy to check that D_{6, 3}(n) < 0 when 7\le n\le 14 . If n\ge 15 , then by (4.23) one gets that

    \begin{align} D_{6, 3}(n) &\le 2\log_3 (n-5)^5-n-7\\ & = \log_3 \frac{(n-5)^{10}}{3^{n-5}}-12\\ & = \log_3G_{5,3}(n-5)-12\\ &\le \log_3 G_{5,3}(10)-12\\ & = \log_3 \frac{10^{10}}{3^{22}} \lt 0. \end{align} (4.26)

    If p\ge 5 , we can obtain that v_p(n-5) = 0 and only one of \{v_p(n), v_p(n-1), v_p(n-2), v_p(n-4)\} can be nonzero. Without loss of generality, assume that v_p(n)\ge 1 and v_p(n-1) = v_p(n-2) = v_p(n-4) = 0 . It follows that \min\{v_p(n), v_p(E_{6, n}\}\le 1 . Hence

    \begin{align} &\sum\limits_{i = 0}^6v_p(n-i)+v_p(E_{6,n})-4v_p(3)-v_p(5)-v_p(7) \le1+\log_pn^5. \end{align} (4.27)

    Obviously, one has D_{6, p}(n) < 0 if n = 7 . Now let n\ge 8 . Then by Lemma 2.2 together with (4.23) and (4.27), we arrive at

    \begin{align} D_{6, p}(n) &\le(p-1)(1+\log_pn^5)-n(p-2)-7\\ & = \log_p\frac{n^{5(p-1)}}{p^{n(p-2)}}+p-8\\ & = \log_pG_{5,p}(n)+p-8\\ &\le \log_pG_{5,p}(8)+p-8\\ & = \log_p\frac{8^{5(p-1)}}{p^{7p-8}} \lt 0 \end{align} (4.28)

    as desired. Thus (4.22), (4.25), (4.26) and (4.28) tell us that v_p((n-6)!S(n, n-6)) < n .

    CASE 7. k = 7 . From Theorem 1.2, we obtain that

    \begin{align*} &v_p(S(n, n-7))\\ & = \sum\limits_{i = 0}^5v_p(n-i)+2v_p(n-6)+2v_p(n-7)\\ &\ \ \ \ \ +v_p(9n^4-198n^3+1563n^2-5182n+6008)-v_p(144)-v_p(8!)\\ & = \sum\limits_{i = 0}^7v_p(n-i)+v_p(n-6)+v_p(n-7)\\ &\ \ \ \ \ +v_p(9n^4-198n^3+1563n^2-5182n+6008)-4v_p(3)-v_p(5)-v_p(7) \end{align*}

    and

    \begin{align} &v_p((n-7)!S(n, n-7))-n\\ & = \frac{n-7-d_p(n-7)}{p-1}+v_p(S(n, n-7))-n\\ &\le \frac{1}{p-1}\Big(n-8+(p-1)\big(v_p(S(n, n-7))-n\big)\Big) : = \frac{D_{7, p}(n)}{p-1} \end{align} (4.29)

    with

    \begin{align} D_{7, p}(n)& = (p-1)v_p(S(n, n-7))-n(p-2)-8\\ & = -n(p-2)-8+(p-1)\Big(\sum\limits_{i = 0}^7v_p(n-i)+v_p(n-6)+v_p(n-7)\\ &\ \ \ \ \ \ \ \ \ \ +v_p(E_{7, n})-4v_p(3)-v_p(5)-v_p(7)\Big) \end{align} (4.30)

    and

    E_{7, n}: = 9n^4-198n^3+1563n^2-5182n+6008.

