betaFeedback

WorksMathematics

Excedances and Deficiencies of Integer Partitions

Abstract

The notion of a fixed point, applied by Blecher and Knopfmacher to integer partitions, can be considered as a permutation statistic. Here we apply two related permutation statistics, excedance and deficiency, to integer partitions. Using analytic and combinatorial techniques, we establish several generating functions and make connections to partition concepts including Dyson's crank statistic, overpartitions, and the Rogers–Ramanujan partitions.

Introduction and background

A partition of a positive integer nn is a weakly decreasing sequence of positive integers whose sum is nn. The positive integers in the sequence are called parts. Let P(n)P(n) be the set of partitions of nn and write p(n)=∣P(n)∣p(n)=|P(n)|. By convention, p(0)=1p(0)=1.

The Durfee square of a partition λ=(λ1,…,λt)\lambda=(\lambda_1,\ldots,\lambda_t) has dimension dd where dd is the greatest index for which λd≥d\lambda_d\ge d. This statistic gives a combinatorial understanding of Jacobi’s result

∑n≥0p(n)qn=1(q;q)∞=∑d≥0qd2(q;q)d2(1)\sum_{n\ge0}p(n)q^n=\frac{1}{(q;q)_\infty}=\sum_{d\ge0}\frac{q^{d^2}}{(q;q)_d^2} \tag*{(1)}

where we use standard qq-series notation; see, for example, Chapter 10 of Andrews, Askey, and Roy [1]. As shown in Figure 1, a partition can be decomposed into its d×dd\times d Durfee square, the conjugate α′\alpha' of a partition α\alpha with largest part at most dd to the right of the Durfee square, and a partition β\beta with largest part at most dd below the Durfee square.

Figure 1. The Durfee square decomposition of a partition.

We will also use the 2ϕ1{}_{2}\phi_{1} series studied by Heine, in particular the identity [1 Equation 10.10.1]:

2ϕ1(a,b;q,xc)=(c/b;q)∞(bx;q)∞(c;q)∞(x;q)∞2ϕ1(abx/c,b;q,c/b).(2){}_{2}\phi_{1}\left(\begin{matrix}a,b;q,x\\c\end{matrix}\right)=\frac{(c/b;q)_{\infty}(bx;q)_{\infty}}{(c;q)_{\infty}(x;q)_{\infty}}{}_{2}\phi_{1}\left(\begin{matrix}abx/c,b;q,c/b\end{matrix}\right). \tag*{(2)}

The crank of a partition, named by Dyson in 1944 but not found until 1988 work of Andrews and Garvan [3], is arguably the most important partition statistic. We will not need its definition, but will use the following generating function for the number of partitions of nn with positive crank. This was established in both Andrews–Newman [4 Theorem 1] and Hopkins–Sellers [8 proof of Theorem 1], 2020 works that independently established a connection between the crank and another partition statistic called the mex.

Proposition 1. A generating function for partitions with non-negative crank is

1(q;q)∞∑n≥0(−1)nq(n+12).\frac{1}{(q;q)_{\infty}}\sum_{n\geq0}(-1)^{n}q^{\binom{n+1}{2}}.

In 2022, Blecher and Knopfmacher applied the pervasive mathematical notion of fixed points to partitions as follows [5].

Definition 2. A partition λ=(λ1,…,λt)\lambda=(\lambda_{1},\ldots,\lambda_{t}) has a fixed point if λi=i\lambda_{i}=i for some index ii.

Note that, due to the ordering convention, a partition has at most one fixed point. Also, if λi=i\lambda_{i}=i, then ii is the size of the Durfee square.

In resolving a conjecture of Blecher and Knopfmacher, Hopkins and Sellers established the following connections between fixed points and the crank [9].

Proposition 3. Given n≥1n\geq1, the number of partitions of nn with a fixed point equals the number of partitions of nn with positive crank.

To refine the notion of fixed point, let f(n,d)f(n,d) be the number of partitions λ\lambda of nn that have a fixed point dd, i.e., λd=d\lambda_{d}=d. Hopkins established the following recurrence [7].

