Introduction
In 1981 Gordon and Hughes studied Ramanujan type congruence families for the distinct partition function, denoted by q ( n ) q(n) q ( n ) , modulo powers of 5 and 7. They [16 Section 1] speculated on the possibility of analogous congruences by powers of 11:
“As far as congruences for q ( n ) q(n) q ( n ) mod 11 γ 11^\gamma 1 1 γ are concerned, the problem is greatly complicated by the fact that the group Γ 0 ( 22 ) \Gamma_0(22) Γ 0 ( 22 ) has genus 2. The computational difficulties seem too formidable to justify pursuing this problem here.”
It is nearly a half-century since this paper was published, and no congruence families have been found associated with Γ 0 ( 22 ) \Gamma_0(22) Γ 0 ( 22 ) or any other congruence subgroup corresponding to a genus 2 modular curve. In this paper, we give an explicit example of such a family and propose a broader class of such genus 2 congruence families. The congruences are exhibited by the 2-elongated plane partition diamond function d 2 ( n ) d_2(n) d 2 ( n ) rather than q ( n ) q(n) q ( n ) ; nevertheless, the associated group is Γ 0 ( 22 ) \Gamma_0(22) Γ 0 ( 22 ) , as Gordon and Hughes had anticipated. Likewise, the difficulties, theoretical and computational, are as formidable as Gordon and Hughes foresaw. The principal theoretical novelty of the proof is the use of localizing functions that are not modular units, i.e., whose divisors are not supported on the cusps. We believe that the methods developed to handle such functions have implications beyond the present family, both for modular congruences and for the ℓ \ell ℓ -adic theory of modular forms more generally.
Background
The theory of modular congruence relationships has been an active area of research since Ramanujan’s groundbreaking work [22 ] on the divisibility properties of the partition function. Consider the original congruence families discovered and proved by Ramanujan, Watson [24 ] , and Atkin [6 ] :
If n , α ∈ N and 24 n ≡ 1 ( m o d ℓ α ) , then p ( n ) ≡ 0 ( m o d ℓ β ) , (1) \text{If } n,\alpha\in\mathbb{N} \text{ and } 24n \equiv1 \pmod{\ell^\alpha}, \text{ then } p(n) \equiv0 \pmod{\ell^\beta}, \tag*{(1)} If n , α ∈ N and 24 n ≡ 1 ( mod ℓ α ) , then p ( n ) ≡ 0 ( mod ℓ β ) , (1)
with
β : = { α if ℓ ∈ { 5 , 11 } , ⌊ α 2 ⌋ + 1 if ℓ = 7. \beta:=
\begin{cases}
\alpha& \text{if } \ell\in\{5,11\},\\
\left\lfloor\frac{\alpha}{2} \right\rfloor+ 1 & \text{if } \ell= 7.
\end{cases} β := { α ⌊ 2 α ⌋ + 1 if ℓ ∈ { 5 , 11 } , if ℓ = 7.
Since the proof of these families, the arithmetic of the partition function has been an active area of research. In his seminal work, Ono [18 ] proved that for every prime ℓ ≥ 5 \ell\ge5 ℓ ≥ 5 there are infinitely many congruences of the form p ( A n + B ) ≡ 0 ( m o d ℓ ) p(An+B) \equiv0 \pmod{\ell} p ( A n + B ) ≡ 0 ( mod ℓ ) , and Ahlgren [1 ] extended this to all moduli coprime to 6 (see also Ahlgren and Ono [4 ] ). Ahlgren and Boylan [3 ] used tools from the theory of Galois representations and modular forms to show that Ramanujan’s partition congruences are the only ones which are of the form p ( ℓ n + β ) ≡ 0 ( m o d ℓ ) p(\ell n+\beta) \equiv0 \pmod{\ell} p ( ℓ n + β ) ≡ 0 ( mod ℓ ) . For prime powers, Folsom, Kent, and Ono [13 ] and Boylan and Webb [10 ] established a strong form of ℓ \ell ℓ -adic stability for the U ℓ U_\ell U ℓ -action. Ahlgren, Allen, and Tang [2 ] proved that congruences like Atkin’s hold modulo every prime ℓ ≥ 5 \ell\ge5 ℓ ≥ 5 .
2010 Mathematics Subject Classification. 11P81, 11P82, 11P83, 30F35.
Key words and phrases. Elongated plane partitions, localization method, modular curve, modular forms, partition congruences, Riemann surface.
This richness of arithmetic properties has been found to exist within the coefficients of many different modular forms. We are particularly interested in ℓ \ell ℓ -adic patterns analogous to (1.1). A striking aspect of the subject is the substantial variability in the difficulty of a given congruence family. Some admit short proofs by methods that were already available to Ramanujan. Others remain open to this day.
One major difficulty lies in the topology of the associated modular curve, e.g., the cusp count and the genus. The most difficult problems generally occur when the associated modular curve has many cusps. Naively speaking, in these cases the action of the relevant U ℓ U_{\ell} U ℓ operator can no longer be controlled within a single, finitely generated space of modular functions, and the ℓ \ell ℓ -adic estimates required to close the induction break down. Families of this kind, such as those arising from traces of singular moduli and Hurwitz class numbers, have so far resisted even the most sophisticated techniques currently available [9 , 17 ] .
Another significant factor in the difficulty of a congruence family lies with the genus. For example, the associated modular curve for (1.1) is simply X 0 ( ℓ ) X_0(\ell) X 0 ( ℓ ) . In all three cases the cusp count is 2. However, for ℓ ∈ { 5 , 7 } \ell\in\{5, 7\} ℓ ∈ { 5 , 7 } , the curve has genus 0, while the curve has genus 1 for ℓ = 11 \ell= 11 ℓ = 11 . Watson, who had not completed the proof of (1.1) for ℓ = 11 \ell= 11 ℓ = 11 , speculated in 1938 that the proof would not be especially difficult, but perhaps “recht langweilig” [24 ] . Upon his completion of the proof for ℓ = 11 \ell= 11 ℓ = 11 in 1967, Atkin [6 Section 5] agreed with Watson’s speculation.
It has long been understood that families associated with curves of positive genus are difficult to treat by conventional techniques. In addition to the aforementioned remarks by Gordon and Hughes, Calegari [12 Section 4.2] has also pointed out that the standard proof methods “become essentially infeasible as soon as [the associated modular curve] has genus > 0 > 0 > 0 ”. The difficulty of a proof notwithstanding, it is interesting to note that no explicit congruence families for genus 2 appear to have been conjectured prior to this work. To the best of our knowledge, the possibility of such a family has been raised in the literature only by Gordon and Hughes [16 ] . To test the expectation that such a family should be accessible to current methods, we undertook a search for a suitable example. A computational search for q ( n ) q(n) q ( n ) , the function originally considered by Gordon and Hughes, produced no such candidate. A natural alternative is provided by the Fourier coefficients of the following eta quotient, whose congruence families have already served as a valuable testing ground for genus 0 and 1 [7 , 8 ] , and which therefore suggested itself as a promising source of a genus 2 example:
D k ( q ) : = ∑ n ≥ 0 d k ( n ) q n = ( q 2 ; q 2 ) ∞ k ( q ; q ) ∞ 3 k + 1 . D_k(q) := \sum_{n\geq0} d_k(n)q^n = \frac{(q^2;q^2)_\infty^k}{(q;q)_\infty^{3k+1}}. D k ( q ) := n ≥ 0 ∑ d k ( n ) q n = ( q ; q ) ∞ 3 k + 1 ( q 2 ; q 2 ) ∞ k .
The Fourier coefficients d k ( n ) d_k(n) d k ( n ) have an elegant combinatorial interpretation, discovered by Andrews and Paule [5 ] in their “Partition Analysis” series of papers. It is among the Fourier coefficients of D 2 ( q ) D_2(q) D 2 ( q ) that we find a relevant family of interest, the main result of this paper:
Theorem 1.1. Let n , α ∈ Z ≥ 1 n,\alpha\in\mathbb{Z}_{\geq1} n , α ∈ Z ≥ 1 such that 8 n ≡ 1 ( m o d 11 α ) 8n \equiv1 \pmod{11^\alpha} 8 n ≡ 1 ( mod 1 1 α ) . Then
d 2 ( n ) ≡ 0 ( m o d 11 α + 1 2 ) . d_2(n) \equiv0 \pmod{11^{\frac{\alpha+1}{2}}}. d 2 ( n ) ≡ 0 ( mod 1 1 2 α + 1 ) .
Localization and difficulty of the proof
Gordon and Hughes’s assessment of the difficulty proves accurate. The genus of X 0 ( 22 ) X_0(22) X 0 ( 22 ) , together with difficulties in its corresponding space of eta quotients, necessarily complicates our attempts at a proof of Theorem 1.1. As in previous approaches (see [7 , 8 ] ) via modular equation manipulations, the idea was straightforward: to study d 2 ( n ) d_2(n) d 2 ( n ) in the progressions relevant to us, we construct a sequence of functions L α L_\alpha L α of the form
L α : = ϕ α ∑ n ≥ 0 8 n ≡ 1 ( m o d 11 α ) d 2 ( n ) q ⌊ n 11 α ⌋ + 1 . L_\alpha:= \phi_\alpha\sum_{\substack{n\geq0\\8n\equiv1\pmod{11^\alpha}}} d_2(n)q^{\left\lfloor\frac{n}{11^\alpha}\right\rfloor+1}. L α := ϕ α n ≥ 0 8 n ≡ 1 ( mod 1 1 α ) ∑ d 2 ( n ) q ⌊ 1 1 α n ⌋ + 1 .
With the right holomorphic prefactors ϕ α \phi_\alpha ϕ α , we can arrange so that L α L_\alpha L α is a modular function over Γ 0 ( 22 ) \Gamma_0(22) Γ 0 ( 22 ) , i.e., a meromorphic function on X 0 ( 22 ) X_0(22) X 0 ( 22 ) with poles only at the cusps. We can construct a pair of modified Atkin U 11 U_{11} U 11 operators U ( i ) U^{(i)} U ( i ) , with i ∈ { 0 , 1 } i \in\{0,1\} i ∈ { 0 , 1 } , such that
U ( a ) ( L α ) = L α + 1 , U^{(a)}(L_\alpha)=L_{\alpha+1}, U ( a ) ( L α ) = L α + 1 ,
with a ≡ α ( m o d 2 ) a \equiv\alpha\pmod2 a ≡ α ( mod 2 ) . We want to place L α L_\alpha L α into a suitable subspace of the meromorphic functions over X 0 ( 22 ) X_0(22) X 0 ( 22 ) , in which the basis functions are subject to certain modular equations. We then want to study how U ( i ) U^{(i)} U ( i ) affects those basis functions. This is done by directly examining a handful of initial cases, and then using the associated modular equations to construct more general relationships. Recall that X 0 ( 22 ) X_0(22) X 0 ( 22 ) is a compact Riemann surface with cusps
{ [ ∞ ] , [ 1 11 ] , [ 1 2 ] , [ 0 ] } . \left\{[\infty],\left[\frac{1}{11}\right],\left[\frac{1}{2}\right],[0]\right\}. { [ ∞ ] , [ 11 1 ] , [ 2 1 ] , [ 0 ] } .
Since X 0 ( 22 ) X_0(22) X 0 ( 22 ) is compact, every non-constant meromorphic function on it has a pole. Generally, we prefer to restrict these poles to the cusps. We consider the space of holomorphic functions over the curve except for a pole (of possibly large order) at the cusp [ 0 ] [0] [ 0 ] , and denote this space as M ( 0 ) ( X 0 ( 22 ) ) \mathcal{M}^{(0)}(X_0(22)) M ( 0 ) ( X 0 ( 22 )) . Using a straightforward computational procedure, we can use the Dedekind eta quotient to construct suitable functions with integer Fourier coefficients, and with orders − 3 -3 − 3 , − 4 -4 − 4 , − 5 -5 − 5 at [ 0 ] [0] [ 0 ] . We denote these functions x , y , z x,y,z x , y , z , respectively. Their explicit form is given below (see (2.3)–(2.5)).
Combinations of these three functions can be used to construct functions of order − n -n − n for n ≥ 6 n \ge6 n ≥ 6 . The only missing orders are − 1 -1 − 1 , − 2 -2 − 2 . By the Weierstrass gap theorem [20 ] exactly two pole orders at [ 0 ] [0] [ 0 ] are not attained, i.e., as x x x , y y y , z z z have orders − 3 -3 − 3 , − 4 -4 − 4 , − 5 -5 − 5 , the gaps are 1 and 2. Thus, we have
M ( 0 ) ( X 0 ( 22 ) ) = C [ x ] ⊕ y C [ x ] ⊕ z C [ x ] . \mathcal{M}^{(0)}(X_0(22))=\mathbb{C}[x]\oplus y\mathbb{C}[x]\oplus z\mathbb{C}[x]. M ( 0 ) ( X 0 ( 22 )) = C [ x ] ⊕ y C [ x ] ⊕ z C [ x ] .
Suppose L α L_\alpha L α has a pole at [ 0 ] [0] [ 0 ] and nowhere else. Then we could express it as a suitable polynomial combination of x x x , y y y , and z z z . However, the poles of L α L_\alpha L α are supported at [ 0 ] [0] [ 0 ] and [ 1 2 ] \left[\frac{1}{2}\right] [ 2 1 ] . In context of the localization method, we can multiply L α L_\alpha L α by some modular eta quotient which lives in M ( 0 ) ( X 0 ( 22 ) ) \mathcal{M}^{(0)}(X_0(22)) M ( 0 ) ( X 0 ( 22 )) , but has positive order at [ 1 2 ] \left[\frac{1}{2}\right] [ 2 1 ] . Indeed, a very natural candidate exists. Consider,1
w : = ( q 2 ; q 2 ) ∞ 11 ( q 11 ; q 11 ) ∞ 11 ( q ; q ) ∞ 11 ( q 22 ; q 22 ) ∞ 11 = 1 + 11 y + 11 2 z . w:=\frac{(q^2;q^2)_\infty^{11}(q^{11};q^{11})_\infty^{11}}{(q;q)_\infty^{11}(q^{22};q^{22})_\infty^{11}}=1+11y+11^2z. w := ( q ; q ) ∞ 11 ( q 22 ; q 22 ) ∞ 11 ( q 2 ; q 2 ) ∞ 11 ( q 11 ; q 11 ) ∞ 11 = 1 + 11 y + 1 1 2 z .
Indeed, w 3 L 1 ∈ M ( 0 ) ( X 0 ( 22 ) ) w^3L_1\in\mathcal{M}^{(0)}(X_0(22)) w 3 L 1 ∈ M ( 0 ) ( X 0 ( 22 )) . We can therefore express L 1 L_1 L 1 as a rational polynomial with summands of the form
y j z k x m ( 1 + 11 y + 11 2 z ) 3 \frac{y^jz^kx^m}{(1+11y+11^2z)^3} ( 1 + 11 y + 1 1 2 z ) 3 y j z k x m
with m ∈ N 0 m\in\mathbb{N}_0 m ∈ N 0 , j , k ∈ { 0 , 1 } j,k\in\{0,1\} j , k ∈ { 0 , 1 } , and j + k ≤ 1 j+k\le1 j + k ≤ 1 .
However, such an approach raises various difficulties upon attempts to apply it to L α L_\alpha L α generally. In the first place, the denominator is governed by a degree 55 modular equation, while x x x is governed by a degree 33 modular equation. As such, we will be forced to examine how U ( i ) U^{(i)} U ( i ) affects 1815 different combinations, with an additional factor of 2 for the value of i i i , and a factor of 3 for the possible presence of y y y and z z z . This immediately forces well over 10 4 10^4 1 0 4 large polynomial relations to be verified. Moreover, the presence of y y y and z z z in w w w ensures that one must tediously remove monomials y a z b y^az^b y a z b with a + b ≥ 2 a+b\ge2 a + b ≥ 2 from resultant representations. Replacing z z z by w w w does not resolve this complication, because such a substitution would incur high powers of 11 in the denominators of coefficients in L α L_\alpha L α . This immediately destroys the utility of L α L_\alpha L α as a witness to divisibility by powers of 11. Another alternative is to work with a C [ w , w − 1 ] \mathbb{C}[w,w^{-1}] C [ w , w − 1 ] module of rank 5. In this case, one still requires the localization technique and needs to verify at least 3 ⋅ 10 4 3\cdot10^4 3 ⋅ 1 0 4 associated relations. Finally, the methods of Paule and Radu [19 ] , and of Garvan, Sellers, and the second author [15 Section 3] do not appear to adapt to the present setting.
These obstacles are consistent with Gordon and Hughes’s original assessment. We instead construct the localizing factor from x x x alone. To remove the poles of L α L_\alpha L α at [ 1 2 ] \left[\frac{1}{2}\right] [ 2 1 ] , note that the expansion of x x x near [ 1 2 ] \left[\frac{1}{2}\right] [ 2 1 ] in terms of the local variable q ^ \widehat{q} q is the following
x ( τ 2 τ + 1 ) = − 2 11 + b q ^ + O ( q ^ 2 ) with b ≠ 0. x\left(\frac{\tau}{2\tau+1}\right)=-\frac{2}{11}+b\widehat{q}+O(\widehat{q}^{2})\quad\text{with }b\ne0. x ( 2 τ + 1 τ ) = − 11 2 + b q + O ( q 2 ) with b = 0.
Thus ord [ 1 / 2 ] ( 2 + 11 x ) = 1 \operatorname{ord}_{[1/2]}(2+11x)=1 ord [ 1/2 ] ( 2 + 11 x ) = 1 and
( 2 + 11 x ) ψ L α ∈ ⨁ v ∈ { x , y , z } v C [ x ] . (2) (2+11x)^\psi L_\alpha\in\bigoplus_{v\in\{x,y,z\}}v\mathbb{C}[x]. \tag*{(2)} ( 2 + 11 x ) ψ L α ∈ v ∈ { x , y , z } ⨁ v C [ x ] . (2)
for an appropriate positive integer power ψ \psi ψ depending on α \alpha α . We can now apply the modular cusp analysis to directly check just 198 expressions for U ( i ) ( y j z k x m ) U^{(i)}(y^jz^kx^m) U ( i ) ( y j z k x m ) . With these relations, we need to examine some additional 6534 cases by algebraic manipulation. This is substantially smaller than the other possibilities considered. As an example, we consider the very first case:
L 1 = ( q 11 ; q 11 ) ∞ 7 ( q 22 ; q 22 ) ∞ 7 ∑ n = 0 ∞ d 2 ( 11 n + 7 ) q n + 2 . L_1=\frac{(q^{11};q^{11})_\infty^7}{(q^{22};q^{22})_\infty^7}\sum_{n=0}^{\infty}d_2(11n+7)q^{n+2}. L 1 = ( q 22 ; q 22 ) ∞ 7 ( q 11 ; q 11 ) ∞ 7 n = 0 ∑ ∞ d 2 ( 11 n + 7 ) q n + 2 .
Next we rewrite L 1 L_1 L 1 as2
L 1 = 11 2 ( 2 + 11 x ) 15 ( M x ( x ) + y M y ( x ) + z M z ( x ) + ( 3 + 7 x ) r L ) , (3) L_1 = \frac{11}{2(2+11x)^{15}}\left(M_x(x)+yM_y(x)+zM_z(x)+(3+7x)r_L\right), \tag*{(3)} L 1 = 2 ( 2 + 11 x ) 15 11 ( M x ( x ) + y M y ( x ) + z M z ( x ) + ( 3 + 7 x ) r L ) , (3)
with
M v ∈ Z [ X ] , v ∈ { x , y , z } , r L : = 1 11 ( x + 3 x 2 + 9 x 3 + 10 y + 10 x y + 2 x 2 y + 7 z + 8 x z ) ∈ Z [ [ q ] ] . M_v\in\mathbb{Z}[X],\quad v\in\{x,y,z\},\quad r_L:=\frac{1}{11}\left(x+3x^2+9x^3+10y+10xy+2x^2y+7z+8xz\right)\in\mathbb{Z}[[q]]. M v ∈ Z [ X ] , v ∈ { x , y , z } , r L := 11 1 ( x + 3 x 2 + 9 x 3 + 10 y + 10 x y + 2 x 2 y + 7 z + 8 x z ) ∈ Z [[ q ]] .
The function r L r_L r L , which we call the anomalous term, plays an important role. Its numerator is divisible by 11, although this is not visible from the individual coefficients. We prove that r L r_L r L is indeed an integer power series by appealing to Sturm’s theorem.
