1 Introduction
The arithmetic properties of partition functions have been a central theme in the theory of partitions since the pioneering work of Ramanujan [Reference Ramanujan6]. His celebrated congruences for the ordinary partition function (see below) initiated a vast literature devoted to the congruence properties of both the ordinary partition function and numerous families of restricted partition functions. More recently, many new partition functions arising from Rogers–Ramanujan-type identities and related q-series have also been shown to satisfy remarkable arithmetic properties.
We recall that a partition of a positive integer n is a nonincreasing sequence of positive integers
$\lambda _1, \lambda _2, \ldots , \lambda _r$
such that
$n=\lambda _1+ \lambda _2+ \cdots + \lambda _r$
. Each
$\lambda _i$
,
$1\leq i\leq r$
, is called a part of the partition. We denote the number of partitions of n by
$p(n)$
, the ordinary partition function.
In the late 1910s, Ramanujan [Reference Ramanujan6] proved the following congruences satisfied by
$p(n)$
for the moduli
$5, 7$
and
$11$
using generating function manipulations: for all
$n\geq 0$
,
These congruences, and others that have since been proven for numerous restricted partition functions, serve as the motivation for this work.
Recently, Dombos [Reference Dombos4] introduced a new Rogers–Ramanujan-type identity involving grounded partitions. One family of partitions arising naturally from this identity consists of partitions whose parts are either divisible by 4 or congruent to
$\pm 1 \pmod {6}$
. We refer the reader to [Reference Dombos4, Theorem 1.6] for more details. Let
$\mathit {dp}(n)$
denote the number of such partitions of n. In light of this definition, the generating function for
$\mathit {dp}(n)$
is given by
where
$(a;q)_\infty := (1-a)(1-aq)(1-aq^{2})\cdots $
is the usual q-Pochhammer symbol.
Since the generating function of
$\mathit {dp}(n)$
has a simple infinite product representation, it is natural to investigate whether
$\mathit {dp}(n)$
satisfies Ramanujan-type congruences analogous to those enjoyed by many other restricted partition functions. To the best of our knowledge, no arithmetic properties of
$\mathit {dp}(n)$
have previously been established.
The purpose of this paper is to initiate the study of the arithmetic properties of the partition function
$\mathit {dp}(n)$
by establishing several infinite families of Ramanujan-type congruences. Our proofs rely entirely on elementary generating function manipulations together with classical 2- and 3-dissection identities, and several well-known q-series identities, avoiding the use of modular forms. In addition to congruences modulo powers of 2 and modulo 3, we obtain an infinite family of congruences modulo 4 indexed by primes in two residue classes. These results are summarised in the following theorems.
Theorem 1.1. For all
$n \geq 0$
,
Theorem 1.2. Let p be a prime such that
$p \equiv 13, 17, 19 \text { or } 23 \pmod {24}$
. Then, for all
$k,m \geq 0$
with
$p \nmid m$
,
Theorem 1.3. For all
$n \geq 0$
,
Theorem 1.4. For all
$n \geq 0$
,
Theorem 1.5. For all
$\alpha \geq 1$
and all
$n \geq 0$
,
2 Preliminaries
Throughout the remainder of this paper, we define
to shorten the notation. We begin by rewriting the generating function (1.1) in terms of the
$f_k$
.
Lemma 2.1. We have
Proof. Initially, we note that
Thus, it follows from (1.1) that
which is (2.1).
The proofs of the above-mentioned main results require a few well-known 2- and 3-dissections.
Lemma 2.2 [Reference da Silva and Sellers3, Lemma 1].
We have
Lemma 2.3 [Reference Toh7, (2.1c)].
We have
Lemma 2.4 [Reference Toh7, (2.1b)].
We have
Lemma 2.5 [Reference Hirschhorn5, (14.3.2)].
We have
Lemma 2.6 [Reference Toh7, (3.1)].
We have
We recall Ramanujan’s theta functions
With the above definitions, we can state the following 3-dissection of
$1/\psi (q)$
.
Lemma 2.7 ([Reference Chern and Hao2, Lemma 2.2]).
We have
Lastly, we require the classical identity of Jacobi (see [Reference Berndt1, Theorem 1.3.9]):
3 Proof of Theorem 1.1
Let
Thus,
$h(n) = (-1)^n\mathit {dp}(n)$
for all n. We note that
Using this identity and (2.1),
Thanks to (2.8),
which, after dividing by q and replacing
$q^3$
by q, yields
Using (2.2), we can 2-dissect the expression above to obtain
Thus,
which proves (1.2) since
$\mathit {dp}(6n+4) = h(6n+4) \equiv 0 \pmod {2}$
.
Working modulo 8,
Thanks to (2.7),
which simplifies to
It follows that
$h(18n+10) \equiv 0 \pmod {4}$
, which completes the proof of (1.3) since
${h(n) = (-1)^n \mathit {dp}(n)}$
.
To prove (1.4), we use (2.7) to extract the terms of the form
$q^{3n+2}$
from (3.1). The resulting congruence is
by (2.9). We note that
$f_6^5/f_{12}$
is a function of
$q^3$
and
$n(n+1)/2$
is never congruent to 2 modulo 3. Hence, once we expand (3.2) as a power series in q, there will be no powers of the form
$q^{3n+2}$
. Therefore, for all
$n \geq 0$
,
which completes the proof of (1.4).
4 Proof of Theorem 1.2
Initially, we use (2.4) to extract the terms of the form
$q^{3n+1}$
from (2.1), which gives
After dividing by q and replacing
$q^3$
by q, we are left with
Now, we use (2.3) to extract the odd parts of (4.1), which gives
Using (2.9), it follows from (4.2) that
Thus,
It follows that
$\mathit {dp}(6n+4) \equiv 0 \pmod {4}$
if
$24n+15$
is not of the form
$6x^2+y^2$
.
Since
$p \equiv 13, 17, 19 \text { or } 23 \pmod {24}$
, it follows that
$ ({-6}/{p}) = -1$
, which implies that
$\nu _p(N)$
is even if N is of the form
$6x^2+y^2$
. Taking
$n = p^{2k+1}m + 5({p^{2k+2}-1})/{8}$
, we have
Therefore,
$\nu _p(24n+15)$
is odd and the result follows by the fact that
5 Proofs of Theorems 1.3–1.5
We now transition to proving Theorems 1.3–1.5 where the focus is on arithmetic properties modulo 3. Let us begin by proving (1.5). It follows from (4.1) that
Thanks to (2.5),
Dividing by q and replacing
$q^3$
by q yields
Finally, we use (2.6) to get
from which (1.5) follows.
We next consider the ‘internal congruence’ given in Theorem 1.4. It follows from (5.1) that
Replacing
$q^3$
by q, we are left with
Thus, it follows by (4.1) that
which completes the proof.
Lastly, we use the previous two theorems to prove Theorem 1.5. We prove this theorem by induction on
$\alpha $
. Note that the case
$\alpha =1$
has already been established earlier in Theorem 1.3. We now use it as our base case in a proof by induction. Thus, we assume that, for some
$\alpha \geq 1$
and all
$n\geq 0$
,
We then want to prove that
We note that
by the induction hypothesis. Finally, by Theorem 1.4,
which concludes the inductive proof.