Proposition 4. For d≥2d\geq2 and n≥d2n\geq d^{2},

f(n,d)=f(n−1+1,d)+f(n−d,d)−f(n−2d+1,d)+f(n−2d+1,d−1).f(n,d)=f(n-1+1,d)+f(n-d,d)-f(n-2d+1,d)+f(n-2d+1,d-1).

Considering fixed points in the context of permutations, partitions without a fixed point correspond to derangements. This setting invites the application of other permutation statistics to integer partitions. In this work, we consider two related permutation statistics on partitions.

Definition 5. An excedance of a partition λ∈P(n)\lambda\in P(n) is an index ii such that λi>i\lambda_{i}>i while a deficiency is an index ii such that λi<i\lambda_{i}<i. For k≥0k\geq0, let exc⁡(n,k)\operatorname{exc}(n,k) be the number of partitions of nn with kk excedances, similarly dfc⁡(n,k)\operatorname{dfc}(n,k) for deficiencies.

The terminology dates at least to Reilly 1979 [11]. Note that every part of a partition is an excedance, fixed point, or deficiency. Further, any excedances are the initial parts of a partition, followed by a possible fixed point, with any deficiencies comprising the subsequent parts.

The next two sections give results on excedances and deficiencies, respectively, using both generating function and combinatorial techniques.

Excedances of integer partitions

The array of Figure 2 gives the number of partitions of nn with kk excedances for small values of nn and kk.

Array of exc(n, k) values
n\kn\backslash k01234
01
11
211
312
414
516
6191
71122
81165
91209
1012516
1113025
12136391
13142562
14149805
1515610910
1616414719
1717219232
1818124954
1919031584
2011003961291

Figure 2. Array of exc⁡(n,k)\operatorname{exc}(n,k) values for 0≤n≤200\leq n\leq20 and 0≤k≤40\leq k\leq4.

Theorem 6. A bivariate generating function for the number of partitions of nn with kk excedances is

∑n,kexc⁡(n,k)wkqn=1+∑d≥1wd−1qd2(1−qd+wqd)(q;q)d2.(3)\sum_{n,k} \operatorname{exc}(n,k)w^kq^n=1+\sum_{d\geq1}\frac{w^{d-1}q^{d^2}(1-q^d+wq^d)}{(q;q)_d^2}. \tag*{(3)}
=1(q;q)∞∑n≥0(wq;q)n(−1)nq(n+12)(1+qn+1)(q;q)n.(4)=\frac{1}{(q;q)_\infty}\sum_{n\geq0}\frac{(wq;q)_n(-1)^nq^{\binom{n+1}{2}}(1+q^{n+1})}{(q;q)_n}. \tag*{(4)}

Proof. A partition λ\lambda with Durfee square size dd has d−1d-1 excedances if λd=d\lambda_d=d, that is, referencing Figure 1, if dd is not a part of the subpartition α′\alpha' to the right of the Durfee square. Otherwise, λ\lambda has dd excedances and dd is a part of α′\alpha' (this case includes the empty partition of 0). A generating function for exc⁡(n,k)\operatorname{exc}(n,k) is then

∑d≥1wd−1qd2(q;q)d−1(q;q)d+∑d≥0wdqd2+d(q;q)d2.(5)\sum_{d\geq1}\frac{w^{d-1}q^{d^2}}{(q;q)_{d-1}(q;q)_d}+\sum_{d\geq0}\frac{w^dq^{d^2+d}}{(q;q)_d^2}. \tag*{(5)}

which simplifies to the desired expression (3).

Working from each term of (5),

∑d≥1wd−1qd2(q;q)d−1(q;q)d=∑d≥0wdqd2+2d+1(q;q)d(q;q)d+1=q1−qlim⁡τ→0ϕ1(q/τ,q/τ;q,τ2qwq2)=q1−q1(q2;q)∞lim⁡τ→0ϕ1(wq,q/τ;q,τqτq2w)=1(q;q)∞∑n≥0(wq;q)n(−1)nq(n+22)(q;q)n,\begin{aligned} \sum_{d\geq1}\frac{w^{d-1}q^{d^2}}{(q;q)_{d-1}(q;q)_d} &=\sum_{d\geq0}\frac{w^dq^{d^2+2d+1}}{(q;q)_d(q;q)_{d+1}}\\ &=\frac{q}{1-q}\lim_{\tau\to0}\phi_1\left(\begin{matrix}q/\tau,q/\tau;q,\tau^2qw\\q^2\end{matrix}\right)\\ &=\frac{q}{1-q}\frac{1}{(q^2;q)_\infty}\lim_{\tau\to0}\phi_1\left(\begin{matrix}wq,q/\tau;q,\tau q\\\tau q^2w\end{matrix}\right)\\ &=\frac{1}{(q;q)_\infty}\sum_{n\geq0}\frac{(wq;q)_n(-1)^nq^{\binom{n+2}{2}}}{(q;q)_n}, \end{aligned}

where we used (2) and, similarly,

∑d≥0wdqd2+d(q;q)d2=lim⁡τ→0ϕ1(q/τ,q/τ;q,τ2wq)=1(q;q)∞lim⁡τ→0ϕ1(wq,q/τ;q,ττqw)=1(q;q)∞∑n≥0(wq;q)n(−1)nq(n+12)(q;q)n,\begin{aligned} \sum_{d\geq0}\frac{w^dq^{d^2+d}}{(q;q)_d^2} &=\lim_{\tau\to0}\phi_1\left(\begin{matrix}q/\tau,q/\tau;q,\tau^2w\\q\end{matrix}\right)\\ &=\frac{1}{(q;q)_\infty}\lim_{\tau\to0}\phi_1\left(\begin{matrix}wq,q/\tau;q,\tau\\\tau qw\end{matrix}\right)\\ &=\frac{1}{(q;q)_\infty}\sum_{n\geq0}\frac{(wq;q)_n(-1)^nq^{\binom{n+1}{2}}}{(q;q)_n}, \end{aligned}

which together give (4). □

The results of Theorem 6 allow us to explain many patterns in the exc⁡(n,k)\operatorname{exc}(n,k) array of Table 2.

diag(Exc, 8)F(9)F(9)
Exc(8, 0)(1, 1, 1, 1, 1, 1, 1, 1)(1, 1, 1, 1, 1, 1, 1, 1, 1)F(9,1)F(9, 1)
Exc(7, 1)(7)(7, 2)F(9,2)F(9, 2)
(6, 1)(6, 2, 1)
(5, 2)(5, 2, 2)
(5, 1, 1)(5, 2, 1, 1)
(4, 2, 1)(4, 2, 2, 1)
(4, 1, 1, 1)(4, 2, 1, 1, 1)
(3, 2, 2)(3, 2, 2, 2)
(3, 2, 1, 1)(3, 2, 2, 1, 1)
(3, 1, 1, 1, 1)(3, 2, 1, 1, 1, 1)
(2, 2, 2, 1)(2, 2, 2, 2, 1)
(2, 2, 1, 1, 1)(2, 2, 2, 1, 1, 1)
(2, 1, 1, 1, 1, 1)(2, 2, 1, 1, 1, 1)
Exc(6, 2)(3, 3)(3, 3, 3)F(9,3)F(9, 3)

Table 2. The n=8n=8 example of the bijection in the second proof of Corollary 10.

Clearly the row sums of the exc⁡(n,k)\operatorname{exc}(n,k) array match p(n)p(n), the number of partitions of nn. Analytically, letting w=1w=1 in (3) gives Jacobi’s (1).

Corollary 7. A generating function for the number of partitions of nn with an even number of excedances is

12(q;q)∞∑n≥0(1−(−q;q)n(q;q)n)(−1)nq(n+12)(1+qn+1)\frac{1}{2(q;q)_\infty}\sum_{n\geq0}\left(1-\frac{(-q;q)_n}{(q;q)_n}\right)(-1)^nq^{\binom{n+1}{2}}(1+q^{n+1})

and an analogue for odd number of excedances is

12(q;q)∞∑n≥0(1+(−q;q)n(q;q)n)(−1)nq(n+12)(1+qn+1)\frac{1}{2(q;q)_\infty}\sum_{n\geq0}\left(1+\frac{(-q;q)_n}{(q;q)_n}\right)(-1)^nq^{\binom{n+1}{2}}(1+q^{n+1})

giving an expression for the difference

1(q;q)∞∑n≥0(−q;q)n(−1)nq(n+12)(1+qn+1)(q;q)n.\frac{1}{(q;q)_\infty}\sum_{n\geq0}\frac{(-q;q)_n(-1)^nq^{\binom{n+1}{2}}(1+q^{n+1})}{(q;q)_n}.

Proof. The excedance parity results follow from standard techniques applied to (4) and setting w=1w=1. The difference expression comes most easily from setting w=−1w=-1 in (4). □

Table 1 provides initial values for these counts. Computations suggest that the difference remains positive for n≥12n\geq12. At the time of writing, these sequences were not contained in the On-Line Encyclopedia of Integer Sequences [10].

nn123456789101112131415
even11111236101726405781110
odd012469121620253037445466
diff.10−1-1−3-3−5-5−7-7−9-9−10-10−10-10−8-8−4-43132744

Table 1. The number of partitions of nn with an even number of excedances, an odd number, and the difference, through n=15n = 15.

The weighted row sum exc⁡(n)=∑kkexc⁡(n,k)\operatorname{exc}(n)=\sum_k k\operatorname{exc}(n,k) gives the total number of excedances in the partitions of nn.

Corollary 8. A generating function for the total number of excedances in the partitions of nn is

1(q;q)∞∑n≥1(−1)n−1qn(n+3)/21−qn.\frac{1}{(q;q)_\infty}\sum_{n\geq1}\frac{(-1)^{n-1}q^{n(n+3)/2}}{1-q^n}.

Proof. The result follows from determining the partial derivative of (4) with respect to ww and evaluating that expression at w=1w=1. □

Values for exc⁡(n)\operatorname{exc}(n) for small nn are given in Table 5 in the next section.

nn123456789101112131415
exc⁡(n)\operatorname{exc}(n)01246111626385780117160224304
dfc⁡(n)\operatorname{dfc}(n)01361119315076116169247349494682

Table 5. The total number of excedances and deficiencies in P(n)P(n) for small nn.

Both the diagonal sums

diag⁡(exc⁡,n)=exc⁡(n,0)+exc⁡(n−1,1)+⋯\operatorname{diag}(\operatorname{exc},n)=\operatorname{exc}(n,0)+\operatorname{exc}(n-1,1)+\cdots

and antidiagonal sums

adiag⁡(exc⁡,n)=exc⁡(n,0)+exc⁡(n+1,1)+⋯\operatorname{adiag}(\operatorname{exc},n)=\operatorname{exc}(n,0)+\operatorname{exc}(n+1,1)+\cdots

(finite due to the shape of the array) have combinatorial interpretations.

Corollary 9. For n≥1n\geq1, the diagonal sum diag⁡(exc⁡,n)\operatorname{diag}(\operatorname{exc},n) equals the number of partitions of n+1n+1 with positive crank.

Proof. Letting w=qw=q in (4) gives the diagonal sums of the exc⁡(n,k)\operatorname{exc}(n,k) array. The resulting expression reduces to

1(q;q)∞∑n≥1(−1)n+1q(n+1)/2.\frac{1}{(q;q)_\infty}\sum_{n\geq1}(-1)^{n+1}q^{(n+1)/2}.

which is the complement of the shifted generating function of Proposition 1, i.e., the expression here is a generating function for the number of partitions of n+1n+1 with positive crank, using crank symmetry [3]. □

Corollary 10. The antidiagonal sum adiag⁡(exc⁡,n)\operatorname{adiag}(\operatorname{exc},n) equals p(n)+p(n−1)p(n)+p(n-1).

Proof. Letting w=1/qw=1/q in (4) gives the antidiagonal sums of the exc⁡(n,k)\operatorname{exc}(n,k) array. The resulting term (1;q)n(1;q)_n in the numerator is nonzero only for n=0n=0 so that the expression reduces to

1+q(q;q)∞,\frac{1+q}{(q;q)_\infty},

a generating function for p(n)+p(n−1)p(n)+p(n-1). □

We also give combinatorial proofs for Corollaries 9 and 10. Let Exc⁡(n,k)\operatorname{Exc}(n,k) be the set of partitions of nn with kk excedances. Write F(n,d)F(n,d) for the set of partitions λ∈P(n)\lambda\in P(n) with a fixed point λd=d\lambda_d=d and F(n)F(n) for the set of partitions in P(n)P(n) with a fixed point [7].

Theorem 11. There is a bijection

Exc⁡(n,k)≅F(n+k+1,k+1).\operatorname{Exc}(n,k) \cong F(n+k+1,k+1).

Proof. Given λ∈Exc⁡(n,k)\lambda\in\operatorname{Exc}(n,k), the kk excedances are necessarily the first kk parts. Note that λk≥k+1\lambda_k \ge k+1 and λk+1≤k+1\lambda_{k+1} \le k+1 (else there are at least k+1k+1 excedances). Assign

λ↦μ=(λ1,…,λk,k+1,λk+1,…),\lambda\mapsto\mu=(\lambda_1,\ldots,\lambda_k,k+1,\lambda_{k+1},\ldots),

a partition of n+k+1n+k+1 with fixed point μk+1=k+1\mu_{k+1}=k+1 so that μ∈F(n+k+1,k+1)\mu\in F(n+k+1,k+1). The reverse map is clear: remove the fixed point from μ∈F(n+k+1,k+1)\mu\in F(n+k+1,k+1) to get λ∈Exc⁡(n,k)\lambda\in\operatorname{Exc}(n,k). □

Examples of the bijection of Theorem 11 are given in Table 2.

Combinatorial proof of Corollary 9. By Theorem 11, the diagonal sum

diag⁡(exc⁡,n)=∑i≥0exc⁡(n−i,i)=∑i≥0f(n+1,i+1)=∑j≥1f(n+1,j),\begin{aligned} \operatorname{diag}(\operatorname{exc},n)&=\sum_{i\geq0}\operatorname{exc}(n-i,i)\\ &=\sum_{i\geq0}f(n+1,i+1)\\ &=\sum_{j\geq1}f(n+1,j), \end{aligned}

the number of partitions of n+1n+1 with a fixed point. By Proposition 3, this matches the number of partitions of n+1n+1 with positive crank. □

Table 2 gives an example of the bijection in this combinatorial proof of Corollary 9 which uses three applications of the bijection of Theorem 11.

Theorem 11 also translates Proposition 4 into a recurrence for the exc⁡(n,k)\operatorname{exc}(n,k) array.

Corollary 12. For k≥1k \geq1 and n≥k2+kn \geq k^2+k,

exc⁡(n,k)=exc⁡(n−k,k)+exc⁡(n−k−1,k)−exc⁡(n−2k−1,k)+exc⁡(n−2k,k−1).\operatorname{exc}(n,k)=\operatorname{exc}(n-k,k)+\operatorname{exc}(n-k-1,k)-\operatorname{exc}(n-2k-1,k)+\operatorname{exc}(n-2k,k-1).

For the next combinatorial proof, note that λ∈Exc⁡(n,d)\lambda\in\operatorname{Exc}(n,d) either has Durfee square size dd with no fixed point, i.e., λd≥d+1\lambda_d \ge d+1, or Durfee square size d+1d+1 with fixed point λd+1=d+1\lambda_{d+1}=d+1.

Combinatorial proof of Corollary 10. We establish a bijection

⋃i≥0Exc⁡(n+i,k+i)≅P(n)∪P(n−1).\bigcup_{i\ge0}\operatorname{Exc}(n+i,k+i) \cong P(n)\cup P(n-1).

For λ∈Exc⁡(n,d)\lambda\in\operatorname{Exc}(n,d) with Durfee square size dd, we know λd≥d+1\lambda_d \ge d+1 and λd+1≤d\lambda_{d+1}\le d so that reducing each excedance by one gives a valid partition

(λ1−1,…,λd−1,λd+1,…)∈P(n−d).(\lambda_1-1,\ldots,\lambda_d-1,\lambda_{d+1},\ldots)\in P(n-d).

For λ∈Exc⁡(n,d)\lambda\in\operatorname{Exc}(n,d) with Durfee square size d+1d+1, we know λd+1=d+1\lambda_{d+1}=d+1 and removing that fixed point leaves

(λ1,…,λd,λd+2,…)∈P(n−d−1).(\lambda_1,\ldots,\lambda_d,\lambda_{d+2},\ldots)\in P(n-d-1).

Applying the map to ⋃Exc⁡(n+i,k+i)\bigcup\operatorname{Exc}(n+i,k+i), the first case gives partitions in P(n)P(n) and the second case gives partitions in P(n−1)P(n-1).

For the reverse map, given μ∈P(n)\mu\in P(n) with Durfee square size dd, increase the first dd parts by one to make λ∈P(n+d)\lambda\in P(n+d) with dd excedances, Durfee square size dd, and no fixed point. Given ν∈P(n−1)\nu\in P(n-1), find the greatest ee such that νe−1≥e\nu_{e-1}\ge e and add a part ee to make a partition in P(n+e−1)P(n+e-1) with fixed point ee and e−1e-1 excedances. The bijection establishes the identity adiag⁡(exc⁡,n)=p(n)+p(n−1)\operatorname{adiag}(\operatorname{exc},n)=p(n)+p(n-1). □

Table 3 gives an example of the bijection in this combinatorial proof of Corollary 10.

adiag⁡(Exc⁡,5)\operatorname{adiag}(\operatorname{Exc},5)P(5)P(5)∪P(4)\cup P(4)
Exc⁡(5,0)\operatorname{Exc}(5,0)(1,1,1,1,1)(1,1,1,1,1)(1,1,1,1)(1,1,1,1)
Exc⁡(6,1)\operatorname{Exc}(6,1)(6)(6)(5)(5)
(5,1)(5,1)(4,1)(4,1)
(4,2)(4,2)(4)(4)
(4,1,1)(4,1,1)(3,1,1)(3,1,1)
(3,2,1)(3,2,1)(3,1)(3,1)
(3,1,1,1)(3,1,1,1)(2,1,1,1)(2,1,1,1)
(2,2,2)(2,2,2)(2,2)(2,2)
(2,2,1,1)(2,2,1,1)(2,1,1)(2,1,1)
(2,1,1,1)(2,1,1,1)(1,1,1,1,1)(1,1,1,1,1)
Exc⁡(7,2)\operatorname{Exc}(7,2)(4,3)(4,3)(3,2)(3,2)
(3,3,1)(3,3,1)(2,2,1)(2,2,1)

Table 3. The n=5n=5 example of the bijection in the second proof of Corollary 10.

Note that, by adding an overlined part 1 to the partitions of P(n−1)P(n-1), Corollary 10 can be cast as a bijection to the restricted overpartitions of nn [6] where only a part 1 may be overlined.

Deficiencies of integer partitions

For permutations, excedances and deficiencies are essentially equivalent, since each excedance of a permutation π\pi is a deficiency of the inverse permutation π−1\pi^{-1} and vice versa. That means that the number of excedances over all permutation of SnS_n equals the number of deficiencies over SnS_n. The situation for partitions is very different. As explained in the proof of Theorem 6, a partition of nn with Durfee square size dd has d−1d-1 or dd excedances while the number of deficiencies, the number of parts in β\beta below the Durfee square as in Figure 1, could be as high as n−d2n-d^2. See Corollary 16 and Table 5 for more details.

The array of Figure 3 gives the number of partitions of nn with kk deficiencies for small values of nn and kk.

Theorem 13. A bivariate generating function for the number of partitions of nn with kk deficiencies is

∑n,kdfc⁡(n,k)wkqn=∑d≥0qd2(q;q)d(wq;q)d(6)\sum_{n,k} \operatorname{dfc}(n,k)w^kq^n=\sum_{d\geq0}\frac{q^{d^2}}{(q;q)_d(wq;q)_d} \tag*{(6)}
=1(wq;q)∞∑n≥0(w−1;q)n(−1)nwnqn(n+1)2(q;q)n.(7)=\frac{1}{(wq;q)_\infty}\sum_{n\geq0}\frac{(w^{-1};q)_n(-1)^nw^nq^{\frac{n(n+1)}{2}}}{(q;q)_n}. \tag*{(7)}

Proof. The number of deficiencies of a partition λ\lambda is exactly the number of parts of the subpartition β\beta underneath the Durfee square, which establishes (6).

n\kn\backslash k01234567891011121314
11
211
3111
42111
522111
6332111
73432111
845532111
9566532111
106887532111
117101097532111
129131312107532111
131016171513107532111
14122022201714107532111
1514242826221814107532111

Table 3 (PDF p. 8). Array of dfc⁡(n,k)\operatorname{dfc}(n,k) values for 0≤n,k≤150\leq n,k\leq15.

Similar to the derivations of Theorem 6, we use (2) here:

∑d≥0qd2(q;q)d(wq;q)d=lim⁡τ→0Φ1(qτ,1τ;q,τ2;wq)=lim⁡τ→0(wτ;q)∞(τ;q)∞(wq;q)∞(τ2;q)∞Φ1(1w,1τ;q,wqτ;τ)=1(wq;q)∞∑n≥0(w−1;q)n(−1)nwnq(n+12)(q;q)n,\begin{aligned} \sum_{d\ge0}\frac{q^{d^2}}{(q;q)_d(wq;q)_d} =\lim_{\tau\to0}\Phi_1\left(\frac{q}{\tau},\frac{1}{\tau};q,\tau^2\mathrel{;}wq\right) \\ &=\lim_{\tau\to0}\frac{(w\tau;q)_\infty(\tau;q)_\infty}{(wq;q)_\infty(\tau^2;q)_\infty} \Phi_1\left(\frac{1}{w},\frac{1}{\tau};q,wq\tau\mathrel{;}\tau\right) \\ &=\frac{1}{(wq;q)_\infty}\sum_{n\ge0}\frac{(w^{-1};q)_n(-1)^n w^n q^{\binom{n+1}{2}}}{(q;q)_n}, \end{aligned}

as desired. □

It is interesting to contrast the rather complicated expressions (4) and (7) to

∑d≥0wdqd2(q;q)d(wq;q)d=1(wq;q)∞\sum_{d\ge0}\frac{w^d q^{d^2}}{(q;q)_d(wq;q)_d}=\frac{1}{(wq;q)_\infty}

where each part is weighted by ww, not just the excedances or deficiencies.

Similar to the previous section, the results of Theorem 13 allow us to explain many patterns in the dfc(n,k)(n,k) array of Table 3.

Corollary 14. The sequence dfc(n,0)(n,0) equals the number of Rogers–Ramanujan partitions of nn, those restricted to parts congruent to 1 or 4 modulo 5.

Proof. Removing the (wq;q)d(wq;q)_d term from the denominator in (6) leaves a generating function for the w=0w=0 sequence, and the resulting

∑d≥0qd2(q;q)d\sum_{d\ge0}\frac{q^{d^2}}{(q;q)_d}

is a known generating function for the described Rogers–Ramanujan partitions [2 p. 52]. □

Corollary 15. A generating function for the number of partitions of nn with an even number of deficiencies is

12(q;q)∞−1(−q;q)∞∑n≥1(−q;q)n−1q(n+12)(q;q)n\frac{1}{2(q;q)_\infty}-\frac{1}{(-q;q)_\infty}\sum_{n\ge1}\frac{(-q;q)_{n-1}q^{\binom{n+1}{2}}}{(q;q)_n}

and an analogue for odd number of deficiencies is

12(q;q)∞+1(−q;q)∞∑n≥1(−q;q)n−1q(n+12)(q;q)n\frac{1}{2(q;q)_\infty}+\frac{1}{(-q;q)_\infty}\sum_{n\ge1}\frac{(-q;q)_{n-1}q^{\binom{n+1}{2}}}{(q;q)_n}

giving an expression for the difference

2(−q;q)∞∑n≥1(−q;q)n−1q(n+12)(q;q)n.\frac{2}{(-q;q)_\infty}\sum_{n\ge1}\frac{(-q;q)_{n-1}q^{\binom{n+1}{2}}}{(q;q)_n}.

Proof. The deficiencies parity results follow from standard techniques applied to (7) and setting w=1w=1. The difference expression comes most easily from setting w=−1w=-1 in (7). □

Table 4 provides initial values for the sequences described in Corollary 15. The even counts match [10 A046682] and odd counts match [10 A000701], while the difference matches the number of self-conjugate partitions [10 A000700].

nn123456789101112131415
even11234681216222940526990
odd01123571014202737496686
diff.101111122223334

Table 4. The number of partitions of nn with an even number of deficiencies, an odd number, and the difference, through n=15n = 15.

The weighted row sum dfc⁡(n)=∑kkdfc⁡(n,k)\operatorname{dfc}(n) = \sum_k k\operatorname{dfc}(n,k) gives the total number of deficiencies in the partitions of nn

Corollary 16. A generating function for the total number of deficiencies in the partitions of nn is

1(q;q)∞∑n≥1qn+(−1)nq(n+12)1−qn.\frac{1}{(q;q)_\infty}\sum_{n\geq1}\frac{q^n+(-1)^nq^{\binom{n+1}{2}}}{1-q^n}.

Proof. The result follows from determining the partial derivative of (7) with respect to ww and evaluating that expression at w=1w = 1. □

Table 5 gives values of exc⁡(n)\operatorname{exc}(n) and dfc⁡(n)\operatorname{dfc}(n) for small nn, informing the discussion at the beginning of this section. The dfc⁡(n)\operatorname{dfc}(n) sequence is [10 A114089] (with this interpretation); the exc⁡(n)\operatorname{exc}(n) sequence is not in the OEIS at the time of writing.

Again, the diagonal sums

diag⁡(dfc⁡,n)=dfc⁡(n,0)+dfc⁡(n−1,1)+⋯\operatorname{diag}(\operatorname{dfc},n)=\operatorname{dfc}(n,0)+\operatorname{dfc}(n-1,1)+\cdots

have a combinatorial interpretation (here the antidiagonal sums are infinite).

Corollary 17. The diagonal sum diag⁡(dfc⁡,n)\operatorname{diag}(\operatorname{dfc},n) equals the number of partitions of nn with no part 2.

Proof. As in Corollary 9, letting w=qw=q in (7) gives the diagonal sums of the dfc⁡(n,k)\operatorname{dfc}(n,k) triangle. The resulting ratio

(1/q;q)n(q;q)n\frac{(1/q;q)_n}{(q;q)_n}

in the summation is nonzero only for n=0n=0 and 1, leading to

1+q(q2;q)∞=1−q2(q;q)∞\frac{1+q}{(q^2;q)_\infty}=\frac{1-q^2}{(q;q)_\infty}

where the latter is a generating function for partitions with no part 2. □

We will focus on more combinatorial aspects of partition deficiencies in future work.

References

  1. [1]G. E. Andrews, R. Askey, R. Roy (1999). Special Functions, Encyclopaedia of Mathematics and its Applications, vol. 71. Cambridge University Press.
  2. [2]G. E. Andrews, K. Eriksson (2004). Integer Partitions. Cambridge University Press.
  3. [3]G. E. Andrews, F. Garvan (1988). Dyson’s crank of a partition. Bull. Amer. Math. Soc. 18: 167–171.DOI
  4. [4]G. E. Andrews, D. Newman (2020). The minimal excludant in integer partitions. J. Integer Seq. 23: 20.2.3.
  5. [5]A. Blecher, A. Knopfmacher (2022). Fixed points and matching points in partitions, Ramanujan J. 58: 23–41.DOI
  6. [6]S. Corteel, J. Lovejoy (2004). Overpartitions. Trans. Amer. Math. Soc. 356: 1623–1635.
  7. [7]B. Hopkins (2024). Refining Blecher and Knopfmacher’s integer partition fixed points. Enumer. Combin. Appl. 4: S2R26.DOI
  8. [8]B. Hopkins, J. A. Sellers (2020). Turning the partition crank. Amer. Math. Monthly 127: 654–657.
  9. [9]B. Hopkins, J. A. Sellers (2024). On Blecher and Knopfmacher’s fixed points for integer partitions, Discrete Math. 347: 113839.DOI
  10. [10]OEIS Foundation Inc. (2026). The On-Line Encyclopedia of Integer Sequences, oeis.org.
  11. [11]J. W. Reilly (1979). A combinatorial solution of two related problems in sequence enumeration. J. Combin. Theory Ser. A 26: 304–307.DOI

Paper details

vlnc:W_zpwyqquu
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 11

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents