Introduction and Statement of Results
A partition λ = ( λ 1 , λ 2 , … , λ k ) \lambda=(\lambda_1,\lambda_2,\ldots,\lambda_k) λ = ( λ 1 , λ 2 , … , λ k ) of a non-negative integer n n n is a non-increasing sequence of positive integers summing to n n n . Let p ( n ) p(n) p ( n ) denote the number of partitions of n n n . Throughout, for ∣ q ∣ < 1 |q|<1 ∣ q ∣ < 1 we write ( a ; q ) ∞ : = ∏ k ≥ 0 ( 1 − a q k ) (a;q)_\infty:=\prod_{k\ge0}(1-aq^k) ( a ; q ) ∞ := ∏ k ≥ 0 ( 1 − a q k ) , and for brevity we set f r : = ( q r ; q r ) ∞ f_r:=(q^r;q^r)_\infty f r := ( q r ; q r ) ∞ for every positive integer r r r . The generating function of p ( n ) p(n) p ( n ) is given by
∑ n ≥ 0 p ( n ) q n = ∏ r ≥ 1 1 1 − q r = 1 f 1 . \sum_{n\ge0}p(n)q^n=\prod_{r\ge1}\frac{1}{1-q^r}=\frac{1}{f_1}. n ≥ 0 ∑ p ( n ) q n = r ≥ 1 ∏ 1 − q r 1 = f 1 1 .
The arithmetic and analytic properties of p ( n ) p(n) p ( n ) have been studied extensively; see, for example, [1 , 14 , 17 , 27 , 29 ] . Grabner and Knopfmacher [15 ] introduced and studied the smallest gap of a partition, namely the least positive integer missing from its parts. In 2019, Andrews and Newman [4 , 5 ] rediscovered this statistic under the name minimal excludant, mex ( λ ) \operatorname{mex}(\lambda) mex ( λ ) . They defined σ mex ( n ) : = ∑ λ ⊢ n mex ( λ ) \sigma^{\operatorname{mex}}(n):=\sum_{\lambda\vdash n}\operatorname{mex}(\lambda) σ mex ( n ) := ∑ λ ⊢ n mex ( λ ) and proved
σ mex ( n ) = D 2 ( n ) , (1) \sigma^{\operatorname{mex}}(n)=D_2(n), \tag*{(1)} σ mex ( n ) = D 2 ( n ) , (1)
where D 2 ( n ) D_2(n) D 2 ( n ) is the number of partitions of n n n into distinct parts using two colours. Ballantine and Merca [7 ] gave a combinatorial proof of (1) . Barman and Singh [8 , 9 ] and da Silva and Sellers [13 ] established congruences and parity results for mex-related partition functions.
An overpartition of n n n is a partition of n n n in which the first occurrence of each distinct part may be overlined. Let p ‾ ( n ) \overline{p}(n) p ( n ) be the number of overpartitions of n n n . For example, p ‾ ( 3 ) = 8 \overline{p}(3)=8 p ( 3 ) = 8 :
3 3 3 , 3 ‾ \overline{3} 3 , 2 + 1 2+1 2 + 1 , 2 ‾ + 1 \overline{2}+1 2 + 1 , 2 + 1 ‾ 2+\overline{1} 2 + 1 , 2 ‾ + 1 ‾ \overline{2}+\overline{1} 2 + 1 , 1 + 1 + 1 1+1+1 1 + 1 + 1 , 1 ‾ + 1 + 1 \overline{1}+1+1 1 + 1 + 1 .
The generating function of p ‾ ( n ) \overline{p}(n) p ( n ) is given by
∑ n ≥ 0 p ‾ ( n ) q n = ∏ r ≥ 1 1 + q r 1 − q r = ( − q ; q ) ∞ ( q ; q ) ∞ = f 2 f 1 2 . \sum_{n\ge0}\overline{p}(n)q^n=\prod_{r\ge1}\frac{1+q^r}{1-q^r}=\frac{(-q;q)_\infty}{(q;q)_\infty}=\frac{f_2}{f_1^2}. n ≥ 0 ∑ p ( n ) q n = r ≥ 1 ∏ 1 − q r 1 + q r = ( q ; q ) ∞ ( − q ; q ) ∞ = f 1 2 f 2 .
Overpartitions were studied systematically by Corteel and Lovejoy [12 ] , and the arithmetic properties of p ( n ) p(n) p ( n ) have been investigated by many authors; see, for instance, [19 , 25 ] . Aricheta and Donato [6 ] defined the minimal excludant of an overpartition π \pi π as the smallest positive integer that is not a part of the non-overlined parts of π \pi π . They proved that σ ‾ m e x ( n ) = D 3 ( n ) \overline{\sigma}\mathrm{mex}(n)=D_3(n) σ mex ( n ) = D 3 ( n ) , the number of partitions of n n n into distinct parts using three colours. They also showed that σ ‾ m e x ( n ) \overline{\sigma}\mathrm{mex}(n) σ mex ( n ) is odd exactly when n n n is a triangular number.
A partition is called ℓ \ell ℓ -regular if none of its parts is divisible by ℓ \ell ℓ . Let p ℓ ( n ) p_\ell(n) p ℓ ( n ) denote the number of ℓ \ell ℓ -regular partitions of n n n . Then the generating function of p ℓ ( n ) p_\ell(n) p ℓ ( n ) is given by
∑ n ≥ 0 p ℓ ( n ) q n = ∏ r ≥ 1 ℓ ∤ r 1 1 − q r = f ℓ f 1 . \sum_{n\geq0}p_\ell(n)q^n=\prod_{\substack{r\geq1\\ \ell\nmid r}}\frac{1}{1-q^r}=\frac{f_\ell}{f_1}. n ≥ 0 ∑ p ℓ ( n ) q n = r ≥ 1 ℓ ∤ r ∏ 1 − q r 1 = f 1 f ℓ .
Divisibility, congruence and parity properties of p ℓ ( n ) p_\ell(n) p ℓ ( n ) have been studied, for example, in [14 , 20 , 24 , 28 ] . Recently Meher [26 ] defined the minimal excludant of an ℓ \ell ℓ -regular partition λ \lambda λ by mex ℓ ( λ ) : = min { r ≥ 1 : ℓ ∤ r , r ∉ λ } \operatorname{mex}_\ell(\lambda):=\min\{r\geq1:\ell\nmid r,\ r\notin\lambda\} mex ℓ ( λ ) := min { r ≥ 1 : ℓ ∤ r , r ∈ / λ } . He found the generating functions of σ ℓ mex ( n ) \sigma_\ell^{\operatorname{mex}}(n) σ ℓ mex ( n ) for ℓ = 2 , 3 \ell=2,3 ℓ = 2 , 3 , together with asymptotic formulas. He showed that σ 2 mex ( n ) \sigma_2^{\operatorname{mex}}(n) σ 2 mex ( n ) is odd exactly at the generalized pentagonal numbers, and that σ 3 mex ( n ) \sigma_3^{\operatorname{mex}}(n) σ 3 mex ( n ) is odd for ≫ log log X \gg\log\log X ≫ log log X values of n ≤ X n\leq X n ≤ X . Motivated by these results, we study the minimal excludant of ℓ \ell ℓ -regular overpartitions.
Let ℓ ≥ 2 \ell\geq2 ℓ ≥ 2 be a positive integer. An overpartition is ℓ \ell ℓ -regular if none of its parts, overlined or not, is divisible by ℓ \ell ℓ . Let P ‾ ℓ ( n ) \overline{\mathcal{P}}_\ell(n) P ℓ ( n ) denote the set of all ℓ \ell ℓ -regular overpartitions of n n n , and p ‾ ℓ ( n ) : = ∣ P ‾ ℓ ( n ) ∣ \overline{p}_\ell(n):=|\overline{\mathcal{P}}_\ell(n)| p ℓ ( n ) := ∣ P ℓ ( n ) ∣ . The overlined parts of an ℓ \ell ℓ -regular overpartition form a partition into distinct parts not divisible by ℓ \ell ℓ , and the non-overlined parts form an ℓ \ell ℓ -regular partition. Since
∏ r ≥ 1 ℓ ∤ r ( 1 + q r ) = ( − q ; q ) ∞ ( − q ℓ ; q ℓ ) ∞ = f 2 f 1 ⋅ f ℓ f 2 ℓ , \prod_{\substack{r\geq1\\ \ell\nmid r}}(1+q^r)=\frac{(-q;q)_\infty}{(-q^\ell;q^\ell)_\infty}=\frac{f_2}{f_1}\cdot\frac{f_\ell}{f_{2\ell}}, r ≥ 1 ℓ ∤ r ∏ ( 1 + q r ) = ( − q ℓ ; q ℓ ) ∞ ( − q ; q ) ∞ = f 1 f 2 ⋅ f 2 ℓ f ℓ ,
the generating function of p ‾ ℓ ( n ) \overline{p}_\ell(n) p ℓ ( n ) is given by
P ‾ ℓ ( q ) : = ∑ n ≥ 0 p ‾ ℓ ( n ) q n = ∏ r ≥ 1 ℓ ∤ r 1 + q r 1 − q r = f 2 f ℓ 2 f 1 2 f 2 ℓ . \overline{P}_\ell(q):=\sum_{n\geq0}\overline{p}_\ell(n)q^n=\prod_{\substack{r\geq1\\ \ell\nmid r}}\frac{1+q^r}{1-q^r}=\frac{f_2f_\ell^2}{f_1^2f_{2\ell}}. P ℓ ( q ) := n ≥ 0 ∑ p ℓ ( n ) q n = r ≥ 1 ℓ ∤ r ∏ 1 − q r 1 + q r = f 1 2 f 2 ℓ f 2 f ℓ 2 .
In particular, P ‾ 2 ( q ) = f 2 3 f 1 2 f 4 \overline{P}_2(q)=\frac{f_2^3}{f_1^2f_4} P 2 ( q ) = f 1 2 f 4 f 2 3 and P ‾ 3 ( q ) = f 2 f 3 2 f 1 2 f 6 \overline{P}_3(q)=\frac{f_2f_3^2}{f_1^2f_6} P 3 ( q ) = f 1 2 f 6 f 2 f 3 2 . The function p ‾ ℓ ( n ) \overline{p}_\ell(n) p ℓ ( n ) was introduced by Lovejoy [23 ] , and its arithmetic properties have been studied in [2 , 11 , 16 , 30 ] . For π ∈ P ‾ ℓ ( n ) \pi\in\overline{\mathcal{P}}_\ell(n) π ∈ P ℓ ( n ) , the ℓ \ell ℓ -regular minimal excludant is defined by
mex ‾ ℓ ( π ) : = min { r ≥ 1 : ℓ ∤ r , r is not a non-overlined part of π } . \overline{\operatorname{mex}}_\ell(\pi):=\min\{r\geq1:\ell\nmid r,\ r\text{ is not a non-overlined part of }\pi\}. mex ℓ ( π ) := min { r ≥ 1 : ℓ ∤ r , r is not a non-overlined part of π } .
In other words, mex ‾ ℓ ( π ) \overline{\operatorname{mex}}_\ell(\pi) mex ℓ ( π ) is the smallest positive integer that is not divisible by ℓ \ell ℓ and does not occur as a non-overlined part of π \pi π . Overlined parts play no role in mex ‾ ℓ ( π ) \overline{\operatorname{mex}}_\ell(\pi) mex ℓ ( π ) . We call a positive integer admissible if it is not divisible by ℓ \ell ℓ . Thus every part of an ℓ \ell ℓ -regular overpartition, overlined or not, is admissible, and mex ‾ ℓ ( π ) \overline{\operatorname{mex}}_\ell(\pi) mex ℓ ( π ) is the smallest admissible integer that does not occur as a non-overlined part of π \pi π . Let m m m be a positive integer such that ℓ ∤ m \ell\nmid m ℓ ∤ m . We define
p ‾ ℓ mex ( m , n ) : = # { π ∈ P ‾ ℓ ( n ) : mex ‾ ℓ ( π ) = m } . \overline{p}^{\operatorname{mex}}_\ell(m,n):=\#\{\pi\in\overline{\mathcal{P}}_\ell(n):\overline{\operatorname{mex}}_\ell(\pi)=m\}. p ℓ mex ( m , n ) := # { π ∈ P ℓ ( n ) : mex ℓ ( π ) = m } .
If ℓ ∣ m \ell\mid m ℓ ∣ m , then p ‾ ℓ mex ( m , n ) = 0 \overline{p}^{\operatorname{mex}}_\ell(m,n)=0 p ℓ mex ( m , n ) = 0 by convention. We consider some examples to understand the above definitions.
Example 1. Let ℓ = 4 \ell=4 ℓ = 4 and n = 6 n=6 n = 6 . The admissible parts are 1 , 2 , 3 , 5 , 6 , 7 , 9 , 10 , 11 , … 1,2,3,5,6,7,9,10,11,\ldots 1 , 2 , 3 , 5 , 6 , 7 , 9 , 10 , 11 , …
(1) Let π = 6 \pi=6 π = 6 . The only non-overlined part is $6$, so 1 1 1 does not occur as a non-overlined part. Hence $\overline{\operatorname{mex}}_4(\pi)=1. (2) Let π = 5 + 1 \pi=5+1 π = 5 + 1 . The non-overlined parts contain 1 but not 2, so mex ‾ 4 ( π ) = 2 \overline{\operatorname{mex}}_4(\pi)=2 mex 4 ( π ) = 2 . On the other hand, for π = 5 + 1 ‾ \pi=5+\overline{1} π = 5 + 1 the part 1 is overlined and hence ignored, so mex ‾ 4 ( π ) = 1 \overline{\operatorname{mex}}_4(\pi)=1 mex 4 ( π ) = 1 .
(3) Let π = 2 + 2 + 1 + 1 \pi=2+2+1+1 π = 2 + 2 + 1 + 1 . The non-overlined parts contain 1, 2 but not 3, so mex ‾ 4 ( π ) = 3 \overline{\operatorname{mex}}_4(\pi)=3 mex 4 ( π ) = 3 .
(4) Let π = 3 ‾ + 2 + 1 \pi=\overline{3}+2+1 π = 3 + 2 + 1 . The non-overlined parts contain 1, 2, 3. The integer 4 is not admissible, and 5 does not occur, hence mex ‾ 4 ( π ) = 5 \overline{\operatorname{mex}}_4(\pi)=5 mex 4 ( π ) = 5 . For π = 3 ‾ + 2 + 1 \pi=\overline{3}+2+1 π = 3 + 2 + 1 , the non-overlined parts are 2, 1, so mex ‾ 4 ( π ) = 3 \overline{\operatorname{mex}}_4(\pi)=3 mex 4 ( π ) = 3 .
(5) Let π = 3 + 2 ‾ + 1 \pi=3+\overline{2}+1 π = 3 + 2 + 1 . The non-overlined parts are 3, 1, so mex ‾ 4 ( π ) = 2 \overline{\operatorname{mex}}_4(\pi)=2 mex 4 ( π ) = 2 .
Our main object of study is the function
σ ‾ ℓ mex ( n ) : = ∑ π ∈ P ℓ ( n ) mex ‾ ℓ ( π ) , σ ‾ ℓ mex ( 0 ) : = 1. \overline{\sigma}^{\operatorname{mex}}_{\ell}(n):=\sum_{\pi\in\mathcal{P}_{\ell}(n)}\overline{\operatorname{mex}}_{\ell}(\pi),\qquad\overline{\sigma}^{\operatorname{mex}}_{\ell}(0):=1. σ ℓ mex ( n ) := π ∈ P ℓ ( n ) ∑ mex ℓ ( π ) , σ ℓ mex ( 0 ) := 1.
the natural analogue of σ ℓ mex ( n ) \sigma^{\operatorname{mex}}_{\ell}(n) σ ℓ mex ( n ) for ℓ \ell ℓ -regular overpartitions. We obtain explicit generating functions for σ ‾ ℓ mex ( n ) \overline{\sigma}^{\operatorname{mex}}_{\ell}(n) σ ℓ mex ( n ) , expressed as eta-quotients, for ℓ = 2 \ell=2 ℓ = 2 and ℓ = 3 \ell=3 ℓ = 3 .
