Abstract
Andrews, Hirschhorn, and Sellers studied the partition function ped(n) which enumerates the number of partitions of n with even parts distinct, and obtained a number of interesting congruences. This paper aims to introduce a partition statistic to investigate the partition function ped(n). We give combinatorial interpretations for some properties of ped(n) including the infinite families of congruences given by Andrews et al.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
A partition \(\lambda \) of a positive integer n is a finite non-increasing sequence of positive integers \(\lambda =(\lambda _1,\ldots , \lambda _l)\) such that \(|\lambda |=\sum _{i=1}^{l}\lambda _i=n\). The Ferrers graph of a partition \(\lambda \) is a set of coordinates in the bottom right quadrant of the plane where the i-th row contains \(\lambda _i\) dots. We denote by \(\lambda '\) the conjugate of \(\lambda \), which is the partition whose graph is obtained by reflecting the Ferrers graph of \(\lambda \) about the main diagonal. For example, we give \(\lambda =(4,4,2,2,1)\) and its conjugate partition \(\lambda '=(5,4,2,2)\) in Fig. 1.
Let p(n) denote the ordinary partition function. The partition statistic crank defined by Andrews and Garvan [2, 6] can be used to provide combinatorial interpretations for Ramanujan’s famous congruences
The crank of a partition \(\lambda \ne (1)\) is defined as follows:
where \(n_1(\lambda )\) denotes the number of parts equal to one in \(\lambda \) and \(\mu (\lambda )\) denotes the number of parts in \(\lambda \) larger than \(n_1(\lambda )\). Let M(m, n) enumerate partitions of n with crank m. It should be pointed out that when \(\lambda =(1)\),
Andrews and Garvan [2, 6] established the generating function of M(m, n) as given by
Here and throughout this paper, \((a;q)_\infty \) stands for the q-shifted factorial
and for any positive integer k,
Let M(m, t, n) denote the number of partitions of n with crank congruent to m modulo t. In 1990, Garvan [7] presented a graceful refinement of the congruence (1.1)
together with the combinatorial interpretation
Let ped(n) be the function that enumerates partitions of n with even parts distinct. Obviously,
The sequence \({ped(n)}_{n\ge 0}\) is well known and can be seen in [11, A001935], as well as other combinatorial interpretations. In 2010, Andrews, Hirschhorn, and Sellers [3] proved the following congruences.
Theorem 1.1
For \(\alpha ,n \ge 0\),
Theorem 1.2
For \(n \ge 0\),
In 2017, Merca [9] provided a simple criterion for deciding the parity of ped(n).
Theorem 1.3
The number of partitions of n with distinct even parts is odd if and only if n is a triangular number.
In this paper, we aim at introducing a partition statistic which we call ped-crank to study the partition function ped(n). Let \(M_{ped}(m,n)\) denote the number of partitions of n with even parts distinct with ped-crank m, and let
The main results of this paper are summarized below.
Theorem 1.4
For \(\alpha ,n \ge 0\),
Any of the following three corollaries deduced from Theorem 1.4 provides a combinatorial interpretation or a refinement of Theorem 1.1. When \(\alpha =0\), combining Corollary 1.5 and Corollary 1.6 refines (1.3). Meanwhile, Corollary 1.7 combinatorially interprets (1.3).
Corollary 1.5
For \(m=0,1,2\) and \(\alpha ,n \ge 0\),
Corollary 1.6
For \(\alpha ,n \ge 0\),
Corollary 1.7
For \(\alpha ,n \ge 0\),
Theorem 1.8
If n cannot be written as a sum of a triangular number and a square of even integer, we have
Moreover, \(M_{ped}(0,n)\) is odd if and only if n is a triangular number.
Theorem 1.9
For \(m=0,1,2,3,4,5\) and \(n\ge 0\),
It is worth mentioning that Theorem 1.8 not only combinatorially interprets but also refines Theorem 1.3, and (1.6) provides a combinatorial interpretation for (1.4).
2 Definition of the ped-crank
In this section, we shall define the ped-crank of partitions with even parts distinct based on Glaisher’s bijection and a modified version \(\varphi \) of the Wright map established by Seo and Yee [10].
We first give a quick overview of Glaisher’s bijection and the Frobenius symbol. Let \(D_n\) denote the set of distinct partitions, and let \(O_n\) denote the set of odd partitions of n respectively. Glaisher’s bijection \(\phi \): \(O_n\rightarrow D_n\) is defined as follows. Let \(\lambda =(1^{m_1}3^{m_3}\ldots )\in O_n\) be an odd partition. For every odd i, let \(\phi (\lambda )\) contain part \(i\cdot 2^r\), if and only if the integer \(m_i\) written in binary has 1 at the r-th position. In the other direction, let \(\psi \): \(D_n\rightarrow O_n\) be defined by an iterative procedure. Start with \(\lambda =(\lambda _1,\lambda _2,\ldots )\in D_n\). Substitute every even part \(\lambda _i\) with two parts \(\lambda _i/2\). Repeat until the resulting partition has no even parts.
The Frobenius symbol of n is a two-rowed array [1, 14]
where \(\alpha _1>\alpha _2>\ldots >\alpha _{\ell }\ge 0\), \(\beta _1>\beta _2>\ldots >\beta _{\ell }\ge 0\) and \(n=|\alpha |+|\beta |+\ell \). If we express an ordinary partition by Ferrers graph, it is easy to see that \(\alpha _i\) form rows to the right of the diagonal and \(\beta _i\) form columns below the diagonal. Thus the Frobenius symbol is another representation of an ordinary partition. For instance, the Frobenius symbol for (8, 7, 4, 3, 1) is
Giving a real number c, we define \(c\lambda \) as the partition whose parts are c times each part of \(\lambda \). For example, let \(\lambda =(4,2,2)\). We have \(4\lambda =(16,8,8)\) and \(\frac{1}{2}\lambda =(2,1,1)\). Suppose \(\mu \) and \(\nu \) are two partitions. Let \(\mu \cup \nu \) denote the partition consisting of all the parts of \(\mu \) and \(\nu \). The definition of ped-crank is given based on the following theorem.
Theorem 2.1
For integer \(k_1\ge -1\), \(k_2\ge 1\), there is a bijection \(\Delta \) between the set of partitions of n with even parts distinct and the set of vector partitions \((\alpha ,\beta ,\gamma )\) with \(|\alpha |+|\beta |+|\gamma |\) equal to n. Here \(\alpha \) is an even partition, \(\beta \) is a partition of the form \((4k_1+1,\ldots ,9,5,1)\) or \((4k_2-1,\ldots ,11,7,3)\) and \(\gamma \) is a distinct even partition.
Proof
The bijection \(\Delta \) can be decomposed into six weight preserving steps.
-
Step 1.
\(\lambda \rightarrow (\omega ,\gamma )\): Start with a partition \(\lambda \) of n with even parts distinct. Split \(\lambda \) into a pair of partitions \((\omega ,\gamma )\) according to the parts odd or even. It is clear that \(\omega \) is an odd partition and \(\gamma \) is a distinct even partition.
-
Step 2.
\((\omega ,\gamma )\rightarrow (\xi ,\gamma )\): By Glaisher’s bijection, let \(\phi (\omega )=\xi \). One can see that \(\xi \) is a distinct partition.
-
Step 3.
\((\xi ,\gamma )\rightarrow (\mu ^1,\mu ^2,\pi ,\gamma )\): Divide \(\xi \) into a triple of partitions \((\mu ^1,\mu ^2,\pi )\) according to the remainder of the parts mod 4. Here \(\mu ^1\ (\mu ^2)\) consist of all the parts congruent to 1(3) mod 4 and \(\pi \) consists of all the even parts of \(\xi \).
-
Step 4.
\((\mu ^1,\mu ^2,\pi ,\gamma )\rightarrow (\mu ^1,\mu ^2,\zeta ,\gamma )\): Let \(\zeta =2\psi (\frac{1}{2}\pi )\) by applying Glaisher’s bijection. Since \(\frac{1}{2}\pi \) is a distinct partition, we can say that \(\zeta \) is a partition with all parts congruent to 2 mod 4.
-
Step 5.
\((\mu ^1,\mu ^2,\zeta ,\gamma )\rightarrow (\eta ,\beta ,\zeta ,\gamma )\): Write \(\mu ^1\) and \(\mu ^2\) as
$$\begin{aligned} \mu ^1&=(4a_1+1,4a_2+1,\ldots ,4a_{s+m}+1),\\ \mu ^2&=(4b_1+3,4b_2+3,\ldots ,4b_{s}+3), \end{aligned}$$where \(a_1>a_2>\cdots >a_{s+m}\ge 0\) and \(b_1>b_2>\cdots >b_s\ge 0\).
-
Case 1.
\(m\ge 0\). Using the bijection \(\varphi \) established by Seo and Yee [10], a Frobenius symbol
$$\begin{aligned} \mu =\left( \begin{array}{llll} a_{1+m} &{}\quad a_{2+m} &{}\cdots &{}a_{s+m} \\ b_1 &{} \quad b_2 &{} \cdots &{} b_s \end{array}\right) \end{aligned}$$and a partition \(\nu =(a_1-m+1,a_2-m+2,\ldots ,a_m)\) can be constructed. Let \(\varphi (\mu ^1,\mu ^2)=(\eta ,\beta )\), where \(\eta =4(\mu \cup \nu )\) and \(\beta =(4(m-1)+1,4(m-2)+1,\ldots ,5,1)\).
-
Case 2.
\(m<0\). Correspondingly, a Frobenius symbol
$$\begin{aligned} \mu =\left( \begin{array}{llll} b_{1-m} &{}\quad b_{2-m} &{} \cdots &{} b_s \\ a_{1} &{} \quad a_{2} &{}\cdots &{}a_{s+m} \end{array}\right) \end{aligned}$$and a partition \(\nu =(b_1+m+1,b_2+m+2,\ldots ,b_{-m})\) can be constructed. Let \(\varphi (\mu ^1,\mu ^2)=(\eta ,\beta )\), where \(\eta =4(\mu \cup \nu )'\) and \(\beta =(4(-m-1)+3,4(-m-2)+3,\ldots ,7,3)\).
-
Case 1.
-
Step 6.
\((\eta ,\beta ,\zeta ,\gamma )\rightarrow (\alpha ,\beta ,\gamma )\): Ultimately, let \(\alpha =\eta \cup \zeta \) and define \(\Delta (\lambda )=(\alpha ,\beta ,\gamma )\).
Furthermore, one sees that the above construction can be reversed. This completes the proof. \(\square \)
An example of the bijection \(\Delta \) is given below.
Example 2.2
[4]
Now we are ready to give the definition of the ped-crank of a partition with even parts distinct under the bijection \(\Delta \).
Definition 2.3
Let \(\lambda \) be a partition with even parts distinct and \(\Delta (\lambda )=(\alpha , \beta , \gamma )\). The ped-crank of \(\lambda \), denoted by \(c_{ped}(\lambda )\), is defined as the crank of \(\frac{1}{2}\alpha \).
3 Generating function of \(M_{ped}(m,n)\)
This section focuses on the generating function of \(M_{ped}(m,n)\).
According to the bijection \(\Delta \), the generating function of \(\beta \) can be derived by using Jacobi’s triple product identity.
Moreover, it is trivial that the generating function of \(\gamma \) is
Since the ped-crank only relies on the even partition \(\alpha \), by (1.2), the generating function of \(M_{ped}(m,n)\) can be given as
By considering the transformation that interchanges z and \(z^{-1}\) in (3.1), we have
Thus, for any positive integer t,
In other words,
4 Preliminaries
In this section we present some results that will be used in Section 5.
Lemma 4.1
Proof
Replacing q by \(-q\) in \((q;q)_\infty \), we have
\(\square \)
Lemma 4.2
( [4, Entry 22, p. 36])
Replacing q by \(-q\) in the above equation, we get
Lemma 4.3
( [4, p. 49])
Theorem 4.4
( [5, Lemma 2.2])
where
Lemma 4.5
( [13, Lemma 2.5])
Lemma 4.6
( [8, p. 5])
Lemma 4.7
( [3, Theorem 3.1])
The following two theorems are crucial for establishing combinatorial interpretations.
Theorem 4.8
For any fixed n, if
then
Proof
By (3.2), we have
Substituting (4.4)–(4.7) into (4.1)–(4.2), we have
Solving system of linear homogeneous equations (4.3), (4.8) and (4.9), we get
Moreover,
which implies
This completes the proof. \(\square \)
The following theorem can be checked similarly.
Theorem 4.9
For any fixed n, if
then
5 Proofs of main results
In this section, we give proofs of our main results. Hereafter we always assume \(\alpha , n \ge 0\) unless specified otherwise.
Proof of Theorem 1.4
Setting \(z=e^{\pi i}=-1\) in (3.1), we get
Let g(q) be a polynomial of q, observing that the coefficient of \(q^n\) in g(q) is zero implies the coefficient of \(q^n\) in \(g(-q)\) is zero and vice versa. Hence we consider the following equation. By Lemma 4.1, replacing q by \(-q\) in (5.1), we have
According to Lemma 4.6, we find that
Extracting those terms associated with powers \(q^{3n+1}\) on both sides of (5.3), then dividing by q and replacing \(q^3\) by q, we arrive at
Since the coefficients of \(q^{3n+1}\) and \(q^{3n+2}\) in (5.4) are both zero, we can conclude that the coefficients of \(q^{9n+4}\) and \(q^{9n+7}\) in (5.2) are both zero. This yields
Extracting the terms involving \(q^{3n}\) in (5.4) and substituting \(q^3\) by q gives
Since 9n has the same parity as n, (5.7) becomes
From (5.2), (5.8) and mathematical induction, it follows that
Comparing (5.9) with (5.2), the following equations can be proved by similar arguments for (5.5)–(5.6), and hence the proof is omitted.
Substituting \(z=e^{\frac{\pi i}{3}}\) into (3.1), by (3.2) and \(e^{\frac{\pi i}{3}}+e^{\frac{5\pi i}{3}}=-(e^{\frac{2\pi i}{3}}+e^{\frac{4\pi i}{3}})=-e^{\pi i}=1\), we see that
By Lemma 4.3, we have
Extracting the terms involving \(q^{3n+1}\) in (5.12), then dividing by q and replacing \(q^3\) by q, we find that
Obviously, the coefficients of \(q^{3n+1}\) and \(q^{3n+2}\) in (5.13) are both zero, which gives
Considering the terms involving \(q^{3n}\) in (5.13), after simplification, we get
Comparing (5.16) with (5.1), according to (5.14)–(5.15) and the proofs of (5.10)–(5.11), a simple deduction shows that
Combining (5.10)–(5.11), (5.17)–(5.18) and Theorem 4.9, Theorem 1.4 follows immediately. \(\square \)
Proof of Corollary 1.5
Corollary 1.5 can be checked easily by (4.6)–(4.7) and Theorem 1.4, hence we omitted the details. \(\square \)
Proof of Corollary 1.6
By (4.6), (4.10) and Theorem 1.4, one can see that
Hence Corollary 1.6 holds.
Proof of Corollary 1.7
By (3.2), we have
Then Corollary 1.7 follows immediately according to (5.10)–(5.11) and the fact that \(ped(n)=M_{ped}(0,2,n)+M_{ped}(1,2,n)\). \(\square \)
Proof of Theorem 1.8
Substituting \(z=e^\frac{\pi i}{2}=i\) into (3.1), by (3.2) and Lemmas 4.2–4.3, we find that
By (3.2), one can see that
In light of (5.20) and the fact that \(M_{ped}(0,n)\) has the same parity as \(M_{ped}(0,4,n)\), Theorem 1.8 holds. \(\square \)
We next aim to prove Theorem 1.9.
Proof of Theorem 1.9
Substituting \(z=e^{\frac{2\pi i}{3}}\) into (3.1), we obtain
Using Lemmas 4.3, 4.6, by (3.2) and the fact that \(1+e^{\frac{2\pi i}{3}}+e^{\frac{4\pi i}{3}}=0\), we get
Extracting those terms associated with powers \(q^{3n+1}\) on both sides of the above equation, then dividing by q and replacing \(q^3\) by q, one can see that
Since
using Lemma 4.4,
after simplification.
Clearly, the coefficient of \(q^{3n+2}\) in (5.23) is zero. We can conclude that
Combining (5.11), (5.18), (5.24) and Theorem 4.8, we complete the proof of Theorem 1.9. \(\square \)
Remark 1
Andrews, Hirschhorn, and Sellers [3] presented an interesting infinite family of congruences modulo 3 as given by
and deduced that
Actually, based on a substantial amount of numerical evidence, we conjecture that the ped-crank can be used to provide a combinatorial interpretation of (5.25), namely
Here, we only prove the case for \(\alpha =1\), and for any \(\alpha >1\), we are not able to provide an elementary proof of this conjecture.
Proof
Extracting the terms involving \(q^{3n}\) in (5.23) and substituting \(q^3\) by q, we obtain
From Lemmas 4.4, 4.5, 4.7, considering the terms involving \(q^{3n+2}\) in (5.27) leads to
That means
\(\square \)
Unfortunately, the ped-crank cannot be employed to interpret (5.26) even for \(\alpha =1\). Hence, it will be interesting to introduce another partition statistic that could combinatorially interpret (5.26).
Remark 2
For ordinary partitions, recall that we define \(M(0,1)=-1,\ \ M(-1,1)=M(1,1)=1\). From the definition of ped-crank, one can see that a similar problem will arise when \(\Delta (\lambda )=((2), \beta , \gamma )\). So we make the following adjustment to the definition of ped-crank. Let
Definition 5.1
Let \(\lambda \) be a partition of n with even parts distinct. The \(c_{mped}(\lambda )\) is given by
where \(c_{ped}(\lambda )\) is the ped-crank of \(\lambda \).
When \(\lambda \in A_n\), an injection from \(A_n\) to \(B_n\) can be constructed by changing \(\alpha \) to \(\emptyset \) and adding 2 to the largest part of \(\gamma \). Another direction is obvious. Hence for any non-negative integer n, there is a bijection between \(A_n\) and \(B_n\). Let \(M_{mped}(m,n)\) denote the number of partitions of n with even parts distinct with \(c_{mped}(\lambda )=m\). By the definition of \(c_{mped}(\lambda )\), one can check \(M_{mped}(m,n)=M_{ped}(m,n)\) for any integer m and non-negative integer n.
For example, if \(\lambda =(6,5,4,1,1,1),\Delta (\lambda )=((2),(5,1),(6,4))\), then \(c_{mped}(\lambda )=1\) and if \(\lambda =(8,5,4,1),\Delta (\lambda )=((0),(5,1),(8,4))\), then \(c_{mped}(\lambda )=-1\).
Table 1 gives the 12 partitions of 7 with even parts distinct. It is easy to check that these partitions are divided into six equinumerous subsets by ped-crank. Moreover,
6 Closing remarks
In 2014, Xia [12] proved the following congruence modulo 4 for ped(n).
Theorem 6.1
[12, Equation (9), Theorem 1] For \(\alpha ,n\ge 0\),
Note that comparing (5.4) with (5.13), a simple deduction gives
Thus
Since for all \(\alpha >0,n\ge 0\), \(3^{2\alpha }n+\frac{3^{2\alpha }-1}{8}\equiv 1\pmod {3}\), a refinement of Theorem 6.1 can be given as
References
Andrews, G.E.: Generalized Frobenius partitions. Mem. Amer. Math. Soc. 49, 301 (1984)
Andrews, G.E., Garvan, F.G.: Dyson’s crank of a partition. Bull. Amer. Math. Soc. 18(2), 167–171 (1988)
Andrews, G.E., Hirschhorn, M.D., Sellers, J.A.: Arithmetic properties of partitions with even parts distinct. Ramanujan J. 23(1), 169–181 (2010)
Berndt, B.C.: Ramanujan’s Notebooks Part III. Springer-Verlag, New York (1991)
Chern, S., Hao, L.J.: Congruences for partition functions related to mock theta functions. Ramanujan J. 48(2), 369–384 (2019)
Garvan, F.G.: New combinatorial interpretations of Ramanujan’s partition congruences mod \(5\), \(7\), \(11\). Trans. Am. Math. Soc. 305(1), 47–77 (1988)
Garvan, F.G.: The crank of partitions mod 8, 9 and 10. Trans. Amer. Math. Soc. 322(1), 79–94 (1990)
Hirschhorn, M.D., Sellers, J.A.: A congruence modulo \(3\) for partitions into distinct non-multiples of four. J. Integer Seq. 17(9), 14–19 (2014)
Merca, M.: New relations for the number of partitions with distinct even parts. J. Number Theory. 176, 1–12 (2017)
Seo, S., Yee, A.J.: Overpartitions and singular overpartitions. Analytic Number Theory, Modular Forms and \(q\)-Hypergeometric Series: In Honor of Krishna Alladi’s 60th Birthday. University of Florida, Gainesville, 693–711 (2016)
Sloane, N.J.A.: The on-line encyclopedia of integer sequences, published electronically at http://oeis.org
Xia, E.X.W.: New infinite families of congruences modulo 8 for partitions with even parts distinct. Electron J. Combin. 21(4), P4–P8 (2014)
Yao, O.X.M., Xia, E.X.W.: New Ramanujan-like congruences modulo powers of 2 and 3 for overpartitions. J. Number Theory. 133(6), 1932–1949 (2013)
Yee, A.J.: Combinatorial proofs of generating function identities for F-partitions. J. Combin. Theory A. 102(1), 217–228 (2003)
Acknowledgements
The author would like to thank the anonymous referee for making valuable comments and suggestions that significantly improved the quality of this paper. This work was supported by the National Natural Science Foundation of China (No. 12101307).
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by Ilse Fischer.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Hao, R.X.J. A partition statistic for partitions with even parts distinct. Monatsh Math 201, 1105–1123 (2023). https://doi.org/10.1007/s00605-022-01816-9
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00605-022-01816-9