The more general result is given below. From this, Theorem 1.1 follows:
Theorem 1.2. Let L α L_\alpha L α be given as (2.1), and
ψ : = ψ ( α ) = ⌈ 11 α + 1 8 ⌉ + 1 − gcd ( α , 2 ) , β : = β ( α ) = ⌈ α + 1 2 ⌉ . (4) \psi:=\psi(\alpha)=\left\lceil\frac{11^{\alpha+1}}{8}\right\rceil+1-\gcd(\alpha,2),\qquad\beta:=\beta(\alpha)=\left\lceil\frac{\alpha+1}{2}\right\rceil. \tag*{(4)} ψ := ψ ( α ) = ⌈ 8 1 1 α + 1 ⌉ + 1 − g cd( α , 2 ) , β := β ( α ) = ⌈ 2 α + 1 ⌉ . (4) There exist sequences ( j α ) α ≥ 1 , ( k α ) α ≥ 1 ∈ Z [ 1 2 ] (j_\alpha)_{\alpha\geq1},(k_\alpha)_{\alpha\geq1}\in\mathbb{Z}\left[\frac{1}{2}\right] ( j α ) α ≥ 1 , ( k α ) α ≥ 1 ∈ Z [ 2 1 ] such that for all α ≥ 1 \alpha\geq1 α ≥ 1 , j 2 α = k 2 α = 0 j_{2\alpha}=k_{2\alpha}=0 j 2 α = k 2 α = 0 , and
( 2 + 11 x ) ψ 11 β L α + ( j α + k α x ) r L ∈ ⨁ v ∈ { x , y , z } v Z [ 1 2 , x ] . \frac{(2+11x)^\psi}{11^\beta}L_\alpha+(j_\alpha+k_\alpha x)r_L\in\bigoplus_{v\in\{x,y,z\}}v\mathbb{Z}\left[\frac{1}{2},x\right]. 1 1 β ( 2 + 11 x ) ψ L α + ( j α + k α x ) r L ∈ v ∈ { x , y , z } ⨁ v Z [ 2 1 , x ] .
It is worthwhile to note that in [7 , 8 ] , the analogous localizing factors are eta quotients, and therefore modular units . But, in this case, 2 + 11 x 2+11x 2 + 11 x is not a modular unit. Consequently, we must work in a larger localization ring, i.e., functions of the form y j k m c ( x ) − d ( 2 + 11 x ) − n y^j k^m c(x)^{-d}(2+11x)^{-n} y j k m c ( x ) − d ( 2 + 11 x ) − n with n , d ≥ 0 n,d\geq0 n , d ≥ 0 , and a certain polynomial c ( x ) c(x) c ( x ) (see § 7 ) induced by the modular equations for x x x . The heart of the proof is to show that, when L α L_\alpha L α is viewed in this ring, the factor c ( x ) d c(x)^d c ( x ) d cancels without destroying the 11-adic structure. This is the novelty and significance of the framework we propose.
Outline
In Section 2.1 we give an exact definition to L α L_\alpha L α , together with some of its key meromorphic properties. We also define the operators U ( i ) U^{(i)} U ( i ) and their effect on L α L_\alpha L α . In § 2.2 we precisely define x , y , z x,y,z x , y , z , as well as the modular equation for x x x , and the resulting recurrence relations needed to treat the associated function spaces. In § 3 we study the action of U ( i ) U^{(i)} U ( i ) on the function spaces and r L r_L r L . In § 3.1 we account for the additional factors induced by the non-unitary behavior of the associated localizing term. We give the complete proof of Theorem 1.2 in § 4 . In § 5 we show how to prove the 198 initial relations required for Theorem 2.3 in § 2.2 , together with the expression for L 1 L_1 L 1 above, and the modular equation for x x x . We end in § 6 with two families of conjectures also over X 0 ( 22 ) X_0(22) X 0 ( 22 ) , and briefly discuss the relevance of proof techniques different from the modular equation approach. § 7 provides explicit descriptions of the constants in Theorem 2.3, as well as the function c ( x ) c(x) c ( x ) mentioned in § 2.2 .
All the supplementary computations are documented in three Mathematica notebooks.3 Moreover, we have also verified these computations with PARI/GP.4
The Function Space of X 0 ( 22 ) X_0(22) X 0 ( 22 )
Construction of L α L_\alpha L α
To construct L α L_\alpha L α , we define, for τ ∈ H \tau\in\mathbb{H} τ ∈ H ,
A : = A ( τ ) : = q 15 D 2 ( q ) D 2 ( q 121 ) = q 15 ( q 2 ; q 2 ) ∞ 2 ( q 121 ; q 121 ) ∞ 7 ( q ; q ) ∞ 7 ( q 242 ; q 242 ) ∞ 2 , \mathcal{A}:=\mathcal{A}(\tau):=q^{15}\frac{D_2(q)}{D_2(q^{121})}=q^{15}\frac{(q^2;q^2)_\infty^2(q^{121};q^{121})_\infty^7}{(q;q)_\infty^7(q^{242};q^{242})_\infty^2}, A := A ( τ ) := q 15 D 2 ( q 121 ) D 2 ( q ) = q 15 ( q ; q ) ∞ 7 ( q 242 ; q 242 ) ∞ 2 ( q 2 ; q 2 ) ∞ 2 ( q 121 ; q 121 ) ∞ 7 ,
U ( 1 ) ( f ) : = U 11 ( f ) , U ( 0 ) ( f ) : = U 11 ( A ⋅ f ) , U ( α ) ( f ) : = U ( i ) ( f ) , α ≡ i ( m o d 2 ) , U^{(1)}(f):=U_{11}(f),\qquad U^{(0)}(f):=U_{11}(\mathcal{A}\cdot f),\qquad U^{(\alpha)}(f):=U^{(i)}(f),\quad\alpha\equiv i\pmod2, U ( 1 ) ( f ) := U 11 ( f ) , U ( 0 ) ( f ) := U 11 ( A ⋅ f ) , U ( α ) ( f ) := U ( i ) ( f ) , α ≡ i ( mod 2 ) ,
and the function sequence L : = ( L α ) α ≥ 1 \mathcal{L}:=(L_\alpha)_{\alpha\geq1} L := ( L α ) α ≥ 1 by
L α ( τ ) : = D 2 ( q 11 ( α ) ) − 1 ⋅ ∑ 8 n ≡ 1 ( m o d 11 α ) d 2 ( n ) q ⌊ n 11 α ⌋ + 1 + ( α ) , (5) L_\alpha(\tau):=D_2\left(q^{11^{(\alpha)}}\right)^{-1}\cdot\sum_{8n\equiv1\pmod{11^\alpha}}d_2(n)q^{\left\lfloor\frac{n}{11^\alpha}\right\rfloor+1+(\alpha)}, \tag*{(5)} L α ( τ ) := D 2 ( q 1 1 ( α ) ) − 1 ⋅ 8 n ≡ 1 ( mod 1 1 α ) ∑ d 2 ( n ) q ⌊ 1 1 α n ⌋ + 1 + ( α ) , (5)
with ( α ) ∈ { 0 , 1 } (\alpha)\in\{0,1\} ( α ) ∈ { 0 , 1 } such that ( α ) ≡ α ( m o d 2 ) (\alpha)\equiv\alpha\pmod2 ( α ) ≡ α ( mod 2 ) . By convention, we set L 0 : = 1 L_0:=1 L 0 := 1 .
Lemma 2.1. For all α ≥ 0 \alpha\ge0 α ≥ 0 , we have
L α + 1 = U ( α ) ( L α ) ∈ M ( X 0 ( 22 ) ) . L_{\alpha+1}=U^{(\alpha)}(L_\alpha)\in\mathcal{M}(X_0(22)). L α + 1 = U ( α ) ( L α ) ∈ M ( X 0 ( 22 )) .
Proof. This follows from a routine computation with eta quotients, see [7 Lemma 4.1] , [8 Lemma 2.7] , and [23 Lemma 2.13] for analogous arguments. □
Since A A A is an eta quotient, it has no zeros or poles in H \mathbb{H} H , hence the poles of L α L_\alpha L α lie at the cusps of X 0 ( 22 ) X_0(22) X 0 ( 22 ) . As noted before,5 the cusps of M ( X 0 ( 22 ) ) \mathcal{M}(X_0(22)) M ( X 0 ( 22 )) are [ 1 ] [1] [ 1 ] , [ 1 2 ] [\frac{1}{2}] [ 2 1 ] , [ 1 11 ] [\frac{1}{11}] [ 11 1 ] , [ 1 22 ] [\frac{1}{22}] [ 22 1 ] . We study the behavior of each L α L_\alpha L α at these cusps.
From (2.1), it is not hard to see that L α L_\alpha L α has positive order at [ 1 22 ] [\frac{1}{22}] [ 22 1 ] . For the remaining three cusps we use bounds given by [21 Definition 38, Theorem 47] . Let us denote the order of a function at the cusp [ a c ] [\frac{a}{c}] [ c a ] as ord a / c \operatorname{ord}_{a/c} ord a / c . Then we have
ord 1 ( L 2 α − 1 ) ≥ − 11 2 α + 1 2 , ord 1 ( L 2 α ) ≥ − 11 2 α + 1 + 11 2 , ord 1 / 11 ( L 2 α − 1 ) ≥ 11 2 α − 2 − 1 2 ⋅ 11 2 α − 1 ≥ 0 , \operatorname{ord}_{1}(L_{2\alpha-1})\ge\frac{-11^{2\alpha}+1}{2},\qquad
\operatorname{ord}_{1}(L_{2\alpha})\ge\frac{-11^{2\alpha+1}+11}{2},\qquad
\operatorname{ord}_{1/11}(L_{2\alpha-1})\ge\frac{11^{2\alpha-2}-1}{2\cdot11^{2\alpha-1}}\ge0, ord 1 ( L 2 α − 1 ) ≥ 2 − 1 1 2 α + 1 , ord 1 ( L 2 α ) ≥ 2 − 1 1 2 α + 1 + 11 , ord 1/11 ( L 2 α − 1 ) ≥ 2 ⋅ 1 1 2 α − 1 1 1 2 α − 2 − 1 ≥ 0 ,
ord 1 / 11 ( L 2 α ) ≥ 11 2 α − 1 2 ⋅ 11 2 α ≥ 0 , \operatorname{ord}_{1/11}(L_{2\alpha})\ge\frac{11^{2\alpha}-1}{2\cdot11^{2\alpha}}\ge0, ord 1/11 ( L 2 α ) ≥ 2 ⋅ 1 1 2 α 1 1 2 α − 1 ≥ 0 ,
and
ord 1 / 2 ( L 2 α − 1 ) ≥ − 11 2 α + 1 8 = − ψ ( 2 α − 1 ) , ord 1 / 2 ( L 2 α ) ≥ − 11 2 α + 1 + 11 8 = − ψ ( 2 α ) . (6) \operatorname{ord}_{1/2}(L_{2\alpha-1})\ge\frac{-11^{2\alpha}+1}{8}=-\psi(2\alpha-1),\qquad
\operatorname{ord}_{1/2}(L_{2\alpha})\ge\frac{-11^{2\alpha+1}+11}{8}=-\psi(2\alpha). \tag*{(6)} ord 1/2 ( L 2 α − 1 ) ≥ 8 − 1 1 2 α + 1 = − ψ ( 2 α − 1 ) , ord 1/2 ( L 2 α ) ≥ 8 − 1 1 2 α + 1 + 11 = − ψ ( 2 α ) . (6)
This proves that L α L_\alpha L α has negative order only at [ 0 ] [0] [ 0 ] and [ 1 2 ] [\frac{1}{2}] [ 2 1 ] , with well-defined lower bounds in both cases. Moreover, (2.2), together with the positive order of 2 + 11 x 2+11x 2 + 11 x at [ 1 2 ] [\frac{1}{2}] [ 2 1 ] , confirms (1.2).
Basis Functions For X 0 ( 22 ) X_0(22) X 0 ( 22 )
We come now to the question of constructing a suitable set of basis functions with which to express L α L_\alpha L α . Related functions x x x , y y y , z z z were already computed with a pole at [ ∞ ] [\infty] [ ∞ ] in [21 Section 4.1] , but we quickly explain the concept. We want functions of orders − 3 -3 − 3 , − 4 -4 − 4 and − 5 -5 − 5 at [ 0 ] [0] [ 0 ] . The space of modular eta quotients over Γ 0 ( 22 ) \Gamma_0(22) Γ 0 ( 22 ) with a pole only at [ 0 ] [0] [ 0 ] is a monoid which is generated by 7 elements. These can be quickly computed with existing algorithmic machinery. Of these elements, 3 of them have order − 5 -5 − 5 at [ 0 ] [0] [ 0 ] , while one has order − 6 -6 − 6 , one has order − 7 -7 − 7 , one with order − 8 -8 − 8 , and one with order − 9 -9 − 9 . Thus we can easily find a suitable function z z z . To get x x x and y y y we can simply compute the expansions of different combinations of order − 5 -5 − 5 eta quotients at [ 0 ] [0] [ 0 ] and subtract off the higher order terms. This can be done in different ways, but one suitable example is the following:
x = 1 44 ( − 7 + 11 ( q 2 ; q 2 ) ∞ 4 ( q 11 ; q 11 ) ∞ 8 ( q ; q ) ∞ 8 ( q 22 ; q 22 ) ∞ 4 − 4 ( q 2 ; q 2 ) ∞ 11 ( q 11 ; q 11 ) ∞ ( q ; q ) ∞ 11 ( q 22 ; q 22 ) ∞ + 132 q 5 ( q 2 ; q 2 ) ∞ 3 ( q 22 ; q 22 ) ∞ 7 ( q ; q ) ∞ 7 ( q 11 ; q 11 ) ∞ 3 ) (7) x=\frac{1}{44}\left(-7+11\frac{(q^2;q^2)_\infty^4(q^{11};q^{11})_\infty^8}{(q;q)_\infty^8(q^{22};q^{22})_\infty^4}-4\frac{(q^2;q^2)_\infty^{11}(q^{11};q^{11})_\infty}{(q;q)_\infty^{11}(q^{22};q^{22})_\infty}+132q^5\frac{(q^2;q^2)_\infty^3(q^{22};q^{22})_\infty^7}{(q;q)_\infty^7(q^{11};q^{11})_\infty^3}\right) \tag*{(7)} x = 44 1 ( − 7 + 11 ( q ; q ) ∞ 8 ( q 22 ; q 22 ) ∞ 4 ( q 2 ; q 2 ) ∞ 4 ( q 11 ; q 11 ) ∞ 8 − 4 ( q ; q ) ∞ 11 ( q 22 ; q 22 ) ∞ ( q 2 ; q 2 ) ∞ 11 ( q 11 ; q 11 ) ∞ + 132 q 5 ( q ; q ) ∞ 7 ( q 11 ; q 11 ) ∞ 3 ( q 2 ; q 2 ) ∞ 3 ( q 22 ; q 22 ) ∞ 7 ) (7)
y = 1 11 ( − 1 + ( q 2 ; q 2 ) ∞ 11 ( q 11 ; q 11 ) ∞ ( q ; q ) ∞ 11 ( q 22 ; q 22 ) ∞ − 121 q 5 ( q 2 ; q 2 ) ∞ 3 ( q 22 ; q 22 ) ∞ 7 ( q ; q ) ∞ 7 ( q 11 ; q 11 ) ∞ 3 ) (8) y=\frac{1}{11}\left(-1+\frac{(q^2;q^2)_\infty^{11}(q^{11};q^{11})_\infty}{(q;q)_\infty^{11}(q^{22};q^{22})_\infty}-121q^5\frac{(q^2;q^2)_\infty^3(q^{22};q^{22})_\infty^7}{(q;q)_\infty^7(q^{11};q^{11})_\infty^3}\right) \tag*{(8)} y = 11 1 ( − 1 + ( q ; q ) ∞ 11 ( q 22 ; q 22 ) ∞ ( q 2 ; q 2 ) ∞ 11 ( q 11 ; q 11 ) ∞ − 121 q 5 ( q ; q ) ∞ 7 ( q 11 ; q 11 ) ∞ 3 ( q 2 ; q 2 ) ∞ 3 ( q 22 ; q 22 ) ∞ 7 ) (8)
z = q 5 ( q 2 ; q 2 ) ∞ 3 ( q 22 ; q 22 ) ∞ 7 ( q ; q ) ∞ 7 ( q 11 ; q 11 ) ∞ 3 . (9) z=q^5\frac{(q^2;q^2)_\infty^3(q^{22};q^{22})_\infty^7}{(q;q)_\infty^7(q^{11};q^{11})_\infty^3}. \tag*{(9)} z = q 5 ( q ; q ) ∞ 7 ( q 11 ; q 11 ) ∞ 3 ( q 2 ; q 2 ) ∞ 3 ( q 22 ; q 22 ) ∞ 7 . (9)
Since ( q ; q ) n 11 ≡ ( q 11 ; q 11 ) n (q;q)_n^{11}\equiv(q^{11};q^{11})_n ( q ; q ) n 11 ≡ ( q 11 ; q 11 ) n (mod 11), the Fourier coefficients of x x x and y y y are integers. Note that r L r_L r L is also a series with integer coefficients, which is verified by checking the first 361 coefficients (Sturm bound) of the weight 120 form f r L : = η ( 22 τ ) 240 r L f_{r_L}:=\eta(22\tau)^{240}r_L f r L := η ( 22 τ ) 240 r L .
We now derive a recurrence for the action of U ( i ) U^{(i)} U ( i ) on powers of x x x . Both x = x ( τ ) x=x(\tau) x = x ( τ ) and x ( 11 τ ) x(11\tau) x ( 11 τ ) are modular functions for Γ 0 ( 242 ) \Gamma_0(242) Γ 0 ( 242 ) . On X 0 ( 242 ) X_0(242) X 0 ( 242 ) , the function x x x has a single pole, of order 33, at [ 0 ] [0] [ 0 ] , whereas x ( 11 τ ) x(11\tau) x ( 11 τ ) has poles only at [ 0 ] [0] [ 0 ] and [ 1 11 ] [\frac{1}{11}] [ 11 1 ] , of total order 33. Hence, by [14 Theorem 8.3] , x x x satisfies a monic polynomial equation of degree 33 over C ( x ( 11 τ ) ) \mathbb{C}(x(11\tau)) C ( x ( 11 τ )) . Since every pole of x x x is also a pole of x ( 11 τ ) x(11\tau) x ( 11 τ ) , the function x x x is integral over C [ x ( 11 τ ) ] \mathbb{C}[x(11\tau)] C [ x ( 11 τ )] , so the coefficients of this equation are polynomials: there exist a j ∈ C [ X ] a_j\in\mathbb{C}[X] a j ∈ C [ X ] , 0 ≤ j ≤ 32 0\le j\le32 0 ≤ j ≤ 32 , such that
x 33 + ∑ j = 0 32 a j ( x ( 11 τ ) ) x j = 0. (10) x^{33}+\sum_{j=0}^{32}a_j(x(11\tau))x^j=0. \tag*{(10)} x 33 + j = 0 ∑ 32 a j ( x ( 11 τ )) x j = 0. (10)
In fact a j ∈ Z [ X ] a_j\in\mathbb{Z}[X] a j ∈ Z [ X ] .6 This is verified directly by multiplying the left-hand side of (3.4) by a suitable eta quotient that moves all poles to [ ∞ ] [\infty] [ ∞ ] , and check that the resulting principal part and constant term vanish (see Section 6).
For a polynomial P P P we write P ( τ ) : = P ( x ( τ ) ) P(\tau):=P(x(\tau)) P ( τ ) := P ( x ( τ )) and thus U ( i ) ( P ( 11 τ ) f ) = P ( τ ) U ( i ) ( f ) U^{(i)}(P(11\tau)f)=P(\tau)U^{(i)}(f) U ( i ) ( P ( 11 τ ) f ) = P ( τ ) U ( i ) ( f ) , which is how a μ ( τ ) a_\mu(\tau) a μ ( τ ) and b ν ( τ ) b_\nu(\tau) b ν ( τ ) arise in Lemma 2.2 below. Setting t : = 2 + 11 x t:=2+11x t := 2 + 11 x , by (2.6) with x = t − 2 11 x=\frac{t-2}{11} x = 11 t − 2 , we get an associated modular equation for 2 + 11 x 2+11x 2 + 11 x that reads
t 33 + ∑ k = 0 32 b k ( x ( 11 τ ) ) t k = 0. (11) t^{33}+\sum_{k=0}^{32}b_k\bigl(x(11\tau)\bigr)t^k=0. \tag*{(11)} t 33 + k = 0 ∑ 32 b k ( x ( 11 τ ) ) t k = 0. (11)
Here b k ∈ Z [ 1 2 , x ] b_k\in\mathbb{Z}[\frac{1}{2},x] b k ∈ Z [ 2 1 , x ] . Precisely,8
b 0 ( τ ) = − 1 2 20 ( 2 + 11 x ) 11 c ( x ) with c ( x ) : = C 1 ( 0 ) + ∑ r = 1 22 C 1 ( r ) 11 ⌊ 11 r + 27 13 ⌋ x r ∈ Z [ x ] , b_0(\tau)=-\frac{1}{2^{20}}(2+11x)^{11}c(x)\quad\text{with}\quad c(x):=C_1(0)+\sum_{r=1}^{22}C_1(r)11^{\left\lfloor\frac{11r+27}{13}\right\rfloor}x^r\in\mathbb{Z}[x], b 0 ( τ ) = − 2 20 1 ( 2 + 11 x ) 11 c ( x ) with c ( x ) := C 1 ( 0 ) + r = 1 ∑ 22 C 1 ( r ) 1 1 ⌊ 13 11 r + 27 ⌋ x r ∈ Z [ x ] ,
C 1 ( 0 ) = 69550975755977284911104 ≡ 4 ( m o d 11 ) . C_1(0)=69550975755977284911104\equiv4\pmod{11}. C 1 ( 0 ) = 69550975755977284911104 ≡ 4 ( mod 11 ) .
Given the form of L α L_\alpha L α in Theorem 1.2, and the linearity of U ( i ) U^{(i)} U ( i ) , we want to study
U ( i ) ( y j z k x m ( 2 + 11 x ) n ) . (12) U^{(i)}\left(\frac{y^jz^kx^m}{(2+11x)^n}\right). \tag*{(12)} U ( i ) ( ( 2 + 11 x ) n y j z k x m ) . (12)
Precisely, we have the following:
Lemma 2.2. For all m , n ∈ Z m,n\in\mathbb{Z} m , n ∈ Z , and i , j , k ∈ { 0 , 1 } i,j,k\in\{0,1\} i , j , k ∈ { 0 , 1 } , with j + k ≤ 1 j+k\leq1 j + k ≤ 1 , we have
U ( i ) ( y j z k x m ( 2 + 11 x ) n ) = − 2 20 ( 2 + 11 x ) 11 c ( x ) ∑ μ = 0 32 ∑ ν = 1 33 a μ ( τ ) b ν ( τ ) U ( i ) ( y j z k x m + μ − 33 ( 2 + 11 x ) n − ν ) . U^{(i)}\left(\frac{y^jz^kx^m}{(2+11x)^n}\right)=\frac{-2^{20}}{(2+11x)^{11}c(x)}\sum_{\mu=0}^{32}\sum_{\nu=1}^{33}a_\mu(\tau)b_\nu(\tau)U^{(i)}\left(\frac{y^jz^kx^{m+\mu-33}}{(2+11x)^{n-\nu}}\right). U ( i ) ( ( 2 + 11 x ) n y j z k x m ) = ( 2 + 11 x ) 11 c ( x ) − 2 20 μ = 0 ∑ 32 ν = 1 ∑ 33 a μ ( τ ) b ν ( τ ) U ( i ) ( ( 2 + 11 x ) n − ν y j z k x m + μ − 33 ) .
Proof. This is immediate from (2.6) and (2.7).
Due to the additional nontrivial factor c ( x ) c(x) c ( x ) , we study more general rational polynomial spaces of the following form:
W n , d ( i ) : = { 1 ( 2 + 11 x ) n c ( x ) d ∑ j + k ≤ 1 y j z k ∑ m ≥ 1 − ( j + k ) s j k ( m ) 11 θ j k i ( m ) x m } . \mathcal{W}_{n,d}^{(i)}:=\left\{\frac{1}{(2+11x)^n c(x)^d}\sum_{j+k\leq1}y^jz^k\sum_{m\geq1-(j+k)}s_{jk}(m)11^{\theta_{jk}^{i}(m)}x^m\right\}. W n , d ( i ) := ⎩ ⎨ ⎧ ( 2 + 11 x ) n c ( x ) d 1 j + k ≤ 1 ∑ y j z k m ≥ 1 − ( j + k ) ∑ s j k ( m ) 1 1 θ j k i ( m ) x m ⎭ ⎬ ⎫ .
Set
θ j k i ( m ) : = max { 0 , ⌊ 11 ( m − 4 i + j + 2 k ) + 3 i − 7 13 ⌋ } . \theta_{jk}^{i}(m):=\max\left\{0,\left\lfloor\frac{11(m-4i+j+2k)+3i-7}{13}\right\rfloor\right\}. θ j k i ( m ) := max { 0 , ⌊ 13 11 ( m − 4 i + j + 2 k ) + 3 i − 7 ⌋ } .
We now give a more useful recurrence:
Theorem 2.3. For each ( i , j , k , γ , δ , m , n ) ∈ N 0 7 (i,j,k,\gamma,\delta,m,n)\in\mathbb{N}_0^7 ( i , j , k , γ , δ , m , n ) ∈ N 0 7 with i , j , k , γ , δ ∈ { 0 , 1 } i,j,k,\gamma,\delta\in\{0,1\} i , j , k , γ , δ ∈ { 0 , 1 } , such that j + k ≤ 1 j+k\leq1 j + k ≤ 1 and γ + δ ≤ 1 \gamma+\delta\leq1 γ + δ ≤ 1 , there exists h γ δ i j k ( m , n , r ) ∈ Z [ 1 2 ] h_{\gamma\delta}^{ijk}(m,n,r)\in\mathbb{Z}[\frac{1}{2}] h γ δ ij k ( m , n , r ) ∈ Z [ 2 1 ] with finite support in r r r , such that
U ( i ) ( y j z k x m ( 2 + 11 x ) n ) = 1 ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n ( k ( 1 − i ) 6 m 4 n L r + ∑ 0 ≤ γ + δ ≤ 1 ∑ r ≥ 1 − ( γ + δ ) h γ δ i j k ( m , n , r ) 11 π γ δ i j k ( m , r ) y γ z δ x r ) . (13) U^{(i)}\left(\frac{y^jz^kx^m}{(2+11x)^n}\right)
=
\frac{1}{(2+11x)^{11n+15(1-i)}c(x)^n}
\left(k(1-i)6^m4^nL_r+\sum_{0\leq\gamma+\delta\leq1}\sum_{r\geq1-(\gamma+\delta)}h_{\gamma\delta}^{ijk}(m,n,r)11^{\pi_{\gamma\delta}^{ijk}(m,r)}y^\gamma z^\delta x^r\right). \tag*{(13)} U ( i ) ( ( 2 + 11 x ) n y j z k x m ) = ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n 1 k ( 1 − i ) 6 m 4 n L r + 0 ≤ γ + δ ≤ 1 ∑ r ≥ 1 − ( γ + δ ) ∑ h γ δ ij k ( m , n , r ) 1 1 π γ δ ij k ( m , r ) y γ z δ x r . (13) with
π γ δ i j k ( m , r ) : = max { 0 , ⌊ 11 r − m + λ γ δ i j k 13 ⌋ } , \pi_{\gamma\delta}^{ijk}(m,r):=\max\left\{0,\left\lfloor\frac{11r-m+\lambda_{\gamma\delta}^{ijk}}{13}\right\rfloor\right\}, π γ δ ij k ( m , r ) := max { 0 , ⌊ 13 11 r − m + λ γ δ ij k ⌋ } , and λ γ δ i j k ∈ Z \lambda_{\gamma\delta}^{ijk}\in\mathbb{Z} λ γ δ ij k ∈ Z given in Section 7.
Proof. We prove by induction. Suppose that (13) holds for m − μ − 1 m-\mu-1 m − μ − 1 and n − ν n-\nu n − ν with m , n ∈ N ≥ 33 m,n\in\mathbb{N}_{\ge33} m , n ∈ N ≥ 33 fixed and 0 ≤ μ ≤ 32 0\leq\mu\leq32 0 ≤ μ ≤ 32 , 1 ≤ ν ≤ 33 1\leq\nu\leq33 1 ≤ ν ≤ 33 . We show that (13) holds for m , n ∈ N 0 m,n\in\mathbb{N}_0 m , n ∈ N 0 . By Lemma 2.2 , we have
U ( i ) ( y j z k x m ( 2 + 11 x ) n ) = − 2 20 ( 2 + 11 x ) 11 c ( x ) ∑ μ = 0 32 ∑ ν = 1 33 a μ ( τ ) b ν ( τ ) ⋅ U ( i ) ( y j z k x m + μ − 33 ( 2 + 11 x ) n − ν ) = − 2 20 ( 2 + 11 x ) 11 c ( x ) ∑ μ = 0 32 ∑ ν = 1 33 a μ ( τ ) b ν ( τ ) 1 ( 2 + 11 x ) 11 ( n − ν ) + 15 ( 1 − i ) c ( x ) n − ν × ( k ( 1 − i ) 6 m + μ − 33 4 n − ν r L + ∑ 0 ≤ γ + δ ≤ 1 r ≥ 0 h γ δ i j k ( m + μ − 33 , n − ν , r ) 11 π γ δ i j k ( m + μ − 33 , r ) y γ z δ x r ) = 1 ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n ( ∑ 0 ≤ μ ≤ 32 1 ≤ ν ≤ 33 W μ , ν ( τ ) 6 m + μ − 33 4 n − ν k ( 1 − i ) r L ) + 1 ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n ∑ 0 ≤ μ ≤ 32 , 1 ≤ ν ≤ 33 0 ≤ γ + δ ≤ 1 , r ≥ 0 W μ , ν ( τ ) h γ δ i j k ( m + μ − 33 , n − ν , r ) 11 π γ δ i j k ( m + μ − 33 , r ) y γ z δ x r . \begin{aligned}
U^{(i)}\left(\frac{y^jz^kx^m}{(2+11x)^n}\right)
&=\frac{-2^{20}}{(2+11x)^{11}c(x)}
\sum_{\mu=0}^{32}\sum_{\nu=1}^{33}a_\mu(\tau)b_\nu(\tau)\cdot U^{(i)}\left(\frac{y^jz^kx^{m+\mu-33}}{(2+11x)^{n-\nu}}\right)\\
&=\frac{-2^{20}}{(2+11x)^{11}c(x)}
\sum_{\mu=0}^{32}\sum_{\nu=1}^{33}a_\mu(\tau)b_\nu(\tau)
\frac{1}{(2+11x)^{11(n-\nu)+15(1-i)}c(x)^{n-\nu}}\\
&\quad\times\left(k(1-i)6^{m+\mu-33}4^{n-\nu}r_L+
\sum_{\substack{0\leq\gamma+\delta\leq1\\r\geq0}}
h_{\gamma\delta}^{ijk}(m+\mu-33,n-\nu,r)11^{\pi_{\gamma\delta}^{ijk}(m+\mu-33,r)}
y^\gamma z^\delta x^r\right)\\
&=\frac{1}{(2+11x)^{11n+15(1-i)}c(x)^n}
\left(\sum_{\substack{0\leq\mu\leq32\\1\leq\nu\leq33}}
W_{\mu,\nu}(\tau)6^{m+\mu-33}4^{n-\nu}k(1-i)r_L\right)\\
&\quad+\frac{1}{(2+11x)^{11n+15(1-i)}c(x)^n}
\sum_{\substack{0\leq\mu\leq32,\ 1\leq\nu\leq33\\0\leq\gamma+\delta\leq1,\ r\geq0}}
W_{\mu,\nu}(\tau)h_{\gamma\delta}^{ijk}(m+\mu-33,n-\nu,r)11^{\pi_{\gamma\delta}^{ijk}(m+\mu-33,r)}
y^\gamma z^\delta x^r.
\end{aligned} U ( i ) ( ( 2 + 11 x ) n y j z k x m ) = ( 2 + 11 x ) 11 c ( x ) − 2 20 μ = 0 ∑ 32 ν = 1 ∑ 33 a μ ( τ ) b ν ( τ ) ⋅ U ( i ) ( ( 2 + 11 x ) n − ν y j z k x m + μ − 33 ) = ( 2 + 11 x ) 11 c ( x ) − 2 20 μ = 0 ∑ 32 ν = 1 ∑ 33 a μ ( τ ) b ν ( τ ) ( 2 + 11 x ) 11 ( n − ν ) + 15 ( 1 − i ) c ( x ) n − ν 1 × k ( 1 − i ) 6 m + μ − 33 4 n − ν r L + 0 ≤ γ + δ ≤ 1 r ≥ 0 ∑ h γ δ ij k ( m + μ − 33 , n − ν , r ) 1 1 π γ δ ij k ( m + μ − 33 , r ) y γ z δ x r = ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n 1 0 ≤ μ ≤ 32 1 ≤ ν ≤ 33 ∑ W μ , ν ( τ ) 6 m + μ − 33 4 n − ν k ( 1 − i ) r L + ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n 1 0 ≤ μ ≤ 32 , 1 ≤ ν ≤ 33 0 ≤ γ + δ ≤ 1 , r ≥ 0 ∑ W μ , ν ( τ ) h γ δ ij k ( m + μ − 33 , n − ν , r ) 1 1 π γ δ ij k ( m + μ − 33 , r ) y γ z δ x r .
where 10
W μ , ν ( τ ) : = − 2 20 a μ ( τ ) b ν ( τ ) ( 2 + 11 x ) 11 ( ν − 1 ) c ( x ) ν − 1 = : ∑ ℓ ≥ 0 w μ , ν ( ℓ ) 11 ϕ ( ℓ , μ ) x ℓ , W_{\mu,\nu}(\tau):=-2^{20}a_\mu(\tau)b_\nu(\tau)(2+11x)^{11(\nu-1)}c(x)^{\nu-1}
=:\sum_{\ell\geq0}w_{\mu,\nu}(\ell)11^{\phi(\ell,\mu)}x^\ell, W μ , ν ( τ ) := − 2 20 a μ ( τ ) b ν ( τ ) ( 2 + 11 x ) 11 ( ν − 1 ) c ( x ) ν − 1 =: ℓ ≥ 0 ∑ w μ , ν ( ℓ ) 1 1 ϕ ( ℓ , μ ) x ℓ ,
ϕ ( ℓ , μ ) : = max { 0 , ⌊ 11 ℓ + μ − 21 13 ⌋ } . \phi(\ell,\mu):=\max\left\{0,\left\lfloor\frac{11\ell+\mu-21}{13}\right\rfloor\right\}. ϕ ( ℓ , μ ) := max { 0 , ⌊ 13 11 ℓ + μ − 21 ⌋ } .
Thus, the above becomes
= 1 ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n ∑ 0 ≤ μ ≤ 32 1 ≤ ν ≤ 33 W μ , ν ( τ ) 6 m + μ − 33 4 n − ν k ( 1 − i ) r L + 1 ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n ∑ 0 ≤ μ ≤ 32 , 1 ≤ ν ≤ 33 0 ≤ γ + δ ≤ 1 , r ≥ 0 , ℓ ≥ 0 w μ , ν ( ℓ ) h γ δ i j k ( m + μ − 33 , n − ν , r ) 11 π γ δ i j k ( m + μ − 33 , r ) + ϕ ( ℓ , μ ) y γ z δ x r + ℓ . (14) \begin{aligned}
&=\frac{1}{(2+11x)^{11n+15(1-i)}c(x)^n}
\sum_{\substack{0\leq\mu\leq32\\1\leq\nu\leq33}}
W_{\mu,\nu}(\tau)6^{m+\mu-33}4^{n-\nu}k(1-i)r_L\\
&\quad+\frac{1}{(2+11x)^{11n+15(1-i)}c(x)^n}
\sum_{\substack{0\leq\mu\leq32,\ 1\leq\nu\leq33\\0\leq\gamma+\delta\leq1,\ r\geq0,\ \ell\geq0}}
w_{\mu,\nu}(\ell)h_{\gamma\delta}^{ijk}(m+\mu-33,n-\nu,r)11^{\pi_{\gamma\delta}^{ijk}(m+\mu-33,r)+\phi(\ell,\mu)}
y^\gamma z^\delta x^{r+\ell}. \tag*{(14)}
\end{aligned} = ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n 1 0 ≤ μ ≤ 32 1 ≤ ν ≤ 33 ∑ W μ , ν ( τ ) 6 m + μ − 33 4 n − ν k ( 1 − i ) r L + ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n 1 0 ≤ μ ≤ 32 , 1 ≤ ν ≤ 33 0 ≤ γ + δ ≤ 1 , r ≥ 0 , ℓ ≥ 0 ∑ w μ , ν ( ℓ ) h γ δ ij k ( m + μ − 33 , n − ν , r ) 1 1 π γ δ ij k ( m + μ − 33 , r ) + ϕ ( ℓ , μ ) y γ z δ x r + ℓ . (14)
Next, we bound
π γ δ i j k ( m + μ − 33 , r ) + ϕ ( ℓ , μ ) ≥ ⌊ 11 r − ( m + μ − 33 ) + λ γ δ i j k 13 ⌋ + ⌊ 11 ℓ + μ − 21 13 ⌋ ≥ ⌊ 11 ( r + ℓ ) − m + λ γ δ i j k + 33 − 21 − 12 13 ⌋ ≥ π γ δ i j k ( m , r + ℓ ) . \begin{aligned}
\pi_{\gamma\delta}^{ijk}(m+\mu-33,r)+\phi(\ell,\mu)
&\geq\left\lfloor\frac{11r-(m+\mu-33)+\lambda_{\gamma\delta}^{ijk}}{13}\right\rfloor
+\left\lfloor\frac{11\ell+\mu-21}{13}\right\rfloor\\
&\geq\left\lfloor\frac{11(r+\ell)-m+\lambda_{\gamma\delta}^{ijk}+33-21-12}{13}\right\rfloor
\geq\pi_{\gamma\delta}^{ijk}(m,r+\ell).
\end{aligned} π γ δ ij k ( m + μ − 33 , r ) + ϕ ( ℓ , μ ) ≥ ⌊ 13 11 r − ( m + μ − 33 ) + λ γ δ ij k ⌋ + ⌊ 13 11 ℓ + μ − 21 ⌋ ≥ ⌊ 13 11 ( r + ℓ ) − m + λ γ δ ij k + 33 − 21 − 12 ⌋ ≥ π γ δ ij k ( m , r + ℓ ) .
This confirms that the second sum in (14) has proper 11 11 11 -adic valuation. Moreover, since ℓ ≥ 0 \ell\geq0 ℓ ≥ 0 , we have
r + ℓ ≥ r ≥ 1 − ( γ + δ ) . r+\ell\geq r\geq1-(\gamma+\delta). r + ℓ ≥ r ≥ 1 − ( γ + δ ) .
Indeed, the induction is complete except for the case that i = 0 i=0 i = 0 and k = 1 k=1 k = 1 , in which case the first sum of (14) remains to be examined. For the rest of the proof, assume i = 0 i=0 i = 0 , k = 1 k=1 k = 1 . Precisely, we now study the following function:
W μ , ν ( τ ) 6 m + μ − 33 4 n − ν r L = − 2 20 6 m + μ − 33 4 n − ν a μ ( τ ) b ν ( τ ) ( 2 + 11 x ) 11 ( ν − 1 ) c ( x ) ν − 1 r L . (15) W_{\mu,\nu}(\tau)6^{m+\mu-33}4^{n-\nu}r_L
=-2^{20}6^{m+\mu-33}4^{n-\nu}a_\mu(\tau)b_\nu(\tau)(2+11x)^{11(\nu-1)}c(x)^{\nu-1}r_L. \tag*{(15)} W μ , ν ( τ ) 6 m + μ − 33 4 n − ν r L = − 2 20 6 m + μ − 33 4 n − ν a μ ( τ ) b ν ( τ ) ( 2 + 11 x ) 11 ( ν − 1 ) c ( x ) ν − 1 r L . (15)
We show 11 that when at least one of μ \mu μ , ν \nu ν is not divisible by 11, (15) expands into a polynomial with an 11-adic lower bound of
max { 0 , ⌊ 11 r − 33 + λ γ δ 001 13 ⌋ } ≥ max { 0 , ⌊ 11 r − m + λ γ δ 001 13 ⌋ } = π γ δ 001 ( m , r ) \max\left\{0,\left\lfloor\frac{11r-33+\lambda_{\gamma\delta}^{001}}{13}\right\rfloor\right\}\ge\max\left\{0,\left\lfloor\frac{11r-m+\lambda_{\gamma\delta}^{001}}{13}\right\rfloor\right\}=\pi_{\gamma\delta}^{001}(m,r) max { 0 , ⌊ 13 11 r − 33 + λ γ δ 001 ⌋ } ≥ max { 0 , ⌊ 13 11 r − m + λ γ δ 001 ⌋ } = π γ δ 001 ( m , r )
for γ γ δ x r \gamma^{\gamma\delta}x^{r} γ γ δ x r (since m ≥ 33 m\ge33 m ≥ 33 ).
The exceptional cases occur when both μ \mu μ and ν \nu ν are multiples of 11, i.e., when the pair ( μ , ν ) = ( 11 λ , 11 κ ) (\mu,\nu)=(11\lambda,11\kappa) ( μ , ν ) = ( 11 λ , 11 κ ) , for 0 ≤ λ ≤ 2 0\le\lambda\le2 0 ≤ λ ≤ 2 , 1 ≤ κ ≤ 3 1\le\kappa\le3 1 ≤ κ ≤ 3 . Combining these terms, we find 12 that
∑ 0 ≤ μ ≤ 32 , 1 ≤ ν ≤ 33 W μ , ν ( τ ) 6 m + μ − 33 4 n − ν r L = f 0 + 6 m − 33 4 n ∑ 0 ≤ μ ≤ 2 , 1 ≤ ν ≤ 3 W 11 μ , 11 ν ( τ ) 6 11 μ 4 − 11 ν r L = f 0 + 6 m − 33 4 n ( f 1 + 1 2 24 ( f 2 + 2 r L ) ) , \begin{aligned}
\sum_{\substack{0\le\mu\le32,\\1\le\nu\le33}}W_{\mu,\nu}(\tau)6^{m+\mu-33}4^{n-\nu}r_L
=f_0+6^{m-33}4^n\sum_{\substack{0\le\mu\le2,\\1\le\nu\le3}}W_{11\mu,11\nu}(\tau)6^{11\mu}4^{-11\nu}r_L \\
&=f_0+6^{m-33}4^n\left(f_1+\frac{1}{2^{24}}(f_2+2r_L)\right),
\end{aligned} 0 ≤ μ ≤ 32 , 1 ≤ ν ≤ 33 ∑ W μ , ν ( τ ) 6 m + μ − 33 4 n − ν r L = f 0 + 6 m − 33 4 n 0 ≤ μ ≤ 2 , 1 ≤ ν ≤ 3 ∑ W 11 μ , 11 ν ( τ ) 6 11 μ 4 − 11 ν r L = f 0 + 6 m − 33 4 n ( f 1 + 2 24 1 ( f 2 + 2 r L ) ) ,
where f 0 f_0 f 0 , f 1 f_1 f 1 , and f 2 f_2 f 2 satisfy the valuation requirements, and can be absorbed into (2.10). Thus, we next study 6 m − 33 4 n 2 23 r L \frac{6^{m-33}4^n}{2^{23}}r_L 2 23 6 m − 33 4 n r L . Using the fact that 2 − 236 m − 33 4 n ≡ 6 m 4 n ( m o d 11 ) 2^{-236^{m-33}4^n}\equiv6^{m}4^n\pmod{11} 2 − 23 6 m − 33 4 n ≡ 6 m 4 n ( mod 11 ) , we have
6 m − 33 4 n 2 23 r L = 6 m 4 n r L + f 3 with f 3 ∈ 11 r L Z [ 1 2 ] , \frac{6^{m-33}4^n}{2^{23}}r_L=6^m4^nr_L+f_3\quad\text{with}\quad f_3\in11r_L\mathbb{Z}\left[\frac{1}{2}\right], 2 23 6 m − 33 4 n r L = 6 m 4 n r L + f 3 with f 3 ∈ 11 r L Z [ 2 1 ] ,
where f 3 f_3 f 3 again has the required 11-adic valuation properties. This concludes 13 the proof of (13) . □
Next we study the denominator of (12) by only varying its powers.
Lemma 2.4. Let i , j , k ∈ { 0 , 1 } i,j,k\in\{0,1\} i , j , k ∈ { 0 , 1 } such that j + k ≤ 1 j+k\le1 j + k ≤ 1 and ( i , k ) ≠ ( 0 , 1 ) (i,k)\ne(0,1) ( i , k ) = ( 0 , 1 ) . Let m , r ∈ N 0 m,r\in\mathbb{N}_0 m , r ∈ N 0 , and n ∈ N ≥ 33 n\in\mathbb{N}_{\ge33} n ∈ N ≥ 33 . We have
h γ δ i j k ( m , n , r ) ≡ h γ δ i j k ( m , n − 11 , r ) + 7 h γ δ i j k ( m , n − 22 , r ) + 9 h γ δ i j k ( m , n − 33 , r ) ( m o d 11 ) . h_{\gamma\delta}^{ijk}(m,n,r)\equiv h_{\gamma\delta}^{ijk}(m,n-11,r)+7h_{\gamma\delta}^{ijk}(m,n-22,r)+9h_{\gamma\delta}^{ijk}(m,n-33,r)\pmod{11}. h γ δ ij k ( m , n , r ) ≡ h γ δ ij k ( m , n − 11 , r ) + 7 h γ δ ij k ( m , n − 22 , r ) + 9 h γ δ ij k ( m , n − 33 , r ) ( mod 11 ) .
Proof. For m m m fixed, by (11) , we have
U ( i ) ( y j z k x m ( 2 + 11 x ) n ) = − 1 b 0 ∑ ν = 1 33 b ν U ( i ) ( y j z k x m ( 2 + 11 x ) n − ν ) = 2 20 ( 2 + 11 x ) 11 c ( x ) ∑ ν = 1 33 b ν U ( i ) ( y j z k x m ( 2 + 11 x ) n − ν ) = 2 20 ( 2 + 11 x ) 11 c ( x ) ∑ ν = 1 33 b ν ( 2 + 11 x ) 11 ( n − ν ) + 15 ( 1 − i ) c ( x ) n − ν ∑ 0 ≤ γ + δ ≤ 1 , r ≥ 1 − ( γ + δ ) h γ δ i j k ( m , n − ν , r ) 11 π γ δ i j k ( m , r ) y γ z δ x r = 1 ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n ∑ ν = 1 33 2 20 b ν ( 2 + 11 x ) 11 ( ν − 1 ) c ( x ) ν − 1 ∑ 0 ≤ γ + δ ≤ 1 , r ≥ 1 − ( γ + δ ) h γ δ i j k ( m , n − ν , r ) 11 π γ δ i j k ( m , r ) y γ z δ x r . (16) \begin{aligned}
U^{(i)}\left(\frac{y^jz^kx^m}{(2+11x)^n}\right)=-\frac{1}{b_0}\sum_{\nu=1}^{33}b_\nu U^{(i)}\left(\frac{y^jz^kx^m}{(2+11x)^{n-\nu}}\right)=\frac{2^{20}}{(2+11x)^{11c(x)}}\sum_{\nu=1}^{33}b_\nu U^{(i)}\left(\frac{y^jz^kx^m}{(2+11x)^{n-\nu}}\right) \\
&=\frac{2^{20}}{(2+11x)^{11c(x)}}\sum_{\nu=1}^{33}\frac{b_\nu}{(2+11x)^{11(n-\nu)+15(1-i)c(x)^{n-\nu}}}\sum_{\substack{0\le\gamma+\delta\le1,\\r\ge1-(\gamma+\delta)}}h_{\gamma\delta}^{ijk}(m,n-\nu,r)11^{\pi_{\gamma\delta}^{ijk}(m,r)}y^\gamma z^\delta x^r \\
&=\frac{1}{(2+11x)^{11n+15(1-i)c(x)^n}}\sum_{\nu=1}^{33}2^{20}b_\nu(2+11x)^{11(\nu-1)}c(x)^{\nu-1}\sum_{\substack{0\le\gamma+\delta\le1,\\r\ge1-(\gamma+\delta)}}h_{\gamma\delta}^{ijk}(m,n-\nu,r)11^{\pi_{\gamma\delta}^{ijk}(m,r)}y^\gamma z^\delta x^r. \tag*{(16)}
\end{aligned} U ( i ) ( ( 2 + 11 x ) n y j z k x m ) = − b 0 1 ν = 1 ∑ 33 b ν U ( i ) ( ( 2 + 11 x ) n − ν y j z k x m ) = ( 2 + 11 x ) 11 c ( x ) 2 20 ν = 1 ∑ 33 b ν U ( i ) ( ( 2 + 11 x ) n − ν y j z k x m ) = ( 2 + 11 x ) 11 c ( x ) 2 20 ν = 1 ∑ 33 ( 2 + 11 x ) 11 ( n − ν ) + 15 ( 1 − i ) c ( x ) n − ν b ν 0 ≤ γ + δ ≤ 1 , r ≥ 1 − ( γ + δ ) ∑ h γ δ ij k ( m , n − ν , r ) 1 1 π γ δ ij k ( m , r ) y γ z δ x r = ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n 1 ν = 1 ∑ 33 2 20 b ν ( 2 + 11 x ) 11 ( ν − 1 ) c ( x ) ν − 1 0 ≤ γ + δ ≤ 1 , r ≥ 1 − ( γ + δ ) ∑ h γ δ ij k ( m , n − ν , r ) 1 1 π γ δ ij k ( m , r ) y γ z δ x r . (16)
Define
Q ( ν , τ ) : = 2 20 b ν ( 2 + 11 x ) 11 ( ν − 1 ) c ( x ) ν − 1 = : ∑ ℓ ≥ 0 Q ν ( ℓ ) 11 κ ( ℓ , ν ) x ℓ , Q(\nu,\tau):=2^{20}b_\nu(2+11x)^{11(\nu-1)}c(x)^{\nu-1}=: \sum_{\ell\ge0}Q_\nu(\ell)11^{\kappa(\ell,\nu)}x^\ell, Q ( ν , τ ) := 2 20 b ν ( 2 + 11 x ) 11 ( ν − 1 ) c ( x ) ν − 1 =: ℓ ≥ 0 ∑ Q ν ( ℓ ) 1 1 κ ( ℓ , ν ) x ℓ ,
with 14
κ ( ℓ , ν ) : = { 0 ν ∈ { 11 , 22 , 33 } and ℓ = 0 , ⌊ 11 ℓ + 25 13 ⌋ otherwise . \kappa(\ell,\nu):=
\begin{cases}
0 & \nu\in\{11,22,33\}\text{ and }\ell=0,\\
\left\lfloor\frac{11\ell+25}{13}\right\rfloor& \text{otherwise}.
\end{cases} κ ( ℓ , ν ) := { 0 ⌊ 13 11 ℓ + 25 ⌋ ν ∈ { 11 , 22 , 33 } and ℓ = 0 , otherwise .
Thus, (2.12) becomes
U ( i ) ( y j z k x m ( 2 + 11 x ) n ) = 1 ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n ∑ 1 ≤ ν ≤ 33 , ℓ ≥ 0 Q ν ( ℓ ) 11 κ ( ℓ , ν ) x ℓ ∑ 0 ≤ γ + δ ≤ 1 , r ≥ 1 − ( γ + δ ) h γ δ i j k ( m , n − ν , r ) 11 π γ δ i j k ( m , r ) y γ z δ x r = 1 ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n ∑ 1 ≤ ν ≤ 33 , ℓ ≥ 0 , 0 ≤ γ + δ ≤ 1 , r ≥ 1 − ( γ + δ ) Q ν ( ℓ ) h γ δ i j k ( m , n − ν , r ) 11 π γ δ i j k ( m , r − ℓ ) + κ ( ℓ , ν ) y γ z δ x r + ℓ . (17) \begin{aligned}
U^{(i)}\left(\frac{y^jz^kx^m}{(2+11x)^n}\right)
=
\frac{1}{(2+11x)^{11n+15(1-i)c(x)^n}}
\sum_{\substack{1\leq\nu\leq33,\\ \ell\geq0}}
Q_\nu(\ell)11^{\kappa(\ell,\nu)}x^\ell
\sum_{\substack{0\leq\gamma+\delta\leq1,\\ r\geq1-(\gamma+\delta)}}
h_{\gamma\delta}^{ijk}(m,n-\nu,r)11^{\pi_{\gamma\delta}^{ijk}(m,r)}y^\gamma z^\delta x^r \\
&=
\frac{1}{(2+11x)^{11n+15(1-i)c(x)^n}}
\sum_{\substack{1\leq\nu\leq33,\\ \ell\geq0,\\ 0\leq\gamma+\delta\leq1,\\ r\geq1-(\gamma+\delta)}}
Q_\nu(\ell)h_{\gamma\delta}^{ijk}(m,n-\nu,r)11^{\pi_{\gamma\delta}^{ijk}(m,r-\ell)+\kappa(\ell,\nu)}y^\gamma z^\delta x^{r+\ell}. \tag*{(17)}
\end{aligned} U ( i ) ( ( 2 + 11 x ) n y j z k x m ) = ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n 1 1 ≤ ν ≤ 33 , ℓ ≥ 0 ∑ Q ν ( ℓ ) 1 1 κ ( ℓ , ν ) x ℓ 0 ≤ γ + δ ≤ 1 , r ≥ 1 − ( γ + δ ) ∑ h γ δ ij k ( m , n − ν , r ) 1 1 π γ δ ij k ( m , r ) y γ z δ x r = ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n 1 1 ≤ ν ≤ 33 , ℓ ≥ 0 , 0 ≤ γ + δ ≤ 1 , r ≥ 1 − ( γ + δ ) ∑ Q ν ( ℓ ) h γ δ ij k ( m , n − ν , r ) 1 1 π γ δ ij k ( m , r − ℓ ) + κ ( ℓ , ν ) y γ z δ x r + ℓ . (17)
On the other hand, since ( i , k ) ≠ ( 0 , 1 ) (i,k)\ne(0,1) ( i , k ) = ( 0 , 1 ) , the anomalous term in (13) vanishes, so (13) expresses the left-hand side of (17) in terms of h γ δ i j k ( m , n , r ) h_{\gamma\delta}^{ijk}(m,n,r) h γ δ ij k ( m , n , r ) . Thus, we obtain
U ( i ) ( y j z k x m ( 2 + 11 x ) n ) = 1 ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n ∑ 0 ≤ γ + δ ≤ 1 , r ≥ 1 − ( γ + δ ) h γ δ i j k ( m , n , r ) 11 π γ δ i j k ( m , r ) y γ z δ x r . (18) U^{(i)}\left(\frac{y^jz^kx^m}{(2+11x)^n}\right)
=
\frac{1}{(2+11x)^{11n+15(1-i)c(x)^n}}
\sum_{\substack{0\leq\gamma+\delta\leq1,\\ r\geq1-(\gamma+\delta)}}
h_{\gamma\delta}^{ijk}(m,n,r)11^{\pi_{\gamma\delta}^{ijk}(m,r)}y^\gamma z^\delta x^r. \tag*{(18)} U ( i ) ( ( 2 + 11 x ) n y j z k x m ) = ( 2 + 11 x ) 11 n + 15 ( 1 − i ) c ( x ) n 1 0 ≤ γ + δ ≤ 1 , r ≥ 1 − ( γ + δ ) ∑ h γ δ ij k ( m , n , r ) 1 1 π γ δ ij k ( m , r ) y γ z δ x r . (18)
As such, (17) and (18) give us a means to study the auxiliary coefficients h γ δ i j k ( m , n , r ) h_{\gamma\delta}^{ijk}(m,n,r) h γ δ ij k ( m , n , r ) .
Comparing the coefficients of y γ z δ x r y^\gamma z^\delta x^r y γ z δ x r in (13) and (17) gives
h γ δ i j k ( m , n , r ) 11 π γ δ i j k ( m , r ) = ∑ ν = 1 33 ∑ ℓ = 0 r Q ν ( ℓ ) h γ δ i j k ( m , n − ν , r − ℓ ) 11 π γ δ i j k ( m , r − ℓ ) + κ ( ℓ , ν ) . h_{\gamma\delta}^{ijk}(m,n,r)11^{\pi_{\gamma\delta}^{ijk}(m,r)}
=
\sum_{\nu=1}^{33}\sum_{\ell=0}^{r}Q_\nu(\ell)h_{\gamma\delta}^{ijk}(m,n-\nu,r-\ell)11^{\pi_{\gamma\delta}^{ijk}(m,r-\ell)+\kappa(\ell,\nu)}. h γ δ ij k ( m , n , r ) 1 1 π γ δ ij k ( m , r ) = ν = 1 ∑ 33 ℓ = 0 ∑ r Q ν ( ℓ ) h γ δ ij k ( m , n − ν , r − ℓ ) 1 1 π γ δ ij k ( m , r − ℓ ) + κ ( ℓ , ν ) .
Unless ℓ = 0 \ell=0 ℓ = 0 and ν ∈ { 11 , 22 , 33 } \nu\in\{11,22,33\} ν ∈ { 11 , 22 , 33 } , we have
π γ δ i j k ( m , r − ℓ ) + κ ( ℓ , ν ) ≥ π γ δ i j k ( m , r ) + 1. \pi_{\gamma\delta}^{ijk}(m,r-\ell)+\kappa(\ell,\nu)\geq\pi_{\gamma\delta}^{ijk}(m,r)+1. π γ δ ij k ( m , r − ℓ ) + κ ( ℓ , ν ) ≥ π γ δ ij k ( m , r ) + 1.
Dividing by 11 π γ δ i j k ( m , r ) 11^{\pi_{\gamma\delta}^{ijk}(m,r)} 1 1 π γ δ ij k ( m , r ) and reducing modulo 11, we obtain
h γ δ i j k ( m , n , r ) ≡ Q 11 ( 0 ) h γ δ i j k ( m , n − 11 , r ) + Q 22 ( 0 ) h γ δ i j k ( m , n − 22 , r ) + Q 33 ( 0 ) h γ δ i j k ( m , n − 33 , r ) ( m o d 11 ) . h_{\gamma\delta}^{ijk}(m,n,r)\equiv Q_{11}(0)h_{\gamma\delta}^{ijk}(m,n-11,r)+Q_{22}(0)h_{\gamma\delta}^{ijk}(m,n-22,r)+Q_{33}(0)h_{\gamma\delta}^{ijk}(m,n-33,r)\pmod{11}. h γ δ ij k ( m , n , r ) ≡ Q 11 ( 0 ) h γ δ ij k ( m , n − 11 , r ) + Q 22 ( 0 ) h γ δ ij k ( m , n − 22 , r ) + Q 33 ( 0 ) h γ δ ij k ( m , n − 33 , r ) ( mod 11 ) .
and the lemma follows from Q 11 ( 0 ) ≡ 1 Q_{11}(0)\equiv1 Q 11 ( 0 ) ≡ 1 , Q 22 ( 0 ) ≡ 7 Q_{22}(0)\equiv7 Q 22 ( 0 ) ≡ 7 , Q 33 ( 0 ) ≡ 9 ( m o d 11 ) Q_{33}(0)\equiv9\pmod{11} Q 33 ( 0 ) ≡ 9 ( mod 11 ) .
Action of the U ( i ) U^{(i)} U ( i ) Operators
We are now ready to study the application of U ( i ) U^{(i)} U ( i ) to the spaces W n , d ( i ) \mathcal{W}_{n,d}^{(i)} W n , d ( i ) .
Lemma 3.1. For m 1 , m 2 ∈ Z m_1,m_2\in\mathbb{Z} m 1 , m 2 ∈ Z and n ∈ N 0 n\in\mathbb{N}_0 n ∈ N 0 ,
U ( 1 ) ( ( m 1 x + m 2 ) r L ( 2 + 11 x ) n ) ∈ W 11 n , n ( 0 ) . U^{(1)}\left(\frac{(m_1x+m_2)r_L}{(2+11x)^n}\right)\in\mathcal{W}_{11n,n}^{(0)}. U ( 1 ) ( ( 2 + 11 x ) n ( m 1 x + m 2 ) r L ) ∈ W 11 n , n ( 0 ) . Proof. Analogous to the proof of Lemma 2.4. By (16) , we need to verify the claim for 0 ≤ n ≤ 32 0\leq n\leq32 0 ≤ n ≤ 32 . Moreover, since U ( 1 ) U^{(1)} U ( 1 ) linear, we only need to consider the cases ( m 1 , m 2 ) ∈ { ( 1 , 0 ) , ( 0 , 1 ) } (m_1,m_2)\in\{(1,0),(0,1)\} ( m 1 , m 2 ) ∈ {( 1 , 0 ) , ( 0 , 1 )} . We verify these cases in Section 3a of the supplement file.
Next, we study action of U ( 1 ) U^{(1)} U ( 1 ) on f f f .
Theorem 3.2. For f ∈ W n , 0 ( 1 ) f\in\mathcal{W}_{n,0}^{(1)} f ∈ W n , 0 ( 1 ) , we have
U ( 1 ) ( f ) ∈ W 11 n , n ( 0 ) . U^{(1)}(f)\in\mathcal{W}_{11n,n}^{(0)}. U ( 1 ) ( f ) ∈ W 11 n , n ( 0 ) . Proof. Let
f : = 1 ( 2 + 11 x ) n ∑ j + k ≤ 1 y j z k ∑ m ≥ 1 − ( j + k ) s j k ( m ) 11 θ j k 1 ( m ) x m , f:=\frac{1}{(2+11x)^n}\sum_{j+k\leq1}y^jz^k\sum_{m\geq1-(j+k)}s_{jk}(m)11^{\theta_{jk}^{1}(m)}x^m, f := ( 2 + 11 x ) n 1 j + k ≤ 1 ∑ y j z k m ≥ 1 − ( j + k ) ∑ s j k ( m ) 1 1 θ j k 1 ( m ) x m ,
where s j k ( m ) ∈ Z [ 1 2 ] s_{jk}(m) \in\mathbb{Z}\left[\frac{1}{2}\right] s j k ( m ) ∈ Z [ 2 1 ] has finite support in m m m . Note that
U ( 1 ) ( f ) = ∑ j + k ≤ 1 ∑ m ≥ 1 − ( j + k ) s j k ( m ) 11 θ j k 1 ( m ) U ( 1 ) ( y j z k x m ( 2 + 11 x ) n ) . U^{(1)}(f)=\sum_{j+k\leq1}\sum_{m\geq1-(j+k)}s_{jk}(m)11^{\theta^1_{jk}(m)}U^{(1)}\left(\frac{y^j z^k x^m}{(2+11x)^n}\right). U ( 1 ) ( f ) = j + k ≤ 1 ∑ m ≥ 1 − ( j + k ) ∑ s j k ( m ) 1 1 θ j k 1 ( m ) U ( 1 ) ( ( 2 + 11 x ) n y j z k x m ) .
Now, by Theorem 2.3,
U^{(1)}(f)=\frac{1}{(2+11x)^{11n}c(x)^n}\sum_{\substack{j+k\leq1\\m\geq1-(j+k)}}\sum_{\substack{0\leq\gamma+\delta\leq1\\r\geq0}}s_{jk}(m)1^{j k}_{\gamma\delta}(m,r)11^{\pi^{j k}_{\gamma\delta}(m,r)+\theta^1_{jk}(m)}y^\gamma\delta^x^r.
Thus, it suffices to show that
π γ δ 1 j k ( m , r ) + θ j k 1 ( m ) ≥ θ γ δ 0 ( r ) . (19) \pi^{1jk}_{\gamma\delta}(m,r)+\theta^1_{jk}(m)\geq\theta^0_{\gamma\delta}(r). \tag*{(19)} π γ δ 1 j k ( m , r ) + θ j k 1 ( m ) ≥ θ γ δ 0 ( r ) . (19)
There exists M : = M ( j , k ) ∈ N M:=M(j,k)\in\mathbb{N} M := M ( j , k ) ∈ N such that for m ≥ M m\geq M m ≥ M , we bound against
π γ δ 1 j k ( m , r ) + θ j k 1 ( m ) ≥ ⌊ 11 r − m + λ γ δ 1 j k 13 ⌋ + ⌊ 11 m + 11 j + 22 k − 48 13 ⌋ ≥ ⌊ 11 r + 10 m + 11 j + 22 k + λ γ δ 1 j k − 60 13 ⌋ ≥ ⌊ 11 r + 11 γ + 22 δ − 7 13 ⌋ + ⌊ 10 m + 11 j + 22 k − 11 γ − 22 δ + λ γ δ 1 j k − 53 13 ⌋ ≥ ⌊ 11 r + 11 γ + 22 δ − 7 13 ⌋ ≥ θ γ δ 0 ( r ) . \begin{aligned}
\pi^{1jk}_{\gamma\delta}(m,r)+\theta^1_{jk}(m)\geq\left\lfloor\frac{11r-m+\lambda^{1jk}_{\gamma\delta}}{13}\right\rfloor+\left\lfloor\frac{11m+11j+22k-48}{13}\right\rfloor\geq\left\lfloor\frac{11r+10m+11j+22k+\lambda^{1jk}_{\gamma\delta}-60}{13}\right\rfloor\\
&\geq\left\lfloor\frac{11r+11\gamma+22\delta-7}{13}\right\rfloor+\left\lfloor\frac{10m+11j+22k-11\gamma-22\delta+\lambda^{1jk}_{\gamma\delta}-53}{13}\right\rfloor\\
&\geq\left\lfloor\frac{11r+11\gamma+22\delta-7}{13}\right\rfloor\geq\theta^0_{\gamma\delta}(r).
\end{aligned} π γ δ 1 j k ( m , r ) + θ j k 1 ( m ) ≥ ⌊ 13 11 r − m + λ γ δ 1 j k ⌋ + ⌊ 13 11 m + 11 j + 22 k − 48 ⌋ ≥ ⌊ 13 11 r + 10 m + 11 j + 22 k + λ γ δ 1 j k − 60 ⌋ ≥ ⌊ 13 11 r + 11 γ + 22 δ − 7 ⌋ + ⌊ 13 10 m + 11 j + 22 k − 11 γ − 22 δ + λ γ δ 1 j k − 53 ⌋ ≥ ⌊ 13 11 r + 11 γ + 22 δ − 7 ⌋ ≥ θ γ δ 0 ( r ) .
In particular,
M = { 6 if j = k = 0 , 5 if j = 1 , 4 if k = 1. M=\begin{cases}
6 & \text{if }j=k=0,\\
5 & \text{if }j=1,\\
4 & \text{if }k=1.
\end{cases} M = ⎩ ⎨ ⎧ 6 5 4 if j = k = 0 , if j = 1 , if k = 1.
For the remaining cases we check the inequalities more directly. Below we check the inequalities for the floor values of
⌊ π γ δ 1 j k ( m , r ) + θ j k 1 ( m ) − θ γ δ 0 ( r ) ⌋ . \left\lfloor\pi^{1jk}_{\gamma\delta}(m,r)+\theta^1_{jk}(m)-\theta^0_{\gamma\delta}(r)\right\rfloor. ⌊ π γ δ 1 j k ( m , r ) + θ j k 1 ( m ) − θ γ δ 0 ( r ) ⌋ .
Precisely, for ( j , k , γ , δ ) = ( 0 , 0 , γ , δ ) (j,k,\gamma,\delta)=(0,0,\gamma,\delta) ( j , k , γ , δ ) = ( 0 , 0 , γ , δ ) and 1 ≤ m ≤ 5 1\leq m\leq5 1 ≤ m ≤ 5 , we have
( 0 , 0 , 0 , 0 ) : ⌊ 11 r − m 13 ⌋ − ⌊ 11 r − 7 13 ⌋ ≥ 0 , (0,0,0,0):\quad\left\lfloor\frac{11r-m}{13}\right\rfloor-\left\lfloor\frac{11r-7}{13}\right\rfloor\geq0, ( 0 , 0 , 0 , 0 ) : ⌊ 13 11 r − m ⌋ − ⌊ 13 11 r − 7 ⌋ ≥ 0 ,
( 0 , 0 , 1 , 0 ) : ⌊ 11 r + 11 − m 13 ⌋ − ⌊ 11 r + 4 13 ⌋ ≥ 0 , (0,0,1,0):\quad\left\lfloor\frac{11r+11-m}{13}\right\rfloor-\left\lfloor\frac{11r+4}{13}\right\rfloor\geq0, ( 0 , 0 , 1 , 0 ) : ⌊ 13 11 r + 11 − m ⌋ − ⌊ 13 11 r + 4 ⌋ ≥ 0 ,
( 0 , 0 , 0 , 1 ) : ⌊ 11 r + 24 − m 13 ⌋ − ⌊ 11 r + 15 13 ⌋ ≥ 0. (0,0,0,1):\quad\left\lfloor\frac{11r+24-m}{13}\right\rfloor-\left\lfloor\frac{11r+15}{13}\right\rfloor\geq0. ( 0 , 0 , 0 , 1 ) : ⌊ 13 11 r + 24 − m ⌋ − ⌊ 13 11 r + 15 ⌋ ≥ 0.
Moreover, for ( j , k , γ , δ ) = ( 1 , 0 , γ , δ ) (j,k,\gamma,\delta)=(1,0,\gamma,\delta) ( j , k , γ , δ ) = ( 1 , 0 , γ , δ ) and 0 ≤ m ≤ 4 0\leq m\leq4 0 ≤ m ≤ 4 , it follows that
( 1 , 0 , 0 , 0 ) : ⌊ 11 r − m − 1 13 ⌋ − ⌊ 11 r − 7 13 ⌋ ≥ 0 , (1,0,0,0):\quad\left\lfloor\frac{11r-m-1}{13}\right\rfloor-\left\lfloor\frac{11r-7}{13}\right\rfloor\geq0, ( 1 , 0 , 0 , 0 ) : ⌊ 13 11 r − m − 1 ⌋ − ⌊ 13 11 r − 7 ⌋ ≥ 0 ,
( 1 , 0 , 1 , 0 ) : ⌊ 11 r + 10 − m 13 ⌋ − ⌊ 11 r + 4 13 ⌋ ≥ 0 , (1,0,1,0):\quad\left\lfloor\frac{11r+10-m}{13}\right\rfloor-\left\lfloor\frac{11r+4}{13}\right\rfloor\geq0, ( 1 , 0 , 1 , 0 ) : ⌊ 13 11 r + 10 − m ⌋ − ⌊ 13 11 r + 4 ⌋ ≥ 0 ,
( 1 , 0 , 0 , 1 ) : ⌊ 11 r + 23 − m 13 ⌋ − ⌊ 11 r + 15 13 ⌋ ≥ 0. (1,0,0,1):\quad\left\lfloor\frac{11r+23-m}{13}\right\rfloor-\left\lfloor\frac{11r+15}{13}\right\rfloor\geq0. ( 1 , 0 , 0 , 1 ) : ⌊ 13 11 r + 23 − m ⌋ − ⌊ 13 11 r + 15 ⌋ ≥ 0.
Finally, for ( j , k , γ , δ ) = ( 0 , 1 , γ , δ ) (j,k,\gamma,\delta)=(0,1,\gamma,\delta) ( j , k , γ , δ ) = ( 0 , 1 , γ , δ ) and 0 ≤ m ≤ 3 0\le m\le3 0 ≤ m ≤ 3 , we obtain
( 0 , 1 , 0 , 0 ) : ⌊ 11 r − m − 3 13 ⌋ − ⌊ 11 r − 7 13 ⌋ ≥ 0 , (0,1,0,0):\quad\left\lfloor\frac{11r-m-3}{13}\right\rfloor-\left\lfloor\frac{11r-7}{13}\right\rfloor\ge0, ( 0 , 1 , 0 , 0 ) : ⌊ 13 11 r − m − 3 ⌋ − ⌊ 13 11 r − 7 ⌋ ≥ 0 ,
( 0 , 1 , 1 , 0 ) : ⌊ 11 r + 8 − m 13 ⌋ − ⌊ 11 r + 4 13 ⌋ ≥ 0 , (0,1,1,0):\quad\left\lfloor\frac{11r+8-m}{13}\right\rfloor-\left\lfloor\frac{11r+4}{13}\right\rfloor\ge0, ( 0 , 1 , 1 , 0 ) : ⌊ 13 11 r + 8 − m ⌋ − ⌊ 13 11 r + 4 ⌋ ≥ 0 ,
( 0 , 1 , 0 , 1 ) : ⌊ 11 r + 19 − m 13 ⌋ − ⌊ 11 r + 15 13 ⌋ ≥ 0. (0,1,0,1):\quad\left\lfloor\frac{11r+19-m}{13}\right\rfloor-\left\lfloor\frac{11r+15}{13}\right\rfloor\ge0. ( 0 , 1 , 0 , 1 ) : ⌊ 13 11 r + 19 − m ⌋ − ⌊ 13 11 r + 15 ⌋ ≥ 0.
This concludes the proof. □
Theorem 3.3. For g ∈ W n , 0 ( 0 ) g\in W_{n,0}^{(0)} g ∈ W n , 0 ( 0 ) with 55 ∣ n 55\mid n 55 ∣ n , there exist m 1 , m 2 ∈ Z m_1,m_2\in\mathbb{Z} m 1 , m 2 ∈ Z such that
1 11 U ( 0 ) ( g ) − ( m 1 x + m 2 ) r L ( 2 + 11 x ) 11 n + 15 c ( x ) n ∈ W 11 n + 15 , n ( 1 ) . \frac{1}{11}U^{(0)}(g)-\frac{(m_1x+m_2)r_L}{(2+11x)^{11n+15}c(x)^n}\in W_{11n+15,n}^{(1)}. 11 1 U ( 0 ) ( g ) − ( 2 + 11 x ) 11 n + 15 c ( x ) n ( m 1 x + m 2 ) r L ∈ W 11 n + 15 , n ( 1 ) . Proof. Let
g : = 1 ( 2 + 11 x ) n ∑ j + k ≤ 1 y j z k ∑ m ≥ 1 − ( j + k ) s j k ( m ) 11 θ j k 0 ( m ) x m , g:=\frac{1}{(2+11x)^n}\sum_{j+k\le1}y^jz^k\sum_{m\ge1-(j+k)}s_{jk}(m)11^{\theta_{jk}^{0}(m)}x^m, g := ( 2 + 11 x ) n 1 j + k ≤ 1 ∑ y j z k m ≥ 1 − ( j + k ) ∑ s j k ( m ) 1 1 θ j k 0 ( m ) x m , with s j k ( m ) ∈ Z [ 1 2 ] s_{jk}(m)\in\mathbb{Z}\left[\frac{1}{2}\right] s j k ( m ) ∈ Z [ 2 1 ] is with finite support. We consider
U ( 0 ) ( g ) = ∑ j + k ≤ 1 ∑ m ≥ 1 − ( j + k ) s j k ( m ) 11 θ j k 0 ( m ) U ( 0 ) ( y j z k x m ( 2 + 11 x ) n ) . U^{(0)}(g)=\sum_{j+k\le1}\sum_{m\ge1-(j+k)}s_{jk}(m)11^{\theta_{jk}^{0}(m)}U^{(0)}\left(\frac{y^jz^kx^m}{(2+11x)^n}\right). U ( 0 ) ( g ) = j + k ≤ 1 ∑ m ≥ 1 − ( j + k ) ∑ s j k ( m ) 1 1 θ j k 0 ( m ) U ( 0 ) ( ( 2 + 11 x ) n y j z k x m ) . Again applying Theorem 2.3,
U ( 0 ) ( g ) = 1 ( 2 + 11 x ) 11 n + 15 c ( x ) n ∑ j + k ≤ 1 , m ≥ 1 − ( j + k ) 0 ≤ γ + δ ≤ 1 , r ≥ 0 s j k ( m ) h γ δ 0 j k ( m , n , r ) 11 π γ δ 0 j k ( m , r ) + θ j k 0 ( m ) y γ z δ x r . U^{(0)}(g)=\frac{1}{(2+11x)^{11n+15}c(x)^n}\sum_{\substack{j+k\le1,\ m\ge1-(j+k)\\0\le\gamma+\delta\le1,\ r\ge0}}s_{jk}(m)h_{\gamma\delta}^{0jk}(m,n,r)11^{\pi_{\gamma\delta}^{0jk}(m,r)+\theta_{jk}^{0}(m)}y^\gamma z^\delta x^r. U ( 0 ) ( g ) = ( 2 + 11 x ) 11 n + 15 c ( x ) n 1 j + k ≤ 1 , m ≥ 1 − ( j + k ) 0 ≤ γ + δ ≤ 1 , r ≥ 0 ∑ s j k ( m ) h γ δ 0 j k ( m , n , r ) 1 1 π γ δ 0 j k ( m , r ) + θ j k 0 ( m ) y γ z δ x r . Next, we show that
π γ δ 0 j k ( m , r ) + θ j k 0 ( m ) ≥ θ γ δ 1 ( r ) + 1. \pi_{\gamma\delta}^{0jk}(m,r)+\theta_{jk}^{0}(m)\ge\theta_{\gamma\delta}^{1}(r)+1. π γ δ 0 j k ( m , r ) + θ j k 0 ( m ) ≥ θ γ δ 1 ( r ) + 1. First, for
10 m + 11 j + 22 k − 11 γ − 22 δ + λ γ δ 0 j k + 16 ≥ 0 , (20) 10m+11j+22k-11\gamma-22\delta+\lambda_{\gamma\delta}^{0jk}+16\ge0, \tag*{(20)} 10 m + 11 j + 22 k − 11 γ − 22 δ + λ γ δ 0 j k + 16 ≥ 0 , (20) we have
π γ δ 0 j k ( m , r ) + θ j k 0 ( m ) ≥ ⌊ 11 r − m + λ γ δ 0 j k 13 ⌋ + ⌊ 11 m + 11 j + 22 k − 7 13 ⌋ ≥ ⌊ 11 r + 10 m + 11 j + 22 k + λ γ δ 0 j k − 19 13 ⌋ ≥ ⌊ 11 r + 11 γ + 22 δ − 48 13 ⌋ + ⌊ 10 m + 11 j + 22 k − 11 γ − 22 δ + λ γ δ 0 j k + 29 13 ⌋ ≥ ⌊ 11 r + 11 γ + 22 δ − 48 13 ⌋ + 1 ≥ θ γ δ 1 ( r ) + 1. \begin{aligned}
\pi_{\gamma\delta}^{0jk}(m,r)+\theta_{jk}^{0}(m)
&\ge\left\lfloor\frac{11r-m+\lambda_{\gamma\delta}^{0jk}}{13}\right\rfloor+\left\lfloor\frac{11m+11j+22k-7}{13}\right\rfloor\\
&\ge\left\lfloor\frac{11r+10m+11j+22k+\lambda_{\gamma\delta}^{0jk}-19}{13}\right\rfloor\\
&\ge\left\lfloor\frac{11r+11\gamma+22\delta-48}{13}\right\rfloor+\left\lfloor\frac{10m+11j+22k-11\gamma-22\delta+\lambda_{\gamma\delta}^{0jk}+29}{13}\right\rfloor\\
&\ge\left\lfloor\frac{11r+11\gamma+22\delta-48}{13}\right\rfloor+1\ge\theta_{\gamma\delta}^{1}(r)+1.
\end{aligned} π γ δ 0 j k ( m , r ) + θ j k 0 ( m ) ≥ ⌊ 13 11 r − m + λ γ δ 0 j k ⌋ + ⌊ 13 11 m + 11 j + 22 k − 7 ⌋ ≥ ⌊ 13 11 r + 10 m + 11 j + 22 k + λ γ δ 0 j k − 19 ⌋ ≥ ⌊ 13 11 r + 11 γ + 22 δ − 48 ⌋ + ⌊ 13 10 m + 11 j + 22 k − 11 γ − 22 δ + λ γ δ 0 j k + 29 ⌋ ≥ ⌊ 13 11 r + 11 γ + 22 δ − 48 ⌋ + 1 ≥ θ γ δ 1 ( r ) + 1. Note that (3.2) holds for m ≥ 2 m\ge2 m ≥ 2 if j = k = 0 j=k=0 j = k = 0 , m ≥ 1 m\ge1 m ≥ 1 if j = 1 j=1 j = 1 and k = 1 k=1 k = 1 . Thus, for the remaining cases, we estimate lower bounds of
π γ δ 0 j k ( m , r ) + θ j k 0 ( m ) − θ γ δ 1 ( r ) − 1. (21) \pi_{\gamma\delta}^{0jk}(m,r)+\theta_{jk}^{0}(m)-\theta_{\gamma\delta}^{1}(r)-1. \tag*{(21)} π γ δ 0 j k ( m , r ) + θ j k 0 ( m ) − θ γ δ 1 ( r ) − 1. (21) First, consider j = k = 0 j=k=0 j = k = 0 . Since m ≥ 1 − ( j + k ) m\ge1-(j+k) m ≥ 1 − ( j + k ) and r ≥ 1 − ( γ + δ ) r\ge1-(\gamma+\delta) r ≥ 1 − ( γ + δ ) , we need to verify for m = 1 m=1 m = 1 . For ( j , k , γ , δ ) = ( 0 , 0 , 0 , 0 ) (j,k,\gamma,\delta)=(0,0,0,0) ( j , k , γ , δ ) = ( 0 , 0 , 0 , 0 ) and r ≥ 5 r\ge5 r ≥ 5 ,
⌊ 11 r − 35 13 ⌋ − ⌊ 11 r − 48 13 ⌋ − 1 ≥ 0. \left\lfloor\frac{11r-35}{13}\right\rfloor-\left\lfloor\frac{11r-48}{13}\right\rfloor-1\ge0. ⌊ 13 11 r − 35 ⌋ − ⌊ 13 11 r − 48 ⌋ − 1 ≥ 0. Next, for ( j , k , γ , δ ) = ( 0 , 0 , 1 , 0 ) (j,k,\gamma,\delta)=(0,0,1,0) ( j , k , γ , δ ) = ( 0 , 0 , 1 , 0 ) , and r ≥ 4 r\ge4 r ≥ 4 ,
⌊ 11 r − 24 13 ⌋ − ⌊ 11 r − 37 13 ⌋ − 1 ≥ 0. \left\lfloor\frac{11r-24}{13}\right\rfloor-\left\lfloor\frac{11r-37}{13}\right\rfloor-1\ge0. ⌊ 13 11 r − 24 ⌋ − ⌊ 13 11 r − 37 ⌋ − 1 ≥ 0.
Moreover, for ( j , k , γ , δ ) = ( 0 , 0 , 0 , 1 ) (j,k,\gamma,\delta)=(0,0,0,1) ( j , k , γ , δ ) = ( 0 , 0 , 0 , 1 ) and r ≥ 3 r\ge3 r ≥ 3 ,
⌊ 11 r − 13 13 ⌋ − ⌊ 11 r − 26 13 ⌋ − 1 ≥ 0. \left\lfloor\frac{11r-13}{13}\right\rfloor-\left\lfloor\frac{11r-26}{13}\right\rfloor-1\ge0. ⌊ 13 11 r − 13 ⌋ − ⌊ 13 11 r − 26 ⌋ − 1 ≥ 0.
Now we consider j = 1 j=1 j = 1 and we only need to verify for m = 0 m=0 m = 0 . For ( j , k , γ , δ ) = ( 1 , 0 , 0 , 0 ) (j,k,\gamma,\delta)=(1,0,0,0) ( j , k , γ , δ ) = ( 1 , 0 , 0 , 0 ) and r ≥ 5 r\ge5 r ≥ 5 ,
⌊ 11 r − 34 13 ⌋ − ⌊ 11 r − 48 13 ⌋ − 1. \left\lfloor\frac{11r-34}{13}\right\rfloor-\left\lfloor\frac{11r-48}{13}\right\rfloor-1. ⌊ 13 11 r − 34 ⌋ − ⌊ 13 11 r − 48 ⌋ − 1.
For ( j , k , γ , δ ) = ( 1 , 0 , 1 , 0 ) (j,k,\gamma,\delta)=(1,0,1,0) ( j , k , γ , δ ) = ( 1 , 0 , 1 , 0 ) and r ≥ 4 r\ge4 r ≥ 4 ,
⌊ 11 r − 23 13 ⌋ − ⌊ 11 r − 37 13 ⌋ − 1. \left\lfloor\frac{11r-23}{13}\right\rfloor-\left\lfloor\frac{11r-37}{13}\right\rfloor-1. ⌊ 13 11 r − 23 ⌋ − ⌊ 13 11 r − 37 ⌋ − 1.
For ( j , k , γ , δ ) = ( 1 , 0 , 0 , 1 ) (j,k,\gamma,\delta)=(1,0,0,1) ( j , k , γ , δ ) = ( 1 , 0 , 0 , 1 ) and r ≥ 3 r\ge3 r ≥ 3 ,
⌊ 11 r − 12 13 ⌋ − ⌊ 11 r − 26 13 ⌋ − 1. \left\lfloor\frac{11r-12}{13}\right\rfloor-\left\lfloor\frac{11r-26}{13}\right\rfloor-1. ⌊ 13 11 r − 12 ⌋ − ⌊ 13 11 r − 26 ⌋ − 1.
Next, consider the case k = 1 k=1 k = 1 and (3.2) holds trivially. Moreover, we rewrite
∑ m ≥ 0 s 01 ( m ) 11 θ 01 ( 0 ) ( m ) 6 m 4 n r L = ∑ m ≥ 0 s 01 ( m ) 11 ⌊ ( 11 m + 15 ) / 13 ⌋ 6 m 4 n r L = ∑ m ≥ 0 s 01 ( m ) 11 ⌊ ( 11 m + 2 ) / 13 ⌋ 6 m 4 n 11 r L . \sum_{m\ge0}s_{01}(m)11^{\theta_{01}^{(0)}(m)}6^{m}4^{n}r_L
=\sum_{m\ge0}s_{01}(m)11^{\lfloor(11m+15)/13\rfloor}6^{m}4^{n}r_L
=\sum_{m\ge0}s_{01}(m)11^{\lfloor(11m+2)/13\rfloor}6^{m}4^{n}11r_L. m ≥ 0 ∑ s 01 ( m ) 1 1 θ 01 ( 0 ) ( m ) 6 m 4 n r L = m ≥ 0 ∑ s 01 ( m ) 1 1 ⌊( 11 m + 15 ) /13 ⌋ 6 m 4 n r L = m ≥ 0 ∑ s 01 ( m ) 1 1 ⌊( 11 m + 2 ) /13 ⌋ 6 m 4 n 11 r L .
Since s 01 ( m ) s_{01}(m) s 01 ( m ) has finite support, this is an integer multiple of 11 r L 11r_L 11 r L .
It remains to analyze the negative powers of 11 arising for the cases j = k = 0 j=k=0 j = k = 0 and j = 1 j=1 j = 1 . With
1 − ( γ + δ ) ≤ r ≤ 4 − ( γ + 2 δ ) , 1-(\gamma+\delta)\le r\le4-(\gamma+2\delta), 1 − ( γ + δ ) ≤ r ≤ 4 − ( γ + 2 δ ) ,
setting the coefficients of y γ z δ x r y^\gamma z^\delta x^r y γ z δ x r as t γ δ ( r ) t_{\gamma\delta}(r) t γ δ ( r ) , we get
t γ δ ( r ) ≡ h γ δ 000 ( 1 , n , r ) s 00 ( 1 ) + h γ δ 010 ( 0 , n , r ) s 10 ( 0 ) ( m o d 11 ) . (22) t_{\gamma\delta}(r)\equiv h_{\gamma\delta}^{000}(1,n,r)s_{00}(1)+h_{\gamma\delta}^{010}(0,n,r)s_{10}(0)\pmod{11}. \tag*{(22)} t γ δ ( r ) ≡ h γ δ 000 ( 1 , n , r ) s 00 ( 1 ) + h γ δ 010 ( 0 , n , r ) s 10 ( 0 ) ( mod 11 ) . (22)
We observe that only s 00 ( 1 ) s_{00}(1) s 00 ( 1 ) and s 10 ( 0 ) s_{10}(0) s 10 ( 0 ) contribute. Thus, we only consider the auxiliary coefficients for U ( 0 ) ( x j y ( 1 − j ) ( 2 + 11 x ) n ) U^{(0)}\left(\frac{x^j y^{(1-j)}}{(2+11x)^n}\right) U ( 0 ) ( ( 2 + 11 x ) n x j y ( 1 − j ) ) with j ∈ { 0 , 1 } j\in\{0,1\} j ∈ { 0 , 1 } . Since 55 ∣ n 55\mid n 55 ∣ n , it is not hard to see that
h 00 000 ( 1 , 11 n , 1 ) ≡ 3 ⋅ 4 n ( m o d 11 ) (23) h_{00}^{000}(1,11n,1)\equiv3\cdot4^n\pmod{11} \tag*{(23)} h 00 000 ( 1 , 11 n , 1 ) ≡ 3 ⋅ 4 n ( mod 11 ) (23)
for 1 ≤ n ≤ 3 1\le n\le3 1 ≤ n ≤ 3 . By Lemma 2.4, we see that
{ h 00 000 ( 1 , 11 n , 1 ) ( m o d 11 ) } n = 1 ∞ = { 3 , 1 , 4 , 5 , 9 , 3 , 1 , 4 , … } . \left\{h_{00}^{000}(1,11n,1)\pmod{11}\right\}_{n=1}^{\infty}=\{3,1,4,5,9,3,1,4,\ldots\}. { h 00 000 ( 1 , 11 n , 1 ) ( mod 11 ) } n = 1 ∞ = { 3 , 1 , 4 , 5 , 9 , 3 , 1 , 4 , … } .
Thus, (3.5) is true for all n ≥ 0 n\ge0 n ≥ 0 because the coefficients are periodic modulo 11 with period 5. Consequently, for 55 ∣ n 55\mid n 55 ∣ n , we have
h 00 000 ( 1 , n , 1 ) ≡ 9 ( m o d 11 ) . h_{00}^{000}(1,n,1)\equiv9\pmod{11}. h 00 000 ( 1 , n , 1 ) ≡ 9 ( mod 11 ) .
So, (3.4) reduces to
t 00 ( 1 ) ≡ 9 s 00 ( 1 ) + 5 s 10 ( 0 ) ( m o d 11 ) , t 00 ( 2 ) ≡ 3 s 00 ( 1 ) + 9 s 10 ( 0 ) ( m o d 11 ) , t 00 ( 3 ) ≡ 9 s 00 ( 1 ) + 5 s 10 ( 0 ) ( m o d 11 ) , t 00 ( 4 ) ≡ 4 s 00 ( 1 ) + s 10 ( 0 ) ( m o d 11 ) , t 10 ( 0 ) ≡ 2 s 00 ( 1 ) + 6 s 10 ( 0 ) ( m o d 11 ) , t 10 ( 1 ) ≡ 4 s 00 ( 1 ) + s 10 ( 0 ) ( m o d 11 ) , t 10 ( 2 ) ≡ 9 s 00 ( 1 ) + 5 s 10 ( 0 ) ( m o d 11 ) , t 10 ( 3 ) ≡ 7 s 00 ( 1 ) + 10 s 10 ( 0 ) ( m o d 11 ) , t 01 ( 0 ) ≡ 8 s 00 ( 1 ) + 2 s 10 ( 0 ) ( m o d 11 ) , t 01 ( 1 ) ≡ 3 s 00 ( 1 ) + 9 s 10 ( 0 ) ( m o d 11 ) , t 01 ( 2 ) ≡ 6 s 00 ( 1 ) + 7 s 10 ( 0 ) ( m o d 11 ) . \begin{aligned}
t_{00}(1)&\equiv9s_{00}(1)+5s_{10}(0)\pmod{11},&
t_{00}(2)&\equiv3s_{00}(1)+9s_{10}(0)\pmod{11},\\
t_{00}(3)&\equiv9s_{00}(1)+5s_{10}(0)\pmod{11},&
t_{00}(4)&\equiv4s_{00}(1)+s_{10}(0)\pmod{11},\\
t_{10}(0)&\equiv2s_{00}(1)+6s_{10}(0)\pmod{11},&
t_{10}(1)&\equiv4s_{00}(1)+s_{10}(0)\pmod{11},\\
t_{10}(2)&\equiv9s_{00}(1)+5s_{10}(0)\pmod{11},&
t_{10}(3)&\equiv7s_{00}(1)+10s_{10}(0)\pmod{11},\\
t_{01}(0)&\equiv8s_{00}(1)+2s_{10}(0)\pmod{11},&
t_{01}(1)&\equiv3s_{00}(1)+9s_{10}(0)\pmod{11},\\
t_{01}(2)&\equiv6s_{00}(1)+7s_{10}(0)\pmod{11}.
\end{aligned} t 00 ( 1 ) t 00 ( 3 ) t 10 ( 0 ) t 10 ( 2 ) t 01 ( 0 ) t 01 ( 2 ) ≡ 9 s 00 ( 1 ) + 5 s 10 ( 0 ) ( mod 11 ) , ≡ 9 s 00 ( 1 ) + 5 s 10 ( 0 ) ( mod 11 ) , ≡ 2 s 00 ( 1 ) + 6 s 10 ( 0 ) ( mod 11 ) , ≡ 9 s 00 ( 1 ) + 5 s 10 ( 0 ) ( mod 11 ) , ≡ 8 s 00 ( 1 ) + 2 s 10 ( 0 ) ( mod 11 ) , ≡ 6 s 00 ( 1 ) + 7 s 10 ( 0 ) ( mod 11 ) . t 00 ( 2 ) t 00 ( 4 ) t 10 ( 1 ) t 10 ( 3 ) t 01 ( 1 ) ≡ 3 s 00 ( 1 ) + 9 s 10 ( 0 ) ( mod 11 ) , ≡ 4 s 00 ( 1 ) + s 10 ( 0 ) ( mod 11 ) , ≡ 4 s 00 ( 1 ) + s 10 ( 0 ) ( mod 11 ) , ≡ 7 s 00 ( 1 ) + 10 s 10 ( 0 ) ( mod 11 ) , ≡ 3 s 00 ( 1 ) + 9 s 10 ( 0 ) ( mod 11 ) ,
These yield that
∑ γ + δ ≤ 1 1 − ( γ + δ ) ≤ r ≤ 4 − ( γ + 2 δ ) t γ δ ( r ) y γ z δ x r = ( 9 x + 3 x 2 + 9 x 3 + 4 x 4 + 2 y + 4 x y + 9 x 2 y + 7 x 3 y + 8 z + 3 x z + 6 x 2 z ) s 00 ( 1 ) \sum_{\substack{\gamma+\delta\le1\\1-(\gamma+\delta)\le r\le4-(\gamma+2\delta)}}t_{\gamma\delta}(r)y^\gamma z^\delta x^r
=
\left(9x+3x^2+9x^3+4x^4+2y+4xy+9x^2y+7x^3y+8z+3xz+6x^2z\right)s_{00}(1) γ + δ ≤ 1 1 − ( γ + δ ) ≤ r ≤ 4 − ( γ + 2 δ ) ∑ t γ δ ( r ) y γ z δ x r = ( 9 x + 3 x 2 + 9 x 3 + 4 x 4 + 2 y + 4 x y + 9 x 2 y + 7 x 3 y + 8 z + 3 x z + 6 x 2 z ) s 00 ( 1 )
+ ( 5 x + 9 x 2 + 5 x 3 + x 4 + 6 y + x y + 5 x 2 y + 10 x 3 y + 2 z + 9 x z + 7 x 2 z ) s 10 ( 0 ) ≡ ( 1 + x ) ( 9 s 00 ( 1 ) + 5 s 10 ( 0 ) ) 11 r L ( m o d 11 ) , \begin{aligned}
\qquad+\left(5x+9x^2+5x^3+x^4+6y+xy+5x^2y+10x^3y+2z+9xz+7x^2z\right)s_{10}(0) \\
&\equiv(1+x)\left(9s_{00}(1)+5s_{10}(0)\right)11r_L\pmod{11},
\end{aligned} + ( 5 x + 9 x 2 + 5 x 3 + x 4 + 6 y + x y + 5 x 2 y + 10 x 3 y + 2 z + 9 x z + 7 x 2 z ) s 10 ( 0 ) ≡ ( 1 + x ) ( 9 s 00 ( 1 ) + 5 s 10 ( 0 ) ) 11 r L ( mod 11 ) ,
and this concludes the proof. □
Cancellation of c ( x ) c(x) c ( x )
Let
F ( j k ) : = ∑ m ≥ 0 s ( m ) 11 θ j k i ( m ) x m . \mathcal{F}^{(jk)} := \sum_{m\ge0} s(m)11^{\theta_{jk}^{i}(m)}x^m. F ( j k ) := m ≥ 0 ∑ s ( m ) 1 1 θ j k i ( m ) x m .
Suppose that c ( x ) n ∣ F ( j k ) ( x ) c(x)^n\mid\mathcal{F}^{(jk)}(x) c ( x ) n ∣ F ( j k ) ( x ) , i.e., there exists a polynomial F 0 ( j k ) \mathcal{F}_0^{(jk)} F 0 ( j k ) such that
F ( j k ) ( x ) = c ( x ) n F 0 ( j k ) . \mathcal{F}^{(jk)}(x)=c(x)^n\mathcal{F}_0^{(jk)}. F ( j k ) ( x ) = c ( x ) n F 0 ( j k ) .
In this section, our aim is to show that F 0 ( j k ) \mathcal{F}_0^{(jk)} F 0 ( j k ) has a similar 11 11 11 -adic form to that of F ( j k ) \mathcal{F}^{(jk)} F ( j k ) .
Lemma 3.4. Let M , m 1 , m 2 ∈ Z M,m_1,m_2\in\mathbb{Z} M , m 1 , m 2 ∈ Z , with gcd ( m 1 m 2 , 11 ) = 1 \gcd(m_1m_2,11)=1 g cd( m 1 m 2 , 11 ) = 1 , and let s ( m ) s(m) s ( m ) be an integer-valued function with finite support. If
c ( x ) n ∣ ( i ( m 1 x + m 2 ) M r L + ∑ j + k ≤ 1 y j z k ∑ m ≥ 0 s ( m ) 11 θ j k i ( m ) x m ) , c(x)^n\mid\left(i(m_1x+m_2)M r_L+\sum_{j+k\le1}y^jz^k\sum_{m\ge0}s(m)11^{\theta_{jk}^{i}(m)}x^m\right), c ( x ) n ∣ i ( m 1 x + m 2 ) M r L + j + k ≤ 1 ∑ y j z k m ≥ 0 ∑ s ( m ) 1 1 θ j k i ( m ) x m , then there exists an integer M ^ ∈ Z \widehat{M}\in\mathbb{Z} M ∈ Z and another integer-valued function s ^ ( m ) \widehat{s}(m) s ( m ) , also with finite support, such that
i M ( m 1 x + m 2 ) r L + ∑ j + k ≤ 1 y j z k ∑ m ≥ 0 s ( m ) 11 θ j k i ( m ) x m c ( x ) n = i M ^ ( m 1 x + m 2 ) r L + ∑ j + k ≤ 1 y j z k ∑ m ≥ 0 s ^ ( m ) 11 θ j k i ( m ) x m . \frac{iM(m_1x+m_2)r_L+\sum_{j+k\le1}y^jz^k\sum_{m\ge0}s(m)11^{\theta_{jk}^{i}(m)}x^m}{c(x)^n}
=i\widehat{M}(m_1x+m_2)r_L+\sum_{j+k\le1}y^jz^k\sum_{m\ge0}\widehat{s}(m)11^{\theta_{jk}^{i}(m)}x^m. c ( x ) n i M ( m 1 x + m 2 ) r L + ∑ j + k ≤ 1 y j z k ∑ m ≥ 0 s ( m ) 1 1 θ j k i ( m ) x m = i M ( m 1 x + m 2 ) r L + j + k ≤ 1 ∑ y j z k m ≥ 0 ∑ s ( m ) 1 1 θ j k i ( m ) x m .
Proof. Suppose that
F 0 ( i j ) : = ∑ m ≥ 0 s 0 ( m ) x m . \mathcal{F}_0^{(ij)}:=\sum_{m\ge0}s_0(m)x^m. F 0 ( ij ) := m ≥ 0 ∑ s 0 ( m ) x m .
We show that s 0 ( m ) s_0(m) s 0 ( m ) is divisible by 11 θ j k i ( m ) 11^{\theta_{jk}^{i}(m)} 1 1 θ j k i ( m ) by induction principle. Since
⌊ 11 m 1 + 27 13 ⌋ + ⌊ 11 m 2 + 27 13 ⌋ ≥ ⌊ 11 ( m 1 + m 2 ) + 27 13 ⌋ , \left\lfloor\frac{11m_1+27}{13}\right\rfloor+\left\lfloor\frac{11m_2+27}{13}\right\rfloor\ge\left\lfloor\frac{11(m_1+m_2)+27}{13}\right\rfloor, ⌊ 13 11 m 1 + 27 ⌋ + ⌊ 13 11 m 2 + 27 ⌋ ≥ ⌊ 13 11 ( m 1 + m 2 ) + 27 ⌋ ,
c ( x ) n c(x)^n c ( x ) n has the form
c ( x ) n = C n ( 0 ) + ∑ r ≥ 1 C n ( r ) 11 ⌊ 11 r + 27 13 ⌋ x r . c(x)^n=C_n(0)+\sum_{r\ge1}C_n(r)11^{\left\lfloor\frac{11r+27}{13}\right\rfloor}x^r. c ( x ) n = C n ( 0 ) + r ≥ 1 ∑ C n ( r ) 1 1 ⌊ 13 11 r + 27 ⌋ x r .
Thus, we obtain
s ( 0 ) 11 θ j k i ( 0 ) = C n ( 0 ) s 0 ( 0 ) . s(0)11^{\theta_{jk}^{i}(0)}=C_n(0)s_0(0). s ( 0 ) 1 1 θ j k i ( 0 ) = C n ( 0 ) s 0 ( 0 ) .
Since C n ( 0 ) = C 1 ( 0 ) n C_n(0)=C_1(0)^n C n ( 0 ) = C 1 ( 0 ) n and 11 ∤ C n ( 0 ) 11\nmid C_n(0) 11 ∤ C n ( 0 ) , there must exists s ^ ( 0 ) ∈ Z \widehat{s}(0)\in\mathbb{Z} s ( 0 ) ∈ Z such that
s 0 ( 0 ) = s ^ ( 0 ) 11 θ j k i ( 0 ) . s_0(0)=\widehat{s}(0)11^{\theta_{jk}^{i}(0)}. s 0 ( 0 ) = s ( 0 ) 1 1 θ j k i ( 0 ) .
This concludes the base case. By induction hypothesis, 11 θ j k i ( r ) ∣ s 0 ( r ) 11^{\theta_{jk}^{i}(r)}\mid s_0(r) 1 1 θ j k i ( r ) ∣ s 0 ( r ) for 0 ≤ r ≤ m − 1 0\le r\le m-1 0 ≤ r ≤ m − 1 and for some fixed m ≥ 1 m\ge1 m ≥ 1 . Then for each r r r , there exists s ^ ( r ) ∈ Z \widehat{s}(r)\in\mathbb{Z} s ( r ) ∈ Z such that
s 0 ( r ) = s ^ ( r ) 11 θ j k i ( r ) . s_0(r)=\widehat{s}(r)11^{\theta_{jk}^{i}(r)}. s 0 ( r ) = s ( r ) 1 1 θ j k i ( r ) .
Assume that, for r ≥ 4 r\ge4 r ≥ 4 ,
θ j k i ( r ) = ⌊ 11 r + μ 13 ⌋ , \theta_{jk}^{i}(r)=\left\lfloor\frac{11r+\mu}{13}\right\rfloor, θ j k i ( r ) = ⌊ 13 11 r + μ ⌋ ,
with μ : = μ ( i , j , k ) ∈ Z \mu:=\mu(i,j,k)\in\mathbb{Z} μ := μ ( i , j , k ) ∈ Z . We then have
s ( m ) 11 θ j k i ( m ) = ∑ r = 1 m − 1 C n ( r ) s 0 ( m − r ) 11 ⌊ 11 r + 27 13 ⌋ + C n ( 0 ) s 0 ( m ) ≥ ∑ r = 1 m − 1 C n ( r ) s ^ ( m − r ) 11 ⌊ 11 r + 14 13 ⌋ + ⌊ 11 ( m − r ) + μ 13 ⌋ + 1 + C n ( 0 ) s 0 ( m ) ≥ ∑ r = 1 m − 1 C n ( r ) s ^ ( m − r ) 11 ⌊ 11 m + μ + 2 13 ⌋ + 1 + C n ( 0 ) s 0 ( m ) = 11 θ j k i ( m ) + 1 + ϵ ∑ r = 1 m − 1 C n ( r ) s ^ ( m − r ) + C n ( 0 ) s 0 ( m ) (24) \begin{aligned}
s(m)11^{\theta_{jk}^{i}(m)}
&=\sum_{r=1}^{m-1}C_n(r)s_0(m-r)11^{\left\lfloor\frac{11r+27}{13}\right\rfloor}+C_n(0)s_0(m)\\
&\ge\sum_{r=1}^{m-1}C_n(r)\widehat{s}(m-r)11^{\left\lfloor\frac{11r+14}{13}\right\rfloor+\left\lfloor\frac{11(m-r)+\mu}{13}\right\rfloor+1}+C_n(0)s_0(m)\\
&\ge\sum_{r=1}^{m-1}C_n(r)\widehat{s}(m-r)11^{\left\lfloor\frac{11m+\mu+2}{13}\right\rfloor+1}+C_n(0)s_0(m)\\
&=11^{\theta_{jk}^{i}(m)+1+\epsilon}\sum_{r=1}^{m-1}C_n(r)\widehat{s}(m-r)+C_n(0)s_0(m) \tag*{(24)}
\end{aligned} s ( m ) 1 1 θ j k i ( m ) = r = 1 ∑ m − 1 C n ( r ) s 0 ( m − r ) 1 1 ⌊ 13 11 r + 27 ⌋ + C n ( 0 ) s 0 ( m ) ≥ r = 1 ∑ m − 1 C n ( r ) s ( m − r ) 1 1 ⌊ 13 11 r + 14 ⌋ + ⌊ 13 11 ( m − r ) + μ ⌋ + 1 + C n ( 0 ) s 0 ( m ) ≥ r = 1 ∑ m − 1 C n ( r ) s ( m − r ) 1 1 ⌊ 13 11 m + μ + 2 ⌋ + 1 + C n ( 0 ) s 0 ( m ) = 1 1 θ j k i ( m ) + 1 + ϵ r = 1 ∑ m − 1 C n ( r ) s ( m − r ) + C n ( 0 ) s 0 ( m ) (24)
with ϵ ≥ 0 \epsilon\ge0 ϵ ≥ 0 . Thus 11 θ j k i ( m ) ∣ s 0 ( m ) 11^{\theta^{i}_{jk}(m)} \mid s_0(m) 1 1 θ j k i ( m ) ∣ s 0 ( m ) and we write
s 0 ( m ) = s ^ ( m ) 11 θ j k i ( m ) . s_0(m)=\hat{s}(m)11^{\theta^{i}_{jk}(m)}. s 0 ( m ) = s ^ ( m ) 1 1 θ j k i ( m ) .
Consequently, with s ^ ( m ) ∈ Z \hat{s}(m)\in\mathbb{Z} s ^ ( m ) ∈ Z , it follows that
F 0 ( j k ) = ∑ m ≥ 0 s ^ ( m ) 11 θ j k i ( m ) x m . \mathcal{F}_0^{(jk)}=\sum_{m\geq0}\hat{s}(m)11^{\theta^{i}_{jk}(m)}x^m. F 0 ( j k ) = m ≥ 0 ∑ s ^ ( m ) 1 1 θ j k i ( m ) x m .
Now, we consider i = 1 i=1 i = 1 and j = k = 0 j=k=0 j = k = 0 . With t ( m ) ∈ Z t(m)\in\mathbb{Z} t ( m ) ∈ Z and gcd ( t ( m ) , 11 ) = 1 \gcd(t(m),11)=1 g cd( t ( m ) , 11 ) = 1 , we assume
s ( 0 ) = 0 and s ( m ) = t ( m ) 11 for 1 ≤ m ≤ 3. s(0)=0\quad\text{and}\quad s(m)=\frac{t(m)}{11}\quad\text{for }1\leq m\leq3. s ( 0 ) = 0 and s ( m ) = 11 t ( m ) for 1 ≤ m ≤ 3.
Following similar estimations done in (24) , for 1 ≤ m ≤ 3 1\leq m\leq3 1 ≤ m ≤ 3 , we have
s 0 ( m ) = s ^ ( m ) 11 11 θ 00 1 ( m ) and s ^ ( m ) ≡ C n ( 0 ) − 1 t ( m ) ( m o d 11 ) . s_0(m)=\frac{\hat{s}(m)}{11}11^{\theta^{1}_{00}(m)}
\quad\text{and}\quad
\hat{s}(m)\equiv C_n(0)^{-1}t(m)\pmod{11}. s 0 ( m ) = 11 s ^ ( m ) 1 1 θ 00 1 ( m ) and s ^ ( m ) ≡ C n ( 0 ) − 1 t ( m ) ( mod 11 ) .
Moreover, s 0 ( m ) s_0(m) s 0 ( m ) is divisible by 11 θ 00 i ( m ) 11^{\theta^{i}_{00}(m)} 1 1 θ 00 i ( m ) for m ≥ 4 m\geq4 m ≥ 4 and can be again shown by following similar arguments in (24) . Analogous argument works as well for i = j = 1 i=j=1 i = j = 1 , or for i = k = 1 i=k=1 i = k = 1 . Let m 1 , m 2 ∈ Z m_1,m_2\in\mathbb{Z} m 1 , m 2 ∈ Z with gcd ( m 1 m 2 , 11 ) = 1 \gcd(m_1m_2,11)=1 g cd( m 1 m 2 , 11 ) = 1 and split ( m 1 x + m 2 ) r L (m_1x+m_2)r_L ( m 1 x + m 2 ) r L as
( m 1 x + m 2 ) r L = r L ( 00 ) + y r L ( 10 ) + z r L ( 01 ) with r L ( j k ) ∈ Q [ x ] . (m_1x+m_2)r_L=r_L^{(00)}+yr_L^{(10)}+zr_L^{(01)}
\quad\text{with}\quad r_L^{(jk)}\in\mathbb{Q}[x]. ( m 1 x + m 2 ) r L = r L ( 00 ) + y r L ( 10 ) + z r L ( 01 ) with r L ( j k ) ∈ Q [ x ] .
Since y y y and z z z are not divisible by positive powers of c ( x ) c(x) c ( x ) , thus, if
c ( x ) n ∣ ( m 1 x + m 2 ) r L + ∑ j + k ≤ 1 y j z k ∑ m ≥ 0 s ( m ) 11 θ j k i ( m ) x m = ∑ j + k ≤ 1 y j z k ( M ⋅ r L ( j k ) + ∑ m ≥ 0 s ( m ) 11 θ j k i ( m ) x m ) c(x)^n\mid(m_1x+m_2)r_L+\sum_{j+k\leq1}y^jz^k\sum_{m\geq0}s(m)11^{\theta^{i}_{jk}(m)}x^m
=\sum_{j+k\leq1}y^jz^k\left(M\cdot r_L^{(jk)}+\sum_{m\geq0}s(m)11^{\theta^{i}_{jk}(m)}x^m\right) c ( x ) n ∣ ( m 1 x + m 2 ) r L + j + k ≤ 1 ∑ y j z k m ≥ 0 ∑ s ( m ) 1 1 θ j k i ( m ) x m = j + k ≤ 1 ∑ y j z k ( M ⋅ r L ( j k ) + m ≥ 0 ∑ s ( m ) 1 1 θ j k i ( m ) x m )
we have
C n ( 0 ) − 1 M ( m 1 x + m 2 ) ( x + 3 x 2 + 9 x 3 + 10 y + 10 x y + 2 x 2 y + 7 z + 8 x z ) 11 ≡ 0 ( m o d ( m 1 + m 2 x ) r L ) . \frac{C_n(0)^{-1}M(m_1x+m_2)(x+3x^2+9x^3+10y+10xy+2x^2y+7z+8xz)}{11}\equiv0\pmod{(m_1+m_2x)r_L}. 11 C n ( 0 ) − 1 M ( m 1 x + m 2 ) ( x + 3 x 2 + 9 x 3 + 10 y + 10 x y + 2 x 2 y + 7 z + 8 x z ) ≡ 0 ( mod ( m 1 + m 2 x ) r L ) .
Proof of Theorem 1.2
Proof of Theorem 1.2. By (1.3), we have
2 11 L 1 − ( 3 + 7 x ) r L ( 2 + 11 x ) 15 ∈ W 15 , 0 ( 1 ) . \frac{2}{11}L_1-\frac{(3+7x)r_L}{(2+11x)^{15}}\in W_{15,0}^{(1)}. 11 2 L 1 − ( 2 + 11 x ) 15 ( 3 + 7 x ) r L ∈ W 15 , 0 ( 1 ) .
From (1.4), ψ ( 1 ) = 15 \psi(1)=15 ψ ( 1 ) = 15 . Now suppose that, for some j α , k α ∈ Z [ 1 2 ] j_\alpha,k_\alpha\in\mathbb{Z}\left[\frac{1}{2}\right] j α , k α ∈ Z [ 2 1 ] , we have
1 11 α L 2 α − 1 − ( j 2 α − 1 + k 2 α − 1 x ) r L ( 2 + 11 x ) ψ ( 2 α − 1 ) = f 2 α − 1 ∈ W ψ ( 2 α − 1 ) , 0 ( 1 ) . \frac{1}{11^\alpha}L_{2\alpha-1}-\frac{(j_{2\alpha-1}+k_{2\alpha-1}x)r_L}{(2+11x)^{\psi(2\alpha-1)}}=f_{2\alpha-1}\in W_{\psi(2\alpha-1),0}^{(1)}. 1 1 α 1 L 2 α − 1 − ( 2 + 11 x ) ψ ( 2 α − 1 ) ( j 2 α − 1 + k 2 α − 1 x ) r L = f 2 α − 1 ∈ W ψ ( 2 α − 1 ) , 0 ( 1 ) .
Combining Lemma 2.1, Lemma 3.1, and Theorem 3.2, we obtain
1 11 α L 2 α = 1 11 α U ( 1 ) ( L 2 α − 1 ) = U ( 1 ) ( ( j 2 α − 1 + k 2 α − 1 x ) r L ( 2 + 11 x ) ψ ( 2 α − 1 ) ) + U ( 1 ) ( f 2 α − 1 ) ∈ W 11 ψ ( 2 α − 1 ) , ψ ( 2 α − 1 ) ( 0 ) . \frac{1}{11^\alpha}L_{2\alpha}
=\frac{1}{11^\alpha}U^{(1)}(L_{2\alpha-1})
=U^{(1)}\left(\frac{(j_{2\alpha-1}+k_{2\alpha-1}x)r_L}{(2+11x)^{\psi(2\alpha-1)}}\right)+U^{(1)}(f_{2\alpha-1})
\in W_{11\psi(2\alpha-1),\psi(2\alpha-1)}^{(0)}. 1 1 α 1 L 2 α = 1 1 α 1 U ( 1 ) ( L 2 α − 1 ) = U ( 1 ) ( ( 2 + 11 x ) ψ ( 2 α − 1 ) ( j 2 α − 1 + k 2 α − 1 x ) r L ) + U ( 1 ) ( f 2 α − 1 ) ∈ W 11 ψ ( 2 α − 1 ) , ψ ( 2 α − 1 ) ( 0 ) .
It readily follows that
11 ψ ( 2 α − 1 ) = 11 ⌊ 11 2 α 8 ⌋ = 11 ( 11 2 α − 1 8 ) = 11 2 α + 1 − 3 8 − 1 = ψ ( 2 α ) . 11\psi(2\alpha-1)=11\left\lfloor\frac{11^{2\alpha}}{8}\right\rfloor
=11\left(\frac{11^{2\alpha}-1}{8}\right)
=\frac{11^{2\alpha+1}-3}{8}-1=\psi(2\alpha). 11 ψ ( 2 α − 1 ) = 11 ⌊ 8 1 1 2 α ⌋ = 11 ( 8 1 1 2 α − 1 ) = 8 1 1 2 α + 1 − 3 − 1 = ψ ( 2 α ) .
Thus, we get
1 11 α L 2 α = W ψ ( 2 α ) , ψ ( 2 α − 1 ) ( 0 ) . \frac{1}{11^\alpha}L_{2\alpha}=W_{\psi(2\alpha),\psi(2\alpha-1)}^{(0)}. 1 1 α 1 L 2 α = W ψ ( 2 α ) , ψ ( 2 α − 1 ) ( 0 ) .
Consequently, there exists
F 2 α ( j k ) = ∑ m ≥ 0 s j k ( m ) 11 θ j k 0 ( m ) x m ∈ Z [ 1 2 , x ] with 1 11 α L 2 α = F 2 α ( 00 ) + y F 2 α ( 10 ) + z F 2 α ( 01 ) ( 2 + 11 x ) ψ ( 2 α ) c ( x ) ψ ( 2 α − 1 ) . \mathcal{F}_{2\alpha}^{(jk)}
=\sum_{m\geq0}s_{jk}(m)11^{\theta^{0}_{jk}(m)}x^m\in\mathbb{Z}\left[\frac{1}{2},x\right]
\quad\text{with}\quad
\frac{1}{11^\alpha}L_{2\alpha}
=\frac{\mathcal{F}_{2\alpha}^{(00)}+y\mathcal{F}_{2\alpha}^{(10)}+z\mathcal{F}_{2\alpha}^{(01)}}{(2+11x)^{\psi(2\alpha)}c(x)^{\psi(2\alpha-1)}}. F 2 α ( j k ) = m ≥ 0 ∑ s j k ( m ) 1 1 θ j k 0 ( m ) x m ∈ Z [ 2 1 , x ] with 1 1 α 1 L 2 α = ( 2 + 11 x ) ψ ( 2 α ) c ( x ) ψ ( 2 α − 1 ) F 2 α ( 00 ) + y F 2 α ( 10 ) + z F 2 α ( 01 ) .
So, by (2) , we have
( 2 + 11 x ) ψ ( 2 α ) 11 α L 2 α = F 2 α ( 00 ) + y F 2 α ( 10 ) + z F 2 α ( 01 ) c ( x ) ψ ( 2 α − 1 ) ∈ ⨁ v ∈ { x , y , z } v Z [ 1 2 , x ] . \frac{(2+11x)^{\psi(2\alpha)}}{11^\alpha}L_{2\alpha}
=\frac{\mathcal{F}_{2\alpha}^{(00)}+y\mathcal{F}_{2\alpha}^{(10)}+z\mathcal{F}_{2\alpha}^{(01)}}{c(x)^{\psi(2\alpha-1)}}\in\bigoplus_{v\in\{x,y,z\}}v\mathbb{Z}\left[\frac{1}{2},x\right]. 1 1 α ( 2 + 11 x ) ψ ( 2 α ) L 2 α = c ( x ) ψ ( 2 α − 1 ) F 2 α ( 00 ) + y F 2 α ( 10 ) + z F 2 α ( 01 ) ∈ v ∈ { x , y , z } ⨁ v Z [ 2 1 , x ] .
whence
c ( x ) ψ ( 2 α − 1 ) ∣ F 2 α ( j k ) . c(x)^{\psi(2\alpha-1)}\mid\mathcal{F}_{2\alpha}^{(jk)}. c ( x ) ψ ( 2 α − 1 ) ∣ F 2 α ( j k ) .
Moreover, by Lemma 3.4 , there exists
F ^ 2 α ( j k ) = ∑ m ≥ 0 s j k ( m ) 11 θ j k ( m ) x m ∈ Z [ 1 2 , x ] with F 2 α ( j k ) c ( x ) ψ ( 2 α − 1 ) = F ^ 2 α ( j k ) . \widehat{\mathcal{F}}_{2\alpha}^{(jk)}
=\sum_{m\geq0}s_{jk}(m)11^{\theta_{jk}(m)}x^m\in\mathbb{Z}\left[\frac{1}{2},x\right]
\quad\text{with}\quad
\frac{\mathcal{F}_{2\alpha}^{(jk)}}{c(x)^{\psi(2\alpha-1)}}=\widehat{\mathcal{F}}_{2\alpha}^{(jk)}. F 2 α ( j k ) = m ≥ 0 ∑ s j k ( m ) 1 1 θ j k ( m ) x m ∈ Z [ 2 1 , x ] with c ( x ) ψ ( 2 α − 1 ) F 2 α ( j k ) = F 2 α ( j k ) .
Thus, we get
( 2 + 11 x ) ψ ( 2 α ) 11 α L 2 α = F ^ 2 α ( 00 ) + y F ^ 2 α ( 10 ) + z F ^ 2 α ( 01 ) ∈ ⨁ v ∈ { x , y , z } v Z [ 1 2 , x ] and 1 11 α L 2 α = W ψ ( 2 α ) , 0 ( 0 ) . \frac{(2+11x)^{\psi(2\alpha)}}{11^\alpha}L_{2\alpha}
=\widehat{\mathcal{F}}_{2\alpha}^{(00)}
+y\widehat{\mathcal{F}}_{2\alpha}^{(10)}
+z\widehat{\mathcal{F}}_{2\alpha}^{(01)}
\in\bigoplus_{v\in\{x,y,z\}}v\mathbb{Z}\left[\frac{1}{2},x\right]
\quad\text{and}\quad
\frac{1}{11^\alpha}L_{2\alpha}=\mathcal{W}^{(0)}_{\psi(2\alpha),0}. 1 1 α ( 2 + 11 x ) ψ ( 2 α ) L 2 α = F 2 α ( 00 ) + y F 2 α ( 10 ) + z F 2 α ( 01 ) ∈ v ∈ { x , y , z } ⨁ v Z [ 2 1 , x ] and 1 1 α 1 L 2 α = W ψ ( 2 α ) , 0 ( 0 ) .
Next, applying U ( 0 ) U^{(0)} U ( 0 ) and by Lemma 3.3, for some j ^ 2 α + 1 , k ^ 2 α + 1 ∈ Z [ 1 2 ] \widehat{j}_{2\alpha+1},\widehat{k}_{2\alpha+1}\in\mathbb{Z}\left[\frac{1}{2}\right] j 2 α + 1 , k 2 α + 1 ∈ Z [ 2 1 ] , we have
1 11 α L 2 α + 1 = 1 11 α U ( 0 ) ( L 2 α ) = 11 f 2 α + 1 + 11 ( j ^ 2 α + 1 + k ^ 2 α + 1 x ) r L ( 2 + 11 x ) 11 ψ ( 2 α ) + 15 c ( x ) ψ ( 2 α ) \frac{1}{11^\alpha}L_{2\alpha+1}
=\frac{1}{11^\alpha}U^{(0)}(L_{2\alpha})
=11f_{2\alpha+1}
+\frac{11(\widehat{j}_{2\alpha+1}+\widehat{k}_{2\alpha+1}x)r_L}{(2+11x)^{11\psi(2\alpha)+15}c(x)^{\psi(2\alpha)}} 1 1 α 1 L 2 α + 1 = 1 1 α 1 U ( 0 ) ( L 2 α ) = 11 f 2 α + 1 + ( 2 + 11 x ) 11 ψ ( 2 α ) + 15 c ( x ) ψ ( 2 α ) 11 ( j 2 α + 1 + k 2 α + 1 x ) r L
with f 2 α + 1 ∈ W 11 ψ ( 2 α ) + 15 , ψ ( 2 α ) ( 0 ) f_{2\alpha+1}\in\mathcal{W}^{(0)}_{11\psi(2\alpha)+15,\psi(2\alpha)} f 2 α + 1 ∈ W 11 ψ ( 2 α ) + 15 , ψ ( 2 α ) ( 0 ) .
Note that
11 ψ ( 2 α ) + 15 = ψ ( 2 α + 1 ) , 11\psi(2\alpha)+15=\psi(2\alpha+1), 11 ψ ( 2 α ) + 15 = ψ ( 2 α + 1 ) ,
and so,
1 11 α + 1 L 2 α + 1 − ( j ^ 2 α + 1 + k ^ 2 α + 1 x ) r L ( 2 + 11 x ) ψ ( 2 α + 1 ) c ( x ) ψ ( 2 α ) = f 2 α + 1 ∈ W ψ ( 2 α + 1 ) , ψ ( 2 α ) ( 1 ) . \frac{1}{11^{\alpha+1}}L_{2\alpha+1}
-\frac{(\widehat{j}_{2\alpha+1}+\widehat{k}_{2\alpha+1}x)r_L}{(2+11x)^{\psi(2\alpha+1)}c(x)^{\psi(2\alpha)}}
=f_{2\alpha+1}\in\mathcal{W}^{(1)}_{\psi(2\alpha+1),\psi(2\alpha)}. 1 1 α + 1 1 L 2 α + 1 − ( 2 + 11 x ) ψ ( 2 α + 1 ) c ( x ) ψ ( 2 α ) ( j 2 α + 1 + k 2 α + 1 x ) r L = f 2 α + 1 ∈ W ψ ( 2 α + 1 ) , ψ ( 2 α ) ( 1 ) .
By (2) , ( 2 + 11 x ) ψ ( 2 α + 1 ) (2+11x)^{\psi(2\alpha+1)} ( 2 + 11 x ) ψ ( 2 α + 1 ) cancels the pole of L 2 α + 1 L_{2\alpha+1} L 2 α + 1 at [ 1 2 ] \left[\frac{1}{2}\right] [ 2 1 ] , the numerator of
f 2 α + 1 + ( j ^ 2 α + 1 + k ^ 2 α + 1 x ) r L ( 2 + 11 x ) ψ ( 2 α + 1 ) c ( x ) ψ ( 2 α ) f_{2\alpha+1}
+\frac{(\widehat{j}_{2\alpha+1}+\widehat{k}_{2\alpha+1}x)r_L}{(2+11x)^{\psi(2\alpha+1)}c(x)^{\psi(2\alpha)}} f 2 α + 1 + ( 2 + 11 x ) ψ ( 2 α + 1 ) c ( x ) ψ ( 2 α ) ( j 2 α + 1 + k 2 α + 1 x ) r L
is divisible by the factor c ( x ) ψ ( 2 α ) c(x)^{\psi(2\alpha)} c ( x ) ψ ( 2 α ) . So, by Lemma 3.4 , there exist j α + 1 , k α + 1 ∈ Z [ 1 2 ] j_{\alpha+1},k_{\alpha+1}\in\mathbb{Z}\left[\frac{1}{2}\right] j α + 1 , k α + 1 ∈ Z [ 2 1 ] such that
1 11 α + 1 L 2 α + 1 − ( j 2 α + 1 + k 2 α + 1 x ) r L ( 2 + 11 x ) ψ ( 2 α + 1 ) ∈ W ψ ( 2 α + 1 ) , 0 ( 1 ) . \frac{1}{11^{\alpha+1}}L_{2\alpha+1}
-\frac{(j_{2\alpha+1}+k_{2\alpha+1}x)r_L}{(2+11x)^{\psi(2\alpha+1)}}
\in\mathcal{W}^{(1)}_{\psi(2\alpha+1),0}. 1 1 α + 1 1 L 2 α + 1 − ( 2 + 11 x ) ψ ( 2 α + 1 ) ( j 2 α + 1 + k 2 α + 1 x ) r L ∈ W ψ ( 2 α + 1 ) , 0 ( 1 ) .
This concludes the proof. □
Proof of Initial Relations for Theorem 2.3
The proofs in § 2 –§ 4 depend on the following finitely many identities between modular functions, all of which we verify by computer:
the modular equation (10) for x x x , and hence (11) for t = 2 + 11 x t=2+11x t = 2 + 11 x ,
the representation (3) of L 1 L_1 L 1 ,
the 198 relations (25) for U ( i ) ( y j z k x m ) U^{(i)}(y^jz^kx^m) U ( i ) ( y j z k x m ) , i , j , k ∈ { 0 , 1 } i,j,k\in\{0,1\} i , j , k ∈ { 0 , 1 } , j + k ≤ 1 j+k\leq1 j + k ≤ 1 , 0 ≤ m ≤ 32 0\leq m\leq32 0 ≤ m ≤ 32 , from which we obtain the 6534 initial cases 0 ≤ m , n ≤ 32 0\leq m,n\leq32 0 ≤ m , n ≤ 32 of Theorem 2.3 ,
the initial cases 0 ≤ n ≤ 32 0\leq n\leq32 0 ≤ n ≤ 32 of Lemma 3.1 .
We now explain how each of these is reduced to a finite computation. Precisely, to complete the proofs, we need to confirm a certain set of initial relations between modular functions. Note that by (11) , we have
U ( i ) ( y j z k x m ( 2 + 11 x ) n ) = − 1 b 0 ∑ ν = 1 33 b ν U ( i ) ( y j z k x m ( 2 + 11 x ) n − ν ) = 2 20 ( 2 + 11 x ) 11 c ( x ) ∑ ν = 1 33 b ν U ( i ) ( y j z k x m ( 2 + 11 x ) n − ν ) . U^{(i)}\left(\frac{y^jz^kx^m}{(2+11x)^n}\right)
=-\frac{1}{b_0}\sum_{\nu=1}^{33}b_\nu U^{(i)}\left(\frac{y^jz^kx^m}{(2+11x)^{n-\nu}}\right)
=\frac{2^{20}}{(2+11x)^{11}c(x)}\sum_{\nu=1}^{33}b_\nu U^{(i)}\left(\frac{y^jz^kx^m}{(2+11x)^{n-\nu}}\right). U ( i ) ( ( 2 + 11 x ) n y j z k x m ) = − b 0 1 ν = 1 ∑ 33 b ν U ( i ) ( ( 2 + 11 x ) n − ν y j z k x m ) = ( 2 + 11 x ) 11 c ( x ) 2 20 ν = 1 ∑ 33 b ν U ( i ) ( ( 2 + 11 x ) n − ν y j z k x m ) .
We can of course extrapolate until n − ν < 0 n-\nu<0 n − ν < 0 , i.e., until we are studying combinations of
U ( i ) ( y j z k x m ( 2 + 11 x ) n ) U^{(i)}\left(y^jz^kx^m(2+11x)^n\right) U ( i ) ( y j z k x m ( 2 + 11 x ) n )
for n > 0 n > 0 n > 0 . This in turn can be expressed through the linearity of U ( i ) U^{(i)} U ( i ) and the binomial expansion ( 2 + 11 x ) n (2+11x)^n ( 2 + 11 x ) n into combinations of
U ( i ) ( y j z k x m ) . (25) U^{(i)}\left(y^jz^kx^m\right). \tag*{(25)} U ( i ) ( y j z k x m ) . (25)
Thus, if we have explicit relations for (25) for i , j , k ∈ { 0 , 1 } i,j,k \in\{0,1\} i , j , k ∈ { 0 , 1 } and 0 ≤ m ≤ 32 0 \le m \le32 0 ≤ m ≤ 32 , then from these we can construct the 6534 6534 6534 relations for U ( i ) ( y j z k x m ( 2 + 11 x ) n ) U^{(i)}\left(\frac{y^jz^kx^m}{(2+11x)^n}\right) U ( i ) ( ( 2 + 11 x ) n y j z k x m ) with 0 ≤ m , n ≤ 32 0 \le m,n \le32 0 ≤ m , n ≤ 32 .
Likewise, the proof of (10) constitutes a polynomial relationship between modular functions that needs to be checked.
Relations for (25) recorded in the Mathematica supplement have the following form:
U ( 1 ) ( y j z k x m ) = f j , k , m ( 1 ) ∈ ⨁ v ∈ { x , y , z } v C [ x ] , (26) U^{(1)}\left(y^jz^kx^m\right)=f_{j,k,m}^{(1)}\in\bigoplus_{v\in\{x,y,z\}}v\mathbb{C}[x], \tag*{(26)} U ( 1 ) ( y j z k x m ) = f j , k , m ( 1 ) ∈ v ∈ { x , y , z } ⨁ v C [ x ] , (26)
( 2 + 11 x ) 15 U ( 0 ) ( y j z k x m ) = f j , k , m ( 0 ) ∈ ⨁ v ∈ { x , y , z } v C [ x ] , (27) (2+11x)^{15}U^{(0)}\left(y^jz^kx^m\right)=f_{j,k,m}^{(0)}\in\bigoplus_{v\in\{x,y,z\}}v\mathbb{C}[x], \tag*{(27)} ( 2 + 11 x ) 15 U ( 0 ) ( y j z k x m ) = f j , k , m ( 0 ) ∈ v ∈ { x , y , z } ⨁ v C [ x ] , (27)
in which f j , k , m ( i ) f_{j,k,m}^{(i)} f j , k , m ( i ) are polynomials, given explicitly. In principle, we could compare the principal parts of ( 2 + 11 x ) 15 ( 1 − i ) U ( i ) ( y j z k x m ) (2+11x)^{15(1-i)}U^{(i)}\left(y^jz^kx^m\right) ( 2 + 11 x ) 15 ( 1 − i ) U ( i ) ( y j z k x m ) and f j , k , m ( i ) f_{j,k,m}^{(i)} f j , k , m ( i ) at [ 0 ] [0] [ 0 ] . However, this is difficult to compute, especially on the left-hand side of the relations. On the other hand, we can multiply both sides of each relation by an eta quotient which annihilates the pole at [ 0 ] [0] [ 0 ] and induces a pole at [ ∞ ] [\infty] [ ∞ ] . The resulting principal parts are much easier to check. Indeed, 1 z \frac{1}{z} z 1 suffices for our purpose. There exists M ∈ N M\in\mathbb{N} M ∈ N such that
U ( i ) ( ( 2 + 11 x ( 11 τ ) ) 15 ( 1 − i ) z ( 11 τ ) M y j z k x m ) , 1 z M f j , k , m ( i ) ∈ M ∞ ( X 0 ( 22 ) ) . U^{(i)}\left(\frac{(2+11x(11\tau))^{15(1-i)}}{z(11\tau)^M}y^jz^kx^m\right),\ \frac{1}{z^M}f_{j,k,m}^{(i)}\in M^\infty(X_0(22)). U ( i ) ( z ( 11 τ ) M ( 2 + 11 x ( 11 τ ) ) 15 ( 1 − i ) y j z k x m ) , z M 1 f j , k , m ( i ) ∈ M ∞ ( X 0 ( 22 )) .
Remembering the definition of U ( i ) U^{(i)} U ( i ) , we need only check that the principal part of
U 11 ( A 1 − i ( 2 + 11 x ( 11 τ ) ) 15 ( 1 − i ) z ( 11 τ ) M y j z k x m ) − 1 z M f j , k , m ( i ) U_{11}\left(\mathcal{A}^{1-i}\frac{(2+11x(11\tau))^{15(1-i)}}{z(11\tau)^M}y^jz^kx^m\right)-\frac{1}{z^M}f_{j,k,m}^{(i)} U 11 ( A 1 − i z ( 11 τ ) M ( 2 + 11 x ( 11 τ ) ) 15 ( 1 − i ) y j z k x m ) − z M 1 f j , k , m ( i )
vanishes. These can be computed in Section 5 of the Mathematica supplement. Note that L 1 L_1 L 1 constitutes the case of (27) for j = k = m = 0 j=k=m=0 j = k = m = 0 . The modular equation (10) can likewise be computed by cusp analysis in the notebook. The calculation of (26) -(27) and (10) is done with PARI/GP.15
Conclusion
We conclude this paper by proposing the following conjecture. For a ∈ Z a\in\mathbb{Z} a ∈ Z define p a [ 1 ] ( n ) p_a^{[1]}(n) p a [ 1 ] ( n ) and p a [ 2 ] ( n ) p_a^{[2]}(n) p a [ 2 ] ( n ) by
P a [ 1 ] ( q ) : = ( − q ; q ) ∞ a ( q ; q ) ∞ = ∑ n ≥ 0 p a [ 1 ] ( n ) q n , P a [ 2 ] ( q ) : = ( − q ; q ) ∞ a ( q ; q ) ∞ = ∑ n ≥ 0 p a [ 2 ] ( n ) q n . P_a^{[1]}(q):=\frac{(-q;q)_\infty^a}{(q;q)_\infty}=\sum_{n\ge0}p_a^{[1]}(n)q^n,\qquad P_a^{[2]}(q):=\frac{(-q;q)_\infty^a}{(q;q)_\infty}=\sum_{n\ge0}p_a^{[2]}(n)q^n. P a [ 1 ] ( q ) := ( q ; q ) ∞ ( − q ; q ) ∞ a = n ≥ 0 ∑ p a [ 1 ] ( n ) q n , P a [ 2 ] ( q ) := ( q ; q ) ∞ ( − q ; q ) ∞ a = n ≥ 0 ∑ p a [ 2 ] ( n ) q n .
Since ( − q ; q ) ∞ = ( q 2 ; q 2 ) ∞ / ( q ; q ) ∞ (-q;q)_\infty=(q^2;q^2)_\infty/(q;q)_\infty ( − q ; q ) ∞ = ( q 2 ; q 2 ) ∞ / ( q ; q ) ∞ , these are eta quotients of level 2 2 2 , and the associated congruence families live on X 0 ( 22 ) X_0(22) X 0 ( 22 ) , the very curve of genus 2 2 2 that we have hitherto studied. Note that P 2 [ 2 ] ( q ) = D 2 ( q ) P_2^{[2]}(q)=D_2(q) P 2 [ 2 ] ( q ) = D 2 ( q ) , so Theorem 1.1 is the case a = 2 a=2 a = 2 of part (2) below.
Conjecture 6.1. Let α ∈ N \alpha\in\mathbb{N} α ∈ N and n ∈ N 0 n\in\mathbb{N}_0 n ∈ N 0 .
If a ≡ 3 , 5 , 7 ( m o d 11 ) a\equiv3,5,7\pmod{11} a ≡ 3 , 5 , 7 ( mod 11 ) , then
p a [ 1 ] ( 11 α n + δ a , α [ 1 ] ) ≡ 0 ( m o d 11 ⌈ α 2 ⌉ ) , where 24 δ a , α [ 1 ] ≡ 1 − a ( m o d 11 α ) . p_a^{[1]}\left(11^\alpha n+\delta_{a,\alpha}^{[1]}\right)\equiv0\pmod{11^{\lceil\frac{\alpha}{2}\rceil}},\qquad\text{where }24\delta_{a,\alpha}^{[1]}\equiv1-a\pmod{11^\alpha}. p a [ 1 ] ( 1 1 α n + δ a , α [ 1 ] ) ≡ 0 ( mod 1 1 ⌈ 2 α ⌉ ) , where 24 δ a , α [ 1 ] ≡ 1 − a ( mod 1 1 α ) . If a ≢ 1 , 3 , 5 ( m o d 11 ) a\not\equiv1,3,5\pmod{11} a ≡ 1 , 3 , 5 ( mod 11 ) , then
p a [ 2 ] ( 11 α n + δ a , α [ 2 ] ) ≡ 0 ( m o d 11 ⌈ α 2 ⌉ ) , where 24 δ a , α [ 2 ] ≡ 5 − a ( m o d 11 α ) . p_a^{[2]}\left(11^\alpha n+\delta_{a,\alpha}^{[2]}\right)\equiv0\pmod{11^{\lceil\frac{\alpha}{2}\rceil}},\qquad\text{where }24\delta_{a,\alpha}^{[2]}\equiv5-a\pmod{11^\alpha}. p a [ 2 ] ( 1 1 α n + δ a , α [ 2 ] ) ≡ 0 ( mod 1 1 ⌈ 2 α ⌉ ) , where 24 δ a , α [ 2 ] ≡ 5 − a ( mod 1 1 α ) .
It is natural to ask which methods could prove Conjecture 6.1 . The classical approach via modular equations, in the tradition of Watson and Atkin and as refined by Paule–Radu and by the localization method, does succeed, as the present paper shows, but only at the cost of a non-unit localizing factor, an auxiliary denominator c ( x ) c(x) c ( x ) that must be properly controlled, and a substantial computation. A more conceptual argument would clearly be highly desirable. The ℓ \ell ℓ -adic Galois-theoretic methods developed by Folsom–Kent–Ono, Boylan–Webb guarantee that congruences of this shape exist. But it is uncertain whether the exponent ⌈ α 2 ⌉ \lceil\frac{\alpha}{2} \rceil ⌈ 2 α ⌉ can be properly accounted for. Moreover, the generating functions involved are weakly holomorphic forms of half-integral weight with an eta multiplier, not Hecke eigenforms to which such techniques most naturally apply. The framework developed in [11 ] based on slopes of U ℓ U_{\ell} U ℓ on over-convergent forms is perhaps the most suggestive, since the growth rate ⌈ α 2 ⌉ \lceil\frac{\alpha}{2} \rceil ⌈ 2 α ⌉ is strongly reminiscent of a slope 1 2 \frac{1}{2} 2 1 . However, this would require packaging these forms in the appropriate component of weight space at a certain tame level.
Appendix
Here we state explicitly the expression for c ( x ) c(x) c ( x ) .
c ( x ) : = 69550975755977284911104 − 71287715497527282651820032 x + 17900193598896240733994540032 x 2 + 2433656434135752614962897574400 x 3 + 135334237711138006616618300344576 x 4 + 4603551043973265247618043536808704 x 5 + 109616062492953370639775636433454464 x 6 + 1954419189636606079380190879818816064 x 7 + 27144272314898816695853907223593334304 x 8 + 301089044785364403767185228909074692504 x 9 + 2710730798554339991175805693674009005740 x 10 + 20011291470089564255540586248512208139082 x 11 + 121821501325881013307089187765394122218837 x 12 + 612703095907075970646910882428597137091912 x 13 + 2541715085550544936276941583396435763817408 x 14 + 8649797863778836226399665129987964376904896 x 15 + 23913154138234188165042955394025759141437184 x 16 + 52880092072078116046416414506583439275205632 x 17 + 91342853380107788221814612586966899809652736 x 18 + 118783323283193765986323689131241918352015360 x 19 + 109398375434451346894431470218757768960606208 x 20 + 63625268819176226667199093217678019620241408 x 21 + 17574243049798037033395145952709018207125504 x 22 . \begin{aligned}
c(x) :={}& 69550975755977284911104-71287715497527282651820032x\\
&+17900193598896240733994540032x^{2}+2433656434135752614962897574400x^{3}\\
&+135334237711138006616618300344576x^{4}+4603551043973265247618043536808704x^{5}\\
&+109616062492953370639775636433454464x^{6}+1954419189636606079380190879818816064x^{7}\\
&+27144272314898816695853907223593334304x^{8}+301089044785364403767185228909074692504x^{9}\\
&+2710730798554339991175805693674009005740x^{10}+20011291470089564255540586248512208139082x^{11}\\
&+121821501325881013307089187765394122218837x^{12}+612703095907075970646910882428597137091912x^{13}\\
&+2541715085550544936276941583396435763817408x^{14}+8649797863778836226399665129987964376904896x^{15}\\
&+23913154138234188165042955394025759141437184x^{16}+52880092072078116046416414506583439275205632x^{17}\\
&+91342853380107788221814612586966899809652736x^{18}+118783323283193765986323689131241918352015360x^{19}\\
&+109398375434451346894431470218757768960606208x^{20}+63625268819176226667199093217678019620241408x^{21}\\
&+17574243049798037033395145952709018207125504x^{22}.
\end{aligned} c ( x ) := 69550975755977284911104 − 71287715497527282651820032 x + 17900193598896240733994540032 x 2 + 2433656434135752614962897574400 x 3 + 135334237711138006616618300344576 x 4 + 4603551043973265247618043536808704 x 5 + 109616062492953370639775636433454464 x 6 + 1954419189636606079380190879818816064 x 7 + 27144272314898816695853907223593334304 x 8 + 301089044785364403767185228909074692504 x 9 + 2710730798554339991175805693674009005740 x 10 + 20011291470089564255540586248512208139082 x 11 + 121821501325881013307089187765394122218837 x 12 + 612703095907075970646910882428597137091912 x 13 + 2541715085550544936276941583396435763817408 x 14 + 8649797863778836226399665129987964376904896 x 15 + 23913154138234188165042955394025759141437184 x 16 + 52880092072078116046416414506583439275205632 x 17 + 91342853380107788221814612586966899809652736 x 18 + 118783323283193765986323689131241918352015360 x 19 + 109398375434451346894431470218757768960606208 x 20 + 63625268819176226667199093217678019620241408 x 21 + 17574243049798037033395145952709018207125504 x 22 .
Here are the initial values of the constants λ γ δ i j k \lambda_{\gamma\delta}^{ijk} λ γ δ ij k from Theorem 2.3 .
λ 00 100 = 0 , λ 00 000 = − 34 λ 10 100 = 11 , λ 10 000 = − 23 λ 01 100 = 24 , λ 01 000 = − 12 , \lambda_{00}^{100}=0,\quad\lambda_{00}^{000}=-34\quad\lambda_{10}^{100}=11,\quad\lambda_{10}^{000}=-23\quad\lambda_{01}^{100}=24,\quad\lambda_{01}^{000}=-12, λ 00 100 = 0 , λ 00 000 = − 34 λ 10 100 = 11 , λ 10 000 = − 23 λ 01 100 = 24 , λ 01 000 = − 12 ,
λ 00 110 = − 1 , λ 00 010 = − 34 λ 10 110 = 10 , λ 10 010 = − 23 λ 01 110 = 23 , λ 01 010 = − 12 , \lambda_{00}^{110}=-1,\quad\lambda_{00}^{010}=-34\quad\lambda_{10}^{110}=10,\quad\lambda_{10}^{010}=-23\quad\lambda_{01}^{110}=23,\quad\lambda_{01}^{010}=-12, λ 00 110 = − 1 , λ 00 010 = − 34 λ 10 110 = 10 , λ 10 010 = − 23 λ 01 110 = 23 , λ 01 010 = − 12 ,
λ 00 101 = − 3 , λ 00 001 = − 38 λ 10 101 = 8 , λ 10 001 = − 27 λ 01 101 = 19 , λ 01 001 = − 16. \lambda_{00}^{101}=-3,\quad\lambda_{00}^{001}=-38\quad\lambda_{10}^{101}=8,\quad\lambda_{10}^{001}=-27\quad\lambda_{01}^{101}=19,\quad\lambda_{01}^{001}=-16. λ 00 101 = − 3 , λ 00 001 = − 38 λ 10 101 = 8 , λ 10 001 = − 27 λ 01 101 = 19 , λ 01 001 = − 16.
Acknowledgments
This research began when the first author was funded by the Austrian Science Fund (FWF): W1214-N15, Project DK6 and currently, he is supported by the European Research Council (ERC) under the European Unions Horizon 2020 research and innovation programme (grant agreement No. 101001179). Nicolas Allen Smoot was funded in whole by the Austrian Science Fund (FWF) Principal Investigator Project (No. 10.55776 / PAT26154525, “Multiplicities Between Modular Congruences”). He would like to thank the Austrian Government and People for their generous support. A personal thanks to Denise and Alexander Teuchtmann for their inspiration during the writing of this paper.