Theorem 1.1. We have
∑ n ≥ 0 σ ‾ 2 mex ( n ) q n = ( q 2 ; q 2 ) ∞ 8 ( q ; q ) ∞ 4 ( q 4 ; q 4 ) ∞ 3 = f 2 8 f 1 4 f 4 3 . (2) \sum_{n\geq0}\overline{\sigma}^{\operatorname{mex}}_{2}(n)q^n=\frac{(q^2;q^2)_{\infty}^{8}}{(q;q)_{\infty}^{4}(q^4;q^4)_{\infty}^{3}}=\frac{f_2^{8}}{f_1^{4}f_4^{3}}. \tag*{(2)} n ≥ 0 ∑ σ 2 mex ( n ) q n = ( q ; q ) ∞ 4 ( q 4 ; q 4 ) ∞ 3 ( q 2 ; q 2 ) ∞ 8 = f 1 4 f 4 3 f 2 8 . (2)
Theorem 1.2. We have
∑ n ≥ 0 σ ‾ 3 mex ( n ) q n = f 2 f 6 4 f 1 2 f 12 2 + q f 2 2 f 3 2 f 12 2 f 1 2 f 6 2 . (3) \sum_{n\geq0}\overline{\sigma}^{\operatorname{mex}}_{3}(n)q^n=\frac{f_2f_6^{4}}{f_1^{2}f_{12}^{2}}+q\frac{f_2^{2}f_3^{2}f_{12}^{2}}{f_1^{2}f_6^{2}}. \tag*{(3)} n ≥ 0 ∑ σ 3 mex ( n ) q n = f 1 2 f 12 2 f 2 f 6 4 + q f 1 2 f 6 2 f 2 2 f 3 2 f 12 2 . (3)
In general, motivated by [31 ] we derive the following.
Theorem 1.3. Let ℓ ≥ 2 \ell\geq2 ℓ ≥ 2 be an integer. If ℓ \ell ℓ is even, then
∑ n ≥ 0 σ ‾ ℓ mex ( n ) q n = f 2 f ℓ 2 f ℓ ( ℓ − 1 ) f 1 2 f 2 ℓ ∑ s = 1 ℓ / 2 q s ( s − 1 ) 2 ( − q ℓ ( ℓ + 2 s − 3 ) 2 ; q ℓ ( ℓ − 1 ) ) ∞ ( − q ℓ ( ℓ − 2 s + 1 ) 2 ; q ℓ ( ℓ − 1 ) ) ∞ . (4) \sum_{n\geq0}\overline{\sigma}^{\operatorname{mex}}_{\ell}(n)q^n=
\frac{f_2f_{\ell}^{2}f_{\ell(\ell-1)}}{f_1^{2}f_{2\ell}}
\sum_{s=1}^{\ell/2}q^{\frac{s(s-1)}{2}}
\left(-q^{\frac{\ell(\ell+2s-3)}{2}};q^{\ell(\ell-1)}\right)_{\infty}
\left(-q^{\frac{\ell(\ell-2s+1)}{2}};q^{\ell(\ell-1)}\right)_{\infty}. \tag*{(4)} n ≥ 0 ∑ σ ℓ mex ( n ) q n = f 1 2 f 2 ℓ f 2 f ℓ 2 f ℓ ( ℓ − 1 ) s = 1 ∑ ℓ /2 q 2 s ( s − 1 ) ( − q 2 ℓ ( ℓ + 2 s − 3 ) ; q ℓ ( ℓ − 1 ) ) ∞ ( − q 2 ℓ ( ℓ − 2 s + 1 ) ; q ℓ ( ℓ − 1 ) ) ∞ . (4) If ℓ \ell ℓ is odd, then
∑ n ≥ 0 σ ‾ ℓ mex ( n ) q n = f 2 f ℓ 2 f ℓ ( ℓ − 1 ) f 1 2 f 2 ℓ [ ∑ s = 1 ( ℓ − 1 ) / 2 q s ( s − 1 ) 2 ( − q ℓ ( ℓ + 2 s − 3 ) 2 ; q ℓ ( ℓ − 1 ) ) ∞ ( − q ℓ ( ℓ − 2 s + 1 ) 2 ; q ℓ ( ℓ − 1 ) ) ∞ + q ℓ 2 − 1 8 f 2 ℓ 2 f ℓ ( ℓ − 1 ) f ℓ 2 f 2 ( ℓ − 1 ) ] . (5) \sum_{n\geq0}\overline{\sigma}^{\operatorname{mex}}_{\ell}(n)q^n=
\frac{f_2f_{\ell}^{2}f_{\ell(\ell-1)}}{f_1^{2}f_{2\ell}}
\left[
\sum_{s=1}^{(\ell-1)/2}q^{\frac{s(s-1)}{2}}
\left(-q^{\frac{\ell(\ell+2s-3)}{2}};q^{\ell(\ell-1)}\right)_{\infty}
\left(-q^{\frac{\ell(\ell-2s+1)}{2}};q^{\ell(\ell-1)}\right)_{\infty}
+q^{\frac{\ell^{2}-1}{8}}\frac{f_{2\ell}^{2}f_{\ell(\ell-1)}}{f_{\ell}^{2}f_{2(\ell-1)}}
\right]. \tag*{(5)} n ≥ 0 ∑ σ ℓ mex ( n ) q n = f 1 2 f 2 ℓ f 2 f ℓ 2 f ℓ ( ℓ − 1 ) s = 1 ∑ ( ℓ − 1 ) /2 q 2 s ( s − 1 ) ( − q 2 ℓ ( ℓ + 2 s − 3 ) ; q ℓ ( ℓ − 1 ) ) ∞ ( − q 2 ℓ ( ℓ − 2 s + 1 ) ; q ℓ ( ℓ − 1 ) ) ∞ + q 8 ℓ 2 − 1 f ℓ 2 f 2 ( ℓ − 1 ) f 2 ℓ 2 f ℓ ( ℓ − 1 ) . (5)
Further, we show that { σ ‾ ℓ mex ( n ) } n ≥ 0 \{\overline{\sigma}^{\operatorname{mex}}_{\ell}(n)\}_{n\geq0} { σ ℓ mex ( n ) } n ≥ 0 is weakly increasing for every ℓ ≥ 2 \ell\geq2 ℓ ≥ 2 . In particular, we have the following result.
Theorem 1.4. The sequence { σ ‾ ℓ mex ( n ) } n ≥ 0 \{\overline{\sigma}^{\operatorname{mex}}_{\ell}(n)\}_{n\geq0} { σ ℓ mex ( n ) } n ≥ 0 is weakly increasing. Moreover, it satisfies the inequality
σ ‾ ℓ mex ( n + 1 ) − σ ‾ ℓ mex ( n ) ≥ p ‾ ℓ ( n ) − p ‾ ℓ ( n − 1 ) . (6) \overline{\sigma}^{\operatorname{mex}}_{\ell}(n+1)-\overline{\sigma}^{\operatorname{mex}}_{\ell}(n)\geq\overline{p}_{\ell}(n)-\overline{p}_{\ell}(n-1). \tag*{(6)} σ ℓ mex ( n + 1 ) − σ ℓ mex ( n ) ≥ p ℓ ( n ) − p ℓ ( n − 1 ) . (6)
Using Ingham’s Tauberian theorem, we establish Hardy–Ramanujan type asymptotic formulas for σ ‾ 2 mex ( n ) \overline{\sigma}^{\operatorname{mex}}_{2}(n) σ 2 mex ( n ) and σ ‾ 3 mex ( n ) \overline{\sigma}^{\operatorname{mex}}_{3}(n) σ 3 mex ( n ) .
Theorem 1.5. We have
σ ‾ 2 mex ( n ) ∼ 1 2 2 n e π n / 2 , as n → ∞ . \overline{\sigma}^{\operatorname{mex}}_{2}(n)\sim\frac{1}{2\sqrt{2n}}e^{\pi\sqrt{n/2}},\qquad\text{as }n\to\infty. σ 2 mex ( n ) ∼ 2 2 n 1 e π n /2 , as n → ∞.
Theorem 1.6. We have
σ ‾ 3 mex ( n ) ∼ 1 4 n e π 2 n / 3 , as n → ∞ . \overline{\sigma}^{\operatorname{mex}}_{3}(n)\sim\frac{1}{4\sqrt{n}}e^{\pi\sqrt{2n/3}},\qquad\text{as }n\to\infty. σ 3 mex ( n ) ∼ 4 n 1 e π 2 n /3 , as n → ∞.
We further determine the parity of σ ‾ 2 m e x ( n ) \overline{\sigma}^{\mathrm{mex}}_{2}(n) σ 2 mex ( n ) and σ ‾ 3 m e x ( n ) \overline{\sigma}^{\mathrm{mex}}_{3}(n) σ 3 mex ( n ) completely, and deduce the number of n ≤ X n \le X n ≤ X for which σ ‾ 3 m e x ( n ) \overline{\sigma}^{\mathrm{mex}}_{3}(n) σ 3 mex ( n ) is odd and even. In particular, we get the following results.
Theorem 1.7. (1) σ ‾ 2 m e x ( n ) \overline{\sigma}^{\mathrm{mex}}_{2}(n) σ 2 mex ( n ) is an even integer for every n ≥ 1 n \ge1 n ≥ 1 .
(2) σ ‾ 3 m e x ( n ) ≡ 1 ( m o d 2 ) \overline{\sigma}^{\mathrm{mex}}_{3}(n) \equiv1 \pmod{2} σ 3 mex ( n ) ≡ 1 ( mod 2 ) if and only if n = 0 n = 0 n = 0 or n = 3 k 2 + 3 k + 1 n = 3k^{2} + 3k + 1 n = 3 k 2 + 3 k + 1 for some non-negative integer k k k .
Theorem 1.8. For every real number X ≥ 1 X \ge1 X ≥ 1 , we have
# { n ≤ X : σ ‾ 3 m e x ( n ) is an odd integer } = X 3 + O ( 1 ) , \#\{n \le X : \overline{\sigma}^{\mathrm{mex}}_{3}(n)\text{ is an odd integer}\} = \sqrt{\frac{X}{3}} + O(1), # { n ≤ X : σ 3 mex ( n ) is an odd integer } = 3 X + O ( 1 ) , while
# { n ≤ X : σ ‾ 3 m e x ( n ) is an even integer } = X − X 3 + O ( 1 ) . \#\{n \le X : \overline{\sigma}^{\mathrm{mex}}_{3}(n)\text{ is an even integer}\} = X - \sqrt{\frac{X}{3}} + O(1). # { n ≤ X : σ 3 mex ( n ) is an even integer } = X − 3 X + O ( 1 ) .
More generally, for every ℓ ≥ 2 \ell\ge2 ℓ ≥ 2 we determine the parity of σ ‾ ℓ m e x ( n ) \overline{\sigma}^{\mathrm{mex}}_{\ell}(n) σ ℓ mex ( n ) completely.
Theorem 1.9. Let ℓ ≥ 2 \ell\ge2 ℓ ≥ 2 . Then σ ‾ ℓ m e x ( n ) \overline{\sigma}^{\mathrm{mex}}_{\ell}(n) σ ℓ mex ( n ) is odd if and only if n = 0 n = 0 n = 0 or
n = R ℓ ( ℓ k + s ) = ℓ ( ℓ − 1 ) 2 k 2 + ( s − 1 ) ℓ k + s ( s − 1 ) 2 n = R_{\ell}(\ell k+s)=\frac{\ell(\ell-1)}{2}k^{2}+(s-1)\ell k+\frac{s(s-1)}{2} n = R ℓ ( ℓ k + s ) = 2 ℓ ( ℓ − 1 ) k 2 + ( s − 1 ) ℓ k + 2 s ( s − 1 ) for some integers k ≥ 0 k \ge0 k ≥ 0 and 2 ≤ s ≤ ℓ − 1 2 \le s \le\ell-1 2 ≤ s ≤ ℓ − 1 . Consequently, for every real number X ≥ 1 X \ge1 X ≥ 1 ,
# { 0 ≤ n ≤ X : σ ‾ ℓ m e x ( n ) is odd } = ( ℓ − 2 ) 2 X ℓ ( ℓ − 1 ) + O ( 1 ) . \#\{0 \le n \le X : \overline{\sigma}^{\mathrm{mex}}_{\ell}(n)\text{ is odd}\}=(\ell-2)\sqrt{\frac{2X}{\ell(\ell-1)}}+O(1). # { 0 ≤ n ≤ X : σ ℓ mex ( n ) is odd } = ( ℓ − 2 ) ℓ ( ℓ − 1 ) 2 X + O ( 1 ) . In particular, for every ℓ ≥ 3 \ell\ge3 ℓ ≥ 3 , σ ‾ ℓ m e x ( n ) \overline{\sigma}^{\mathrm{mex}}_{\ell}(n) σ ℓ mex ( n ) is odd for infinitely many n n n , and the number of such n ≤ X n \le X n ≤ X is ≫ X \gg\sqrt{X} ≫ X .
Finally, we prove the following congruences.
Theorem 1.10. For every non-negative integer n n n ,
(1) σ ‾ 2 m e x ( 3 n + 2 ) ≡ 0 ( m o d 3 ) \overline{\sigma}^{\mathrm{mex}}_{2}(3n+2) \equiv0 \pmod{3} σ 2 mex ( 3 n + 2 ) ≡ 0 ( mod 3 ) ;
(2) for n ≥ 1 n \ge1 n ≥ 1 , σ ‾ 2 m e x ( n ) ≡ 2 ( m o d 4 ) \overline{\sigma}^{\mathrm{mex}}_{2}(n) \equiv2 \pmod{4} σ 2 mex ( n ) ≡ 2 ( mod 4 ) if n = 2 k 2 n=2k^{2} n = 2 k 2 for some k ≥ 1 k \ge1 k ≥ 1 , and σ ‾ 2 m e x ( n ) ≡ 0 ( m o d 4 ) \overline{\sigma}^{\mathrm{mex}}_{2}(n) \equiv0 \pmod{4} σ 2 mex ( n ) ≡ 0 ( mod 4 ) ;
(3)
∑ n ≥ 0 σ ‾ 2 m e x ( 2 n + 1 ) q n = 4 f 2 4 f 1 2 f 4 2 . \sum_{n\ge0}\overline{\sigma}^{\mathrm{mex}}_{2}(2n+1)q^{n}=4\frac{f_{2}^{4}}{f_{1}^{2}f_{4}^{2}}. n ≥ 0 ∑ σ 2 mex ( 2 n + 1 ) q n = 4 f 1 2 f 4 2 f 2 4 . Consequently, for odd m m m , σ ‾ 2 m e x ( m ) ≡ 4 ( m o d 8 ) \overline{\sigma}^{\mathrm{mex}}_{2}(m) \equiv4 \pmod{8} σ 2 mex ( m ) ≡ 4 ( mod 8 ) if m m m is a perfect square, and σ ‾ 2 m e x ( m ) ≡ 0 ( m o d 8 ) \overline{\sigma}^{\mathrm{mex}}_{2}(m) \equiv0 \pmod{8} σ 2 mex ( m ) ≡ 0 ( mod 8 ) ;
(4) for n ≥ 1 n \ge1 n ≥ 1 , σ ‾ 3 m e x ( 3 n ) ≡ 2 ( m o d 4 ) \overline{\sigma}^{\mathrm{mex}}_{3}(3n) \equiv2 \pmod{4} σ 3 mex ( 3 n ) ≡ 2 ( mod 4 ) if n = 3 k 2 n=3k^{2} n = 3 k 2 for some k ≥ 1 k \ge1 k ≥ 1 , and σ ‾ 3 m e x ( 3 n ) ≡ 0 ( m o d 4 ) \overline{\sigma}^{\mathrm{mex}}_{3}(3n) \equiv0 \pmod{4} σ 3 mex ( 3 n ) ≡ 0 ( mod 4 ) otherwise. In particular σ ‾ 3 m e x ( 9 n + 3 ) ≡ σ ‾ 3 m e x ( 9 n + 6 ) ≡ 0 ( m o d 4 ) \overline{\sigma}^{\mathrm{mex}}_{3}(9n+3) \equiv\overline{\sigma}^{\mathrm{mex}}_{3}(9n+6) \equiv0 \pmod{4} σ 3 mex ( 9 n + 3 ) ≡ σ 3 mex ( 9 n + 6 ) ≡ 0 ( mod 4 ) .
Preliminaries
Lemma 2.1. We have
∑ t ∈ Z q t 2 = ( q 2 ; q 2 ) ∞ 5 ( q ; q ) ∞ 2 ( q 4 ; q 4 ) ∞ 2 = f 2 5 f 1 2 f 4 2 . \sum_{t\in\mathbb{Z}}q^{t^{2}}=\frac{(q^{2};q^{2})_{\infty}^{5}}{(q;q)_{\infty}^{2}(q^{4};q^{4})_{\infty}^{2}}=\frac{f_{2}^{5}}{f_{1}^{2}f_{4}^{2}}. t ∈ Z ∑ q t 2 = ( q ; q ) ∞ 2 ( q 4 ; q 4 ) ∞ 2 ( q 2 ; q 2 ) ∞ 5 = f 1 2 f 4 2 f 2 5 . Proof. See [18 Eq. (1.5.1)] . □
Lemma 2.2. For z ≠ 0 z\ne0 z = 0 and ∣ Q ∣ < 1 \lvert Q\rvert<1 ∣ Q ∣ < 1 , we have
∑ r = − ∞ ∞ Q r 2 z r = ( − Q z ; Q 2 ) ∞ ( − Q z ; Q 2 ) ∞ ( Q 2 ; Q 2 ) ∞ . (7) \sum_{r=-\infty}^{\infty}Q^{r^{2}}z^{r}=(-Qz;Q^{2})_{\infty}\left(-\frac{Q}{z};Q^{2}\right)_{\infty}(Q^{2};Q^{2})_{\infty}. \tag*{(7)} r = − ∞ ∑ ∞ Q r 2 z r = ( − Q z ; Q 2 ) ∞ ( − z Q ; Q 2 ) ∞ ( Q 2 ; Q 2 ) ∞ . (7) Proof. See [3 Theorem 2.8] . □
Lemma 2.3 ([21 ] ). Let B ( q ) : = ∑ n = 0 ∞ b ( n ) q n B(q) := \sum_{n=0}^{\infty} b(n)q^n B ( q ) := ∑ n = 0 ∞ b ( n ) q n be a power series with radius of convergence 1. Assume that { b ( n ) } \{b(n)\} { b ( n )} is a weakly increasing sequence of non-negative real numbers. If there are constants β , γ ∈ R \beta,\gamma\in\mathbb{R} β , γ ∈ R and μ > 0 \mu> 0 μ > 0 such that
B ( e − y ) ∼ β y γ e μ / y , as y → 0 + , B(e^{-y}) \sim\beta y^\gamma e^{\mu/y}, \qquad\text{as } y \to0^+, B ( e − y ) ∼ β y γ e μ / y , as y → 0 + , then
b ( n ) ∼ β μ 2 γ + 1 4 2 π n 2 γ + 3 4 e 2 μ n , as n → ∞ . b(n) \sim\frac{\beta\mu^{\frac{2\gamma+1}{4}}}{2\sqrt{\pi}n^{\frac{2\gamma+3}{4}}}e^{2\sqrt{\mu n}}, \qquad\text{as } n \to\infty. b ( n ) ∼ 2 π n 4 2 γ + 3 β μ 4 2 γ + 1 e 2 μ n , as n → ∞.
Lemma 2.4. For any positive integers u u u and j j j , and a prime number p p p , we have
( q u ; q u ) ∞ p j ≡ ( q p u ; q p u ) ∞ p j − 1 ( m o d p j ) . (8) (q^{u};q^{u})_{\infty}^{p^j} \equiv(q^{pu};q^{pu})_{\infty}^{p^{j-1}} \pmod{p^j}. \tag*{(8)} ( q u ; q u ) ∞ p j ≡ ( q p u ; q p u ) ∞ p j − 1 ( mod p j ) . (8)
Lemma 2.5. We have
f 1 3 = ( q ; q ) ∞ 3 = ∑ n = 0 ∞ ( − 1 ) n ( 2 n + 1 ) q n ( n + 1 ) 2 . (9) f_1^3=(q;q)_{\infty}^3=\sum_{n=0}^{\infty}(-1)^n(2n+1)q^{\frac{n(n+1)}{2}}. \tag*{(9)} f 1 3 = ( q ; q ) ∞ 3 = n = 0 ∑ ∞ ( − 1 ) n ( 2 n + 1 ) q 2 n ( n + 1 ) . (9) Proof. See [18 Eq. (1.7.1)] . □
Lemma 2.6. Let k ≥ 0 k \geq0 k ≥ 0 and 1 ≤ s ≤ ℓ − 1 1 \leq s \leq\ell-1 1 ≤ s ≤ ℓ − 1 . For a positive integer ℓ ≥ 2 \ell\geq2 ℓ ≥ 2 , let
R ℓ ( m ) = ∑ 1 ≤ j < m ℓ ∤ j j . R_{\ell}(m)=\sum_{\substack{1\leq j<m\\ \ell\nmid j}}j. R ℓ ( m ) = 1 ≤ j < m ℓ ∤ j ∑ j . Then
R ℓ ( ℓ k + s ) = ℓ ( ℓ − 1 ) 2 k 2 + ( s − 1 ) ℓ k + s ( s − 1 ) 2 . R_{\ell}(\ell k+s)=\frac{\ell(\ell-1)}{2}k^2+(s-1)\ell k+\frac{s(s-1)}{2}. R ℓ ( ℓ k + s ) = 2 ℓ ( ℓ − 1 ) k 2 + ( s − 1 ) ℓ k + 2 s ( s − 1 ) . Proof. See [31 Lemma 2.4] . □
Lemma 2.7. Let ℓ ≥ 2 \ell\geq2 ℓ ≥ 2 be an integer. Then we have two cases. If ℓ \ell ℓ is even, then
∑ n ≥ 0 σ ℓ max ( n ) q n = f ℓ f ℓ ( ℓ − 1 ) f 1 ∑ s = 1 ℓ / 2 q s ( s − 1 ) 2 ( − q ℓ ( ℓ + 2 s − 3 ) 2 ; q ℓ ( ℓ − 1 ) ) ∞ ( − q ℓ ( ℓ − 2 s + 1 ) 2 ; q ℓ ( ℓ − 1 ) ) ∞ . \sum_{n\geq0}\sigma_{\ell}^{\max}(n)q^n=\frac{f_{\ell}f_{\ell(\ell-1)}}{f_1}\sum_{s=1}^{\ell/2}q^{\frac{s(s-1)}{2}}\left(-q^{\frac{\ell(\ell+2s-3)}{2}};q^{\ell(\ell-1)}\right)_{\infty}\left(-q^{\frac{\ell(\ell-2s+1)}{2}};q^{\ell(\ell-1)}\right)_{\infty}. n ≥ 0 ∑ σ ℓ m a x ( n ) q n = f 1 f ℓ f ℓ ( ℓ − 1 ) s = 1 ∑ ℓ /2 q 2 s ( s − 1 ) ( − q 2 ℓ ( ℓ + 2 s − 3 ) ; q ℓ ( ℓ − 1 ) ) ∞ ( − q 2 ℓ ( ℓ − 2 s + 1 ) ; q ℓ ( ℓ − 1 ) ) ∞ . If ℓ \ell ℓ is odd, then
∑ n ≥ 0 σ ℓ max ( n ) q n = f ℓ f ℓ ( ℓ − 1 ) f 1 [ ∑ s = 1 ( ℓ − 1 ) / 2 q s ( s − 1 ) 2 ( − q ℓ ( ℓ + 2 s − 3 ) 2 ; q ℓ ( ℓ − 1 ) ) ∞ ( − q ℓ ( ℓ − 2 s + 1 ) 2 ; q ℓ ( ℓ − 1 ) ) ∞ + q ℓ 2 − 1 8 f 2 ℓ 2 f ℓ ( ℓ − 1 ) f ℓ 2 f ℓ ( ℓ − 1 ) ] . \sum_{n\geq0}\sigma_{\ell}^{\max}(n)q^n=\frac{f_{\ell}f_{\ell(\ell-1)}}{f_1}\left[\sum_{s=1}^{(\ell-1)/2}q^{\frac{s(s-1)}{2}}\left(-q^{\frac{\ell(\ell+2s-3)}{2}};q^{\ell(\ell-1)}\right)_{\infty}\left(-q^{\frac{\ell(\ell-2s+1)}{2}};q^{\ell(\ell-1)}\right)_{\infty}+q^{\frac{\ell^2-1}{8}}\frac{f_{2\ell}^2f_{\ell(\ell-1)}}{f_{\ell}^2f_{\ell(\ell-1)}}\right]. n ≥ 0 ∑ σ ℓ m a x ( n ) q n = f 1 f ℓ f ℓ ( ℓ − 1 ) s = 1 ∑ ( ℓ − 1 ) /2 q 2 s ( s − 1 ) ( − q 2 ℓ ( ℓ + 2 s − 3 ) ; q ℓ ( ℓ − 1 ) ) ∞ ( − q 2 ℓ ( ℓ − 2 s + 1 ) ; q ℓ ( ℓ − 1 ) ) ∞ + q 8 ℓ 2 − 1 f ℓ 2 f ℓ ( ℓ − 1 ) f 2 ℓ 2 f ℓ ( ℓ − 1 ) . Proof. See [31 Theorem 1.2] . □
Lemma 2.8. We have
1 f 1 4 = f 4 14 f 2 14 f 8 4 + 4 q f 4 2 f 8 4 f 2 10 . (10) \frac{1}{f_1^4}=\frac{f_4^{14}}{f_2^{14}f_8^4}+4q\frac{f_4^2f_8^4}{f_2^{10}}. \tag*{(10)} f 1 4 1 = f 2 14 f 8 4 f 4 14 + 4 q f 2 10 f 4 2 f 8 4 . (10) f 2 2 f 1 = f 6 f 9 2 f 3 f 18 + q f 18 2 f 9 . (11) \frac{f_2^2}{f_1}=\frac{f_6f_9^2}{f_3f_{18}}+q\frac{f_{18}^2}{f_9}. \tag*{(11)} f 1 f 2 2 = f 3 f 18 f 6 f 9 2 + q f 9 f 18 2 . (11) f 2 2 f 1 = f 6 f 9 4 f 3 2 f 18 + 2 q 2 f 6 3 f 9 3 f 3 7 + 4 q 2 f 6 2 f 18 3 f 3 5 . (12) \frac{f_2^2}{f_1}=\frac{f_6f_9^4}{f_3^2f_{18}}+2q^2\frac{f_6^3f_9^3}{f_3^7}+4q^2\frac{f_6^2f_{18}^3}{f_3^5}. \tag*{(12)} f 1 f 2 2 = f 3 2 f 18 f 6 f 9 4 + 2 q 2 f 3 7 f 6 3 f 9 3 + 4 q 2 f 3 5 f 6 2 f 18 3 . (12) Proof. See [18 ] . □
Proofs
Proposition 1. For a positive integer ℓ ≥ 2 \ell\ge2 ℓ ≥ 2 , and
R ℓ ( m ) = ∑ 1 ≤ j < m ℓ ∤ j j , R_{\ell}(m)=\sum_{\substack{1\le j<m\\ \ell\nmid j}}j, R ℓ ( m ) = 1 ≤ j < m ℓ ∤ j ∑ j , we have
∑ m ≥ 1 ℓ ∤ m ∑ n = 0 ∞ p ‾ ℓ mex ( m , n ) z m q n = f 2 ℓ f ℓ 2 f 1 2 f 2 ℓ ∑ m ≥ 1 ℓ ∤ m z m q R ℓ ( m ) ( 1 − q m ) . (13) \sum_{\substack{m\ge1\\ \ell\nmid m}}\sum_{n=0}^{\infty}\overline{p}_{\ell}^{\operatorname{mex}}(m,n)z^{m}q^{n}
=\frac{f_{2\ell}f_{\ell}^{2}}{f_{1}^{2}f_{2\ell}}\sum_{\substack{m\ge1\\ \ell\nmid m}}z^{m}q^{R_{\ell}(m)}(1-q^{m}). \tag*{(13)} m ≥ 1 ℓ ∤ m ∑ n = 0 ∑ ∞ p ℓ mex ( m , n ) z m q n = f 1 2 f 2 ℓ f 2 ℓ f ℓ 2 m ≥ 1 ℓ ∤ m ∑ z m q R ℓ ( m ) ( 1 − q m ) . (13)
Proof. Suppose mex ‾ ℓ ( π ) = m \overline{\operatorname{mex}}_{\ell}(\pi)=m mex ℓ ( π ) = m , ℓ ∤ m \ell\nmid m ℓ ∤ m . Then
every admissible integer 1 ≤ r < m 1\le r<m 1 ≤ r < m , ℓ ∤ r \ell\nmid r ℓ ∤ r , must occur at least once as a non-overlined part;
the part m m m is absent from the non-overlined parts;
every admissible integer greater than m m m may occur as a non-overlined part with arbitrary multiplicity;
the overlined parts form an arbitrary partition into distinct admissible parts.
The compulsory contribution of the non-overlined parts below m m m is q R ℓ ( m ) ∏ r < m ℓ ∤ r 1 1 − q r q^{R_{\ell}(m)}\prod_{\substack{r<m\\ \ell\nmid r}}\frac{1}{1-q^{r}} q R ℓ ( m ) ∏ r < m ℓ ∤ r 1 − q r 1 . The unrestricted larger admissible non-overlined parts contribute ∏ r > m ℓ ∤ r 1 1 − q r \prod_{\substack{r>m\\ \ell\nmid r}}\frac{1}{1-q^{r}} ∏ r > m ℓ ∤ r 1 − q r 1 , and the overlined parts contribute ∏ r ≥ 1 ℓ ∤ r ( 1 + q r ) \prod_{\substack{r\ge1\\ \ell\nmid r}}(1+q^{r}) ∏ r ≥ 1 ℓ ∤ r ( 1 + q r ) . The only admissible r r r missing from the denominators is r = m r=m r = m . Thus
∑ n = 0 ∞ p ‾ ℓ mex ( m , n ) q n = q R ℓ ( m ) ( 1 − q m ) ∏ r ≥ 1 ℓ ∤ r 1 1 − q r ∏ r ≥ 1 ℓ ∤ r ( 1 + q r ) , ℓ ∤ m . (14) \sum_{n=0}^{\infty}\overline{p}_{\ell}^{\operatorname{mex}}(m,n)q^{n}
=q^{R_{\ell}(m)}(1-q^{m})\prod_{\substack{r\ge1\\ \ell\nmid r}}\frac{1}{1-q^{r}}\prod_{\substack{r\ge1\\ \ell\nmid r}}(1+q^{r}),\qquad\ell\nmid m. \tag*{(14)} n = 0 ∑ ∞ p ℓ mex ( m , n ) q n = q R ℓ ( m ) ( 1 − q m ) r ≥ 1 ℓ ∤ r ∏ 1 − q r 1 r ≥ 1 ℓ ∤ r ∏ ( 1 + q r ) , ℓ ∤ m . (14)
By the generating function of p ‾ ℓ ( n ) \overline{p}_{\ell}(n) p ℓ ( n ) given in § 1 ,
∑ n = 0 ∞ p ‾ ℓ mex ( m , n ) q n = q R ℓ ( m ) f 2 ℓ f ℓ 2 f 1 2 f 2 ℓ ( 1 − q m ) , ℓ ∤ m . (15) \sum_{n=0}^{\infty}\overline{p}_{\ell}^{\operatorname{mex}}(m,n)q^{n}
=q^{R_{\ell}(m)}\frac{f_{2\ell}f_{\ell}^{2}}{f_{1}^{2}f_{2\ell}}(1-q^{m}),\qquad\ell\nmid m. \tag*{(15)} n = 0 ∑ ∞ p ℓ mex ( m , n ) q n = q R ℓ ( m ) f 1 2 f 2 ℓ f 2 ℓ f ℓ 2 ( 1 − q m ) , ℓ ∤ m . (15)
Consequently, multiplying (15) by z m z^{m} z m and summing over m ≥ 1 m\ge1 m ≥ 1 , ℓ ∤ m \ell\nmid m ℓ ∤ m , we have (13) . □ \square □
Proof of Theorem 1.1 and 1.2.
Proof of Theorem 1.1. Let p ‾ 2 mex ( r , n ) \overline{p}_{2}^{\operatorname{mex}}(r,n) p 2 mex ( r , n ) be the number of 2-regular overpartitions π \pi π of n n n such that mex ‾ 2 ( π ) = 2 r − 1 \overline{\operatorname{mex}}_{2}(\pi)=2r-1 mex 2 ( π ) = 2 r − 1 . We have
H 2 ( z , q ) : = ∑ n ≥ 0 ∑ r ≥ 1 p ‾ 2 mex ( r , n ) z 2 r − 1 q n = ∑ r = 1 ∞ z 2 r − 1 ∑ n = 0 ∞ p ‾ 2 mex ( r , n ) q n . (16) H_{2}(z,q):=\sum_{n\ge0}\sum_{r\ge1}\overline{p}_{2}^{\operatorname{mex}}(r,n)z^{2r-1}q^{n}
=\sum_{r=1}^{\infty}z^{2r-1}\sum_{n=0}^{\infty}\overline{p}_{2}^{\operatorname{mex}}(r,n)q^{n}. \tag*{(16)} H 2 ( z , q ) := n ≥ 0 ∑ r ≥ 1 ∑ p 2 mex ( r , n ) z 2 r − 1 q n = r = 1 ∑ ∞ z 2 r − 1 n = 0 ∑ ∞ p 2 mex ( r , n ) q n . (16)
Since mex ‾ 2 ( π ) = 2 r − 1 \overline{\operatorname{mex}}_{2}(\pi)=2r-1 mex 2 ( π ) = 2 r − 1 , the compulsory non-overlined parts are 1 , 3 , 5 , … , 2 r − 3 1,3,5,\ldots,2r-3 1 , 3 , 5 , … , 2 r − 3 , and their sum is
1 + 3 + ⋯ + ( 2 r − 3 ) = ( r − 1 ) 2 . 1+3+\cdots+(2r-3)=(r-1)^{2}. 1 + 3 + ⋯ + ( 2 r − 3 ) = ( r − 1 ) 2 .
Hence these compulsory parts contribute q ( r − 1 ) 2 ∏ j = 1 r − 1 1 1 − q 2 j − 1 q^{(r-1)^{2}}\prod_{j=1}^{r-1}\frac{1}{1-q^{2j-1}} q ( r − 1 ) 2 ∏ j = 1 r − 1 1 − q 2 j − 1 1 . The non-overlined part 2 r − 1 2r-1 2 r − 1 is excluded, while every larger odd part may occur as a non-overlined part with arbitrary multiplicity. The overlined parts form a partition into distinct odd parts, which contributes ( − q ; q 2 ) ∞ (-q;q^{2})_{\infty} ( − q ; q 2 ) ∞ . Thus,
∑ n = 0 ∞ p ‾ 2 mex ( r , n ) q n = q ( r − 1 ) 2 ∏ j = 1 r − 1 1 1 − q 2 j − 1 ∏ j = r + 1 ∞ 1 1 − q 2 j − 1 ( − q ; q 2 ) ∞ = q ( r − 1 ) 2 ( 1 − q 2 r − 1 ) ( − q ; q 2 ) ∞ ( q ; q 2 ) ∞ . (17) \sum_{n=0}^{\infty}\overline{p}_{2}^{\operatorname{mex}}(r,n)q^{n}
=q^{(r-1)^{2}}\prod_{j=1}^{r-1}\frac{1}{1-q^{2j-1}}\prod_{j=r+1}^{\infty}\frac{1}{1-q^{2j-1}}(-q;q^{2})_{\infty}
=\frac{q^{(r-1)^{2}}(1-q^{2r-1})(-q;q^{2})_{\infty}}{(q;q^{2})_{\infty}}. \tag*{(17)} n = 0 ∑ ∞ p 2 mex ( r , n ) q n = q ( r − 1 ) 2 j = 1 ∏ r − 1 1 − q 2 j − 1 1 j = r + 1 ∏ ∞ 1 − q 2 j − 1 1 ( − q ; q 2 ) ∞ = ( q ; q 2 ) ∞ q ( r − 1 ) 2 ( 1 − q 2 r − 1 ) ( − q ; q 2 ) ∞ . (17)
Combining (16) and (17) , we get
H 2 ( z , q ) = ( − q ; q 2 ) ∞ ( q ; q 2 ) ∞ ∑ r = 1 ∞ z 2 r − 1 q ( r − 1 ) 2 ( 1 − q 2 r − 1 ) . (18) H_2(z,q)=\frac{(-q;q^2)_\infty}{(q;q^2)_\infty}\sum_{r=1}^{\infty}z^{2r-1}q^{(r-1)^2}(1-q^{2r-1}). \tag*{(18)} H 2 ( z , q ) = ( q ; q 2 ) ∞ ( − q ; q 2 ) ∞ r = 1 ∑ ∞ z 2 r − 1 q ( r − 1 ) 2 ( 1 − q 2 r − 1 ) . (18)
Differentiating H 2 ( z , q ) H_2(z,q) H 2 ( z , q ) with respect to z z z and putting z = 1 z=1 z = 1 , we get
∑ n ≥ 0 σ ‾ 2 max ( n ) q n = ∂ ∂ z H 2 ( z , q ) ∣ z = 1 = ( − q ; q 2 ) ∞ ( q ; q 2 ) ∞ ∑ r = 1 ∞ ( 2 r − 1 ) q ( r − 1 ) 2 ( 1 − q 2 r − 1 ) = ( − q ; q 2 ) ∞ ( q ; q 2 ) ∞ [ ∑ r = 1 ∞ ( 2 r − 1 ) q ( r − 1 ) 2 − ∑ r = 1 ∞ ( 2 r − 1 ) q ( r − 1 ) 2 q 2 r − 1 ] = ( − q ; q 2 ) ∞ ( q ; q 2 ) ∞ [ ∑ t = 0 ∞ ( 2 t + 1 ) q t 2 − ∑ t = 0 ∞ ( 2 t + 1 ) q ( t + 1 ) 2 ] = ( − q ; q 2 ) ∞ ( q ; q 2 ) ∞ [ ∑ t = 0 ∞ ( 2 t + 1 ) q t 2 − ∑ t = 1 ∞ ( 2 t − 1 ) q t 2 ] = ( − q ; q 2 ) ∞ ( q ; q 2 ) ∞ [ 1 + 2 ∑ t = 1 ∞ q t 2 ] = ( − q ; q 2 ) ∞ ( q ; q 2 ) ∞ ∑ t ∈ Z q t 2 . (19) \begin{aligned}
\sum_{n\geq0}\overline{\sigma}_2^{\max}(n)q^n
&=\left.\frac{\partial}{\partial z}H_2(z,q)\right|_{z=1}\\
&=\frac{(-q;q^2)_\infty}{(q;q^2)_\infty}\sum_{r=1}^{\infty}(2r-1)q^{(r-1)^2}(1-q^{2r-1})\\
&=\frac{(-q;q^2)_\infty}{(q;q^2)_\infty}\left[\sum_{r=1}^{\infty}(2r-1)q^{(r-1)^2}-\sum_{r=1}^{\infty}(2r-1)q^{(r-1)^2}q^{2r-1}\right]\\
&=\frac{(-q;q^2)_\infty}{(q;q^2)_\infty}\left[\sum_{t=0}^{\infty}(2t+1)q^{t^2}-\sum_{t=0}^{\infty}(2t+1)q^{(t+1)^2}\right]\\
&=\frac{(-q;q^2)_\infty}{(q;q^2)_\infty}\left[\sum_{t=0}^{\infty}(2t+1)q^{t^2}-\sum_{t=1}^{\infty}(2t-1)q^{t^2}\right]\\
&=\frac{(-q;q^2)_\infty}{(q;q^2)_\infty}\left[1+2\sum_{t=1}^{\infty}q^{t^2}\right]\\
&=\frac{(-q;q^2)_\infty}{(q;q^2)_\infty}\sum_{t\in\mathbb{Z}}q^{t^2}. \tag*{(19)}
\end{aligned} n ≥ 0 ∑ σ 2 m a x ( n ) q n = ∂ z ∂ H 2 ( z , q ) z = 1 = ( q ; q 2 ) ∞ ( − q ; q 2 ) ∞ r = 1 ∑ ∞ ( 2 r − 1 ) q ( r − 1 ) 2 ( 1 − q 2 r − 1 ) = ( q ; q 2 ) ∞ ( − q ; q 2 ) ∞ [ r = 1 ∑ ∞ ( 2 r − 1 ) q ( r − 1 ) 2 − r = 1 ∑ ∞ ( 2 r − 1 ) q ( r − 1 ) 2 q 2 r − 1 ] = ( q ; q 2 ) ∞ ( − q ; q 2 ) ∞ [ t = 0 ∑ ∞ ( 2 t + 1 ) q t 2 − t = 0 ∑ ∞ ( 2 t + 1 ) q ( t + 1 ) 2 ] = ( q ; q 2 ) ∞ ( − q ; q 2 ) ∞ [ t = 0 ∑ ∞ ( 2 t + 1 ) q t 2 − t = 1 ∑ ∞ ( 2 t − 1 ) q t 2 ] = ( q ; q 2 ) ∞ ( − q ; q 2 ) ∞ [ 1 + 2 t = 1 ∑ ∞ q t 2 ] = ( q ; q 2 ) ∞ ( − q ; q 2 ) ∞ t ∈ Z ∑ q t 2 . (19)
Since ( − q ; q 2 ) ∞ = f 2 2 f 1 f 4 (-q;q^2)_\infty=\frac{f_2^2}{f_1f_4} ( − q ; q 2 ) ∞ = f 1 f 4 f 2 2 and ( q ; q 2 ) ∞ = f 1 f 2 (q;q^2)_\infty=\frac{f_1}{f_2} ( q ; q 2 ) ∞ = f 2 f 1 , Lemma 2.1 gives
∑ n ≥ 0 σ ‾ 2 max ( n ) q n = f 2 3 f 1 2 f 4 ⋅ f 2 5 f 1 2 f 2 2 = f 2 8 f 1 4 f 4 3 . (20) \sum_{n\geq0}\overline{\sigma}_2^{\max}(n)q^n=\frac{f_2^3}{f_1^2f_4}\cdot\frac{f_2^5}{f_1^2f_2^2}=\frac{f_2^8}{f_1^4f_4^3}. \tag*{(20)} n ≥ 0 ∑ σ 2 m a x ( n ) q n = f 1 2 f 4 f 2 3 ⋅ f 1 2 f 2 2 f 2 5 = f 1 4 f 4 3 f 2 8 . (20)
In order to prove Theorem 1.2, we need the following proposition.
Proof of Theorem 1.2. From Proposition 1, we have
H 3 ( z , q ) : = ∑ n ≥ 0 m ≥ 1 3 ∣ m p 3 max ( m , n ) z m q n = f 2 f 3 2 f 1 2 f 6 ∑ m ≥ 1 3 ∣ m z m q R 3 ( m ) ( 1 − q m ) . (21) H_3(z,q):=\sum_{\substack{n\geq0\\m\geq1\\3\mid m}}p_3^{\max}(m,n)z^mq^n=\frac{f_2f_3^2}{f_1^2f_6}\sum_{\substack{m\geq1\\3\mid m}}z^mq^{R_3(m)}(1-q^m). \tag*{(21)} H 3 ( z , q ) := n ≥ 0 m ≥ 1 3 ∣ m ∑ p 3 m a x ( m , n ) z m q n = f 1 2 f 6 f 2 f 3 2 m ≥ 1 3 ∣ m ∑ z m q R 3 ( m ) ( 1 − q m ) . (21)
Using Lemma 2.6 in (21) , we get
H 3 ( z , q ) = f 2 f 3 2 f 1 2 f 6 [ ∑ r ≥ 0 z 3 r + 1 q 3 r 2 ( 1 − q 3 r + 1 ) + ∑ r ≥ 0 z 3 r + 2 q 3 r 2 + 3 r + 1 ( 1 − q 3 r + 2 ) ] . (22) H_3(z,q)=\frac{f_2f_3^2}{f_1^2f_6}\left[\sum_{r\geq0}z^{3r+1}q^{3r^2}(1-q^{3r+1})+\sum_{r\geq0}z^{3r+2}q^{3r^2+3r+1}(1-q^{3r+2})\right]. \tag*{(22)} H 3 ( z , q ) = f 1 2 f 6 f 2 f 3 2 [ r ≥ 0 ∑ z 3 r + 1 q 3 r 2 ( 1 − q 3 r + 1 ) + r ≥ 0 ∑ z 3 r + 2 q 3 r 2 + 3 r + 1 ( 1 − q 3 r + 2 ) ] . (22)
Differentiating (22) with respect to z z z and putting z = 1 z=1 z = 1 , we obtain
∑ n ≥ 0 σ ‾ 3 max ( n ) q n = ∂ ∂ z H 3 ( z , q ) ∣ z = 1 = f 2 f 3 2 f 1 2 f 6 N 3 ( q ) . (23) \sum_{n\geq0}\overline{\sigma}_3^{\max}(n)q^n=\left.\frac{\partial}{\partial z}H_3(z,q)\right|_{z=1}=\frac{f_2f_3^2}{f_1^2f_6}N_3(q). \tag*{(23)} n ≥ 0 ∑ σ 3 m a x ( n ) q n = ∂ z ∂ H 3 ( z , q ) z = 1 = f 1 2 f 6 f 2 f 3 2 N 3 ( q ) . (23)
where
N 3 ( q ) = ∑ r ≥ 0 ( 3 r + 1 ) q 3 r 2 ( 1 − q 3 r + 1 ) + ∑ r ≥ 0 ( 3 r + 2 ) q 3 r 2 + 3 r + 1 ( 1 − q 3 r + 2 ) = ∑ r ≥ 0 ( 3 r + 1 ) q 3 r 2 + ∑ r ≥ 0 q 3 r 2 + 3 r + 1 − ∑ r ≥ 0 ( 3 r + 2 ) q 3 ( r + 1 ) 2 = ∑ r ≥ 0 ( 3 r + 1 ) q 3 r 2 − ∑ r ≥ 1 ( 3 r − 1 ) q 3 r 2 + ∑ r ≥ 0 q 3 r 2 + 3 r + 1 . (24) \begin{aligned}
N_3(q)&=\sum_{r\ge0}(3r+1)q^{3r^2}(1-q^{3r+1})+\sum_{r\ge0}(3r+2)q^{3r^2+3r+1}(1-q^{3r+2})\\
&=\sum_{r\ge0}(3r+1)q^{3r^2}+\sum_{r\ge0}q^{3r^2+3r+1}-\sum_{r\ge0}(3r+2)q^{3(r+1)^2}\\
&=\sum_{r\ge0}(3r+1)q^{3r^2}-\sum_{r\ge1}(3r-1)q^{3r^2}+\sum_{r\ge0}q^{3r^2+3r+1}. \tag*{(24)}
\end{aligned} N 3 ( q ) = r ≥ 0 ∑ ( 3 r + 1 ) q 3 r 2 ( 1 − q 3 r + 1 ) + r ≥ 0 ∑ ( 3 r + 2 ) q 3 r 2 + 3 r + 1 ( 1 − q 3 r + 2 ) = r ≥ 0 ∑ ( 3 r + 1 ) q 3 r 2 + r ≥ 0 ∑ q 3 r 2 + 3 r + 1 − r ≥ 0 ∑ ( 3 r + 2 ) q 3 ( r + 1 ) 2 = r ≥ 0 ∑ ( 3 r + 1 ) q 3 r 2 − r ≥ 1 ∑ ( 3 r − 1 ) q 3 r 2 + r ≥ 0 ∑ q 3 r 2 + 3 r + 1 . (24)
= 1 + 2 ∑ r ≥ 1 q 3 r 2 + ∑ r ≥ 0 q 3 r 2 + 3 r + 1 = ∑ r ∈ Z q 3 r 2 + ∑ r ≥ 0 q 3 r 2 + 3 r + 1 . (25) =1+2\sum_{r\ge1}q^{3r^2}+\sum_{r\ge0}q^{3r^2+3r+1}=\sum_{r\in\mathbb{Z}}q^{3r^2}+\sum_{r\ge0}q^{3r^2+3r+1}. \tag*{(25)} = 1 + 2 r ≥ 1 ∑ q 3 r 2 + r ≥ 0 ∑ q 3 r 2 + 3 r + 1 = r ∈ Z ∑ q 3 r 2 + r ≥ 0 ∑ q 3 r 2 + 3 r + 1 . (25)
Here, in the second line we combined − ( 3 r + 1 ) q 3 r 2 + 3 r + 1 + ( 3 r + 2 ) q 3 r 2 + 3 r + 1 = q 3 r 2 + 3 r + 1 -(3r+1)q^{3r^2+3r+1}+(3r+2)q^{3r^2+3r+1}=q^{3r^2+3r+1} − ( 3 r + 1 ) q 3 r 2 + 3 r + 1 + ( 3 r + 2 ) q 3 r 2 + 3 r + 1 = q 3 r 2 + 3 r + 1 , and used 3 r 2 + 3 r + 1 + ( 3 r + 2 ) = 3 ( r + 1 ) 2 3r^2+3r+1+(3r+2)=3(r+1)^2 3 r 2 + 3 r + 1 + ( 3 r + 2 ) = 3 ( r + 1 ) 2 .
Note that, for any real number r r r ,
( − r − 1 ) 2 + ( − r − 1 ) = r 2 + r . (26) (-r-1)^2+(-r-1)=r^2+r. \tag*{(26)} ( − r − 1 ) 2 + ( − r − 1 ) = r 2 + r . (26)
Using (26) , we get
∑ r ≥ 0 q 3 r 2 + 3 r + 1 = 1 2 ∑ r ∈ Z q 3 r 2 + 3 r + 1 = q 2 ∑ r = − ∞ ∞ ( q 3 ) r 2 ( q 3 ) r . (27) \sum_{r\ge0}q^{3r^2+3r+1}=\frac{1}{2}\sum_{r\in\mathbb{Z}}q^{3r^2+3r+1}=\frac{q}{2}\sum_{r=-\infty}^{\infty}(q^3)^{r^2}(q^3)^r. \tag*{(27)} r ≥ 0 ∑ q 3 r 2 + 3 r + 1 = 2 1 r ∈ Z ∑ q 3 r 2 + 3 r + 1 = 2 q r = − ∞ ∑ ∞ ( q 3 ) r 2 ( q 3 ) r . (27)
Applying Q = q 3 Q=q^3 Q = q 3 and z = q 3 z=q^3 z = q 3 in Lemma 2.2, we have
∑ r = − ∞ ∞ q 3 r 2 + 3 r = ( − q 6 ; q 6 ) ∞ ( − 1 ; q 6 ) ∞ ( q 6 ; q 6 ) ∞ = 2 ( − q 6 ; q 6 ) ∞ 2 ( q 6 ; q 6 ) ∞ , (28) \sum_{r=-\infty}^{\infty}q^{3r^2+3r}=(-q^6;q^6)_\infty(-1;q^6)_\infty(q^6;q^6)_\infty=2(-q^6;q^6)_\infty^2(q^6;q^6)_\infty, \tag*{(28)} r = − ∞ ∑ ∞ q 3 r 2 + 3 r = ( − q 6 ; q 6 ) ∞ ( − 1 ; q 6 ) ∞ ( q 6 ; q 6 ) ∞ = 2 ( − q 6 ; q 6 ) ∞ 2 ( q 6 ; q 6 ) ∞ , (28)
since ( − 1 ; q 6 ) ∞ = 2 ( − q 6 ; q 6 ) ∞ (-1;q^6)_\infty=2(-q^6;q^6)_\infty ( − 1 ; q 6 ) ∞ = 2 ( − q 6 ; q 6 ) ∞ . Further observe that
( − q 6 ; q 6 ) ∞ = ( q 12 ; q 12 ) ∞ ( q 6 ; q 6 ) ∞ . (29) (-q^6;q^6)_\infty=\frac{(q^{12};q^{12})_\infty}{(q^6;q^6)_\infty}. \tag*{(29)} ( − q 6 ; q 6 ) ∞ = ( q 6 ; q 6 ) ∞ ( q 12 ; q 12 ) ∞ . (29)
Using (28) and (29) in (27) , we get
∑ r ≥ 0 q 3 r 2 + 3 r + 1 = q ( q 12 ; q 12 ) ∞ 2 ( q 6 ; q 6 ) ∞ . (30) \sum_{r\ge0}q^{3r^2+3r+1}=q\frac{(q^{12};q^{12})_\infty^2}{(q^6;q^6)_\infty}. \tag*{(30)} r ≥ 0 ∑ q 3 r 2 + 3 r + 1 = q ( q 6 ; q 6 ) ∞ ( q 12 ; q 12 ) ∞ 2 . (30)
Also, using Lemma 2.1 with q → q 3 q\to q^3 q → q 3 , we have
∑ r ∈ Z q 3 r 2 = ( q 6 ; q 6 ) ∞ 5 ( q 3 ; q 3 ) ∞ 2 ( q 12 ; q 12 ) ∞ 2 . (31) \sum_{r\in\mathbb{Z}}q^{3r^2}=\frac{(q^6;q^6)_\infty^5}{(q^3;q^3)_\infty^2(q^{12};q^{12})_\infty^2}. \tag*{(31)} r ∈ Z ∑ q 3 r 2 = ( q 3 ; q 3 ) ∞ 2 ( q 12 ; q 12 ) ∞ 2 ( q 6 ; q 6 ) ∞ 5 . (31)
Thus, by (23) , (25) , (30) and (31) ,
∑ n ≥ 0 σ ‾ 3 max ( n ) q n = f 2 f 3 2 f 1 2 f 6 [ f 6 5 f 3 2 f 2 2 + q f 2 2 f 6 2 ] = f 2 f 6 4 f 1 2 f 12 + q f 2 f 3 2 f 12 2 f 1 2 f 6 2 . (32) \begin{aligned}
\sum_{n\ge0}\overline{\sigma}_3^{\max}(n)q^n
&=\frac{f_2f_3^2}{f_1^2f_6}\left[\frac{f_6^5}{f_3^2f_2^2}+q\frac{f_2^2}{f_6^2}\right]\\
&=\frac{f_2f_6^4}{f_1^2f_{12}}+q\frac{f_2f_3^2f_{12}^2}{f_1^2f_6^2}. \tag*{(32)}
\end{aligned} n ≥ 0 ∑ σ 3 m a x ( n ) q n = f 1 2 f 6 f 2 f 3 2 [ f 3 2 f 2 2 f 6 5 + q f 6 2 f 2 2 ] = f 1 2 f 12 f 2 f 6 4 + q f 1 2 f 6 2 f 2 f 3 2 f 12 2 . (32)
Proof of Theorem 1.3.
Proof. Differentiating (13) with respect to z z z and putting z = 1 z=1 z = 1 , we get
∑ n ≥ 0 σ ℓ mex ‾ ( n ) q n = f 2 f ℓ f 1 2 f 2 ℓ ∑ m ≥ 1 ℓ ∤ m m q R ℓ ( m ) ( 1 − q m ) . (33) \sum_{n\geq0}\sigma_{\ell}^{\overline{\operatorname{mex}}}(n)q^n=\frac{f_2f_{\ell}}{f_1^2f_{2\ell}}\sum_{\substack{m\geq1\\ \ell\nmid m}}m q^{R_{\ell}(m)}(1-q^m). \tag*{(33)} n ≥ 0 ∑ σ ℓ mex ( n ) q n = f 1 2 f 2 ℓ f 2 f ℓ m ≥ 1 ℓ ∤ m ∑ m q R ℓ ( m ) ( 1 − q m ) . (33)
For ℓ \ell ℓ -regular partitions there are no overlined parts, so the argument in the proof of Proposition 1 , with the factor
∏ r ≥ 1 ℓ ∤ r ( 1 + q r ) \prod_{\substack{r\geq1\\ \ell\nmid r}}(1+q^r) r ≥ 1 ℓ ∤ r ∏ ( 1 + q r )
is removed. Therefore
∑ n ≥ 0 σ ℓ mex ‾ ( n ) q n = f ℓ f 1 ∑ m ≥ 1 ℓ ∤ m m q R ℓ ( m ) ( 1 − q m ) . \sum_{n\geq0}\sigma_{\ell}^{\overline{\operatorname{mex}}}(n)q^n=\frac{f_{\ell}}{f_1}\sum_{\substack{m\geq1\\ \ell\nmid m}}m q^{R_{\ell}(m)}(1-q^m). n ≥ 0 ∑ σ ℓ mex ( n ) q n = f 1 f ℓ m ≥ 1 ℓ ∤ m ∑ m q R ℓ ( m ) ( 1 − q m ) .
Comparing with (33) , we obtain
∑ n ≥ 0 σ ℓ mex ‾ ( n ) q n = f 2 f ℓ f 1 f 2 ℓ ∑ n ≥ 0 σ ℓ mex ( n ) q n . (34) \sum_{n\geq0}\sigma_{\ell}^{\overline{\operatorname{mex}}}(n)q^n=\frac{f_2f_{\ell}}{f_1f_{2\ell}}\sum_{n\geq0}\sigma_{\ell}^{\operatorname{mex}}(n)q^n. \tag*{(34)} n ≥ 0 ∑ σ ℓ mex ( n ) q n = f 1 f 2 ℓ f 2 f ℓ n ≥ 0 ∑ σ ℓ mex ( n ) q n . (34)
Substituting Lemma 2.7 in (34) , and using
f 2 f ℓ f 1 f 2 ℓ ⋅ f ℓ f ℓ ( ℓ − 1 ) f 1 = f 2 f ℓ 2 f ℓ ( ℓ − 1 ) f 1 2 f 2 ℓ , \frac{f_2f_{\ell}}{f_1f_{2\ell}}\cdot\frac{f_{\ell}f_{\ell(\ell-1)}}{f_1}=\frac{f_2f_{\ell}^2f_{\ell(\ell-1)}}{f_1^2f_{2\ell}}, f 1 f 2 ℓ f 2 f ℓ ⋅ f 1 f ℓ f ℓ ( ℓ − 1 ) = f 1 2 f 2 ℓ f 2 f ℓ 2 f ℓ ( ℓ − 1 ) ,
we obtain (4) for even ℓ \ell ℓ and (5) for odd ℓ \ell ℓ . □
Proof of Theorem 1.4. Consider the map ϕ : P ‾ ℓ ( n ) → P ‾ ℓ ( n + 1 ) \phi:\overline{\mathcal{P}}_{\ell}(n)\to\overline{\mathcal{P}}_{\ell}(n+1) ϕ : P ℓ ( n ) → P ℓ ( n + 1 ) defined by ϕ ( π ) = π ∪ { 1 } \phi(\pi)=\pi\cup\{1\} ϕ ( π ) = π ∪ { 1 } , where the new part 1 1 1 is non-overlined. Since 1 1 1 is never divisible by ℓ \ell ℓ , ϕ ( π ) \phi(\pi) ϕ ( π ) is still an ℓ \ell ℓ -regular overpartition, and ∣ ϕ ( π ) ∣ = ∣ π ∣ + 1 = n + 1 |\phi(\pi)|=|\pi|+1=n+1 ∣ ϕ ( π ) ∣ = ∣ π ∣ + 1 = n + 1 . If π \pi π already contains 1 1 1 , the new part is simply an additional non-overlined 1 1 1 , which is allowed. The map is injective: if ϕ ( π 1 ) = ϕ ( π 2 ) \phi(\pi_1)=\phi(\pi_2) ϕ ( π 1 ) = ϕ ( π 2 ) , then removing one non-overlined 1 1 1 from both gives π 1 = π 2 \pi_1=\pi_2 π 1 = π 2 .
Now we consider two cases. Firstly, suppose that 1 1 1 is a non-overlined part of π \pi π . Then mex ‾ ℓ ( π ) > 1 \overline{\operatorname{mex}}_{\ell}(\pi)>1 mex ℓ ( π ) > 1 . Adding another non-overlined 1 1 1 does not change whether any other part occurs as a non-overlined part. Hence mex ‾ ℓ ( ϕ ( π ) ) = mex ‾ ℓ ( π ) \overline{\operatorname{mex}}_{\ell}(\phi(\pi))=\overline{\operatorname{mex}}_{\ell}(\pi) mex ℓ ( ϕ ( π )) = mex ℓ ( π ) . Secondly, suppose that 1 1 1 is not a non-overlined part of π \pi π . Then mex ‾ ℓ ( π ) = 1 \overline{\operatorname{mex}}_{\ell}(\pi)=1 mex ℓ ( π ) = 1 , and after adding 1 1 1 , 1 1 1 is a non-overlined part of ϕ ( π ) \phi(\pi) ϕ ( π ) . Therefore mex ‾ ℓ ( ϕ ( π ) ) \overline{\operatorname{mex}}_{\ell}(\phi(\pi)) mex ℓ ( ϕ ( π )) is the smallest positive integer other than 1 1 1 that is not divisible by ℓ \ell ℓ and does not occur as a non-overlined part. Thus mex ‾ ℓ ( ϕ ( π ) ) > 1 = mex ‾ ℓ ( π ) \overline{\operatorname{mex}}_{\ell}(\phi(\pi))>1=\overline{\operatorname{mex}}_{\ell}(\pi) mex ℓ ( ϕ ( π )) > 1 = mex ℓ ( π ) . In both cases mex ‾ ℓ ( ϕ ( π ) ) ≥ mex ‾ ℓ ( π ) \overline{\operatorname{mex}}_{\ell}(\phi(\pi))\geq\overline{\operatorname{mex}}_{\ell}(\pi) mex ℓ ( ϕ ( π )) ≥ mex ℓ ( π ) . Since ϕ \phi ϕ is injective, we have
σ ℓ mex ‾ ( n + 1 ) = ∑ μ ∈ P ‾ ℓ ( n + 1 ) mex ‾ ℓ ( μ ) ≥ ∑ π ∈ P ‾ ℓ ( n ) mex ‾ ℓ ( ϕ ( π ) ) ≥ ∑ π ∈ P ‾ ℓ ( n ) mex ‾ ℓ ( π ) = σ ℓ mex ‾ ( n ) . \sigma_{\ell}^{\overline{\operatorname{mex}}}(n+1)=\sum_{\mu\in\overline{\mathcal{P}}_{\ell}(n+1)}\overline{\operatorname{mex}}_{\ell}(\mu)\geq\sum_{\pi\in\overline{\mathcal{P}}_{\ell}(n)}\overline{\operatorname{mex}}_{\ell}(\phi(\pi))\geq\sum_{\pi\in\overline{\mathcal{P}}_{\ell}(n)}\overline{\operatorname{mex}}_{\ell}(\pi)=\sigma_{\ell}^{\overline{\operatorname{mex}}}(n). σ ℓ mex ( n + 1 ) = μ ∈ P ℓ ( n + 1 ) ∑ mex ℓ ( μ ) ≥ π ∈ P ℓ ( n ) ∑ mex ℓ ( ϕ ( π )) ≥ π ∈ P ℓ ( n ) ∑ mex ℓ ( π ) = σ ℓ mex ( n ) .
Hence σ ℓ mex ‾ ( n + 1 ) ≥ σ ℓ mex ‾ ( n ) \sigma_{\ell}^{\overline{\operatorname{mex}}}(n+1)\geq\sigma_{\ell}^{\overline{\operatorname{mex}}}(n) σ ℓ mex ( n + 1 ) ≥ σ ℓ mex ( n ) for all n ≥ 0 n\geq0 n ≥ 0 . Therefore, the sequence { σ ℓ mex ‾ ( n ) } n ≥ 0 \{\sigma_{\ell}^{\overline{\operatorname{mex}}}(n)\}_{n\geq0} { σ ℓ mex ( n ) } n ≥ 0 is weakly increasing for every ℓ ≥ 2 \ell\geq2 ℓ ≥ 2 .
Every overpartition π ∈ P ‾ ℓ ( n ) \pi\in\overline{\mathcal{P}}_{\ell}(n) π ∈ P ℓ ( n ) without a non-overlined 1 1 1 satisfies mex ‾ ℓ ( π ) = 1 \overline{\operatorname{mex}}_{\ell}(\pi)=1 mex ℓ ( π ) = 1 . After adding a part 1 1 1 , its mex becomes strictly larger, i.e. mex ‾ ℓ ( ϕ ( π ) ) > 1 = mex ‾ ℓ ( π ) \overline{\operatorname{mex}}_{\ell}(\phi(\pi))>1=\overline{\operatorname{mex}}_{\ell}(\pi) mex ℓ ( ϕ ( π )) > 1 = mex ℓ ( π ) . If 1 1 1 is a non-overlined part of π \pi π , then mex ‾ ℓ ( ϕ ( π ) ) = mex ‾ ℓ ( π ) \overline{\operatorname{mex}}_{\ell}(\phi(\pi))=\overline{\operatorname{mex}}_{\ell}(\pi) mex ℓ ( ϕ ( π )) = mex ℓ ( π ) . Hence
σ ℓ mex ‾ ( n + 1 ) − σ ℓ mex ‾ ( n ) ≥ # { π ∈ P ‾ ℓ ( n ) : 1 is not a non-overlined part of π } . (35) \sigma_{\ell}^{\overline{\operatorname{mex}}}(n+1)-\sigma_{\ell}^{\overline{\operatorname{mex}}}(n)\geq\#\{\pi\in\overline{\mathcal{P}}_{\ell}(n):1\text{ is not a non-overlined part of }\pi\}. \tag*{(35)} σ ℓ mex ( n + 1 ) − σ ℓ mex ( n ) ≥ # { π ∈ P ℓ ( n ) : 1 is not a non-overlined part of π } . (35)
Now consider the overpartitions in P ‾ ℓ ( n ) \overline{\mathcal{P}}_{\ell}(n) P ℓ ( n ) that contain at least one non-overlined part 1 1 1 . Removing one such 1 1 1 gives a bijection with P ‾ ℓ ( n − 1 ) \overline{\mathcal{P}}_{\ell}(n-1) P ℓ ( n − 1 ) . Therefore,
# { π ∈ P ‾ ℓ ( n ) : 1 is not a non-overlined part of π } = p ‾ ℓ ( n ) − p ‾ ℓ ( n − 1 ) . (36) \#\{\pi\in\overline{\mathcal{P}}_{\ell}(n):1\text{ is not a non-overlined part of }\pi\}=\overline{p}_{\ell}(n)-\overline{p}_{\ell}(n-1). \tag*{(36)} # { π ∈ P ℓ ( n ) : 1 is not a non-overlined part of π } = p ℓ ( n ) − p ℓ ( n − 1 ) . (36)
Comparing (35) and (36) , we get σ 2 max ‾ ( n + 1 ) − σ 2 max ‾ ( n ) ≥ p ‾ ℓ ( n ) − p ‾ ℓ ( n − 1 ) \sigma_2^{\overline{\max}}(n+1)-\sigma_2^{\overline{\max}}(n)\geq\overline{p}_{\ell}(n)-\overline{p}_{\ell}(n-1) σ 2 m a x ( n + 1 ) − σ 2 m a x ( n ) ≥ p ℓ ( n ) − p ℓ ( n − 1 ) .
Proof of Theorem 1.5. Let
G ( q ) = ∑ n ≥ 0 σ 2 max ‾ ( n ) q n = ( q 2 ; q 2 ) ∞ 8 ( q ; q ) ∞ 4 ( q 4 ; q 4 ) ∞ 3 . G(q)=\sum_{n\geq0}\sigma_2^{\overline{\max}}(n)q^n=\frac{(q^2;q^2)_\infty^8}{(q;q)_\infty^4(q^4;q^4)_\infty^3}. G ( q ) = n ≥ 0 ∑ σ 2 m a x ( n ) q n = ( q ; q ) ∞ 4 ( q 4 ; q 4 ) ∞ 3 ( q 2 ; q 2 ) ∞ 8 .
Applying the transformation formula for Dedekind’s eta function [22 ] , we have
( e − y ; e − y ) ∞ ∼ 2 π y exp ( − π 2 6 y ) , y → 0 + . (37) (e^{-y};e^{-y})_\infty\sim\sqrt{\frac{2\pi}{y}}\exp\left(-\frac{\pi^2}{6y}\right),\qquad y\to0^+. \tag*{(37)} ( e − y ; e − y ) ∞ ∼ y 2 π exp ( − 6 y π 2 ) , y → 0 + . (37)
Thus
( e − 2 y ; e − 2 y ) ∞ 4 ∼ ( π y ) 4 e − 2 π 2 3 y , ( e − y ; e − y ) ∞ 4 ∼ ( 2 π y ) 2 e − 2 π 2 3 y , (e^{-2y};e^{-2y})_\infty^4\sim\left(\frac{\pi}{y}\right)^4e^{-\frac{2\pi^2}{3y}},\qquad(e^{-y};e^{-y})_\infty^4\sim\left(\frac{2\pi}{y}\right)^2e^{-\frac{2\pi^2}{3y}}, ( e − 2 y ; e − 2 y ) ∞ 4 ∼ ( y π ) 4 e − 3 y 2 π 2 , ( e − y ; e − y ) ∞ 4 ∼ ( y 2 π ) 2 e − 3 y 2 π 2 ,
and
( e − 4 y ; e − 4 y ) ∞ 3 ∼ ( π 2 y ) 3 / 2 e − π 2 8 y . (e^{-4y};e^{-4y})_\infty^3\sim\left(\frac{\pi}{2y}\right)^{3/2}e^{-\frac{\pi^2}{8y}}. ( e − 4 y ; e − 4 y ) ∞ 3 ∼ ( 2 y π ) 3/2 e − 8 y π 2 .
Therefore,
G ( e − y ) ∼ ( π y ) 4 ( 2 π y ) 2 ( π 2 y ) 3 / 2 e − 2 π 2 3 y + 2 π 2 3 y + π 2 8 y = π 2 y − 1 / 2 e π 2 8 y . G(e^{-y})\sim\frac{\left(\frac{\pi}{y}\right)^4}{\left(\frac{2\pi}{y}\right)^2\left(\frac{\pi}{2y}\right)^{3/2}}e^{-\frac{2\pi^2}{3y}+\frac{2\pi^2}{3y}+\frac{\pi^2}{8y}}=\sqrt{\frac{\pi}{2}}y^{-1/2}e^{\frac{\pi^2}{8y}}. G ( e − y ) ∼ ( y 2 π ) 2 ( 2 y π ) 3/2 ( y π ) 4 e − 3 y 2 π 2 + 3 y 2 π 2 + 8 y π 2 = 2 π y − 1/2 e 8 y π 2 .
From Theorem 1.4 , we know that the sequence { σ 2 max ‾ ( n ) } \{\sigma_2^{\overline{\max}}(n)\} { σ 2 m a x ( n )} is weakly increasing. Thus, by applying Lemma 2.3 with parameters β = π 2 \beta=\sqrt{\frac{\pi}{2}} β = 2 π , γ = − 1 2 \gamma=-\frac{1}{2} γ = − 2 1 , μ = π 2 8 \mu=\frac{\pi^2}{8} μ = 8 π 2 , we obtain
σ 2 max ‾ ( n ) ∼ π / 2 2 π n e π 2 n 8 = 1 2 2 n e π n / 2 , as n → ∞ . \sigma_2^{\overline{\max}}(n)\sim\frac{\sqrt{\pi/2}}{2\sqrt{\pi n}}e^{\sqrt{\frac{\pi^2n}{8}}}=\frac{1}{2\sqrt{2n}}e^{\pi\sqrt{n/2}},\qquad\text{as }n\to\infty. σ 2 m a x ( n ) ∼ 2 π n π /2 e 8 π 2 n = 2 2 n 1 e π n /2 , as n → ∞.
Proof of Theorem 1.6. From (32) , we have
∑ n ≥ 0 σ 3 max ‾ ( n ) q n = ( q 2 ; q 2 ) ∞ ( q 6 ; q 6 ) ∞ 4 ( q ; q ) ∞ 2 ( q 12 ; q 12 ) ∞ 2 + q ( q 2 ; q 2 ) ∞ ( q 3 ; q 3 ) ∞ 2 ( q 12 ; q 12 ) ∞ 2 ( q ; q ) ∞ 2 ( q 6 ; q 6 ) ∞ 2 . \sum_{n\geq0}\sigma_3^{\overline{\max}}(n)q^n=\frac{(q^2;q^2)_\infty(q^6;q^6)_\infty^4}{(q;q)_\infty^2(q^{12};q^{12})_\infty^2}+q\frac{(q^2;q^2)_\infty(q^3;q^3)_\infty^2(q^{12};q^{12})_\infty^2}{(q;q)_\infty^2(q^6;q^6)_\infty^2}. n ≥ 0 ∑ σ 3 m a x ( n ) q n = ( q ; q ) ∞ 2 ( q 12 ; q 12 ) ∞ 2 ( q 2 ; q 2 ) ∞ ( q 6 ; q 6 ) ∞ 4 + q ( q ; q ) ∞ 2 ( q 6 ; q 6 ) ∞ 2 ( q 2 ; q 2 ) ∞ ( q 3 ; q 3 ) ∞ 2 ( q 12 ; q 12 ) ∞ 2 .
Let
S 1 ( q ) : = ( q 2 ; q 2 ) ∞ ( q 6 ; q 6 ) ∞ 4 ( q ; q ) ∞ 2 ( q 12 ; q 12 ) ∞ 2 . S_1(q):=\frac{(q^2;q^2)_\infty(q^6;q^6)_\infty^4}{(q;q)_\infty^2(q^{12};q^{12})_\infty^2}. S 1 ( q ) := ( q ; q ) ∞ 2 ( q 12 ; q 12 ) ∞ 2 ( q 2 ; q 2 ) ∞ ( q 6 ; q 6 ) ∞ 4 .
Using (37) , we get
( e − 2 y ; e − 2 y ) ∞ ( e − 6 y ; e − 6 y ) ∞ 4 ∼ π y ( π 3 y ) 2 e − π 2 12 y − π 2 9 y , (e^{-2y};e^{-2y})_\infty(e^{-6y};e^{-6y})_\infty^4\sim\sqrt{\frac{\pi}{y}}\left(\frac{\pi}{3y}\right)^2e^{-\frac{\pi^2}{12y}-\frac{\pi^2}{9y}}, ( e − 2 y ; e − 2 y ) ∞ ( e − 6 y ; e − 6 y ) ∞ 4 ∼ y π ( 3 y π ) 2 e − 12 y π 2 − 9 y π 2 ,
and
( e − y ; e − y ) ∞ 2 ( e − 12 y ; e − 12 y ) ∞ 2 ∼ 2 π y ⋅ π 6 y e − π 2 3 y − π 2 36 y . (e^{-y};e^{-y})_\infty^2(e^{-12y};e^{-12y})_\infty^2\sim\frac{2\pi}{y}\cdot\frac{\pi}{6y}e^{-\frac{\pi^2}{3y}-\frac{\pi^2}{36y}}. ( e − y ; e − y ) ∞ 2 ( e − 12 y ; e − 12 y ) ∞ 2 ∼ y 2 π ⋅ 6 y π e − 3 y π 2 − 36 y π 2 .
Since 1 12 − 1 9 + 1 3 + 1 36 = 1 6 \frac{1}{12}-\frac{1}{9}+\frac{1}{3}+\frac{1}{36}=\frac{1}{6} 12 1 − 9 1 + 3 1 + 36 1 = 6 1 , we obtain
S 1 ( e − y ) ∼ π 3 y − 1 / 2 e π 2 6 y . S_1(e^{-y})\sim\frac{\sqrt{\pi}}{3}y^{-1/2}e^{\frac{\pi^2}{6y}}. S 1 ( e − y ) ∼ 3 π y − 1/2 e 6 y π 2 .
Similarly, let
S 2 ( q ) : = q ( q 2 ; q 2 ) ∞ ( q 3 ; q 3 ) ∞ 2 ( q 12 ; q 12 ) ∞ 2 ( q ; q ) ∞ 2 ( q 6 ; q 6 ) ∞ 2 . S_2(q):=q\frac{(q^2;q^2)_\infty(q^3;q^3)_\infty^2(q^{12};q^{12})_\infty^2}{(q;q)_\infty^2(q^6;q^6)_\infty^2}. S 2 ( q ) := q ( q ; q ) ∞ 2 ( q 6 ; q 6 ) ∞ 2 ( q 2 ; q 2 ) ∞ ( q 3 ; q 3 ) ∞ 2 ( q 12 ; q 12 ) ∞ 2 .
Using (37) , we get
( e − 2 y ; e − 2 y ) ∞ ( e − 3 y ; e − 3 y ) ∞ 2 ( e − 12 y ; e − 12 y ) ∞ 2 ∼ π y ⋅ 2 π 3 y ⋅ π 6 y e − π 2 12 y − π 2 9 y − π 2 36 y , (e^{-2y};e^{-2y})_\infty(e^{-3y};e^{-3y})_\infty^2(e^{-12y};e^{-12y})_\infty^2\sim\sqrt{\frac{\pi}{y}}\cdot\frac{2\pi}{3y}\cdot\frac{\pi}{6y}e^{-\frac{\pi^2}{12y}-\frac{\pi^2}{9y}-\frac{\pi^2}{36y}}, ( e − 2 y ; e − 2 y ) ∞ ( e − 3 y ; e − 3 y ) ∞ 2 ( e − 12 y ; e − 12 y ) ∞ 2 ∼ y π ⋅ 3 y 2 π ⋅ 6 y π e − 12 y π 2 − 9 y π 2 − 36 y π 2 ,
and
( e − y ; e − y ) ∞ 2 ( e − 6 y ; e − 6 y ) ∞ 2 ∼ 2 π y ⋅ π 3 y e − π 2 3 y − π 2 18 y . \left(e^{-y};e^{-y}\right)_{\infty}^{2}\left(e^{-6y};e^{-6y}\right)_{\infty}^{2}\sim\frac{2\pi}{y}\cdot\frac{\pi}{3y}e^{-\frac{\pi^{2}}{3y}-\frac{\pi^{2}}{18y}}. ( e − y ; e − y ) ∞ 2 ( e − 6 y ; e − 6 y ) ∞ 2 ∼ y 2 π ⋅ 3 y π e − 3 y π 2 − 18 y π 2 .
Since − 1 12 − 1 9 − 1 36 + 1 3 + 1 18 = 1 6 -\frac{1}{12}-\frac{1}{9}-\frac{1}{36}+\frac{1}{3}+\frac{1}{18}=\frac{1}{6} − 12 1 − 9 1 − 36 1 + 3 1 + 18 1 = 6 1 , and since q → 1 q\to1 q → 1 as y → 0 + y\to0^{+} y → 0 + , we get
S 2 ( e − y ) ∼ π 6 y − 1 / 2 e π 2 6 y . S_{2}\left(e^{-y}\right)\sim\frac{\sqrt{\pi}}{6}y^{-1/2}e^{\frac{\pi^{2}}{6y}}. S 2 ( e − y ) ∼ 6 π y − 1/2 e 6 y π 2 .
Note that both terms have the same exponential growth. Consequently,
S 1 ( e − y ) + S 2 ( e − y ) = ∑ n ≥ 0 σ ‾ 3 max ( n ) e − n y ∼ ( π 3 + π 6 ) y − 1 / 2 e π 2 6 y = π 2 y − 1 / 2 e π 2 6 y . S_{1}\left(e^{-y}\right)+S_{2}\left(e^{-y}\right)=\sum_{n\geq0}\overline{\sigma}_{3}^{\max}(n)e^{-ny}\sim\left(\frac{\sqrt{\pi}}{3}+\frac{\sqrt{\pi}}{6}\right)y^{-1/2}e^{\frac{\pi^{2}}{6y}}=\frac{\sqrt{\pi}}{2}y^{-1/2}e^{\frac{\pi^{2}}{6y}}. S 1 ( e − y ) + S 2 ( e − y ) = n ≥ 0 ∑ σ 3 m a x ( n ) e − n y ∼ ( 3 π + 6 π ) y − 1/2 e 6 y π 2 = 2 π y − 1/2 e 6 y π 2 .
Applying Lemma 2.3 with parameters β = π 2 \beta=\frac{\sqrt{\pi}}{2} β = 2 π , γ = − 1 2 \gamma=-\frac{1}{2} γ = − 2 1 , μ = π 2 6 \mu=\frac{\pi^{2}}{6} μ = 6 π 2 , we get
σ ‾ 3 max ( n ) ∼ π / 2 2 π n e 2 π 2 n 6 = 1 4 n e π 2 n / 3 , as n → ∞ . \overline{\sigma}_{3}^{\max}(n)\sim\frac{\sqrt{\pi}/2}{2\sqrt{\pi n}}e^{2\sqrt{\frac{\pi^{2}n}{6}}}=\frac{1}{4\sqrt{n}}e^{\pi\sqrt{2n/3}},\qquad\text{as }n\to\infty. σ 3 m a x ( n ) ∼ 2 π n π /2 e 2 6 π 2 n = 4 n 1 e π 2 n /3 , as n → ∞.
Proof of Theorems 1.7, 1.8 and 1.9.
In order to prove Theorem 1.7, we need the following result.
Proposition 2. For every ℓ ≥ 2 \ell\geq2 ℓ ≥ 2 , we have
P ‾ ℓ ( q ) = f 2 f ℓ 2 f 1 2 f 2 ℓ ≡ 1 ( m o d 2 ) . (38) \overline{P}_{\ell}(q)=\frac{f_{2}f_{\ell}^{2}}{f_{1}^{2}f_{2\ell}}\equiv1\pmod{2}. \tag*{(38)} P ℓ ( q ) = f 1 2 f 2 ℓ f 2 f ℓ 2 ≡ 1 ( mod 2 ) . (38)
Proof. By Lemma 2.4 with p = 2 p=2 p = 2 and j = 1 j=1 j = 1 , f 1 2 ≡ f 2 ( m o d 2 ) f_{1}^{2}\equiv f_{2}\pmod{2} f 1 2 ≡ f 2 ( mod 2 ) and f ℓ 2 ≡ f 2 ℓ ( m o d 2 ) f_{\ell}^{2}\equiv f_{2\ell}\pmod{2} f ℓ 2 ≡ f 2 ℓ ( mod 2 ) . Hence
f 2 f ℓ 2 f 1 2 f 2 ℓ ≡ f 2 f 2 ℓ f 2 f 2 ℓ = 1 ( m o d 2 ) . \frac{f_{2}f_{\ell}^{2}}{f_{1}^{2}f_{2\ell}}\equiv\frac{f_{2}f_{2\ell}}{f_{2}f_{2\ell}}=1\pmod{2}. f 1 2 f 2 ℓ f 2 f ℓ 2 ≡ f 2 f 2 ℓ f 2 f 2 ℓ = 1 ( mod 2 ) .
Proof of Theorem 1.7. (1) From (19) and (20) , we have
∑ n = 0 ∞ σ ‾ 2 max ( n ) q n = P ‾ 2 ( q ) ∑ t ∈ Z q t 2 = P ‾ 2 ( q ) ( 1 + 2 ∑ t ≥ 1 q t 2 ) . (39) \sum_{n=0}^{\infty}\overline{\sigma}_{2}^{\max}(n)q^{n}=\overline{P}_{2}(q)\sum_{t\in\mathbb{Z}}q^{t^{2}}=\overline{P}_{2}(q)\left(1+2\sum_{t\geq1}q^{t^{2}}\right). \tag*{(39)} n = 0 ∑ ∞ σ 2 m a x ( n ) q n = P 2 ( q ) t ∈ Z ∑ q t 2 = P 2 ( q ) ( 1 + 2 t ≥ 1 ∑ q t 2 ) . (39)
By Proposition 2,
∑ n = 0 ∞ σ ‾ 2 max ( n ) q n ≡ 1 ( m o d 2 ) . \sum_{n=0}^{\infty}\overline{\sigma}_{2}^{\max}(n)q^{n}\equiv1\pmod{2}. n = 0 ∑ ∞ σ 2 m a x ( n ) q n ≡ 1 ( mod 2 ) .
Hence σ ‾ 2 max ( n ) \overline{\sigma}_{2}^{\max}(n) σ 2 m a x ( n ) is even for every n ≥ 1 n\geq1 n ≥ 1 .
(2) From (23) , (25) and Proposition 2, we have
∑ n = 0 ∞ σ ‾ 3 max ( n ) q n ≡ ∑ r ∈ Z q 3 r 2 + ∑ r ≥ 0 q 3 r 2 + 3 r + 1 ≡ 1 + ∑ r ≥ 0 q 3 r 2 + 3 r + 1 ( m o d 2 ) , (40) \sum_{n=0}^{\infty}\overline{\sigma}_{3}^{\max}(n)q^{n}\equiv\sum_{r\in\mathbb{Z}}q^{3r^{2}}+\sum_{r\geq0}q^{3r^{2}+3r+1}\equiv1+\sum_{r\geq0}q^{3r^{2}+3r+1}\pmod{2}, \tag*{(40)} n = 0 ∑ ∞ σ 3 m a x ( n ) q n ≡ r ∈ Z ∑ q 3 r 2 + r ≥ 0 ∑ q 3 r 2 + 3 r + 1 ≡ 1 + r ≥ 0 ∑ q 3 r 2 + 3 r + 1 ( mod 2 ) , (40)
since ∑ r ∈ Z q 3 r 2 = 1 + 2 ∑ r ≥ 1 q 3 r 2 \sum_{r\in\mathbb{Z}}q^{3r^{2}}=1+2\sum_{r\geq1}q^{3r^{2}} ∑ r ∈ Z q 3 r 2 = 1 + 2 ∑ r ≥ 1 q 3 r 2 . The exponents 3 r 2 + 3 r + 1 3r^{2}+3r+1 3 r 2 + 3 r + 1 (r ≥ 0 r\geq0 r ≥ 0 ) are distinct positive integers. Comparing the coefficients of q n q^{n} q n on both sides of (40) , we obtain
σ ‾ 3 max ( n ) ≡ 1 ( m o d 2 ) ⟺ n = 0 or n = 3 k 2 + 3 k + 1 for some integer k ≥ 0. \overline{\sigma}_{3}^{\max}(n)\equiv1\pmod{2}\Longleftrightarrow n=0\text{ or }n=3k^{2}+3k+1\text{ for some integer }k\geq0. σ 3 m a x ( n ) ≡ 1 ( mod 2 ) ⟺ n = 0 or n = 3 k 2 + 3 k + 1 for some integer k ≥ 0.
Proof of Theorem 1.8. For a real number X ≥ 1 X\geq1 X ≥ 1 , we define
M 1 ( X ) : = # { n ≤ X : σ ‾ 3 max ( n ) is odd } , M 2 ( X ) : = # { n ≤ X : σ ‾ 3 max ( n ) is even } . M_{1}(X):=\#\{n\leq X:\overline{\sigma}_{3}^{\max}(n)\text{ is odd}\},\qquad M_{2}(X):=\#\{n\leq X:\overline{\sigma}_{3}^{\max}(n)\text{ is even}\}. M 1 ( X ) := # { n ≤ X : σ 3 m a x ( n ) is odd } , M 2 ( X ) := # { n ≤ X : σ 3 m a x ( n ) is even } .
Note that, for any real number k ≥ 0 k\geq0 k ≥ 0 ,
3 k 2 + 3 k + 1 ≤ X ⟺ k ≤ − 3 + 12 X − 3 6 . 3k^{2}+3k+1\leq X\Longleftrightarrow k\leq\frac{-3+\sqrt{12X-3}}{6}. 3 k 2 + 3 k + 1 ≤ X ⟺ k ≤ 6 − 3 + 12 X − 3 .
Using Theorem 1.7 (2), we get
M 1 ( X ) = ⌊ − 3 + 12 X − 3 6 ⌋ + 1 = 12 X − 3 6 + O ( 1 ) . (41) M_1(X)=\left\lfloor\frac{-3+\sqrt{12X-3}}{6}\right\rfloor+1=\frac{\sqrt{12X-3}}{6}+O(1). \tag*{(41)} M 1 ( X ) = ⌊ 6 − 3 + 12 X − 3 ⌋ + 1 = 6 12 X − 3 + O ( 1 ) . (41)
For large x x x , we have x + a = x + a 2 x + O ( x − 3 / 2 ) \sqrt{x+a}=\sqrt{x}+\frac{a}{2\sqrt{x}}+O(x^{-3/2}) x + a = x + 2 x a + O ( x − 3/2 ) . In particular,
12 X − 3 6 = 12 X 6 + O ( X − 1 / 2 ) = X 3 + O ( X − 1 / 2 ) . \frac{\sqrt{12X-3}}{6}=\frac{\sqrt{12X}}{6}+O(X^{-1/2})=\sqrt{\frac{X}{3}}+O(X^{-1/2}). 6 12 X − 3 = 6 12 X + O ( X − 1/2 ) = 3 X + O ( X − 1/2 ) .
Thus
M 1 ( X ) = X 3 + O ( 1 ) . M_1(X)=\sqrt{\frac{X}{3}}+O(1). M 1 ( X ) = 3 X + O ( 1 ) .
Therefore,
M 2 ( X ) = ⌊ X ⌋ − M 1 ( X ) = X − X 3 + O ( 1 ) . M_2(X)=\lfloor X\rfloor-M_1(X)=X-\sqrt{\frac{X}{3}}+O(1). M 2 ( X ) = ⌊ X ⌋ − M 1 ( X ) = X − 3 X + O ( 1 ) .
Proof of Theorem 1.9 . Throughout the proof, ℓ ≥ 2 \ell\ge2 ℓ ≥ 2 is fixed and we write R ( m ) = R ℓ ( m ) R(m)=R_\ell(m) R ( m ) = R ℓ ( m ) . Let
1 = m 1 < m 2 < m 3 < ⋯ 1=m_1<m_2<m_3<\cdots 1 = m 1 < m 2 < m 3 < ⋯
be the positive integers not divisible by ℓ \ell ℓ , listed in increasing order.
By (33) ,
∑ n ≥ 0 σ ℓ max ‾ ( n ) q n = P ‾ ℓ ( q ) S ℓ ( q ) , S ℓ ( q ) : = ∑ m ≥ 1 ℓ ∤ m m q R ( m ) ( 1 − q m ) = ∑ j ≥ 1 m j q R ( m j ) ( 1 − q m j ) . (42) \sum_{n\ge0}\sigma_\ell^{\overline{\max}}(n)q^n=\overline{P}_\ell(q)S_\ell(q),\qquad S_\ell(q):=\sum_{\substack{m\ge1\\ \ell\nmid m}}mq^{R(m)}(1-q^m)=\sum_{j\ge1}m_jq^{R(m_j)}(1-q^{m_j}). \tag*{(42)} n ≥ 0 ∑ σ ℓ m a x ( n ) q n = P ℓ ( q ) S ℓ ( q ) , S ℓ ( q ) := m ≥ 1 ℓ ∤ m ∑ m q R ( m ) ( 1 − q m ) = j ≥ 1 ∑ m j q R ( m j ) ( 1 − q m j ) . (42)
By Proposition 2, P ‾ ℓ ( q ) ≡ 1 ( m o d 2 ) \overline{P}_\ell(q)\equiv1\pmod{2} P ℓ ( q ) ≡ 1 ( mod 2 ) , and therefore
∑ n ≥ 0 σ ℓ max ‾ ( n ) q n ≡ S ℓ ( q ) ( m o d 2 ) . (43) \sum_{n\ge0}\sigma_\ell^{\overline{\max}}(n)q^n\equiv S_\ell(q)\pmod{2}. \tag*{(43)} n ≥ 0 ∑ σ ℓ m a x ( n ) q n ≡ S ℓ ( q ) ( mod 2 ) . (43)
Next, for every j ≥ 1 j\ge1 j ≥ 1 , the integer m j m_j m j is the only positive integer in the interval [ m j , m j + 1 ) [m_j,m_{j+1}) [ m j , m j + 1 ) that is not divisible by ℓ \ell ℓ . Hence, by the definition of R R R ,
R ( m j + 1 ) = ∑ 1 ≤ i < m j + 1 ℓ ∤ i i = ∑ 1 ≤ i < m j ℓ ∤ i i + m j = R ( m j ) + m j ( j ≥ 1 ) . (44) R(m_{j+1})=\sum_{\substack{1\le i<m_{j+1}\\ \ell\nmid i}}i=\sum_{\substack{1\le i<m_j\\ \ell\nmid i}}i+m_j=R(m_j)+m_j\qquad(j\ge1). \tag*{(44)} R ( m j + 1 ) = 1 ≤ i < m j + 1 ℓ ∤ i ∑ i = 1 ≤ i < m j ℓ ∤ i ∑ i + m j = R ( m j ) + m j ( j ≥ 1 ) . (44)
Moreover, R ( m 1 ) = R ( 1 ) = 0 R(m_1)=R(1)=0 R ( m 1 ) = R ( 1 ) = 0 , and the sequence { R ( m j ) } j ≥ 1 \{R(m_j)\}_{j\ge1} { R ( m j ) } j ≥ 1 is strictly increasing. Using (44) in (42) , we obtain
S ℓ ( q ) = ∑ j ≥ 1 m j q R ( m j ) − ∑ j ≥ 1 m j q R ( m j ) + m j = ∑ j ≥ 1 m j q R ( m j ) − ∑ j ≥ 1 m j q R ( m j + 1 ) = ∑ j ≥ 1 m j q R ( m j ) − ∑ j ≥ 2 m j − 1 q R ( m j ) = m 1 q R ( m 1 ) + ∑ j ≥ 2 ( m j − m j − 1 ) q R ( m j ) . \begin{aligned}
S_\ell(q)
&=\sum_{j\ge1}m_jq^{R(m_j)}-\sum_{j\ge1}m_jq^{R(m_j)+m_j}
=\sum_{j\ge1}m_jq^{R(m_j)}-\sum_{j\ge1}m_jq^{R(m_{j+1})}\\
&=\sum_{j\ge1}m_jq^{R(m_j)}-\sum_{j\ge2}m_{j-1}q^{R(m_j)}
=m_1q^{R(m_1)}+\sum_{j\ge2}(m_j-m_{j-1})q^{R(m_j)}.
\end{aligned} S ℓ ( q ) = j ≥ 1 ∑ m j q R ( m j ) − j ≥ 1 ∑ m j q R ( m j ) + m j = j ≥ 1 ∑ m j q R ( m j ) − j ≥ 1 ∑ m j q R ( m j + 1 ) = j ≥ 1 ∑ m j q R ( m j ) − j ≥ 2 ∑ m j − 1 q R ( m j ) = m 1 q R ( m 1 ) + j ≥ 2 ∑ ( m j − m j − 1 ) q R ( m j ) .
This gives
S ℓ ( q ) = 1 + ∑ j ≥ 2 ( m j − m j − 1 ) q R ( m j ) . (45) S_\ell(q)=1+\sum_{j\ge2}(m_j-m_{j-1})q^{R(m_j)}. \tag*{(45)} S ℓ ( q ) = 1 + j ≥ 2 ∑ ( m j − m j − 1 ) q R ( m j ) . (45)
We now determine the differences m j − m j − 1 m_j-m_{j-1} m j − m j − 1 . Let j ≥ 2 j\ge2 j ≥ 2 . If m j ≢ 1 ( m o d ℓ ) m_j\not\equiv1\pmod{\ell} m j ≡ 1 ( mod ℓ ) , then ℓ ∤ m j − 1 \ell\nmid m_j-1 ℓ ∤ m j − 1 , so m j − 1 = m j − 1 m_{j-1}=m_j-1 m j − 1 = m j − 1 and m j − m j − 1 = 1 m_j-m_{j-1}=1 m j − m j − 1 = 1 . If m j ≡ 1 ( m o d ℓ ) m_j\equiv1\pmod{\ell} m j ≡ 1 ( mod ℓ ) , then ℓ ∣ m j − 1 \ell\mid m_j-1 ℓ ∣ m j − 1 , while m j − 2 ≡ − 1 ≢ 0 ( m o d ℓ ) m_j-2\equiv-1\not\equiv0\pmod{\ell} m j − 2 ≡ − 1 ≡ 0 ( mod ℓ ) since ℓ ≥ 2 \ell\ge2 ℓ ≥ 2 . Hence m j − 1 = m j − 2 m_{j-1}=m_j-2 m j − 1 = m j − 2 and m j − m j − 1 = 2 m_j-m_{j-1}=2 m j − m j − 1 = 2 . Therefore
m j − m j − 1 ≡ { 1 ( m o d 2 ) , m j ≢ 1 ( m o d ℓ ) , 0 ( m o d 2 ) , m j ≡ 1 ( m o d ℓ ) . m_j-m_{j-1}\equiv
\begin{cases}
1\pmod{2},&m_j\not\equiv1\pmod{\ell},\\
0\pmod{2},&m_j\equiv1\pmod{\ell}.
\end{cases} m j − m j − 1 ≡ { 1 ( mod 2 ) , 0 ( mod 2 ) , m j ≡ 1 ( mod ℓ ) , m j ≡ 1 ( mod ℓ ) .
Combining this with (43) and (45) , we get
∑ n ≥ 0 σ ℓ m e x ‾ ( n ) q n ≡ 1 + ∑ j ≥ 2 m j ≢ 1 ( m o d ℓ ) q R ( m j ) ( m o d 2 ) . (46) \sum_{n\geq0}\sigma_{\ell}^{\overline{\mathrm{mex}}}(n)q^n \equiv1+\sum_{\substack{j\geq2\\m_j\not\equiv1\pmod{\ell}}}q^{R(m_j)}\pmod{2}. \tag*{(46)} n ≥ 0 ∑ σ ℓ mex ( n ) q n ≡ 1 + j ≥ 2 m j ≡ 1 ( mod ℓ ) ∑ q R ( m j ) ( mod 2 ) . (46)
Since the exponents R ( m j ) R(m_j) R ( m j ) , j ≥ 1 j\geq1 j ≥ 1 , are pairwise distinct and R ( m j ) > 0 R(m_j)>0 R ( m j ) > 0 for j ≥ 2 j\geq2 j ≥ 2 , no two terms on the right-hand side of (46) have the same power of q q q . Comparing the coefficients of q n q^n q n on both sides of (46) , we conclude that σ ℓ m e x ‾ ( n ) \sigma_{\ell}^{\overline{\mathrm{mex}}}(n) σ ℓ mex ( n ) is odd if and only if n = 0 n=0 n = 0 or n = R ( m j ) n=R(m_j) n = R ( m j ) for some j ≥ 2 j\geq2 j ≥ 2 with m j ≢ 1 ( m o d ℓ ) m_j\not\equiv1\pmod{\ell} m j ≡ 1 ( mod ℓ ) .
Note that the integers m j m_j m j with j ≥ 2 j\geq2 j ≥ 2 and m j ≢ 1 ( m o d ℓ ) m_j\not\equiv1\pmod{\ell} m j ≡ 1 ( mod ℓ ) are precisely the integers ℓ k + s \ell k+s ℓ k + s with k ≥ 0 k\geq0 k ≥ 0 and 2 ≤ s ≤ ℓ − 1 2\leq s\leq\ell-1 2 ≤ s ≤ ℓ − 1 . By Lemma 2.6,
R ( ℓ k + s ) = ℓ ( ℓ − 1 ) 2 k 2 + ( s − 1 ) ℓ k + s ( s − 1 ) 2 . R(\ell k+s)=\frac{\ell(\ell-1)}{2}k^2+(s-1)\ell k+\frac{s(s-1)}{2}. R ( ℓ k + s ) = 2 ℓ ( ℓ − 1 ) k 2 + ( s − 1 ) ℓ k + 2 s ( s − 1 ) .
This proves the first assertion.
It remains to count the odd values. Fixing s ∈ { 2 , … , ℓ − 1 } s\in\{2,\ldots,\ell-1\} s ∈ { 2 , … , ℓ − 1 } and putting
a = ℓ ( ℓ − 1 ) 2 , b = ( s − 1 ) ℓ , c = s ( s − 1 ) 2 , a=\frac{\ell(\ell-1)}{2},\qquad b=(s-1)\ell,\qquad c=\frac{s(s-1)}{2}, a = 2 ℓ ( ℓ − 1 ) , b = ( s − 1 ) ℓ , c = 2 s ( s − 1 ) ,
we see that the map k ↦ R ( ℓ k + s ) = a k 2 + b k + c k\mapsto R(\ell k+s)=ak^2+bk+c k ↦ R ( ℓ k + s ) = a k 2 + bk + c is strictly increasing on k ≥ 0 k\geq0 k ≥ 0 . For X ≥ c X\geq c X ≥ c , the inequality a k 2 + b k + c ≤ X ak^2+bk+c\leq X a k 2 + bk + c ≤ X with k ≥ 0 k\geq0 k ≥ 0 is equivalent to 0 ≤ k ≤ κ s ( X ) 0\leq k\leq\kappa_s(X) 0 ≤ k ≤ κ s ( X ) , where
κ s ( X ) = − b + b 2 − 4 a c + 4 a X 2 a . \kappa_s(X)=\frac{-b+\sqrt{b^2-4ac+4aX}}{2a}. κ s ( X ) = 2 a − b + b 2 − 4 a c + 4 a X .
Hence the number N s ( X ) N_s(X) N s ( X ) of integers k ≥ 0 k\geq0 k ≥ 0 with R ( ℓ k + s ) ≤ X R(\ell k+s)\leq X R ( ℓ k + s ) ≤ X equals ⌊ κ s ( X ) ⌋ + 1 \lfloor\kappa_s(X)\rfloor+1 ⌊ κ s ( X )⌋ + 1 for X ≥ c X\geq c X ≥ c , and N s ( X ) = 0 N_s(X)=0 N s ( X ) = 0 for X < c X<c X < c . Since b 2 − 4 a c b^2-4ac b 2 − 4 a c does not depend on X X X , we have
b 2 − 4 a c + 4 a X = 4 a X + O ( X − 1 / 2 ) , \sqrt{b^2-4ac+4aX}=\sqrt{4aX}+O(X^{-1/2}), b 2 − 4 a c + 4 a X = 4 a X + O ( X − 1/2 ) ,
and therefore
N s ( X ) = 4 a X 2 a + O ( 1 ) = X a + O ( 1 ) = 2 X ℓ ( ℓ − 1 ) + O ( 1 ) . N_s(X)=\frac{\sqrt{4aX}}{2a}+O(1)=\sqrt{\frac{X}{a}}+O(1)=\sqrt{\frac{2X}{\ell(\ell-1)}}+O(1). N s ( X ) = 2 a 4 a X + O ( 1 ) = a X + O ( 1 ) = ℓ ( ℓ − 1 ) 2 X + O ( 1 ) .
By the first part of the theorem, the values R ( ℓ k + s ) R(\ell k+s) R ( ℓ k + s ) for different pairs ( k , s ) (k,s) ( k , s ) are distinct, and they are distinct from 0 0 0 . Hence
# { 0 ≤ n ≤ X : σ ℓ m e x ‾ ( n ) is odd } = 1 + ∑ s = 2 ℓ − 1 N s ( X ) = ( ℓ − 2 ) 2 X ℓ ( ℓ − 1 ) + O ( 1 ) . \#\{0\leq n\leq X:\sigma_{\ell}^{\overline{\mathrm{mex}}}(n)\text{ is odd}\}=1+\sum_{s=2}^{\ell-1}N_s(X)=(\ell-2)\sqrt{\frac{2X}{\ell(\ell-1)}}+O(1). # { 0 ≤ n ≤ X : σ ℓ mex ( n ) is odd } = 1 + s = 2 ∑ ℓ − 1 N s ( X ) = ( ℓ − 2 ) ℓ ( ℓ − 1 ) 2 X + O ( 1 ) .
For ℓ ≥ 3 \ell\geq3 ℓ ≥ 3 the main term is a positive multiple of X \sqrt{X} X , so σ ℓ m e x ‾ ( n ) \sigma_{\ell}^{\overline{\mathrm{mex}}}(n) σ ℓ mex ( n ) is odd for infinitely many n n n , and the number of such n ≤ X n\leq X n ≤ X is ≫ X \gg\sqrt{X} ≫ X . This completes the proof. □
Proof of Theorem 1.10.
Proof. (1) By Lemma 2.4 with p = 3 p=3 p = 3 and j = 1 j=1 j = 1 , we have f 1 3 ≡ f 3 f_1^3\equiv f_3 f 1 3 ≡ f 3 , f 2 6 ≡ f 6 2 f_2^6\equiv f_6^2 f 2 6 ≡ f 6 2 and f 3 4 ≡ f 12 ( m o d 3 ) f_3^4\equiv f_{12}\pmod{3} f 3 4 ≡ f 12 ( mod 3 ) . Hence, by Theorem 1.1,
∑ n ≥ 0 σ 2 m e x ‾ ( n ) q n = f 6 2 f 1 4 f 3 4 = f 6 2 f 3 f 12 ⋅ f 2 2 f 1 ( m o d 3 ) . \sum_{n\geq0}\sigma_2^{\overline{\mathrm{mex}}}(n)q^n=\frac{f_6^2}{f_1^4f_3^4}=\frac{f_6^2}{f_3f_{12}}\cdot\frac{f_2^2}{f_1}\pmod{3}. n ≥ 0 ∑ σ 2 mex ( n ) q n = f 1 4 f 3 4 f 6 2 = f 3 f 12 f 6 2 ⋅ f 1 f 2 2 ( mod 3 ) .
Using (11) , we get
∑ n ≥ 0 σ 2 m e x ‾ ( n ) q n = f 6 3 f 9 2 f 3 2 f 12 f 18 + q f 6 2 f 18 2 f 3 f 9 f 12 ( m o d 3 ) . (47) \sum_{n\geq0}\sigma_2^{\overline{\mathrm{mex}}}(n)q^n=\frac{f_6^3f_9^2}{f_3^2f_{12}f_{18}}+q\frac{f_6^2f_{18}^2}{f_3f_9f_{12}}\pmod{3}. \tag*{(47)} n ≥ 0 ∑ σ 2 mex ( n ) q n = f 3 2 f 12 f 18 f 6 3 f 9 2 + q f 3 f 9 f 12 f 6 2 f 18 2 ( mod 3 ) . (47)
The right-hand side of (47) contains no term q 3 n + 2 q^{3n+2} q 3 n + 2 . Therefore σ 2 m e x ‾ ( 3 n + 2 ) ≡ 0 ( m o d 3 ) \sigma_2^{\overline{\mathrm{mex}}}(3n+2)\equiv0\pmod{3} σ 2 mex ( 3 n + 2 ) ≡ 0 ( mod 3 ) . (2) Substituting (10) in Theorem 1.1, we obtain
∑ n ≥ 0 σ ‾ 2 max ( n ) q n = f 4 11 f 2 6 f 8 4 + 4 q f 8 4 f 2 2 f 4 2 . (48) \sum_{n\ge0}\overline{\sigma}^{\max}_2(n)q^n=\frac{f_4^{11}}{f_2^6f_8^4}+4q\frac{f_8^4}{f_2^2f_4^2}. \tag*{(48)} n ≥ 0 ∑ σ 2 m a x ( n ) q n = f 2 6 f 8 4 f 4 11 + 4 q f 2 2 f 4 2 f 8 4 . (48)
Extracting the terms with even powers of q q q and replacing q 2 q^2 q 2 by q q q , we get
∑ n ≥ 0 σ ‾ 2 max ( 2 n ) q n = f 2 11 f 1 6 f 4 4 = f 2 5 f 1 2 f 4 2 ⋅ f 2 6 f 1 4 f 4 2 . \sum_{n\ge0}\overline{\sigma}^{\max}_2(2n)q^n=\frac{f_2^{11}}{f_1^6f_4^4}=\frac{f_2^5}{f_1^2f_4^2}\cdot\frac{f_2^6}{f_1^4f_4^2}. n ≥ 0 ∑ σ 2 m a x ( 2 n ) q n = f 1 6 f 4 4 f 2 11 = f 1 2 f 4 2 f 2 5 ⋅ f 1 4 f 4 2 f 2 6 .
By Lemma 2.4 with p = 2 p=2 p = 2 and j = 2 j=2 j = 2 , f 1 4 ≡ f 2 2 f_1^4\equiv f_2^2 f 1 4 ≡ f 2 2 and f 2 4 ≡ f 4 2 f_2^4\equiv f_4^2 f 2 4 ≡ f 4 2 (mod 4), so f 2 6 f 1 4 f 4 2 ≡ f 2 4 f 4 2 ≡ 1 \frac{f_2^6}{f_1^4f_4^2}\equiv\frac{f_2^4}{f_4^2}\equiv1 f 1 4 f 4 2 f 2 6 ≡ f 4 2 f 2 4 ≡ 1 (mod 4). Hence, by Lemma 2.1,
∑ n ≥ 0 σ ‾ 2 max ( 2 n ) q n ≡ f 2 5 f 1 2 f 4 2 = ∑ k ∈ Z q k 2 ≡ 1 + 2 ∑ k ≥ 1 q k 2 ( m o d 4 ) . (49) \sum_{n\ge0}\overline{\sigma}^{\max}_2(2n)q^n\equiv\frac{f_2^5}{f_1^2f_4^2}=\sum_{k\in\mathbb{Z}}q^{k^2}\equiv1+2\sum_{k\ge1}q^{k^2}\pmod4. \tag*{(49)} n ≥ 0 ∑ σ 2 m a x ( 2 n ) q n ≡ f 1 2 f 4 2 f 2 5 = k ∈ Z ∑ q k 2 ≡ 1 + 2 k ≥ 1 ∑ q k 2 ( mod 4 ) . (49)
Also, by (48) , σ ‾ 2 max ( 2 n + 1 ) ≡ 0 ( m o d 4 ) \overline{\sigma}^{\max}_2(2n+1)\equiv0\pmod4 σ 2 m a x ( 2 n + 1 ) ≡ 0 ( mod 4 ) . Therefore, for n ≥ 1 n\ge1 n ≥ 1 , σ ‾ 2 max ( n ) ≡ 2 ( m o d 4 ) \overline{\sigma}^{\max}_2(n)\equiv2\pmod4 σ 2 m a x ( n ) ≡ 2 ( mod 4 ) if n = 2 k 2 n=2k^2 n = 2 k 2 for some k ≥ 1 k\ge1 k ≥ 1 , and σ ‾ 2 max ( n ) ≡ 0 ( m o d 4 ) \overline{\sigma}^{\max}_2(n)\equiv0\pmod4 σ 2 m a x ( n ) ≡ 0 ( mod 4 ) otherwise.
(3) Extracting the terms with odd powers of q q q in (48) , dividing by q q q and replacing q 2 q^2 q 2 by q q q , we get
∑ n ≥ 0 σ ‾ 2 max ( 2 n + 1 ) q n = 4 f 4 4 f 1 2 f 2 2 . (50) \sum_{n\ge0}\overline{\sigma}^{\max}_2(2n+1)q^n=4\frac{f_4^4}{f_1^2f_2^2}. \tag*{(50)} n ≥ 0 ∑ σ 2 m a x ( 2 n + 1 ) q n = 4 f 1 2 f 2 2 f 4 4 . (50)
By Lemma 2.4 , f 1 2 ≡ f 2 f_1^2\equiv f_2 f 1 2 ≡ f 2 and f 2 2 ≡ f 4 f_2^2\equiv f_4 f 2 2 ≡ f 4 (mod 2), so by Lemma 2.5 ,
f 4 4 f 1 2 f 2 ≡ f 4 3 = ∑ k ≥ 0 ( − 1 ) k ( 2 k + 1 ) q 2 k ( k + 1 ) ≡ ∑ k ≥ 0 q 2 k ( k + 1 ) ( m o d 2 ) . \frac{f_4^4}{f_1^2f_2}\equiv f_4^3=\sum_{k\ge0}(-1)^k(2k+1)q^{2k(k+1)}\equiv\sum_{k\ge0}q^{2k(k+1)}\pmod2. f 1 2 f 2 f 4 4 ≡ f 4 3 = k ≥ 0 ∑ ( − 1 ) k ( 2 k + 1 ) q 2 k ( k + 1 ) ≡ k ≥ 0 ∑ q 2 k ( k + 1 ) ( mod 2 ) .
Hence ∑ n σ ‾ 2 max ( 2 n + 1 ) q n ≡ 4 ∑ k ≥ 0 q 2 k ( k + 1 ) ( m o d 8 ) \sum_n\overline{\sigma}^{\max}_2(2n+1)q^n\equiv4\sum_{k\ge0}q^{2k(k+1)}\pmod8 ∑ n σ 2 m a x ( 2 n + 1 ) q n ≡ 4 ∑ k ≥ 0 q 2 k ( k + 1 ) ( mod 8 ) . Since n = 2 k ( k + 1 ) n=2k(k+1) n = 2 k ( k + 1 ) if and only if 2 n + 1 = ( 2 k + 1 ) 2 2n+1=(2k+1)^2 2 n + 1 = ( 2 k + 1 ) 2 , the claim follows.
(4) By Theorem 1.2,
∑ n ≥ 0 σ ‾ 3 max ( n ) q n = f 6 4 f 12 2 ⋅ f 2 f 1 + q f 3 2 f 12 2 f 6 2 ⋅ f 2 f 1 . \sum_{n\ge0}\overline{\sigma}^{\max}_3(n)q^n=\frac{f_6^4}{f_{12}^2}\cdot\frac{f_2}{f_1}+q\frac{f_3^2f_{12}^2}{f_6^2}\cdot\frac{f_2}{f_1}. n ≥ 0 ∑ σ 3 m a x ( n ) q n = f 12 2 f 6 4 ⋅ f 1 f 2 + q f 6 2 f 3 2 f 12 2 ⋅ f 1 f 2 .
Substituting (12) , extracting the terms with powers q 3 n q^{3n} q 3 n and replacing q 3 q^3 q 3 by q q q , we get
∑ n ≥ 0 σ ‾ 3 max ( 3 n ) q n = f 2 8 f 6 6 f 1 8 f 4 2 f 6 3 + 4 q f 4 2 f 6 3 f 1 4 . \sum_{n\ge0}\overline{\sigma}^{\max}_3(3n)q^n=\frac{f_2^8f_6^6}{f_1^8f_4^2f_6^3}+4q\frac{f_4^2f_6^3}{f_1^4}. n ≥ 0 ∑ σ 3 m a x ( 3 n ) q n = f 1 8 f 4 2 f 6 3 f 2 8 f 6 6 + 4 q f 1 4 f 4 2 f 6 3 .
By Lemma 2.4 with p = 2 p=2 p = 2 and j = 2 j=2 j = 2 , f 1 8 ≡ f 2 4 f_1^8\equiv f_2^4 f 1 8 ≡ f 2 4 , f 4 4 ≡ f 2 2 f_4^4\equiv f_2^2 f 4 4 ≡ f 2 2 and f 3 4 ≡ f 6 2 f_3^4\equiv f_6^2 f 3 4 ≡ f 6 2 (mod 4). Hence, by Lemma 2.1 with q q q replaced by − q 3 -q^3 − q 3 (note that ( − q ; − q ) ∞ = f 3 2 f 1 f 4 (-q;-q)_\infty=\frac{f_3^2}{f_1f_4} ( − q ; − q ) ∞ = f 1 f 4 f 3 2 ),
∑ n ≥ 0 σ ‾ 3 max ( 3 n ) q n ≡ f 3 2 f 6 = ∑ k ∈ Z ( − 1 ) k q 3 k 2 ≡ 1 + 2 ∑ k ≥ 1 q 3 k 2 ( m o d 4 ) . \sum_{n\ge0}\overline{\sigma}^{\max}_3(3n)q^n\equiv\frac{f_3^2}{f_6}=\sum_{k\in\mathbb{Z}}(-1)^kq^{3k^2}\equiv1+2\sum_{k\ge1}q^{3k^2}\pmod4. n ≥ 0 ∑ σ 3 m a x ( 3 n ) q n ≡ f 6 f 3 2 = k ∈ Z ∑ ( − 1 ) k q 3 k 2 ≡ 1 + 2 k ≥ 1 ∑ q 3 k 2 ( mod 4 ) .
Therefore, for n ≥ 1 n\ge1 n ≥ 1 , σ ‾ 3 max ( 3 n ) ≡ 2 ( m o d 4 ) \overline{\sigma}^{\max}_3(3n)\equiv2\pmod4 σ 3 m a x ( 3 n ) ≡ 2 ( mod 4 ) if n = 3 k 2 n=3k^2 n = 3 k 2 for some k ≥ 1 k\ge1 k ≥ 1 , and σ ‾ 3 max ( 3 n ) ≡ 0 ( m o d 4 ) \overline{\sigma}^{\max}_3(3n)\equiv0\pmod4 σ 3 m a x ( 3 n ) ≡ 0 ( mod 4 ) otherwise. Since 3 n + 1 3n+1 3 n + 1 and 3 n + 2 3n+2 3 n + 2 are never of the form 3 k 2 3k^2 3 k 2 , we get σ ‾ 3 max ( 9 n + 3 ) ≡ σ ‾ 3 max ( 9 n + 6 ) ≡ 0 ( m o d 4 ) \overline{\sigma}^{\max}_3(9n+3)\equiv\overline{\sigma}^{\max}_3(9n+6)\equiv0\pmod4 σ 3 m a x ( 9 n + 3 ) ≡ σ 3 m a x ( 9 n + 6 ) ≡ 0 ( mod 4 ) . □
Data availability statement: There is no data associated with our manuscript.
References
Sourav Bhowmick, Department of Mathematics, National Institute of Technology, Raipur, Chhattisgarh 492010.
Email address: souravbhowmick578@gmail.com, sbhowmick.phd2025@nitrr.ac.in
Nabin Kumar Meher, Department of Mathematics, National Institute of Technology, Raipur, Chhattisgarh 492010.
Email address: mehernabin@gmail.com, nkmeher.maths@nitrr.ac.in