    In what follows, we show that D_{7, p}(n) < 0 . If v_p(E_{7, n}) = 0 , then by Lemma 2.3 we derive that

    \begin{align*} &\sum\limits_{i = 0}^7v_p(n-i)+v_p(n-6)+v_p(n-7)+v_p(E_{7, n})-4v_p(3)-v_p(5)-v_p(7) \le1+ 3\log_p n \end{align*}

    since v_p(n-6)+v_p(n-7)\le \log_p n . It follows from (4.30) that

    \begin{align} D_{7, p}(n) &\le (p-1)(1+3\log_pn)-n(p-2)-8\\ & = \log_p\frac{n^{3(p-1)}}{p^{n(p-2)}}+p-9\\ & = \log_pG_{3,p}(n)+p-9\\ &\le\log_pG_{3,p}(8)+p-9\\ & = (p-1)\log_p\frac{8^3}{p^7} \lt 0. \end{align} (4.31)

    Now let v_p(E_{7, n})\ge 1 . Note that

    \begin{align*} E_{7, n}& = 9n^4-198n^3+1563n^2-5182n+6008\\ & = 9 (n - 1)^4 - 162 (n - 1)^3 + 1023 (n - 1)^2 - 2614 (n - 1) + 2200\\ & = 9 (n - 2)^4 - 126 (n - 2)^3 + 591 (n - 2)^2 - 1018 (n - 2) + 456\\ & = 9 (n - 3)^4 - 90 (n - 3)^3 + 267 (n - 3)^2 - 178 (n - 3) - 88\\ & = 9 (n - 4)^4 - 54 (n - 4)^3 + 51 (n - 4)^2 + 122 (n - 4) - 80\\ & = 9 (n - 5)^4 - 18 (n - 5)^3 - 57 (n - 5)^2 + 98 (n - 5) + 48\\ & = 9 (n - 6)^4 + 18 (n - 6)^3 - 57 (n - 6)^2 - 34 (n - 6) + 80\\ & = 9 (n - 7)^4 + 54 (n - 7)^3 + 51 (n - 7)^2 - 58 (n - 7) + 16 \end{align*}

    and 6008 = 2^3\cdot751 , 2200 = 2^3\cdot 5^2\cdot 11 , 456 = 2^3\cdot3\cdot 19 , 88 = 2^3\cdot11 , 80 = 2^4\cdot5 , 48 = 2^4\cdot3 , 16 = 2^4 . Thus one deduces that v_p(n-7) = 0 . We divide rest of the proof into the following subcases.

    CASE 7.1. p = 3 . Then we have v_3(n) = v_3(n-1) = v_3(n-3) = v_3(n-4) = v_3(n-6) = 0 . Moreover, one can derive that either 1\le v_3(E_{7, n})\le 2+\log_3n^4 with v_3(n-2) = v_3(n-5) = 1 or v_3(E_{7, n}) = 1 with \max\{v_3(n-2), v_3(n-5)\}\ge 2 and \min\{v_3(n-2), v_3(n-5)\} = 1 . This implies that

    \begin{align} &\sum\limits_{i = 0}^5v_3(n-i)+2v_3(n-6)+2v_3(n-7)+v_3(E_{7, n})\\ & = v_3(n-2)+v_3(n-5)+v_3(E_{7, n})\\ &\le 4+\log_3 n^4. \end{align} (4.32)

    It follows from (4.30) and (4.32) that

    \begin{align} D_{7, 3}(n) &\le 2(4+\log_3 n^4-4)-n-8\\ & = \log_3\frac{n^{8}}{3^{n+8}}\\ & = \log_3 G_{4,3}(n)-8\\ &\le \log_3 G_{4,3}(8)-8\\ & = \log_3\frac{8^{8}}{3^{16}} \lt 0. \end{align} (4.33)

    CASE 7.2. p = 5 . One notes that v_5(n) = v_5(n-2) = v_5(n-3) = v_5(n-5) = 0 and at most two terms of \{v_5(n-1), v_5(n-4), v_5(n-6)\} can be nonzero. Moreover, if v_5(n-4)\ge 1 and v_5(n-1) = v_5(n-6) = 0 , then we can deduce that

    v_5(n-1)+v_5(n-4)+2v_5(n-6) = v_5(n-4)\le \log_5 (n-4)\le 2+2\log_5 (n-5);

    and if v_5(n-4) = 0 with v_5(n-1)\ge 1 and v_5(n-6)\ge 1 , then we get that \min\{v_5(n-1), v_5(n-6)\} = 1 , which infers that

    v_5(n-1)+v_5(n-4)+2v_5(n-6) = v_5(n-1)+2v_5(n-6)\le 2+2\log_5 (n-5).

    Together with v_5(E_{7, n})\le\log_5(n-5)^4 , one obtains that

    \begin{align*} &\sum\limits_{i = 0}^5v_5(n-i)+2v_5(n-6)+2v_5(n-7)+v_5(E_{7, n})\\ & = v_5(n-1)+v_5(n-4)+2v_5(n-6)+v_5(E_{7, n})\\ &\le 2+\log_5(n-5)^6. \end{align*}

    It is easy to check that D_{7, 5}(n)\le 4(2+\log_5(n-5)^6)-3n-8 = 4\log_5(n-5)^6-3n < 0 for 8\le n\le 10 . Now let n\ge 11 . Then it follows from Lemma 2.2 that

    \begin{align} D_{7, 5}(n) &\le 4\Big(2+\log_5(n-5)^6\Big)-3n-8\\ & = \log_5\frac{(n-5)^{24}}{5^{3(n-5)}}-15\\ & = \log_5G_{6,5}(n-5)-15\\ &\le \log_5G_{6,5}(6)-15\\ & = \log_5\frac{6^{24}}{5^{33}} \lt 0. \end{align} (4.34)

    CASE 7.3. p\ge 7 . One has v_p(n-4) = v_p(n-5) = v_p(n-6) = 0 and only one of \{v_p(n), v_p(n-1), v_p(n-2), v_p(n-3)\} can be nonzero. This infers that

    \begin{align} &\sum\limits_{i = 0}^5v_p(n-i)+2v_p(n-6)+2v_p(n-7)+v_p(E_{7, n})\\ & = \sum\limits_{i = 0}^3v_p(n-i)+v_p(E_{7, n})\\ &\le \log_p n^5. \end{align} (4.35)

    It follows from (4.30) and (4.35) that

    \begin{align} D_{7, p}(n) &\le (p-1)\log_p n^5-n(p-2)-8\\ & = \log_p\frac{n^{5(p-1)}}{p^{n(p-2)}}-8\\ & = \log_pG_{5,p}(n)-8\\ &\le \log_pG_{5,p}(8)-8\\ & = (p-1)\log_p \frac{8^5}{p^8} \lt 0. \end{align} (4.36)

    Hence (4.29), (4.31), (4.33), (4.34) and (4.36) give us that v_p((n-7)!S(n, n-7)) < n .

    This completes the proof of Theorem 1.4.

    From Theorem 1.2, we deduce the following interesting result.

    Proposition 5.1. Let p be a prime. Let n and a be positive integers with (a, p) = 1 .

    (i). One has

    v_p(S(ap^n, ap^n-1)) = \left\{\begin{array}{lc} n-1,& {\it if}\ p = 2, \\ n, & {\it if}\ p \gt 2. \end{array}\right.

    (ii). If n\ge 2 , then

    v_p(S(ap^n, ap^n-2)) = \left\{\begin{array}{lc} n-2,& {\it if}\ p = 2, \\ n-1,& {\it if}\ p = 3, \\ n+1,& {\it if}\ p = 5, \\ n, & {\it if}\ p \gt 5 \end{array}\right.

    and

    v_p(S(ap^n, ap^n-3)) = \left\{\begin{array}{lc} n-2,& {\it if}\ p = 2, \\ n+1,& {\it if}\ p = 3, \\ n, & {\it if}\ p \gt 3. \end{array}\right.

    (iii). If n\ge 3 , then

    \begin{align} v_p(S(ap^n, ap^n-4)) = \left\{\begin{array}{ll} n-3,& {\it if}\ p = 2, \\ n-1,& {\it if}\ p\in \{3,5\}, \\ n+1,& {\it if}\ p = 251, \\ n, & {\it otherwise}, \end{array}\right. \end{align} (5.1)
    \begin{align} v_p(S(ap^n, ap^n-5)) = \left\{\begin{array}{ll} n-2,& {\it if}\ p = 2, \\ n-1,& {\it if}\ p = 3, \\ n+1,& {\it if}\ p\in\{5,19\}, \\ n, & {\it otherwise}, \end{array}\right. \end{align} (5.2)
    \begin{align} v_p(S(ap^n, ap^n-6)) = \left\{\begin{array}{ll} n-3,& {\it if}\ p = 2, \\ n-2,& {\it if}\ p = 3, \\ n-1,& {\it if}\ p = 7, \\ n+1,& {\it if}\ p = 19087, \\ n, & {\it otherwise} \end{array}\right. \end{align} (5.3)

    and

    \begin{align} v_p(S(ap^n, ap^n-7)) = \left\{\begin{array}{ll} n-3,& {\it if}\ p = 2, \\ n-1,& {\it if}\ p = 3, \\ n+1,& {\it if}\ p\in \{7,751\}, \\ n, & {\it otherwise}. \end{array}\right. \end{align} (5.4)

    Proof. Since n\ge 1 and (a, p) = 1 , one may deduce that v_p(ap^n) = n and v_p(ap^n-1) = 0 .

    (ⅰ). From Theorem 1.2, we derive that

    \begin{align*} v_p(S(ap^n, ap^n-1)) = v_p(ap^n)+v_p(ap^n-1)-v_p(2) = n-v_p(2). \end{align*}

    Thus the truth of (ⅰ) follows immediately.

    (ⅱ). Let n\ge 2 . One gets that v_p(ap^n-2) = v_p(2) and v_p(ap^n-3) = v_p(3) . Note that v_p(3ap^n-5) = v_p(5) . By Theorem 1.2, we then obtain that

    \begin{align} v_p(S(ap^n, ap^n-2)) & = \sum\limits_{i = 0}^2v_p(ap^n-i)+v_p(3ap^n-5)-v_p(4)-v_p(3!)\\ & = n+v_p(2)+v_p(5)-v_p(4)-v_p(3!)\\ & = n+v_p(5)-v_p(4)-v_p(3) \end{align} (5.5)

    and

    \begin{align} v_p(S(ap^n, ap^n-3)) & = \sum\limits_{i = 0}^3v_p(ap^n-i)+v_p(ap^n-2)+v_p(ap^n-3)-v_p(2)-v_p(4!)\\ & = n+2v_p(2)+2v_p(3)-v_p(2)-v_p(4!)\\ & = n+v_p(3)-v_p(4). \end{align} (5.6)

    Thus the truth of (ⅱ) follows immediately from (5.5) and (5.6).

    (ⅲ). Let n\ge 3 . It is easy to see that v_p(ap^n-4) = v_p(4) , v_p(ap^n-5) = v_p(5) , v_p(ap^n-6) = v_p(6) and v_p(ap^n-7) = v_p(7) . From Theorem 1.2, one can deduce that

    \begin{align} &v_p(S(ap^n, ap^n-4))\\ & = \sum\limits_{i = 0}^4v_p(ap^n-i) +v_p(15a^3p^{3n}-150a^2p^{2n}+485ap^n-502)-v_p(48)-v_p(5!)\\ & = n+v_p(2)+v_p(3)+v_p(4)-v_p(48)-v_p(5!) +v_p(15a^3p^{3n}-150a^2p^{2n}+485ap^n-502)\\ & = n-v_p(48)-v_p(5)+v_p(15a^3p^{3n}-150a^2p^{2n}+485ap^n-502). \end{align} (5.7)

    Note that 48 = 2^4\cdot 3 and 502 = 2\cdot 251 . Thus we derive that

    \begin{align*} v_p(15a^3p^{3n}-150a^2p^{2n}+485ap^n-502) = \left\{\begin{array}{ll} 1,& {\rm if}\ p\in \{2,251\}, \\ 0, & {\rm otherwise}.\end{array}\right. \end{align*}

    Together with (5.7), one then gets the truth of (5.1).

    Subsequently, by Theorem 1.2 we obtain that

    \begin{align} &v_p(S(ap^n, ap^n-5))\\ & = \sum\limits_{i = 0}^5v_p(ap^n-i)+v_p(ap^n-4)+v_p(ap^n-5)\\ &\ \ \ \ \ +v_p(3a^2p^{2n}-23ap^n+38)-v_p(16)-v_p(6!)\\ & = n+v_p(2)+v_p(3)+2v_p(4)+2v_p(5)-v_p(16)-v_p(6!)+v_p(3a^2p^{2n}-23ap^n+38)\\ & = n+v_p(5)-v_p(6)-v_p(4)+v_p(3a^2p^{2n}-23ap^n+38). \end{align} (5.8)

    Note that 38 = 2\cdot 19 . It follows that

    \begin{align*} v_p(3a^2p^{2n}-23ap^n+38) = \left\{\begin{array}{ll} 1, & {\rm if}\ p\in \{2,19\}, \\ 0, & {\rm otherwise}.\end{array}\right. \end{align*}

    Together with (5.8), one then obtains the truth of (5.2).

    Furthermore, from Theorem 1.2 we get that

    \begin{align} &v_p(S(ap^n, ap^n-6))\\ & = \sum\limits_{i = 0}^{6}v_p(ap^n-i)-v_p(576)-v_p(7!)\\ &\ \ \ \ \ +v_p(63a^5p^{5n}-1575a^4p^{4n}+15435a^3p^{3n} -73801a^2p^{2n}+171150ap^n-152696)\\ & = n-v_p(576)-v_p(7)\\ &\ \ \ \ \ +v_p(63a^5p^{5n}-1575a^4p^{4n}+15435a^3p^{3n} -73801a^2p^{2n}+171150ap^n-152696). \end{align} (5.9)

    Since 152696 = 2^3\cdot 19087 and v_2(171150) = 1 and n\ge 3 , one has

    \begin{align*} v_p(63a^5p^{5n}-1575a^4p^{4n}+15435a^3p^{3n} -73801a^2p^{2n}+171150ap^n-152696) & = \left\{\begin{array}{ll} 3, & {\rm if}\ p = 2, \\ 1, & {\rm if}\ p = 19087, \\ 0, & {\rm otherwise}.\end{array}\right. \end{align*}

    Note that 576 = 2^6\cdot3^2 . Together with (5.9), one derives the truth of (5.3).

    Finally, by Theorem 1.2 we can deduce that

    \begin{align} &v_p(S(ap^n, ap^n-7))\\ & = \sum\limits_{i = 0}^{7}v_p(ap^n-i)+v_p(ap^n-6)+v_p(ap^n-7)-v_p(144)-v_p(8!)\\ &\ \ \ \ \ +v_p(9a^4p^{4n}-198a^3p^{3n}+1563a^2p^{2n}-5182ap^n+6008)\\ & = n+v_p(7)-v_p(24)-v_p(8) +v_p(9a^4p^{4n}-198a^3p^{3n}+1563a^2p^{2n}-5182ap^n+6008). \end{align} (5.10)

    Since 6008 = 2^3\cdot 751 and v_2(5182) = 1 and n\ge 3 , one has

    \begin{align*} &v_p(9a^4p^{4n}-198a^3p^{3n}+1563a^2p^{2n}-5182ap^n+6008) = \left\{\begin{array}{ll} 3, & {\rm if}\ p = 2, \\ 1, & {\rm if}\ p = 751, \\ 0, & {\rm otherwise}.\end{array}\right. \end{align*}

    Note that 24 = 2^3\cdot3 . Together with (5.10), one gets the truth of (5.4).

    This completes the proof of Proposition 5.1.

    Moreover, we can obtain the following result.

    Proposition 5.2. Let p be a prime.

    (i). One has

    v_p(S(p, p-1)) = \left\{\begin{array}{cc} 0,& {\it if}\ p = 2, \\ 1, & {\it if}\ p \gt 2. \end{array}\right.

    (ii). If p\ge 3 , then

    v_p(S(p, p-2)) = \left\{\begin{array}{cc} 0,& {\it if}\ p = 3, \\ 2,& {\it if}\ p = 5, \\ 1, & {\it if}\ p \gt 5. \end{array}\right.

    (iii). If p\ge 5 , then

    v_p(S(p, p-3)) = 1

    and

    v_p(S(p, p-4)) = \left\{\begin{array}{ll} 0,& {\it if}\ p = 5, \\ 2,& {\it if}\ p = 251, \\ 1, & {\it otherwise}. \end{array}\right.

    (iv). If p\ge 7 , then

    v_p(S(p, p-5)) = \left\{\begin{array}{cc} 2,& {\it if}\ p = 19, \\ 1, & {\it if}\ p\ne19 \end{array}\right.

    and

    v_p(S(p, p-6)) = \left\{\begin{array}{cl} 0,& {\it if}\ p = 7, \\ 2,& {\it if}\ p = 19087, \\ 1, & {\it otherwise}. \end{array}\right.

    (v). If p\ge 11 , then

    v_p(S(p, p-7)) = \left\{\begin{array}{cc} 2,& {\it if}\ p = 751, \\ 1, & {\it if}\ p\ne751. \end{array}\right.

    (vi). For any odd prime p , one has

    v_p(S(p^2, p^2-4)) = \left\{\begin{array}{ll} 1,& {\it if}\ p\in \{3,5\}, \\ 3,& {\it if}\ p = 251, \\ 2, & {\it otherwise}, \end{array}\right.
    v_p(S(p^2, p^2-5)) = \left\{\begin{array}{ll} 1,& {\it if}\ p = 3, \\ 3,& {\it if}\ p\in\{5,19\}, \\ 2, & {\it otherwise}, \end{array}\right.
    v_p(S(p^2, p^2-6)) = \left\{\begin{array}{ll} 0,& {\it if}\ p = 3, \\ 1,& {\it if}\ p = 7, \\ 3,& {\it if}\ p = 19087, \\ 2, & {\it otherwise} \end{array}\right.

    and

    v_p(S(p^2, p^2-7)) = \left\{\begin{array}{ll} 1,& {\it if}\ p = 3, \\ 3,& {\it if}\ p\in\{7,751\}, \\ 2, & {\it otherwise}. \end{array}\right.

    Proof. It is easy to check that parts (ⅰ) and (ⅱ) are true, so we omit the details.

    (ⅲ). Since p\ge 5 , by Theorem 1.2, we get that

    v_p(S(p, p-3)) = v_p(p)+v_p(p-1)+2v_p(p-2)+2v_p(p-3)-v_p(2)-v_p(4!) = 1

    and

    \begin{align} v_p(S(p, p-4)) & = \sum\limits_{i = 0}^4v_p(p-i)+v_p(15p^3-150p^2+485p-502)-v_p(48)-v_p(5!)\\ & = 1+v_p(15p^3-150p^2+485p-502)-v_p(5). \end{align} (5.11)

    Note that 502 = 2\cdot 251 and 485\not\equiv 2 \pmod{251} . So one has

    \begin{align*} v_p(15p^3-150p^2+485p-502) = \left\{\begin{array}{ll} 1,& {\rm if}\ p = 251, \\ 0, & {\rm otherwise}.\end{array}\right. \end{align*}

    Together with (5.11) we arrive at the truth of part (ⅲ).

    (ⅳ). Let p\ge 7 . Since 576 = 2^6\cdot3^2 , from Theorem 1.2, one gets that

    \begin{align} v_p(S(p, p-5)) & = \sum\limits_{i = 0}^5v_p(p-i)+v_p(p-4)+v_p(p-5) +v_p(3p^2-23p+38)-v_p(16)-v_p(6!)\\ & = 1+v_p(3p^2-23p+38) \end{align} (5.12)

    and

    \begin{align} &v_p(S(p, p-6))\\ & = \sum\limits_{i = 0}^6v_p(p-i) + v_p(63p^5-1575p^4+15435p^3-73801p^2+171150p-152696)\\ &\ \ \ \ -v_p(576)-v_p(7!)\\ & = 1-v_p(7)+v_p(63p^5-1575p^4+15435p^3-73801p^2+171150p-152696). \end{align} (5.13)

    Note that 38 = 2\cdot19 and 23\not\equiv2\pmod{19} , 152696 = 2^3\cdot 19087 and 171150\not\equiv 2^3\pmod{19087} . It infers that

    \begin{align*} v_p(3p^2-23p+38) = \left\{\begin{array}{ll} 1, & {\rm if}\ p = 19, \\ 0, & {\rm otherwise}\end{array}\right. \end{align*}

    and

    \begin{align*} v_p(63p^5-1575p^4+15435p^3-73801p^2+171150p-152696) = \left\{\begin{array}{ll} 1, & {\rm if}\ p = 19087, \\ 0, & {\rm otherwise}.\end{array}\right. \end{align*}

    By (5.12) and (5.13), we can deduce the truth of part (iv).

    (ⅴ). Since p\ge 11 and 144 = 2^4\cdot3^2 , it follows from Theorem 1.2 that

    \begin{align} v_p(S(p, p-7)) & = \sum\limits_{i = 0}^7v_p(p-i)+v_p(p-6)+v_p(p-7)\\ &\ \ \ \ \ +v_p(9p^4-198p^3+1563p^2-5182p+6008)-v_p(144)-v_p(8!)\\ & = 1+v_p(9p^4-198p^3+1563p^2-5182p+6008). \end{align} (5.14)

    By 6008 = 2^3\cdot 751 and 5182\not\equiv 2^3\pmod{751} , one knows that

    \begin{align*} v_p(9p^4-198p^3+1563p^2-5182p+6008) = \left\{\begin{array}{ll} 1, & {\rm if}\ p = 751, \\ 0, & {\rm otherwise}.\end{array}\right. \end{align*}

    Thus the truth of part (ⅴ) follows immediately from (5.14).

    (ⅵ). Let p\ge 3 . By Theorem 1.2, we get that

    \begin{align} v_p(S(p^2, p^2-4)) & = \sum\limits_{i = 0}^4v_p(p^2-i)+v_p(15p^6-150p^4+485p^2-502)-v_p(48)-v_p(5!)\\ & = 2+v_p(251)-v_p(3)-v_p(5), \end{align} (5.15)
    \begin{align} v_p(S(p^2, p^2-5)) & = \sum\limits_{i = 0}^5v_p(p^2-i)+v_p(p^2-4)+v_p(p^2-5) +v_p(3p^4-23p^2+38)-v_p(16)-v_p(6!)\\ & = 2+v_p(5)+v_p(19)-v_p(3), \end{align} (5.16)
    \begin{align} &v_p(S(p^2, p^2-6))\\ & = \sum\limits_{i = 0}^6v_p(p^2-i) + v_p(63p^{10}-1575p^{8}+15435p^{6}-73801p^{4}+171150p^2-152696)\\ &\ \ \ \ -v_p(576)-v_p(7!)\\ & = 2+v_p(3)+v_p(5)+v_p(6)+v_p(152696)-v_p(576)-v_p(7!)\\ & = 2+v_p(19087)-2v_p(3)-v_p(7) \end{align} (5.17)

    and

    \begin{align} v_p(S(p^2, p^2-7)) & = \sum\limits_{i = 0}^7v_p(p^2-i)+v_p(p^2-6)+v_p(p^2-7)\\ &\ \ \ \ \ +v_p(9p^{8}-198p^{6}+1563p^4-5182p^2+6008)-v_p(144)-v_p(8!)\\ & = 2+v_p(3)+v_p(5)+2v_p(6)+2v_p(7)+v_p(6008)-v_p(144)-v_p(8!)\\ & = 2+v_p(7)+v_p(751)-v_p(3). \end{align} (5.18)

    Hence the truth of part (ⅵ) follows from (5.15) to (5.18).

    This finishes the proof of Proposition 5.2.

    Let p be a prime number. Let n and k be positive integers. The computation of the exact p -adic valuation of Stirling numbers is difficult. In Theorem 1.1, we use 2-associated Stirling number S_2(n, k) to represent a formula to calculate the p -adic valuation of Stirling numbers of the second kind S(n, k) . The formula of v_p(S(n, n-k)) depends on the value of S_2(i, i-k) , where k+2\le i\le 2k . From this, we arrive at a formula to compute v_p(S(n, n-k)) , which enables us to show that v_p((n-k)!S(n, n-k)) < n with 0\le k\le \min\{7, n-1\} and p\ge 3 . Let 1\le k\le 7 and a be a positive integer with (a, p) = 1 . For n\ge 3 , by Proposition 5.1 we know that n-3\le v_p(S(ap^n, ap^n-k))\le n+1 , and v_p(S(ap^n, ap^n-k)) = n holds if p > 19087 . Moreover, for any prime number p with p > 19087 , Propositions 5.1 and 5.2 also imply that v_p(S(p, p-k)) = 1 and v_p(S(p^2, p^2-k)) = 2 .

    The authors would like to thank the anonymous referees for careful reading of the manuscript and helpful comments and suggestions.

    We declare that we have no conflict of interest.



    [1] L. Carlitz, Congruences for generalized Bell and Stirling numbers, Duke Math. J., 22 (1955), 193-205. doi: 10.1215/S0012-7094-55-02220-1
    [2] Y. H. Kwong, Minimum periods of S (n, k) modulo M, Fibonacci Quart., 27 (1989), 217-221.
    [3] O. Y. Chan, D. Manna, Congruences for Stirling numbers of the second kind, Contemp. Math., 517 (2010), 97-111. doi: 10.1090/conm/517/10135
    [4] F. Clarke, Hensel's lemma and the divisibility by primes of Stirling-like numbers, J. Number Theory, 52 (1995), 69-84. doi: 10.1006/jnth.1995.1056
    [5] D. M. Davis, Divisibility by 2 of Stirling-like numbers, Proc. Amer. Math. Soc., 110 (1990), 597-600.
    [6] S. F. Hong, J. R. Zhao, W. Zhao, The 2-adic valuations of Stirling numbers of the second kind, Int. J. Number Theory, 8 (2012), 1057-1066. doi: 10.1142/S1793042112500625
    [7] J. R. Zhao, S. F. Hong, W. Zhao, Divisibility by 2 of Stirling numbers of the second kind and their differences, J. Number Theory, 140 (2014), 324-348. doi: 10.1016/j.jnt.2014.01.005
    [8] W. Zhao, J. R. Zhao, S. F. Hong, The 2-adic valuations of differences of Stirling numbers of the second kind, J. Number Theory, 153 (2015), 309-320. doi: 10.1016/j.jnt.2015.01.016
    [9] T. Amdeberhan, D. Manna, V. Moll, The 2-adic valuation of Stirling numbers, Experiment. Math., 17 (2008), 69-82. doi: 10.1080/10586458.2008.10129026
    [10] T. Lengyel, On the 2-adic order of Stirling numbers of the second kind and their differences, DMTCS Proc. AK, (2009), 561-572.
    [11] S. F. Hong, On the p-adic behaviors of Stirling numbers of the first and second kinds, RIMS Kokyuroku Bessatsu, to appear.
    [12] S. F. Hong, M. Qiu, On the p-adic properties of Stirling numbers of the first kind, Acta Math. Hungar., in press.
    [13] T. Komatsu, P. Young, Exact p-adic valuations of Stirling numbers of the first kind, J. Number Theory, 177 (2017), 20-27. doi: 10.1016/j.jnt.2017.01.023
    [14] T. Lengyel, On the divisibility by 2 of Stirling numbers of the second kind, Fibonacci Quart., 32 (1994), 194-201.
    [15] M. Qiu, S. F. Hong, 2-Adic valuations of Stirling numbers of the first kind, Int. J. Number Theory, 15 (2019), 1827-1855. doi: 10.1142/S1793042119501021
    [16] S. D. Wannermacker, On 2-adic orders of Stirling numbers of the second kind, Integers, 5 (2005), A21, 7.
    [17] L. Comtet, Advanced Combinatorics: The Art of Finite and Infinite Expansions, Revised and Enlarged Edition, D. Reidel Publishing Co., Dordrecht and Boston, 1974.
    [18] N. Koblitz, p-Adic Numbers, p-Adic Analysis and Zeta-Functions, 2Eds., GTM 58, SpringerVerlag, New York, 1984.
  • Reader Comments
  • © 2020 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(4526) PDF downloads(381) Cited by(0)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog