This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Global properties of a Hecke ring associated with the Heisenberg Lie algebra

Fumitake Hyodo
Abstract

This study concerns (not necessarily commutative) Hecke rings associated with certain algebras and describes a formal Dirichlet series with coefficients in the Hecke rings, which can be used to generalize Shimura’s series. Considering the case of the Heisenberg Lie algebra, an analog of the identity for Shimura’s series derived employing the rationality theorem, presented by Hecke and Tamagawa, is established. Moreover, this analog recovers the explicit formula for the pro-isomorphic zeta function of the Heisenberg Lie algebra shown by Grunewald, Segal and Smith.

000 2020 Mathematics Subject Classification. Primary 20C08; Secondary 20G25, 20G30, 11F03, 11M41, 20E07.

1 Introduction

This study concerns Hecke rings introduced by Shimura [15]. A classical study of Hecke rings is the work by Hecke [6] and Tamagawa [18] on the Hecke rings associated with the general linear groups. They showed that these Hecke rings are commutative polynomial rings. Furthermore, they defined formal power series with coefficients in these Hecke rings, and showed their rationality. The results of this work are summarized in [17, Chapter 3], where formal Dirichlet series with coefficients in these Hecke rings were further introduced. Andrianov [1], Hina–Sugano [7], Satake [12], and Shimura [16] studied Hecke rings associated with classical groups, wherein they further developed the work of Hecke [6] and Tamagawa [18]. In addition, other studies were conducted on the Hecke rings associated with Jacobi and Chevalley groups by Dulinsky [4] and Iwahori-Matsumoto [11], respectively.

As mentioned above, various studies have been carried out on Hecke rings. However, the class of Hecke rings defined by Shimura is vast, and only a small part of it has been studied to date.

From now on, an algebra implies an abelian group with a bi-additive product (e.g., an associative algebra, a Lie algebra). Let LL be an algebra that is free of finite rank as an abelian group. Our previous work [9] introduced the Hecke rings RLR_{L} and R^L\widehat{R}_{L} associated with LL. For the definition, see Section 2. In this study, we deal with the formal Dirichlet series DL(s)D_{L}(s) and D^L(s)\widehat{D}_{L}(s) with coefficients in RLR_{L} and R^L\widehat{R}_{L}, respectively, which are defined in Section 3.

The first result of this study is to show that the Euler product formula for D^L(s)\widehat{D}_{L}(s) holds, and to give a sufficient condition for DL(s)D_{L}(s) to have the Euler product expansion (cf. Theorems 3.1 and 3.2).

If L=rL=\mathbb{Z}^{r} is the free abelian Lie algebra of rank rr, the Hecke ring RrR_{\mathbb{Z}^{r}} and R^r\widehat{R}_{\mathbb{Z}^{r}} coincide with those treated by Hecke [6] and Tamagawa [18]. Further, the formal Dirichlet series Dr(s)D_{\mathbb{Z}^{r}}(s) and D^r(s)\widehat{D}_{\mathbb{Z}^{r}}(s) equal those treated in [17, Chapter 3]. Thus, it can be said that our study generalizes their study. We discuss them in Section 4.

Denote by \mathcal{H} the Heisenberg Lie algebra, that is, the free nilpotent Lie algebra of class 22 on two generators. The second result of this study is the establishment of identities for D(s)D_{\mathcal{H}}(s) and D^(s)\widehat{D}_{\mathcal{H}}(s), which is the primary result of this study. Let 𝜽^=(θ^p)p\widehat{\boldsymbol{\theta}}=(\widehat{\theta}_{p})_{p} be a family of indeterminates indexed by all prime numbers pp. The key idea for stating our main theorem involves regarding R^\widehat{R}_{\mathcal{H}} as a module over the polynomial ring R^2[𝜽^]\widehat{R}_{\mathbb{Z}^{2}}[\widehat{\boldsymbol{\theta}}]. The main theorem is as follows:

Theorem 1.1 (Theorem 5.12).

There exists a formal Dirichlet series I^2(𝛉^;s)\widehat{I}_{2}(\widehat{\boldsymbol{\theta}};s) with coefficients in R^2[𝛉^]\widehat{R}_{\mathbb{Z}^{2}}[\widehat{\boldsymbol{\theta}}] satisfying the following identity:

I^2(𝜽^;s)D(s)=I^2(𝜽^;s)D^(s)=1.\widehat{I}_{2}(\widehat{\boldsymbol{\theta}};s)\cdot D_{\mathcal{H}}(s)=\widehat{I}_{2}(\widehat{\boldsymbol{\theta}};s)\cdot\widehat{D}_{\mathcal{H}}(s)=1.

It is worth noting that this theorem is similar to Shimura’s Theorem 4.5 for the case r=2r=2. At the conclusion of Section 5.2, we establish that Theorem 1.1 recovers Shimura’s Theorem for r=2r=2 via the endomorphism ϕ^\widehat{\phi} introduced in Definition 5.10.

The proof is essentially done by using some results of our previous study [8] which is described in Section 5.1. There is no great difficulty in proving the claims stated in this study. Rather, it is important to note a natural generalization of series of [6, 17, 18], and a concise identity given for a formal Dirichlet series whose coefficients are not always commutative (cf. Remark 5.14).

In [8, 9] and this study, the case of the Heisenberg Lie algebra is considered as a first step. The author expects that many new Hecke rings will appear in the class of the Hecke rings of this study. Further study of these Hecke rings is now in progress by the author. In [10], the author investigated the Euler factor of D^L(s)\widehat{D}_{L}(s) at each prime number in the case where LL is a higher Heisenberg Lie algebra.

Tamagawa [18], by using his Hecke theory, further investigated certain zeta functions, and proved that each of them is an entire function and has a functional equation. However, even in the Hecke rings associated with the Heisenberg Lie algebra, no analogue has been found. Further research is needed to find such applications to number theory.

It should be mentioned that our series DL(s)D_{L}(s) and D^L(s)\widehat{D}_{L}(s) are related to the zeta functions of groups and rings introduced by Grunewald, Segal and Smith [5]. Let GG be a torsion-free finitely generated nilpotent group, and G^\widehat{G} its profinite completion. Denote by 𝒮ni(G)(resp.𝒮n(G))\mathcal{S}^{i}_{n}(G)\ (\text{resp.}\ \mathcal{S}^{\wedge}_{n}(G)) the family of subgroups HH of GG of index nn such that there is an isomorphism HGH\cong G of groups (resp.H^G^of topological groups)(\text{resp.}\ \widehat{H}\cong\widehat{G}\ \text{of topological groups}). The zeta functions ζGi(s)\zeta^{i}_{G}(s) and ζG(s)\zeta^{\wedge}_{G}(s) of GG were defined in [5] as follows:

ζGi(s)=n>0#𝒮ni(G)ns,ζG(s)=n>0#𝒮n(G)ns,\zeta^{i}_{G}(s)=\sum_{n>0}\#\mathcal{S}^{i}_{n}(G)n^{-s},\quad\zeta^{\wedge}_{G}(s)=\sum_{n>0}\#\mathcal{S}^{\wedge}_{n}(G)n^{-s},

where ss is a complex variable.

As an analogue of them, one can define the zeta functions of LL. Let ^\widehat{\mathbb{Z}} be the profinite completion of \mathbb{Z}, and set L^=L^\widehat{L}=L\otimes\widehat{\mathbb{Z}}. Denote by 𝒮ni(L)(resp.𝒮n(L))\mathcal{S}^{i}_{n}(L)\ (\text{resp.}\ \mathcal{S}^{\wedge}_{n}(L)) the family of subalgebras MM of LL of index nn such that there is an isomorphism MLM\cong L as algebras (resp.M^L^as algebras over ^)(\text{resp.}\ M\otimes\widehat{\mathbb{Z}}\cong\widehat{L}\ \text{as algebras over $\widehat{\mathbb{Z}}$}). We set ani(L)=#𝒮ni(L)a_{n}^{i}(L)=\#\mathcal{S}^{i}_{n}(L) and an(L)=#𝒮n(L)a_{n}^{\wedge}(L)=\#\mathcal{S}^{\wedge}_{n}(L) for each nn. The zeta functions ζLi(s)\zeta^{i}_{L}(s) and ζL(s)\zeta_{L}^{\wedge}(s) of LL are defined as follows:

ζLi(s)=n>0ani(L)ns,ζL(s)=n>0an(L)ns.\zeta^{i}_{L}(s)=\sum_{n>0}a_{n}^{i}(L)n^{-s},\quad\zeta_{L}^{\wedge}(s)=\sum_{n>0}a_{n}^{\wedge}(L)n^{-s}.

The zeta function ζL(s)\zeta_{L}^{\wedge}(s) was also introduced in [5], and is called pro-isomorphic zeta function ζL(s)\zeta_{L}^{\wedge}(s) of LL in [2] and [3]. Although there are few papers on ζLi(s)\zeta^{i}_{L}(s), it is a natural analogue of ζGi(s)\zeta^{i}_{G}(s). We call ζLi(s)\zeta^{i}_{L}(s) the isomorphic zeta function of LL.

As we mention in Section 6, ζLi(s)\zeta^{i}_{L}(s) and ζL(s)\zeta_{L}^{\wedge}(s) equal the coefficient-wise images of DL(s)D_{L}(s) and D^L(s)\widehat{D}_{L}(s) under the degree maps on RLR_{L} and R^L\widehat{R}_{L}, respectively. For the definition of the degree map on a Hecke ring, see Section 2. Moreover, at the end of Section 6, we prove that our Theorem 1.1 derives the explicit formulae for ζi(s)\zeta^{i}_{\mathcal{H}}(s) and ζ(s)\zeta_{\mathcal{H}}^{\wedge}(s) via the degree map on R^\widehat{R}_{\mathcal{H}} as follows:

ζi(s)=ζ(s)=ζ(2s2)ζ(2s3),\zeta^{i}_{\mathcal{H}}(s)=\zeta_{\mathcal{H}}^{\wedge}(s)=\zeta(2s-2)\zeta(2s-3),

where ζ(s)\zeta(s) is the Riemann zeta function.

This identity is essentially due to Grunewald, Segal and Smith [5, Theorem 7.6]. Precisely, for the free nilpotent group 𝔉=𝔉c,g\mathfrak{F}=\mathfrak{F}_{c,g} of class cc on gg-generators, the identity ζ𝔉i(s)=ζ𝔉(s)\zeta^{i}_{\mathfrak{F}}(s)=\zeta_{\mathfrak{F}}^{\wedge}(s) and the explicit formulae for them were obtained. For the free nilpotent Lie algebra =c,g\mathcal{F}=\mathcal{F}_{c,g} of class cc on gg-generators, by an argument essentially equivalent to that of [5], Berman, Glazer, and Schein proved in [2, Theorem 5.1] that ζ(s)\zeta_{\mathcal{F}}^{\wedge}(s) equals ζ𝔉i(s)\zeta^{i}_{\mathfrak{F}}(s) and ζ𝔉(s)\zeta^{\wedge}_{\mathfrak{F}}(s) (cf. Theorem 7.1). In Proposition 7.4, the equality ζ(s)=ζi(s)\zeta^{\wedge}_{\mathcal{F}}(s)=\zeta^{i}_{\mathcal{F}}(s) is verified in a similar way as in the proof of [5, Theorem 7.6]. As a result, the equality ζi(s)=ζ(s)=ζ𝔉(s)=ζ𝔉i(s)\zeta^{i}_{\mathcal{F}}(s)=\zeta_{\mathcal{F}}^{\wedge}(s)=\zeta_{\mathfrak{F}}^{\wedge}(s)=\zeta^{i}_{\mathfrak{F}}(s) holds.

Write \mathfrak{H} for the Heisenberg group 𝔉2,2\mathfrak{F}_{2,2}, and focus on the identity ζi(s)=ζ(s)=ζ(s)=ζi(s)\zeta^{i}_{\mathcal{H}}(s)=\zeta_{\mathcal{H}}^{\wedge}(s)=\zeta^{\wedge}_{\mathfrak{H}}(s)=\zeta^{i}_{\mathfrak{H}}(s). Another generalization of the identity ζ(s)=ζ(s)\zeta_{\mathcal{H}}^{\wedge}(s)=\zeta^{\wedge}_{\mathfrak{H}}(s) is known. Suppose that the nilpotent class of GG is 22. Define the Lie algebra L(G)L(G) as (G/Z)Z(G/Z)\oplus Z with the usual Lie bracket operation induced by the commutator in GG, where ZZ is the center of GG. Then, we have L()=L(\mathfrak{H})=\mathcal{H}, and the identity ζG(s)=ζL(G)(s)\zeta^{\wedge}_{G}(s)=\zeta^{\wedge}_{L(G)}(s) is known to hold (cf. [2, Section 1.1] or [13, Section 1.2.2]). On the other hand, ζGi(s)=ζG(s)\zeta^{i}_{G}(s)=\zeta_{G}^{\wedge}(s) does not hold in general. Indeed, Theorems 7.1 and 7.3 of [5] provide a counter-example. The equality ζGi(s)=ζL(G)i(s)\zeta^{i}_{G}(s)=\zeta^{i}_{L(G)}(s) is proved in Corollary 7.6, and thus ζL(G)(s)=ζL(G)i(s)\zeta_{L(G)}^{\wedge}(s)=\zeta^{i}_{L(G)}(s) does not always hold.

For pro-isomorphic zeta functions of Lie algebras, Berman, Glazer, and Schein [2] further investigated. The explicit formula for ζL(s)\zeta_{L}^{\wedge}(s) was shown in [2, Section 5], specifically for LL belonging to a certain class of Lie algebras over the integer rings of number fields. So far, we have not found any formulae for DL(s)D_{L}(s) and D^L(s)\widehat{D}_{L}(s) that recover their formulae except for this study.

The contents of this paper are organized as follows. In Section 2, we review the Hecke rings RLR_{L} and R^L\widehat{R}_{L}. In Section 3, the formal series DL(s)D_{L}(s) and D^L(s)\widehat{D}_{L}(s) are introduced. In Section 4, the case of L=rL=\mathbb{Z}^{r} is considered. In Section 5, we study the series D(s)D_{\mathcal{H}}(s) and D^(s)\widehat{D}_{\mathcal{H}}(s), and prove our main theorem. In Section 6, our series DL(s)D_{L}(s) and D^L(s)\widehat{D}_{L}(s) are related to the isomorphic zeta function ζLi(s)\zeta^{i}_{L}(s) and the pro-isomorphic zeta function ζL(s)\zeta_{L}^{\wedge}(s) of LL, respectively. Subsequently, we prove that our main theorem also recovers the explicit formulae for ζi(s)\zeta^{i}_{\mathcal{H}}(s) and ζ(s)\zeta_{\mathcal{H}}^{\wedge}(s). Finally, in Section 7, we observe the isomorphic zeta functions in the cases of the free nilpotent Lie algebras and class-2 nilpotent Lie algebras.

2 Hecke rings associated with algebras

First, we briefly recall the definition of Hecke rings and their degree maps. For more details, refer to [17, Chapter 3]. Let GG be a group, Δ\Delta be a submonoid of GG, and Γ\Gamma be a subgroup of Δ\Delta. We assume that the pair (Γ,Δ)(\Gamma,\Delta) is a double finite pair; that is, for all AΔA\in\Delta, Γ\ΓAΓ\Gamma\backslash\Gamma A\Gamma and ΓAΓ/Γ\Gamma A\Gamma/\Gamma are finite sets. Then, one can define the Hecke ring R=R(Γ,Δ)R=R(\Gamma,\Delta) associated with the pair (Γ,Δ)(\Gamma,\Delta) as follows:

  • The underlying abelian group is the free abelian group on the set Γ\Δ/Γ\Gamma\backslash\Delta/\Gamma.

  • For all A,BΔA,B\in\Delta, the product of ΓAΓ\Gamma A\Gamma and ΓBΓ\Gamma B\Gamma is defined to be

    ΓCΓΓ\Δ/Γ#{ΓβΓ\ΓBΓ|Cβ1ΓAΓ}ΓCΓ.\sum_{\Gamma C\Gamma\in\Gamma\backslash\Delta/\Gamma}\#\{\Gamma\beta\in\Gamma\backslash\Gamma B\Gamma\ |\ C\beta^{-1}\in\Gamma A\Gamma\}\cdot\Gamma C\Gamma.

For every AΔA\in\Delta, write TΓ,Δ(A)T_{\Gamma,\Delta}(A) for the element ΓAΓ\Gamma A\Gamma of RR. We define the degree map on RR to be the additive map degR:R\deg\!{R}:R\to\mathbb{Z} such that TΓ,Δ(A)degR=#Γ\ΓAΓT_{\Gamma,\Delta}(A)^{\deg\!{R}}=\#\Gamma\backslash\Gamma A\Gamma for every AΔA\in\Delta. Notably, it is known that degR\deg\!{R} forms a ring homomorphism.

Let pp be a prime number, and let LL be as in Section 1. We next recall the Hecke rings associated with LL introduced in [9]. Fix a \mathbb{Z}-basis of LL, and let rr be the rank of LL. Then, Autalg(L){\rm Aut}_{\mathbb{Q}}^{alg}(L\otimes\mathbb{Q}), Autpalg(Lp){\rm Aut}_{\mathbb{Q}_{p}}^{alg}(L\otimes\mathbb{Q}_{p}), Endalg(L){\rm End}_{\mathbb{Z}}^{alg}(L), and Endpalg(Lp){\rm End}_{\mathbb{Z}_{p}}^{alg}(L\otimes\mathbb{Z}_{p}) are all identified with subsets of Mr(p)M_{r}(\mathbb{Q}_{p}). In [9, Section 2], the following notation was introduced:

GL=Autalg(L),GLp=Autpalg(Lp),ΔL=Endalg(L)GL,ΔLp=Endpalg(Lp)GLp,ΓL=Autalg(L),ΓLp=Autpalg(Lp).\begin{array}[]{llllll}G_{L}&=&{\rm Aut}_{\mathbb{Q}}^{alg}(L\otimes\mathbb{Q}),&G_{L_{p}}&=&{\rm Aut}_{\mathbb{Q}_{p}}^{alg}(L\otimes\mathbb{Q}_{p}),\\ \rule{0.0pt}{12.91663pt}\Delta_{L}&=&{\rm End}_{\mathbb{Z}}^{alg}(L)\cap G_{L},&\Delta_{L_{p}}&=&{\rm End}_{\mathbb{Z}_{p}}^{alg}(L\otimes\mathbb{Z}_{p})\cap G_{L_{p}},\\ \rule{0.0pt}{12.91663pt}\Gamma_{\!L}&=&{\rm Aut}_{\mathbb{Z}}^{alg}(L),&\Gamma_{\!L_{p}}&=&{\rm Aut}_{\mathbb{Z}_{p}}^{alg}(L\otimes\mathbb{Z}_{p}).\end{array}

The global Hecke rings RLR_{L} and the local Hecke ring RLpR_{L_{p}} are the Hecke rings with respect to (ΓL,ΔL)(\Gamma_{\!L},\Delta_{L}) and (ΓLp,ΔLp)(\Gamma_{\!L_{p}},\Delta_{L_{p}}), respectively.

The other global Hecke ring R^L\widehat{R}_{L} was introduced in [9, Section 3]. Define the group G^L\widehat{G}_{L} to be the restricted direct product of GLpG_{L_{p}} relative to ΓLp\Gamma_{\!L_{p}} for all prime numbers pp, that is, the set of elements (αp)p(\alpha_{p})_{p} of pGLp\prod_{p}G_{L_{p}} such that αpΓLp\alpha_{p}\in\Gamma_{\!L_{p}} for almost all pp. The monoid Δ^L\widehat{\Delta}_{L} and the group Γ^L\widehat{\Gamma}_{\!L} denote G^LpΔLp\widehat{G}_{L}\cap\prod_{p}\Delta_{L_{p}} and pΓLp\prod_{p}\Gamma_{\!L_{p}}, respectively. Then, we write R^L\widehat{R}_{L} for the Hecke ring with respect to (Γ^L,Δ^L)(\widehat{\Gamma}_{\!L},\widehat{\Delta}_{L}).

Section 3 of [9] described relations among these Hecke rings. The local Hecke ring RLpR_{L_{p}} is related to the global Hecke ring R^L\widehat{R}_{L} as follows:

Proposition 2.1 ([9, Proposition 3.1]).

The following assertions hold:

  1. 1.

    The local Hecke ring RLpR_{L_{p}} is regarded as a subring of R^L\widehat{R}_{L} by the map induced by the natural inclusion of ΔLp\Delta_{L_{p}} into Δ^L\widehat{\Delta}_{L}.

  2. 2.

    For each prime number qq with pqp\not=q, the local Hecke rings RLpR_{L_{p}} and RLqR_{L_{q}} commute with each other in R^L\widehat{R}_{L}.

  3. 3.

    R^L\widehat{R}_{L} is generated by the family of local Hecke rings {RLp}p\{R_{L_{p}}\}_{p} as a ring.

For simplicity, we set

TLp=TΓLp,ΔLp,TL=TΓL,ΔL,T^L=TΓ^L,Δ^L.T_{L_{p}}=T_{\Gamma_{\!L_{p}},\Delta_{L_{p}}},\quad T_{L}=T_{\Gamma_{\!L},\Delta_{L}},\quad\widehat{T}_{L}=T_{\widehat{\Gamma}_{\!L},\widehat{\Delta}_{L}}.

Then, we have TLp(α)=T^L(α)T_{L_{p}}(\alpha)=\widehat{T}_{L}(\alpha) in R^L\widehat{R}_{L} for each αΔLp\alpha\in\Delta_{L_{p}}. Let us relate the global Hecke rings R^L\widehat{R}_{L} and RLR_{L}. The map ηL\eta_{L} denotes the diagonal embedding of ΔL\Delta_{L} into pΔLp\prod_{p}\Delta_{L_{p}}. Then, we define the additive map ηL:R^LRL\eta^{*}_{L}:\widehat{R}_{L}\to R_{L} given by T^L(α^)βTL(β)\widehat{T}_{L}(\widehat{\alpha})\mapsto\sum_{\beta}T_{L}(\beta), where β\beta runs through a complete system of representatives of ΓL\ηL1(Γ^Lα^Γ^L)/ΓL\Gamma_{\!L}\backslash\eta_{L}^{-1}(\widehat{\Gamma}_{\!L}\widehat{\alpha}\widehat{\Gamma}_{\!L})/\Gamma_{\!L}. Let us denote by ηL:ΓL\ΔLΓ^L\Δ^L\eta_{*L}:\Gamma_{\!L}\backslash\Delta_{L}\to\widehat{\Gamma}_{\!L}\backslash\widehat{\Delta}_{L} the map induced by ηL\eta_{L}. Then, the two global Hecke rings are related as follows:

Lemma 2.2 ([9, Lemma 3.2]).

If the map ηL\eta_{*L} is bijective, then ηL\eta_{L}^{*} is multiplicative and injective.

From now on, we regard R^L\widehat{R}_{L} as a subring of RLR_{L} if ηL\eta_{*L} is bijective.

In the rest of this section, we relate R^L\widehat{R}_{L} to the automorphism group of L^\widehat{L}.

Proposition 2.3.

Let 𝔸f\mathbb{A}_{\rm f} be the ring of finite adeles over \mathbb{Q}, and set 𝐐=pp\mathbf{Q}=\prod_{p}\mathbb{Q}_{p}. Then, the objects G^L\widehat{G}_{L}, Δ^L\widehat{\Delta}_{L}, and Γ^L\widehat{\Gamma}_{\!L} satisfy the following identities as subsets of Mr(𝐐)M_{r}(\mathbf{Q}):

G^L=Aut𝔸falg(L𝔸f),Δ^L=End^alg(L^)G^L,Γ^L=Aut^alg(L^).\begin{array}[]{llllll}\widehat{G}_{L}={\rm Aut}_{\mathbb{A}_{\rm f}}^{alg}(L\otimes\mathbb{A}_{\rm f}),&\widehat{\Delta}_{L}={\rm End}_{\widehat{\mathbb{Z}}}^{alg}(\widehat{L})\cap\widehat{G}_{L},&\widehat{\Gamma}_{\!L}={\rm Aut}_{\widehat{\mathbb{Z}}}^{alg}(\widehat{L}).\end{array}
Proof.

The second and third identities are straightforward consequences of the fact that L^\widehat{L} equals p(Lp)\prod_{p}(L\otimes\mathbb{Z}_{p}). Let us prove the first identity. Denote by GLr(𝐐)GL^{\prime}_{r}(\mathbf{Q}) the restricted direct product of GLr(p)GL_{r}(\mathbb{Q}_{p}) relative to GLr(p)GL_{r}(\mathbb{Z}_{p}) for all prime numbers pp. Then, it is easy to see that GLr(𝐐)GL^{\prime}_{r}(\mathbf{Q}) coincides with GLr(𝔸f)GL_{r}(\mathbb{A}_{\rm f}). And the group G^L\widehat{G}_{L}, by definition, equals the intersection of GLr(𝐐)GL^{\prime}_{r}(\mathbf{Q}) and pGLp\prod_{p}G_{L_{p}}. Thus, we have

G^L=GLr(𝔸f)pGLp.\widehat{G}_{L}=GL_{r}(\mathbb{A}_{\rm f})\cap\prod_{p}G_{L_{p}}.

Since L𝐐L\otimes\mathbf{Q} is identified with p(Lp)\prod_{p}(L\otimes\mathbb{Q}_{p}), the group pGLp\prod_{p}G_{L_{p}} coincides with Aut𝐐alg(L𝐐){\rm Aut}_{\mathbf{Q}}^{alg}(L\otimes\mathbf{Q}). Hence, we have

GLr(𝔸f)pGLp=GLr(𝔸f)Aut𝐐alg(L𝐐)=Aut𝔸falg(L𝔸f).GL_{r}(\mathbb{A}_{\rm f})\cap\prod_{p}G_{L_{p}}=GL_{r}(\mathbb{A}_{\rm f})\cap{\rm Aut}_{\mathbf{Q}}^{alg}(L\otimes\mathbf{Q})={\rm Aut}_{\mathbb{A}_{\rm f}}^{alg}(L\otimes\mathbb{A}_{\rm f}).

This implies the first identity. ∎

3 Formal power series and formal Dirichlet series associated with algebras

Let pp and LL be as in the previous section. We set Lp=LpL_{p}=L\otimes\mathbb{Z}_{p}. In this section, the formal series PLp(X)P_{L_{p}}(X), DL(s)D_{L}(s), and D^L(s)\widehat{D}_{L}(s) are defined. Subsequently, their relationship is described. For a positive integer nn and a nonnegative integer kk, we introduce the following notation:

𝒜^L(n)\displaystyle\widehat{\mathcal{A}}_{L}(n) ={α^Δ^L|[L^:L^α^]=n},𝒜L(n)={αΔL|[L:Lα]=n},\displaystyle=\left\{\widehat{\alpha}\in\widehat{\Delta}_{L}\ \left|\ [\widehat{L}:\widehat{L}^{\widehat{\alpha}}]=n\right.\right\},\ \mathcal{A}_{L}(n)=\left\{\alpha\in\Delta_{L}\ \left|\ [L:L^{\alpha}]=n\right.\right\},
𝒜Lp(pk)\displaystyle\mathcal{A}_{L_{p}}(p^{k}) ={αΔLp|[Lp:Lpα]=pk},\displaystyle=\left\{\alpha\in\Delta_{L_{p}}\ \left|\ [L_{p}:L_{p}^{\alpha}]=p^{k}\right.\right\},

where LpαL_{p}^{\alpha} is the image of LpL_{p} under the endomorphism α\alpha. Additionally, L^α^\widehat{L}^{\widehat{\alpha}} and LαL^{\alpha} are defined in a similar manner. Note that each element of Δ^L\widehat{\Delta}_{L} is regarded as an element of End^alg(L^){\rm End}_{\widehat{\mathbb{Z}}}^{alg}(\widehat{L}) by Proposition 2.3.

Now, the formal power series PLp(X)P_{L_{p}}(X) is introduced. We define

TLp(pk)=αTLp(α),T_{L_{p}}(p^{k})=\sum_{\alpha}T_{L_{p}}(\alpha),\\

where α\alpha runs through a complete system of representatives of ΓLp\𝒜Lp(pk)/ΓLp\Gamma_{\!L_{p}}\backslash\mathcal{A}_{L_{p}}(p^{k})/\Gamma_{\!L_{p}}. The formal power series PLp(X)P_{L_{p}}(X) is defined as the generating function of the sequence {TLp(pk)}k\{T_{L_{p}}(p^{k})\}_{k}; that is,

PLp(X)=k0TLp(pk)Xk.P_{L_{p}}(X)=\sum_{k\geq 0}T_{L_{p}}(p^{k})X^{k}.

Next, the formal Dirichlet series D^L(s)\widehat{D}_{L}(s) and DL(s)D_{L}(s) are defined. We set

T^L(n)=α^T^L(α^),TL(n)=αTL(α),\widehat{T}_{L}(n)=\sum_{\widehat{\alpha}}\widehat{T}_{L}(\widehat{\alpha}),\quad T_{L}(n)=\sum_{\alpha}T_{L}(\alpha),

where α^\widehat{\alpha} (resp.α)(\text{resp.}\ \alpha) runs through a complete system of representatives of Γ^L\𝒜^L(n)/Γ^L\widehat{\Gamma}_{\!L}\backslash\widehat{\mathcal{A}}_{L}(n)/\widehat{\Gamma}_{\!L} (resp.ΓL\𝒜L(n)/ΓL)(\text{resp.}\ \Gamma_{\!L}\backslash\mathcal{A}_{L}(n)/\Gamma_{\!L}). The formal Dirichlet series D^L(s)\widehat{D}_{L}(s) and DL(s)D_{L}(s) are the generating functions of the sequences of {T^L(n)}n\{\widehat{T}_{L}(n)\}_{n} and {TL(n)}n\{T_{L}(n)\}_{n}, respectively; that is,

D^L(s)=n>0T^L(n)ns,DL(s)=n>0TL(n)ns.\widehat{D}_{L}(s)=\sum_{n>0}\widehat{T}_{L}(n)n^{-s},\quad D_{L}(s)=\sum_{n>0}T_{L}(n)n^{-s}.

Next, PLp(X)P_{L_{p}}(X) is related to D^L(s)\widehat{D}_{L}(s). For each element α^\widehat{\alpha} of Δ^L\widehat{\Delta}_{L}, let αp\alpha_{p} denote its ΔLp\Delta_{L_{p}}component. Then, T^L(α^)=pTLp(αp)\widehat{T}_{L}(\widehat{\alpha})=\prod_{p}T_{L_{p}}(\alpha_{p}) is obtained, where pp runs over all prime numbers. Here, this infinite product is meaningful since its terms commute with each other according to Proposition 2.1, and almost all of them are equal to 11. Consequently, the following theorem is proven:

Theorem 3.1.

The sequence {T^L(n)}n\{\widehat{T}_{L}(n)\}_{n} is multiplicative, and the Euler product formula for D^L(s)\widehat{D}_{L}(s) holds; that is,

D^L(s)=pPLp(ps),\widehat{D}_{L}(s)=\prod_{p}P_{L_{p}}(p^{-s}),

where pp runs through all prime numbers.

Proof.

It is easy to see that TLp(pk)=T^L(pk)T_{L_{p}}(p^{k})=\widehat{T}_{L}(p^{k}) in R^L\widehat{R}_{L}. Since L^\widehat{L} is isomorphic to pLp\prod_{p}L_{p}, it follows that [L^:L^α^]=p[Lp:Lpαp][\widehat{L}:\widehat{L}^{\widehat{\alpha}}]=\prod_{p}[L_{p}:L_{p}^{\alpha_{p}}] for each α^Δ^L\widehat{\alpha}\in\widehat{\Delta}_{L}. Hence, we have

𝒜^L(n)=p𝒜Lp(pvp(n)),\widehat{\mathcal{A}}_{L}(n)=\prod_{p}\mathcal{A}_{L_{p}}(p^{v_{p}(n)}),

where vpv_{p} is the pp-adic valuation. This proves the theorem. ∎

Finally, D^L(s)\widehat{D}_{L}(s) is related to DL(s)D_{L}(s) using the additive map ηL:R^LRL\eta_{L}^{*}:\widehat{R}_{L}\to R_{L}. It is evident that ηL\eta_{L}^{*} maps T^L(n)\widehat{T}_{L}(n) to TL(n)T_{L}(n) for each positive integer nn. Thus, the Euler product formula for DL(s)D_{L}(s) is proven.

Theorem 3.2.

If the map ηL\eta_{*L} is bijective, then the sequence {TL(n)}n\{T_{L}(n)\}_{n} is multiplicative, and the Euler product formula for DL(s)D_{L}(s) holds; that is,

DL(s)=pPLp(ps).D_{L}(s)=\prod_{p}P_{L_{p}}(p^{-s}).
Proof.

By assumption, R^L\widehat{R}_{L} is considered as a subring of RLR_{L}. Since TL(n)=T^L(n)T_{L}(n)=\widehat{T}_{L}(n) and DL(s)=D^L(s)D_{L}(s)=\widehat{D}_{L}(s), Theorem 3.1 implies the desired result. ∎

4 Case of the free abelian Lie algebra r\mathbb{Z}^{r}

Using the notations in Section 3, the theory of the Hecke ring with general linear groups as reported by Hecke [6], Shimura [17], and Tamagawa [18] is considered.

Let rr be a positive integer. Clearly, GrG_{\mathbb{Z}^{r}} and GprG_{{\mathbb{Z}^{r}_{p}}} are identified with GLr()GL_{r}(\mathbb{Q}) and GLr(p)GL_{r}(\mathbb{Q}_{p}), respectively. Similarly, we have Δr=Mr()GLr()\Delta_{\mathbb{Z}^{r}}=M_{r}(\mathbb{Z})\cap GL_{r}(\mathbb{Q}), Δpr=Mr(p)GLr(p)\Delta_{\mathbb{Z}^{r}_{p}}=M_{r}(\mathbb{Z}_{p})\cap GL_{r}(\mathbb{Q}_{p}), Γr=GLr()\Gamma_{\!\mathbb{Z}^{r}}=GL_{r}(\mathbb{Z}), and, Γpr=GLr(p)\Gamma_{\!\mathbb{Z}^{r}_{p}}=GL_{r}(\mathbb{Z}_{p}). Thus, the Hecke rings RrR_{\mathbb{Z}^{r}} and RprR_{\mathbb{Z}^{r}_{p}} coincide with the Hecke rings treated in [6], [17], and [18]. Furthermore, the Hecke ring R^r\widehat{R}_{\mathbb{Z}^{r}} is identified with RrR_{\mathbb{Z}^{r}} as follows:

Proposition 4.1.

The map ηr:R^rRr\eta_{\mathbb{Z}^{r}}^{*}:\widehat{R}_{\mathbb{Z}^{r}}\to R_{\mathbb{Z}^{r}} is an isomorphism.

Proof.

Lemma 3.3 of [9] implies that ηr\eta_{\mathbb{Z}^{r}}^{*} is an injective homomorphism. Moreover, the map Γr\Δr/ΓrΓ^r\Δ^r/Γ^r\Gamma_{\!\mathbb{Z}^{r}}\backslash\Delta_{\mathbb{Z}^{r}}/\Gamma_{\!\mathbb{Z}^{r}}\to\widehat{\Gamma}_{\!\mathbb{Z}^{r}}\backslash\widehat{\Delta}_{\mathbb{Z}^{r}}/\widehat{\Gamma}_{\!\mathbb{Z}^{r}} induced by ηr\eta_{\mathbb{Z}^{r}}, is bijective according to the elementary divisor theorem. Note that, in [9], r\mathbb{Z}^{r} is defined as the ring of the direct sum of rr-copies of \mathbb{Z}, which is incorrect. It is correct to define r\mathbb{Z}^{r} as the abelian free Lie algebra of rank rr, as in the present study. ∎

Certainly, the formal power series PprP_{\mathbb{Z}^{r}_{p}}(X) equals the local Hecke series treated in [6] and [18]. The following theorem was proved:

Theorem 4.2 ([6, Satz 14], [18, Theorem 3]).

Let

Tr,p(i)=Γprdiag[1,,1,p,,pi]ΓprT_{r,p}^{(i)}=\Gamma_{\!\mathbb{Z}^{r}_{p}}{\rm diag}[1,...,1,\overbrace{p,...,p}^{i}]\Gamma_{\!\mathbb{Z}^{r}_{p}}

for each ii with 1ir1\leq i\leq r. Then, the following assertions hold:

  1. 1.

    RprR_{\mathbb{Z}^{r}_{p}} is the polynomial ring over \mathbb{Z} in variables Tr,p(i)T_{r,p}^{(i)} with 1ir1\leq i\leq r.

  2. 2.

    The series Ppr(X)P_{\mathbb{Z}^{r}_{p}}(X) is a rational function over RrR_{\mathbb{Z}^{r}}, more precisely,

    fr,p(X)Ppr(X)=1,f_{r,p}(X)P_{\mathbb{Z}^{r}_{p}}(X)=1,

    where fr,p(X)=i=0r(1)ipi(i1)/2Tr,p(i)Xif_{r,p}(X)=\sum_{i=0}^{r}(-1)^{i}p^{i(i-1)/2}T_{r,p}^{(i)}X^{i}. Particularly,

    f2,p(X)=1T2,p(1)X+pT2,p(2)X2.f_{2,p}(X)=1-T_{2,p}^{(1)}X+pT_{2,p}^{(2)}X^{2}.
Remark 4.3.

Theorem 4.2 in the case r=2r=2 was proved in [6]. For arbitrary rr, it was demonstrated in [18].

The series Dr(s)D_{\mathbb{Z}^{r}}(s) is none other than the formal Dirichlet series treated in [17, Chapter 3]. Since ηr\eta_{*{\mathbb{Z}^{r}}} is bijective, the following theorem is obtained:

Theorem 4.4.

The Euler product formulae for Dr(s)D_{\mathbb{Z}^{r}}(s) and D^r(s)\widehat{D}_{\mathbb{Z}^{r}}(s) hold; i.e.,

Dr(s)=D^r(s)=pPpr(ps),D_{\mathbb{Z}^{r}}(s)=\widehat{D}_{\mathbb{Z}^{r}}(s)=\prod_{p}P_{\mathbb{Z}^{r}_{p}}(p^{-s}),

where pp runs through all prime numbers.

Proof.

It is an immediate consequence of Theorems 3.1 and 3.2. ∎

Therefore, the following theorem is obtained:

Theorem 4.5 ([17, Theorem 3.21]).

Define Ir(s)I_{r}(s) to be the infinite product pfr,p(ps)\prod_{p}f_{r,p}(p^{-s}). Then, the following is obtained:

Ir(s)Dr(s)=Ir(s)D^r(s)=1.I_{r}(s)D_{\mathbb{Z}^{r}}(s)=I_{r}(s)\widehat{D}_{\mathbb{Z}^{r}}(s)=1.
Proof.

This follows from Theorems 4.2 and 4.4. ∎

5 Case of the Heisenberg Lie algebra

This section studies the proposed series in the case of the Heisenberg Lie algebra \mathcal{H}.

5.1 Local properties

Let us recall the main theorem of [8]. For an element AA of Gp2G_{{\mathbb{Z}^{2}_{p}}} and an element 𝐚{\bf a} of p2\mathbb{Q}_{p}^{2}, denote by (A,𝐚)(A,{\bf a}) the element (A𝐚0 0|A|)\begin{pmatrix}A&{\bf a}\\ 0\ 0&|A|\\ \end{pmatrix} of GL3(p)GL_{3}(\mathbb{Q}_{p}), where |A||A| means the determinant of the matrix AA. Fix a system {x1,x2}\{x_{1},x_{2}\} of free generators of \mathcal{H}. Then, the set {x1,x2,[x1,x2]}\{x_{1},x_{2},[x_{1},x_{2}]\} forms a basis of \mathcal{H}. Hence, the group GpG_{\mathcal{H}_{p}} is identified with the following subset of GL3(p)GL_{3}(\mathbb{Q}_{p}):

{(A,𝐚)|AGp2,𝐚p2}.\left\{(A,{\bf a})\ \left|\ A\in G_{{\mathbb{Z}^{2}_{p}}},\ {\bf a}\in\mathbb{Q}_{p}^{2}\right.\right\}.

In addition, an element (A,𝐚)(A,{\bf a}) of GpG_{\mathcal{H}_{p}} is contained in Δp\Delta_{\mathcal{H}_{p}} (resp. Γp\Gamma_{\!\mathcal{H}_{p}}) if and only if AA is in Δp2\Delta_{\mathbb{Z}^{2}_{p}} (resp. Γp2\Gamma_{\!\mathbb{Z}^{2}_{p}}), and 𝐚{\bf a} is in p2\mathbb{Z}_{p}^{2}.

The following three ring homomorphisms ss, ϕ\phi, and θ\theta were introduced in [8, Section 6]:

Definition 5.1.

For simplicity, we put deg=degRp\deg=\deg\!{R_{\mathcal{H}_{p}}}. The ring homomorphisms s:Rp2Rps:R_{\mathbb{Z}^{2}_{p}}\to R_{\mathcal{H}_{p}}, ϕ:RpRp2\phi:R_{\mathcal{H}_{p}}\to R_{\mathbb{Z}^{2}_{p}}, and θ:RpRp\theta:R_{\mathcal{H}_{p}}\to R_{\mathcal{H}_{p}} are defined by

Tp2(A)s\displaystyle T_{\mathbb{Z}^{2}_{p}}(A)^{s} =Tp(A,𝟎)for each AΔp2,\displaystyle=T_{\mathcal{H}_{p}}(A,{\bf 0})\ \text{for each $A\in\Delta_{\mathbb{Z}^{2}_{p}}$},
Tp(A,𝐚)ϕ\displaystyle T_{\mathcal{H}_{p}}(A,{\bf a})^{\phi} =Tp(A,𝐚)degTp(A,𝟎)degTp2(A)for each (A,𝐚)Δp,\displaystyle=\frac{T_{\mathcal{H}_{p}}(A,{\bf a})^{\deg}}{T_{\mathcal{H}_{p}}(A,{\bf 0})^{\deg}}T_{\mathbb{Z}^{2}_{p}}(A)\ \text{for each $(A,{\bf a})\in\Delta_{\mathcal{H}_{p}}$},
Tp(A,𝐚)θ\displaystyle T_{\mathcal{H}_{p}}(A,{\bf a})^{\theta} =Tp(A,𝐚)degTp(A,p𝐚)degTp(A,p𝐚)for each (A,𝐚)Δp.\displaystyle=\frac{T_{\mathcal{H}_{p}}(A,{\bf a})^{\deg}}{T_{\mathcal{H}_{p}}(A,p{\bf a})^{\deg}}T_{\mathcal{H}_{p}}(A,p{\bf a})\ \text{for each $(A,{\bf a})\in\Delta_{\mathcal{H}_{p}}$}.
Remark 5.2.

Although the multiplicativity of ss, ϕ\phi, and θ\theta is not obvious by the definition, it was proved in [8, Section 6].

Some relations among the three ring homomorphisms are introduced.

Proposition 5.3.

The ring homomorphisms ss, ϕ\phi, and θ\theta satisfy the following properties:

ϕs=idRp2,θs=s,ϕθ=ϕ.\phi\circ s=id_{R_{\mathbb{Z}^{2}_{p}}},\quad\theta\circ s=s,\quad\phi\circ\theta=\phi.
Proof.

It is an easy consequence of Definition 5.1. ∎

Our previous work [8] defined the element T2(pk)T_{2}(p^{k}) of RpR_{\mathcal{H}_{p}} for each nonnegative integer kk as follows:

T2(pk)=(A,𝐚)Tp(A,𝐚),T_{2}(p^{k})=\sum_{(A,{\bf a})}T_{\mathcal{H}_{p}}(A,{\bf a}),

where (A,𝐚)(A,{\bf a}) runs through a complete system of representatives of Γp\Δp/Γp\Gamma_{\!\mathcal{H}_{p}}\backslash\Delta_{\mathcal{H}_{p}}/\Gamma_{\!\mathcal{H}_{p}} satisfying vp(|A|)=kv_{p}(|A|)=k. The formal power series D2,2(X)D_{2,2}(X) was defined as the generating function of the sequence {T2(pk)}k\{T_{2}(p^{k})\}_{k}; that is,

D2,2(X)=k0T2(pk)Xk.D_{2,2}(X)=\sum_{k\geq 0}T_{2}(p^{k})X^{k}.

The main theorem of our previous work [8] is as follows:

Theorem 5.4 ([8, Theorem 7.8]).

Let T2,p(1)T_{2,p}^{(1)} and T2,p(2)T_{2,p}^{(2)} be as in Theorem 4.2. For simplicity, let us set Tp(1,p)=T2,p(1)T_{p}(1,p)=T_{2,p}^{(1)} and Tp(p,p)=T2,p(2)T_{p}(p,p)=T_{2,p}^{(2)}. Define Y=pXY=pX. Then, D2,2(X)D_{2,2}(X) satisfies the following identity:

D2,2(X)θ2Tp(1,p)sD2,2(X)θY+pTp(p,p)sD2,2(X)Y2=1,D_{2,2}(X)^{\theta^{2}}-T_{p}(1,p)^{s}D_{2,2}(X)^{\theta}Y+pT_{p}(p,p)^{s}D_{2,2}(X)Y^{2}=1,

where D2,2(X)θD_{2,2}(X)^{\theta} is the coefficient-wise image of D2,2(X)D_{2,2}(X) under θ\theta, and D2,2(X)θ2D_{2,2}(X)^{\theta^{2}} is defined similarly.


The sequences {T2(pk)}k0\{T_{2}(p^{k})\}_{k\geq 0} and {Tp(pk)}k0\{T_{\mathcal{H}_{p}}(p^{k})\}_{k\geq 0} are related as follows:

Proposition 5.5.

Tp(p2k)=T2(pk)T_{\mathcal{H}_{p}}(p^{2k})=T_{2}(p^{k}) and Tp(p2k+1)=0T_{\mathcal{H}_{p}}(p^{2k+1})=0 for each kk.

Proof.

It is evident that vp([p:p(A,𝐚)])=2vp(|A|)v_{p}([\mathcal{H}_{p}:\mathcal{H}_{p}^{(A,{\bf a})}])=2v_{p}(|A|) for every (A,𝐚)Δp(A,{\bf a})\in\Delta_{\mathcal{H}_{p}}. This completes the proof. ∎

The relation between D2,2(X)D_{2,2}(X) and Pp(X)P_{\mathcal{H}_{p}}(X) is described as follows:

Corollary 5.6.

D2,2(X2)=Pp(X)D_{2,2}(X^{2})=P_{\mathcal{H}_{p}}(X).

Proof.

It is an immediate consequence of the proposition above. ∎

The Hecke ring RpR_{\mathcal{H}_{p}} forms a ring over Rp2R_{\mathbb{Z}^{2}_{p}} via the ring homomorphism ss. Moreover, owing to the second identity of Proposition 5.3, θ\theta is a ring homomorphism over Rp2R_{\mathbb{Z}^{2}_{p}}. Thus, RpR_{\mathcal{H}_{p}} is a module (not a ring!) over the polynomial ring Rp2[θ]R_{\mathbb{Z}^{2}_{p}}[\theta] in one variable θ\theta. Further, the maps ss, ϕ\phi, and θ\theta depend on pp. Subsequently, we set sp=ss_{p}=s, ϕp=ϕ\phi_{p}=\phi, and θp=θ\theta_{p}=\theta. Therefore, Theorem 5.4 can be rewritten as follows:

Theorem 5.7.

Let f2,p(X)f_{2,p}(X) be as in Theorem 4.2, and let us keep the notation of Theorem 5.4. Then, Pp(X)P_{\mathcal{H}_{p}}(X) satisfies the following identity:

g2,p(θp;pX2)Pp(X)=1,g_{2,p}(\theta_{p};pX^{2})P_{\mathcal{H}_{p}}(X)=1,

where

g2,p(θp;X)=θp2f2,p(X/θp)=θp2Tp(1,p)θpX+pTp(p,p)X2.g_{2,p}(\theta_{p};X)=\theta_{p}^{2}\cdot f_{2,p}(X/\theta_{p})=\theta_{p}^{2}-T_{p}(1,p)\theta_{p}X+pT_{p}(p,p)X^{2}.
Proof.

Clear. ∎

We have just introduced the three ring homomorphism, of which ϕp\phi_{p} has not been used so far. In fact, it has been shown that ϕp\phi_{p} plays a role establishing the relationship between Pp(X)P_{\mathcal{H}_{p}}(X) and Pp2(X)P_{\mathbb{Z}^{2}_{p}}(X) as follows:

Theorem 5.8 ([8, Theorem 7.5]).

Pp(X)ϕp=Pp2(pX2).P_{\mathcal{H}_{p}}(X)^{\phi_{p}}=P_{\mathbb{Z}^{2}_{p}}(pX^{2}).

5.2 Global properties

In this subsection, the Dirichlet series D(s)D_{\mathcal{H}}(s) and D^(s)\widehat{D}_{\mathcal{H}}(s) are considered. Since the bijectivity of η\eta_{*\mathcal{H}} was proved in [9, Lemma 3.4], the map η\eta_{\mathcal{H}}^{*} is an injective ring homomorphism. Moreover, the nonsurjectivity of η\eta_{\mathcal{H}}^{*} was shown in [9, Section 4]. Hence, the global Hecke ring R^\widehat{R}_{\mathcal{H}} is a proper subring of RR_{\mathcal{H}}. However, the following theorem can be obtained:

Theorem 5.9.

The Euler product formulae for D(s)D_{\mathcal{H}}(s) and D^(s)\widehat{D}_{\mathcal{H}}(s) hold; that is,

D(s)=D^(s)=pPp(ps).D_{\mathcal{H}}(s)=\widehat{D}_{\mathcal{H}}(s)=\prod_{p}P_{\mathcal{H}_{p}}(p^{-s}).
Proof.

It is an immediate consequence of Theorems 3.1 and 3.2. ∎

The ring homomorphisms s^\widehat{s}, ϕ^\widehat{\phi}, and θ^p\widehat{\theta}_{p} are defined as follows:

Definition 5.10.

The ring homomorphisms s^:R^2R^\widehat{s}:\widehat{R}_{\mathbb{Z}^{2}}\to\widehat{R}_{\mathcal{H}} and ϕ^:R^R^2\widehat{\phi}:\widehat{R}_{\mathcal{H}}\to\widehat{R}_{\mathbb{Z}^{2}} are defined by

T^2(A^)s^=pTp2(Ap)sp for each A^Δ^2,\displaystyle\widehat{T}_{\mathbb{Z}^{2}}(\widehat{A})^{\widehat{s}}=\prod_{p}T_{\mathbb{Z}^{2}_{p}}(A_{p})^{s_{p}}\text{ for each $\widehat{A}\in\widehat{\Delta}_{\mathbb{Z}^{2}}$},
T^(α^)ϕ^=pTp(αp)ϕp for each α^Δ^,\displaystyle\widehat{T}_{\mathcal{H}}(\widehat{\alpha})^{\widehat{\phi}}=\prod_{p}T_{\mathcal{H}_{p}}(\alpha_{p})^{\phi_{p}}\text{ for each $\widehat{\alpha}\in\widehat{\Delta}_{\mathcal{H}}$},

where ApA_{p} (resp.αp)(\text{resp.}\ \alpha_{p}) is Δp2\Delta_{\mathbb{Z}^{2}_{p}} (resp.Δp)(\text{resp.}\ \Delta_{\mathcal{H}_{p}}) component of A^\widehat{A} (resp.α^)(\text{resp.}\ \widehat{\alpha}) for each pp.

The ring homomorphism θ^p:R^R^\widehat{\theta}_{p}:\widehat{R}_{\mathcal{H}}\to\widehat{R}_{\mathcal{H}} is defined by

T^(α^)θ^p=Tp(αp)θpqpTq(αq) for each α^Δ^.\widehat{T}_{\mathcal{H}}(\widehat{\alpha})^{\widehat{\theta}_{p}}=T_{\mathcal{H}_{p}}(\alpha_{p})^{\theta_{p}}\cdot\prod_{q\not=p}T_{\mathcal{H}_{q}}(\alpha_{q})\ \text{ for each $\widehat{\alpha}\in\widehat{\Delta}_{\mathcal{H}}$}.

Consequently, the following proposition is obtained:

Proposition 5.11.

The following equalities hold:

  1. 1.

    ϕ^s^=idR^2\widehat{\phi}\circ\widehat{s}=id_{\widehat{R}_{\mathbb{Z}^{2}}},

  2. 2.

    θ^ps^=s^\widehat{\theta}_{p}\circ\widehat{s}=\widehat{s} and ϕ^θ^p=ϕ^\widehat{\phi}\circ\widehat{\theta}_{p}=\widehat{\phi} for each pp,

  3. 3.

    θ^pθ^q=θ^qθ^p\widehat{\theta}_{p}\circ\widehat{\theta}_{q}=\widehat{\theta}_{q}\circ\widehat{\theta}_{p} for any two prime numbers pp, qq.

Proof.

It is an easy consequence of Proposition 5.3. ∎

From the proposition above, it is evident that the Hecke ring R^\widehat{R}_{\mathcal{H}} is a ring over R^2\widehat{R}_{\mathbb{Z}^{2}} by s^\widehat{s}, and that θ^p\widehat{\theta}_{p} is a ring homomorphism over R^2\widehat{R}_{\mathbb{Z}^{2}} for each pp. Set 𝜽^=(θ^p)p\widehat{\boldsymbol{\theta}}=(\widehat{\theta}_{p})_{p}, and let R^2[𝜽^]\widehat{R}_{\mathbb{Z}^{2}}[\widehat{\boldsymbol{\theta}}] be the polynomial ring over R^2\widehat{R}_{\mathbb{Z}^{2}} in infinitely many variables 𝜽^\widehat{\boldsymbol{\theta}}. Then, the Hecke ring R^\widehat{R}_{\mathcal{H}} is an R^2[𝜽^]\widehat{R}_{\mathbb{Z}^{2}}[\widehat{\boldsymbol{\theta}}]-module.

Now, the following theorem is proven, analogous to Theorem 4.5:

Theorem 5.12.

I^2(𝜽^;s)\widehat{I}_{2}(\widehat{\boldsymbol{\theta}};s) is defined as the infinite product

pg2,p(θ^p;p12s).\prod_{p}g_{2,p}(\widehat{\theta}_{p};p^{1-2s}).

Then, the following is obtained:

I^2(𝜽^;s)D(s)=I^2(𝜽^;s)D^(s)=1.\widehat{I}_{2}(\widehat{\boldsymbol{\theta}};s)\cdot D_{\mathcal{H}}(s)=\widehat{I}_{2}(\widehat{\boldsymbol{\theta}};s)\cdot\widehat{D}_{\mathcal{H}}(s)=1.
Proof.

Theorem 5.9 implies that

D^(s)=pPp(ps).\widehat{D}_{\mathcal{H}}(s)=\prod_{p}P_{\mathcal{H}_{p}}(p^{-s}).

Let us fix a prime number pp. Then, the following is obtained:

D^(s)θ^p=Pp(ps)θpqpPq(qs).\widehat{D}_{\mathcal{H}}(s)^{\widehat{\theta}_{p}}=P_{\mathcal{H}_{p}}(p^{-s})^{\theta_{p}}\cdot\prod_{q\not=p}P_{\mathcal{H}_{q}}(q^{-s}).

In addition, RpR_{\mathcal{H}_{p}} and RqR_{\mathcal{H}_{q}} commute with each other in R^\widehat{R}_{\mathcal{H}} for any prime number qq different from pp. Hence, for each element 𝔞p\mathfrak{a}_{p} of Rp2R_{\mathbb{Z}^{2}_{p}}, we have

𝔞p(D^(s))=(𝔞pPp(ps))qpPq(qs).\mathfrak{a}_{p}\cdot(\widehat{D}_{\mathcal{H}}(s))=\left(\mathfrak{a}_{p}\cdot P_{\mathcal{H}_{p}}(p^{-s})\right)\cdot\prod_{q\not=p}P_{\mathcal{H}_{q}}(q^{-s}).

Therefore,

I^2(𝜽^;s)D^(s)=p(g2,p(θp;p12s)Pp(ps)).\widehat{I}_{2}(\widehat{\boldsymbol{\theta}};s)\cdot\widehat{D}_{\mathcal{H}}(s)=\prod_{p}\left(g_{2,p}(\theta_{p};p^{1-2s})\cdot P_{\mathcal{H}_{p}}(p^{-s})\right).

Subsequently, Theorem 5.7 implies that the right-hand side of the equality above is 11, which completes the proof. ∎

In the remainder of this section, it is shown that Theorem 5.12 recovers Shimura’s Theorem 4.5. Let ψ^:R^2[𝜽^]R^2\widehat{\psi}:\widehat{R}_{\mathbb{Z}^{2}}[\widehat{\boldsymbol{\theta}}]\to\widehat{R}_{\mathbb{Z}^{2}} be the ring homomorphism over R^2\widehat{R}_{\mathbb{Z}^{2}} satisfying (θ^p)ψ^=1(\widehat{\theta}_{p})^{\widehat{\psi}}=1 for all pp. Then ϕ^\widehat{\phi} and ψ^\widehat{\psi} are compatible; that is,

(𝔞𝔄)ϕ^=𝔞ψ^𝔄ϕ^for any 𝔞R^2[𝜽^] and any 𝔄R^,(\mathfrak{a}\cdot\mathfrak{A})^{\widehat{\phi}}=\mathfrak{a}^{\widehat{\psi}}\cdot\mathfrak{A}^{\widehat{\phi}}\quad\text{for any $\mathfrak{a}\in\widehat{R}_{\mathbb{Z}^{2}}[\widehat{\boldsymbol{\theta}}]$ and any $\mathfrak{A}\in\widehat{R}_{\mathcal{H}}$},

which follows from Proposition 5.11. Consequently, the following proposition is proven:

Proposition 5.13.

The following identities hold:

  1. 1.

    I^2(𝜽^;s)ψ^=I2(2s1)\widehat{I}_{2}(\widehat{\boldsymbol{\theta}};s)^{\widehat{\psi}}=I_{2}(2s-1),

  2. 2.

    D^(s)ϕ^=D^2(2s1)\widehat{D}_{\mathcal{H}}(s)^{\widehat{\phi}}=\widehat{D}_{\mathbb{Z}^{2}}(2s-1).

Proof.

It is evident that g2,p(θp;pX2)ψ^=g2,p(1;pX2)=f2,p(pX2)g_{2,p}(\theta_{p};pX^{2})^{\widehat{\psi}}=g_{2,p}(1;pX^{2})=f_{2,p}(pX^{2}), which implies the first equality. The second one follows from Theorems 5.8, 5.9 and 4.4. ∎

Therefore, it is concluded that Theorem 5.12 recovers Theorem 4.5 when r=2r=2: The map ϕ^\widehat{\phi} is applied to the equality in Theorem 5.12. Then, Proposition 5.13 and the compatibility of ϕ^\widehat{\phi} and ψ^\widehat{\psi} imply that

I2(2s1)D^2(2s1)=1.I_{2}(2s-1)\widehat{D}_{\mathbb{Z}^{2}}(2s-1)=1.
Remark 5.14.

As shown in Theorem 7.3 of [8], the coefficients of the Hecke series D2,2(X)D_{2,2}(X) are not necessarily commutative. Thus, neither are those of Pp(X)P_{\mathcal{H}_{p}}(X) and D(s)D_{\mathcal{H}}(s).

6 Zeta functions of algebras

Let us return to the case where LL is as in Section 1. In this section, our series DL(s)D_{L}(s) and D^L(s)\widehat{D}_{L}(s) are related to the isomorphic zeta function ζLi(s)\zeta^{i}_{L}(s) and pro-isomorphic zeta function ζL(s)\zeta_{L}^{\wedge}(s) of LL, respectively.

Denote by 𝒮n(L)\mathcal{S}^{\prime\wedge}_{n}(L) the family of ^\widehat{\mathbb{Z}}-subalgebras \mathcal{M} of L^\widehat{L} of index nn such that there is an isomorphism L^\mathcal{M}\cong\widehat{L} of algebras over ^\widehat{\mathbb{Z}}. Then, the maps ΓL\𝒜L(n)𝒮ni(L)\Gamma_{\!L}\backslash\mathcal{A}_{L}(n)\to\mathcal{S}^{i}_{n}(L) given by ΓLαLα\Gamma_{\!L}\alpha\mapsto L^{\alpha} and Γ^L\𝒜^L(n)𝒮n(L)\widehat{\Gamma}_{\!L}\backslash\widehat{\mathcal{A}}_{L}(n)\to\mathcal{S}^{\prime\wedge}_{n}(L) given by Γ^Lα^L^α^\widehat{\Gamma}_{\!L}\widehat{\alpha}\mapsto\widehat{L}^{\widehat{\alpha}} are both bijective. Since LL is free of finite rank as an abelian group, one verifies in a similar way as in the proof of Proposition 1.2 of [5] that the maps 𝒮n(L)𝒮n(L)\mathcal{S}^{\wedge}_{n}(L)\to\mathcal{S}^{\prime\wedge}_{n}(L) defined by MM^M\mapsto M\otimes\widehat{\mathbb{Z}} and 𝒮n(L)𝒮n(L)\mathcal{S}^{\prime\wedge}_{n}(L)\to\mathcal{S}^{\wedge}_{n}(L) defined by L\mathcal{M}\mapsto\mathcal{M}\cap L are inverse to each other, in particular, one has #𝒮n(L)=#𝒮n(L)=an(L)\#\mathcal{S}^{\prime\wedge}_{n}(L)=\#\mathcal{S}^{\wedge}_{n}(L)=a_{n}^{\wedge}(L). Hence, it follows that, for each nn,

TL(n)degRL=ani(L),T^L(n)degR^L=an(L).T_{L}(n)^{\deg\!{R_{L}}}=a_{n}^{i}(L),\quad\widehat{T}_{L}(n)^{\deg\!{\widehat{R}_{L}}}=a_{n}^{\wedge}(L).

Thus, DL(s)D_{L}(s) and D^L(s)\widehat{D}_{L}(s) are related to ζLi(s)\zeta^{i}_{L}(s) and ζL(s)\zeta_{L}^{\wedge}(s) as follows:

DL(s)degRL=ζLi(s),D^L(s)degR^L=ζL(s).D_{L}(s)^{\deg\!{R_{L}}}=\zeta^{i}_{L}(s),\quad\widehat{D}_{L}(s)^{\deg\!{\widehat{R}_{L}}}=\zeta_{L}^{\wedge}(s).

By definition, we have degR^L|RLp=degRLp\deg\!{\widehat{R}_{L}}|_{R_{L_{p}}}=\deg\!{R_{L_{p}}}. Moreover, degR^L\deg\!{\widehat{R}_{L}} and degRL\deg\!{R_{L}} are related as follows:

Proposition 6.1.

If the map ηL\eta_{*L} is bijective, then we have degR^L=degRLηL\deg\!{\widehat{R}_{L}}=\deg\!{R_{L}}\circ\eta_{L}^{*}, that is, degRL|R^L=degR^L\deg\!{R_{L}}|_{\widehat{R}_{L}}=\deg\!{\widehat{R}_{L}}.

Proof.

Since ηL\eta_{*L} is bijective, so is

ηL|ΓL\ηL1(Γ^Lα^Γ^L):ΓL\ηL1(Γ^Lα^Γ^L)Γ^L\Γ^Lα^Γ^L.\eta_{*L}|_{\Gamma_{\!L}\backslash\eta_{L}^{-1}(\widehat{\Gamma}_{\!L}\widehat{\alpha}\widehat{\Gamma}_{\!L})}:\Gamma_{\!L}\backslash\eta_{L}^{-1}(\widehat{\Gamma}_{\!L}\widehat{\alpha}\widehat{\Gamma}_{\!L})\to\widehat{\Gamma}_{\!L}\backslash\widehat{\Gamma}_{\!L}\widehat{\alpha}\widehat{\Gamma}_{\!L}.

Thus, we have #Γ^L\Γ^Lα^Γ^L=#ΓL\ηL1(Γ^Lα^Γ^L),\#\widehat{\Gamma}_{\!L}\backslash\widehat{\Gamma}_{\!L}\widehat{\alpha}\widehat{\Gamma}_{\!L}=\#\Gamma_{\!L}\backslash\eta_{L}^{-1}(\widehat{\Gamma}_{\!L}\widehat{\alpha}\widehat{\Gamma}_{\!L}), which completes the proof. ∎

The above proposition implies the following corollary:

Corollary 6.2.

In order that ζL(s)=ζLi(s)\zeta_{L}^{\wedge}(s)=\zeta^{i}_{L}(s), it is necessary and sufficient that ηL\eta_{*L} is bijective.

Proof.

Sufficiency is an easy consequence of the above proposition and the equality DL(s)=D^L(s)D_{L}(s)=\widehat{D}_{L}(s). Let us prove necessity. Since ηL\eta_{*L} is injective, so is ηL|ΓL\𝒜L(n)\eta_{*L}|_{\Gamma_{\!L}\backslash\mathcal{A}_{L}(n)} for each nn. If ζL(s)=ζLi(s)\zeta_{L}^{\wedge}(s)=\zeta^{i}_{L}(s), then we have #ΓL\𝒜L(n)=#Γ^L\𝒜^L(n)\#\Gamma_{\!L}\backslash\mathcal{A}_{L}(n)=\#\widehat{\Gamma}_{\!L}\backslash\widehat{\mathcal{A}}_{L}(n) for each nn, and thus, ηL\eta_{*L} is bijective, ∎

Next, the case L=rL=\mathbb{Z}^{r} is considered. We make use of the following identity shown by Tamagawa [18]:

Theorem 6.3 ([18, Corollary]).

Let fr,p(X)f_{r,p}(X) be as in Theorem 4.2. Then, the following identity holds:

fr,p(X)degR^r=0kr1(1pkX).f_{r,p}(X)^{\deg\!{\widehat{R}_{\mathbb{Z}^{r}}}}=\prod_{0\leq k\leq r-1}(1-p^{k}X).

This theorem derives the following identity:

Corollary 6.4.

Let Ir(s)I_{r}(s) be as in Theorem 4.5. Then, the following identity holds:

Ir(s)degRr=Ir(s)degR^r=0kr1ζ(sk)1,I_{r}(s)^{\deg\!{R_{\mathbb{Z}^{r}}}}=I_{r}(s)^{\deg\!{\widehat{R}_{\mathbb{Z}^{r}}}}=\prod_{0\leq k\leq r-1}\zeta(s-k)^{-1},

where ζ(s)\zeta(s) is the Riemann zeta function.

Proof.

This follows from the multiplicativity of degR^r\deg\!{\widehat{R}_{\mathbb{Z}^{r}}} and the above theorem. ∎

Theorem 4.5 and Corollary 6.4 recover the explicit formulae for ζr(s)\zeta_{\mathbb{Z}^{r}}^{\wedge}(s) and ζri(s)\zeta^{i}_{\mathbb{Z}^{r}}(s) proved in [5].

Corollary 6.5 ([5, Proposition 1.1]).

The following identity holds:

ζri(s)=ζr(s)=0kr1ζ(sk).\zeta^{i}_{\mathbb{Z}^{r}}(s)=\zeta_{\mathbb{Z}^{r}}^{\wedge}(s)=\prod_{0\leq k\leq r-1}\zeta(s-k).
Proof.

It is immediately verified by applying the maps degRr\deg\!{R_{\mathbb{Z}^{r}}} and degR^r\deg\!{\widehat{R}_{\mathbb{Z}^{r}}} to the identity of Theorem 4.5. ∎

Next, the case L=L=\mathcal{H} is investigated. Let ψ:[𝜽^]\psi:\mathbb{Z}[\widehat{\boldsymbol{\theta}}]\to\mathbb{Z} be the ring homomorphism satisfying (θ^p)ψ=1(\widehat{\theta}_{p})^{\psi}=1 for all pp. Then, the coefficient-wise images of g2,p(θ^p;X)g_{2,p}(\widehat{\theta}_{p};X) and I^2(𝜽^;s)\widehat{I}_{2}(\widehat{\boldsymbol{\theta}};s) under the homomorphism (degR^s^)ψ:R^2[𝜽^](\deg\!{\widehat{R}_{\mathcal{H}}}\circ\widehat{s})\otimes\psi:\widehat{R}_{\mathbb{Z}^{2}}[\widehat{\boldsymbol{\theta}}]\to\mathbb{Z} are as follows:

Proposition 6.6.

The following identities hold:

g2,p(θ^p;X)(degR^s^)ψ\displaystyle g_{2,p}(\widehat{\theta}_{p};X)^{(\deg\!{\widehat{R}_{\mathcal{H}}}\circ\widehat{s})\otimes\psi} =(1p2X2)(1p3X2),\displaystyle=(1-p^{2}X^{2})(1-p^{3}X^{2}),
I^2(𝜽^;s)(degR^s^)ψ\displaystyle\widehat{I}_{2}(\widehat{\boldsymbol{\theta}};s)^{(\deg\!{\widehat{R}_{\mathcal{H}}}\circ\widehat{s})\otimes\psi} =ζ(2s2)1ζ(2s3)1.\displaystyle=\zeta(2s-2)^{-1}\zeta(2s-3)^{-1}.
Proof.

By the identities (1)-(a) and (2)-(a) of [8, Proposition 5.4], we have

Tp(1,p)degR^s^=p2(1+p1),Tp(p,p)degR^s^=p2,T_{p}(1,p)^{\deg\!{\widehat{R}_{\mathcal{H}}}\circ\widehat{s}}=p^{2}(1+p^{-1}),\quad T_{p}(p,p)^{\deg\!{\widehat{R}_{\mathcal{H}}}\circ\widehat{s}}=p^{2},

which implies the first identity. The second is easily derived by the first one. ∎

Remark 6.7.

degR^s^\deg\!{\widehat{R}_{\mathcal{H}}}\circ\widehat{s} and degR^2\deg\!{\widehat{R}_{\mathbb{Z}^{2}}} are slightly different: For each AΔ^2A\in\widehat{\Delta}_{\mathbb{Z}^{2}}, it follows from [8, Proposition 5.4] and [17, Theorem 3.24] that

T^2(A^)degR^s^=[^2:(^2)A]T^2(A^)degR^2.\widehat{T}_{\mathbb{Z}^{2}}(\widehat{A})^{\deg\!{\widehat{R}_{\mathcal{H}}}\circ\widehat{s}}=[\widehat{\mathbb{Z}}^{2}:(\widehat{\mathbb{Z}}^{2})^{A}]\cdot\widehat{T}_{\mathbb{Z}^{2}}(\widehat{A})^{\deg\!{\widehat{R}_{\mathbb{Z}^{2}}}}.

Since degR^θ^p=degR^\deg\!{\widehat{R}_{\mathcal{H}}}\circ\widehat{\theta}_{p}=\deg\!{\widehat{R}_{\mathcal{H}}}, the ring homomorphism (degR^s^)ψ(\deg\!{\widehat{R}_{\mathcal{H}}}\circ\widehat{s}\ )\otimes\psi and degR^\deg\!{\widehat{R}_{\mathcal{H}}} are compatible. Therefore, the explicit formulae for ζi(s)\zeta^{i}_{\mathcal{H}}(s) and ζ(s)\zeta_{\mathcal{H}}^{\wedge}(s) shown in [5] and [2] are recovered.

Corollary 6.8 ([5, Theorem 7.6], [2, Theorem 5.1]).

The following identity holds:

ζi(s)=ζ(s)=ζ(2s2)ζ(2s3).\zeta^{i}_{\mathcal{H}}(s)=\zeta_{\mathcal{H}}^{\wedge}(s)=\zeta(2s-2)\zeta(2s-3).
Proof.

It is an easy consequence of Proposition 6.6 and Theorem 5.12. ∎

7 Isomorphic zeta functions

In this section, we observe the isomorphic zeta functions in the cases of the free nilpotent Lie algebras and class-2 nilpotent Lie algebras.

7.1 Case of the free nilpotent Lie algebras

Let 𝔉=𝔉c,g\mathfrak{F}=\mathfrak{F}_{c,g} (resp. =c,g\mathcal{F}=\mathcal{F}_{c,g}) be the free nilpotent group (resp. Lie algebra) of class cc on gg-generators. For a nilpotent Lie algebra LL, denote by γi(L)\gamma_{i}(L) the ii-th term of its lower central series, and set L𝑎𝑏=L/[L,L]L^{\it ab}=L/[L,L]. In this section, we describe the explicit formulae for the zeta functions of 𝔉\mathfrak{F} and \mathcal{F}. As mentioned in the introduction, they are essentially due to [5]. However, other literature does not deal with isomorphic zeta functions of algebras, and it is necessary to give a detailed proof of the explicit formula for ζi(s)\zeta^{i}_{\mathcal{F}}(s).

In the case c=1c=1, we have 𝔉c,g=c,g=g\mathfrak{F}_{c,g}=\mathcal{F}_{c,g}=\mathbb{Z}^{g}. Hence, the explicit formulae were obtained in Corollary 6.5. In the following, suppose that c2c\geq 2. The explicit formula for ζ(s)\zeta_{\mathcal{F}}^{\wedge}(s) was proved in [2], and those for ζ𝔉i(s)\zeta^{i}_{\mathfrak{F}}(s) and ζ𝔉(s)\zeta_{\mathfrak{F}}^{\wedge}(s) were established in [5]:

Theorem 7.1 ([5, Theorem 7.6], [2, Theorem 5.1]).

Let mim_{i} be the rank of γi()/γi+1()\gamma_{i}(\mathcal{F})/\gamma_{i+1}(\mathcal{F}) for each 1ic1\leq i\leq c. Define α=1gi=1cimi\alpha=\frac{1}{g}\sum_{i=1}^{c}im_{i}, and β=i=2cimi\beta=\sum_{i=2}^{c}im_{i}. Then, we have

ζ𝔉i(s)=ζ𝔉(s)=ζ(s)=j=0g1ζ(αsβj).\zeta^{i}_{\mathfrak{F}}(s)=\zeta_{\mathfrak{F}}^{\wedge}(s)=\zeta_{\mathcal{F}}^{\wedge}(s)=\prod_{j=0}^{g-1}\zeta(\alpha s-\beta-j).
Remark 7.2.

The following formula for mim_{i} was established in [20, Satz 3]:

mi=1ij|iμ(j)gi/j,m_{i}=\frac{1}{i}\sum_{j|i}\mu(j)g^{i/j},

where μ\mu is the Möbius function.

Therefore, it remains to consider ζi(s)\zeta^{i}_{\mathcal{F}}(s). We prepare the following lemma which is a Lie algebra analogue of [19, Theorem 1.8]. The proof imitates the one of [14, Chapter 1, Exercise 7]:

Lemma 7.3.

Let LL be a nilpotent Lie algebra, and let MM be a subalgebra of LL. If L=M+[L,L]L=M+[L,L], then L=ML=M.

Proof.

For each i0i\geq 0, denote by ZiZ_{i} the ii-th upper central series of LL. Assume that MM was a proper subalgebra of LL. Then, there exists i>0i>0 such that Zi+M=LZ_{i}+M=L and Zi1+MLZ_{i-1}+M\subsetneq L. Since [L,L]=[Zi+M,Zi+M]Zi1+M[L,L]=[Z_{i}+M,Z_{i}+M]\subset Z_{i-1}+M, we have L=M+[L,L]Zi1+ML=M+[L,L]\subset Z_{i-1}+M, which is a contradiction. ∎

Now, we prove the explicit formula for ζi(s)\zeta^{i}_{\mathcal{F}}(s).

Proposition 7.4.

Let us keep the notation of Theorem 7.1. Then, the following identity holds:

ζi(s)=ζ(s)=j=0g1ζ(αsβj).\zeta^{i}_{\mathcal{F}}(s)=\zeta_{\mathcal{F}}^{\wedge}(s)=\prod_{j=0}^{g-1}\zeta(\alpha s-\beta-j).
Proof.

Theorem 7.1 reduces us to proving that ζi(s)=ζ(s)\zeta^{i}_{\mathcal{F}}(s)=\zeta_{\mathcal{F}}^{\wedge}(s). Let MM be an element of 𝒮n()\mathcal{S}^{\wedge}_{n}(\mathcal{F}) with a positive integer nn. It is sufficient to show that MM and \mathcal{F} are isomorphic as Lie algebras. Clearly, ab^\mathcal{F}^{\mathrm{ab}}\otimes\widehat{\mathbb{Z}} and Mab^M^{\mathrm{ab}}\otimes\widehat{\mathbb{Z}} are isomorphic to (^)ab(\widehat{\mathcal{F}})^{\mathrm{ab}} and (M^)ab(\widehat{M})^{\mathrm{ab}} as ^\widehat{\mathbb{Z}}-modules, respectively. By assumption, (^)ab(\widehat{\mathcal{F}})^{\mathrm{ab}} and (M^)ab(\widehat{M})^{\mathrm{ab}} are isomorphic as ^\widehat{\mathbb{Z}}-modules. Hence, Mab^M^{\mathrm{ab}}\otimes\widehat{\mathbb{Z}} is isomorphic to ab^=^g\mathcal{F}^{\mathrm{ab}}\otimes\widehat{\mathbb{Z}}=\widehat{\mathbb{Z}}^{g} as ^\widehat{\mathbb{Z}}-modules. Since MabM^{\mathrm{ab}} is finitely generated abelian group, it follows from the fundamental theorem of finitely generated abelian groups that MabM^{\mathrm{ab}} is isomorphic to g\mathbb{Z}^{g}. Hence, we can take elements x1,,xgx_{1},...,x_{g} of MM such that their images under the canonical projection MMabM\to M^{\mathrm{ab}} form a basis of MabM^{\mathrm{ab}}, and it follows from Lemma 7.3 that x1,,xgx_{1},...,x_{g} generate MM. By the universal property of \mathcal{F}, there exists a surjective homomorphism φ:M\varphi:\mathcal{F}\to M, and the induced map φid^:^M^\varphi\otimes id_{\widehat{\mathbb{Z}}}:\widehat{\mathcal{F}}\to\widehat{M} is also surjective. Since ^\widehat{\mathcal{F}} and M^\widehat{M} are isomorphic as algebras over ^\widehat{\mathbb{Z}}, the ranks of them over ^\widehat{\mathbb{Z}} are the same. Therefore, φid^\varphi\otimes id_{\widehat{\mathbb{Z}}} is an isomorphism. The faithful flatness of ^\widehat{\mathbb{Z}} over \mathbb{Z} implies the bijectivity of φ\varphi, which completes the proof. ∎

7.2 Case of class-2 nilpotent Lie algebras

Suppose that LL is a nilpotent Lie algebra of class 22. By the class-two Lie correspondence of [3, Section 3.1], there exists a unique torsion-free finitely generated nilpotent group GG of class 22 up to isomorphism such that LL is isomorphic to L(G)=(G/Z)ZL(G)=(G/Z)\oplus Z as Lie algebras, where ZZ is the center of GG. Hence, we may identify LL with L(G)L(G), and ZZ is regarded as the center of LL by the map zZ(1,z)Lz\in Z\mapsto(1,z)\in L. In this subsection, the equality ζGi(s)=ζLi(s)\zeta_{G}^{i}(s)=\zeta_{L}^{i}(s) is verified. To show this, Proposition 7.5 below is essential. The proposition is mentioned in many references without proof, for example, [5, Section 4],[13, Section 1.2.2], and [2, Section 2.1]. Although a proof is proposed in [3, Proposition 3.1], it is incorrect (cf. Remark 7.15). Therefore, it would be worthwhile to give a precise proof in this study.

Proposition 7.5.

Let nn be a positive integer. Denote by 𝒮n(G)\mathcal{S}_{n}(G) (resp. 𝒮n(L)\mathcal{S}_{n}(L)) the set of subgroups (resp. subalgebras) of GG (resp. LL) of index nn. Then, there exists a bijection fn:𝒮n(G)𝒮n(L)f_{n}:\mathcal{S}_{n}(G)\to\mathcal{S}_{n}(L) such that, for each H𝒮n(G)H\in\mathcal{S}_{n}(G), its image is isomorphic to L(H)L(H) as Lie algebras. In particular, we have #𝒮n(G)=#𝒮n(L)\#\mathcal{S}_{n}(G)=\#\mathcal{S}_{n}(L).

Before proving the proposition, we deduce the following corollary which is our purpose of this subsection:

Corollary 7.6.

The equalities ζGi(s)=ζLi(s)\zeta^{i}_{G}(s)=\zeta^{i}_{L}(s) and ζG(s)=ζL(s)\zeta_{G}^{\wedge}(s)=\zeta_{L}^{\wedge}(s) hold.

Proof.

Let H𝒮n(G)H\in\mathcal{S}_{n}(G), and put M=fn(H)M=f_{n}(H). Then, ML(H)M\cong L(H). By the class-two Lie correspondence of [3, Section 3.1], we see that H𝒮ni(G)H\in\mathcal{S}^{i}_{n}(G) if and only if M𝒮ni(L)M\in\mathcal{S}^{i}_{n}(L). Thus, we have #𝒮ni(G)=#𝒮ni(L)\#\mathcal{S}^{i}_{n}(G)=\#\mathcal{S}^{i}_{n}(L). Since L(G^)=L^L(\widehat{G})=L\otimes\widehat{\mathbb{Z}} and L(H^)=L(H)^L(\widehat{H})=L(H)\otimes\widehat{\mathbb{Z}}, it follows from the local class-two Lie correspondence of [3, Section 3.1] that H𝒮n(G)H\in\mathcal{S}^{\wedge}_{n}(G) if and only if M𝒮n(L)M\in\mathcal{S}^{\wedge}_{n}(L). Thus, we have #𝒮n(G)=#𝒮n(L)\#\mathcal{S}^{\wedge}_{n}(G)=\#\mathcal{S}^{\wedge}_{n}(L), which completes the proof. ∎

Remark 7.7.

It is mentioned without proof in [2, Section 1.1] and [13, Section 1.2.2] that the equality ζG(s)=ζL(s)\zeta_{G}^{\wedge}(s)=\zeta_{L}^{\wedge}(s) holds.

In order to prove Proposition 7.5, we introduce some notation. Let πG\pi_{G} (resp. πL\pi_{L}) be the canonical projection GG/ZG\to G/Z (resp. LG/ZL\to G/Z). Denote by φ\varphi the map G/Z×G/ZZG/Z\times G/Z\to Z given by (xZ,yZ)[x,y]=x1y1xy(xZ,yZ)\mapsto[x,y]=x^{-1}y^{-1}xy. Since GG is of nilpotent class 22, its derived group [G,G][G,G] is contained in ZZ. Hence, G/ZG/Z is abelian group, and we have [xy,z]=[x,z][y,z],[x,yz]=[x,y][x,z][xy,z]=[x,z][y,z],[x,yz]=[x,y][x,z] for any x,y,zGx,y,z\in G. This implies that φ\varphi is a \mathbb{Z}-bilinear form.

Further, let AA and BB be finite-index subgroups of ZZ and G/ZG/Z, respectively. Denote by 𝒮G(A,B)\mathcal{S}_{G}(A,B) the set of subgroups HH of GG such that HZ=AH\cap Z=A and πG(H)=B\pi_{G}(H)=B. Similarly, denote by 𝒮L(A,B)\mathcal{S}_{L}(A,B) the set of subalgebras MM of LL such that MZ=AM\cap Z=A and πL(M)=B\pi_{L}(M)=B. For any H𝒮G(A,B)H\in\mathcal{S}_{G}(A,B) and M𝒮L(A,B)M\in\mathcal{S}_{L}(A,B), we have [G:H]=[L:M]=[Z:A][G/Z:B][G:H]=[L:M]=[Z:A][G/Z:B]. Hence, to prove Proposition 7.5, it is sufficient to show the following lemma:

Lemma 7.8.

There exists a bijection fA,B:𝒮G(A,B)𝒮L(A,B)f_{A,B}:\mathcal{S}_{G}(A,B)\to\mathcal{S}_{L}(A,B) such that, for each H𝒮G(A,B)H\in\mathcal{S}_{G}(A,B), its image is isomorphic to L(H)L(H) as Lie algebras.

To prove this lemma, we need to study 𝒮G(A,B)\mathcal{S}_{G}(A,B) and 𝒮L(A,B)\mathcal{S}_{L}(A,B). First, the following three lemmas are shown. Although they hold in general for arbitrary torsion-free nilpotent groups (cf. Remark 7.12), we provide a direct proof here for self-containedness:

Lemma 7.9.

Let x,yGx,y\in G, and let nn be a positive integer. If [x,yn]=1[x,y^{n}]=1, then [x,y]=1[x,y]=1.

Proof.

Since φ\varphi is bilinear, we have 1=[x,yn]=[x,y]n1=[x,y^{n}]=[x,y]^{n}. Since GG is torsion-free, we have [x,y]=1[x,y]=1. ∎

Lemma 7.10.

G/ZG/Z is torsion-free.

Proof.

Let x,yGx,y\in G, and suppose that ynZy^{n}\in Z for some n>0n>0. By Lemma 7.9, we have [x,y]=1[x,y]=1, which derives that yZy\in Z. Thus, G/ZG/Z is torsion-free. ∎

Lemma 7.11.

For each finite-index subgroup HH of GG, the center ZHZ_{H} of HH equals ZHZ\cap H.

Proof.

It is immediately verified that ZHZHZ\cap H\subset Z_{H}. Let xx be an element of ZHZ_{H}, and let yy be an element of GG. Since HH is of finite index, there exists a positive integer nn such that ynHy^{n}\in H. Hence, we have [x,yn]=1[x,y^{n}]=1, which implies [x,y]=1[x,y]=1 by Lemma 7.9. Thus, we have ZHZHZ_{H}\subset Z\cap H. ∎

Remark 7.12.

Lemmas 7.9, 7.10, and 7.11 hold for an arbitrary torsion-free nilpotent group GG^{\prime}. Let x,yGx,y\in G^{\prime}, and let nn be a positive integer. Suppose that [x,yn]=1[x,y^{n}]=1. Then, we have (x1yx)n=yn(x^{-1}yx)^{n}=y^{n}. According to a result of Chernikov (cf. [19, Theorem 4.10]), the map wGwnGw\in G^{\prime}\mapsto w^{n}\in G^{\prime} is injective, which implies that [x,y]=1[x,y]=1. Thus, Lemma 7.9 holds for GG^{\prime}. By using this, Lemmas 7.10 and 7.11 for GG^{\prime} are verified.

By Lemma 7.10, G/ZG/Z is a free abelian group of finite rank. Let dd denote this rank. We next relate (Z/A)d(Z/A)^{d} to 𝒮G(A,B)\mathcal{S}_{G}(A,B) and 𝒮L(A,B)\mathcal{S}_{L}(A,B). Fix a subset 𝔟={b~i}i=1d\mathfrak{b}=\{\tilde{b}_{i}\}_{i=1}^{d} of GG such that its image under πG\pi_{G} forms a basis of BB. Set bi=πG(b~i)b_{i}=\pi_{G}(\tilde{b}_{i}) for each ii. For an element Ξ=(ξi)i\Xi=(\xi_{i})_{i} of (Z/A)d(Z/A)^{d}, take an element (zi)i(z_{i})_{i} of ZdZ^{d} satisfying ξi=ziA\xi_{i}=z_{i}A for each ii. Further, define H𝔟(Ξ)H^{\mathfrak{b}}(\Xi) (resp. M𝔟(Ξ)M^{\mathfrak{b}}(\Xi)) to be the subgroup (resp. subalgebra) of GG (resp. LL) generated by AA and {b~izi}i\{\tilde{b}_{i}z_{i}\}_{i} (resp. {(bi,zi)}i\{(b_{i},z_{i})\}_{i}). Since AZA\subset Z, the group H𝔟(Ξ)H^{\mathfrak{b}}(\Xi) and the Lie algebra M𝔟(Ξ)M^{\mathfrak{b}}(\Xi) are independent of the choice of (zi)i(z_{i})_{i}.

If φ(B,B)A\varphi(B,B)\subset A, then, [b~i,b~j][\tilde{b}_{i},\tilde{b}_{j}] and [(bi,zi),(bj,zj)][(b_{i},z_{i}),(b_{j},z_{j})] are contained in AA for any ii, jj. Since BB is a free abelian group, each element of H𝔟(Ξ)H^{\mathfrak{b}}(\Xi) (resp. M𝔟(Ξ)M^{\mathfrak{b}}(\Xi)) can be written uniquely as a product (resp. sum) ai(b~izi)nia\cdot\prod_{i}(\tilde{b}_{i}z_{i})^{n_{i}} (resp. (1,a)+ini(bi,zi)(1,a)+\sum_{i}n_{i}(b_{i},z_{i})), where (ni)id(n_{i})_{i}\in\mathbb{Z}^{d} and aAa\in A. Hence, H𝔟(Ξ)𝒮G(A,B)H^{\mathfrak{b}}(\Xi)\in\mathcal{S}_{G}(A,B), and M𝔟(Ξ)𝒮L(A,B)M^{\mathfrak{b}}(\Xi)\in\mathcal{S}_{L}(A,B). Moreover, BAB\oplus A is a subalgebra of LL, and a unique isomorphism BAM𝔟(Ξ)B\oplus A\to M^{\mathfrak{b}}(\Xi) of Lie algebras is determined by the inclusion AM𝔟(Ξ)A\subset M^{\mathfrak{b}}(\Xi) and the additive map BM𝔟(Ξ)B\to M^{\mathfrak{b}}(\Xi) defined by bi(bi,zi)b_{i}\mapsto(b_{i},z_{i}) for each ii.

Lemma 7.13.

The following assertions hold:

  1. 1.

    The following three conditions are equivalent:

    • 𝒮G(A,B)\mathcal{S}_{G}(A,B)\not=\emptyset,

    • 𝒮L(A,B)\mathcal{S}_{L}(A,B)\not=\emptyset,

    • φ(B,B)A\varphi(B,B)\subset A.

  2. 2.

    Suppose that φ(B,B)A\varphi(B,B)\subset A, and let Ξ\Xi be an element of (Z/A)d(Z/A)^{d}. Then, H𝔟(Ξ)𝒮G(A,B)H^{\mathfrak{b}}(\Xi)\in\mathcal{S}_{G}(A,B), and M𝔟(Ξ)𝒮L(A,B)M^{\mathfrak{b}}(\Xi)\in\mathcal{S}_{L}(A,B). Moreover, L(H𝔟(Ξ))L(H^{\mathfrak{b}}(\Xi)) and M𝔟(Ξ)M^{\mathfrak{b}}(\Xi) are isomorphic to the subalgebra BAB\oplus A of LL as Lie algebras.

  3. 3.

    If φ(B,B)A\varphi(B,B)\subset A, then the maps λG𝔟:(Z/A)d𝒮G(A,B)\lambda_{G}^{\mathfrak{b}}:(Z/A)^{d}\to\mathcal{S}_{G}(A,B) given by ΞH𝔟(Ξ)\Xi\mapsto H^{\mathfrak{b}}(\Xi), and λL𝔟:(Z/A)d𝒮L(A,B)\lambda_{L}^{\mathfrak{b}}:(Z/A)^{d}\to\mathcal{S}_{L}(A,B) given by ΞM𝔟(Ξ)\Xi\mapsto M^{\mathfrak{b}}(\Xi) are bijective. In particular, we have #𝒮G(A,B)=#𝒮L(A,B)=[Z:A]d\#\mathcal{S}_{G}(A,B)=\#\mathcal{S}_{L}(A,B)=[Z:A]^{d}.

Proof.

1. If there exists H𝒮G(A,B)H\in\mathcal{S}_{G}(A,B), then H/AH/A is isomorphic to the abelian group BB by the canonical projection G/AG/ZG/A\to G/Z. Hence, φ(B,B)=[H,H]A\varphi(B,B)=[H,H]\subset A. Similarly, we have φ(B,B)A\varphi(B,B)\subset A if 𝒮L(A,B)\mathcal{S}_{L}(A,B)\not=\emptyset. Conversely, if φ(B,B)A\varphi(B,B)\subset A, then we have H𝔟(𝟏)𝒮G(A,B)H^{\mathfrak{b}}(\mathbf{1})\in\mathcal{S}_{G}(A,B), and M𝔟(𝟏)𝒮L(A,B)M^{\mathfrak{b}}(\mathbf{1})\in\mathcal{S}_{L}(A,B), where 𝟏\mathbf{1} is the identity element of (Z/A)d(Z/A)^{d}.

2. It is sufficient to prove that L(H𝔟(Ξ))=BAL(H^{\mathfrak{b}}(\Xi))=B\oplus A, which is an easy consequence of Lemma 7.11.

3. Let H𝒮G(A,B)H\in\mathcal{S}_{G}(A,B). Then, H/AH/A is isomorphic to BB by the canonical projection G/AG/ZG/A\to G/Z. Hence, there exists a unique element Ξ=(ξi)i\Xi=(\xi_{i})_{i} of (Z/A)d(Z/A)^{d} such that {(b~iA)ξi}i\{(\tilde{b}_{i}A)\cdot\xi_{i}\}_{i} forms a basis of H/AH/A, and we have H=H𝔟(Ξ)H=H^{\mathfrak{b}}(\Xi). Therefore, the map λG𝔟\lambda_{G}^{\mathfrak{b}} is bijective. In a similar way, for each M𝒮L(A,B)M\in\mathcal{S}_{L}(A,B), there exists a unique element Ξ=(ξi)i\Xi=(\xi_{i})_{i} of (Z/A)d(Z/A)^{d} such that the subset{(bi,ξi)}i\{(b_{i},\xi_{i})\}_{i} of (G/Z)(Z/A)=L/A(G/Z)\oplus(Z/A)=L/A forms a basis of M/AM/A, and we have M=M𝔟(Ξ)M=M^{\mathfrak{b}}(\Xi). Thus, λL𝔟\lambda_{L}^{\mathfrak{b}} is also bijective. ∎

Now, a proof of Lemma 7.8 is obtained:

Proof of Lemma 7.8.

It follows from Lemma 7.13 that 𝒮G(A,B)=𝒮L(A,B)=\mathcal{S}_{G}(A,B)=\mathcal{S}_{L}(A,B)=\emptyset if φ(B,B)A\varphi(B,B)\not\subset A. If φ(B,B)A\varphi(B,B)\subset A, then fA,B=λL𝔟(λG𝔟)1f_{A,B}=\lambda_{L}^{\mathfrak{b}}\circ(\lambda_{G}^{\mathfrak{b}})^{-1} has the desired property by Assertions 2 and 3 of Lemma 7.13. ∎

In the rest of this subsection, we give an example of Lemmas 7.8 and 7.13. Subsequently, using this, we remark on Proposition 3.1 of [3].

Example 7.14.

Consider the case where GG is the Heisenberg group \mathfrak{H} with a free generating set {xG,yG}\{x_{\scalebox{0.5}{$G$}},y_{\scalebox{0.5}{$G$}}\}. Set 𝔟={xGyG,yG2}\mathfrak{b}=\{x_{\scalebox{0.5}{$G$}}y_{\scalebox{0.5}{$G$}},y_{\scalebox{0.5}{$G$}}^{2}\}, B=πG(xGyG)B=\langle\pi_{G}(x_{\scalebox{0.5}{$G$}}y_{\scalebox{0.5}{$G$}}), πG(yG2)\pi_{G}(y_{\scalebox{0.5}{$G$}}^{2})\rangle, and A=zG2A=\langle z_{\scalebox{0.5}{$G$}}^{2}\rangle, where zG=[xG,yG]=xG1yG1xGyGz_{\scalebox{0.5}{$G$}}=[x_{\scalebox{0.5}{$G$}},y_{\scalebox{0.5}{$G$}}]=x_{\scalebox{0.5}{$G$}}^{-1}y_{\scalebox{0.5}{$G$}}^{-1}x_{\scalebox{0.5}{$G$}}y_{\scalebox{0.5}{$G$}}. Then, we have [G/Z:B]=[Z:A]=2[G/Z:B]=[Z:A]=2, and hence, 𝒮G(A,B)𝒮4(G)\mathcal{S}_{G}(A,B)\subset\mathcal{S}_{4}(G), 𝒮L(A,B)𝒮4(L)\mathcal{S}_{L}(A,B)\subset\mathcal{S}_{4}(L). Since φ(B,B)=A\varphi(B,B)=A, it follows from Lemma 7.13 that #𝒮G(A,B)=#𝒮L(A,B)=4\#\mathcal{S}_{G}(A,B)=\#\mathcal{S}_{L}(A,B)=4.

Now, L=L(G)L=L(G) is the Heisenberg Lie algebra \mathcal{H}, and generated by xL=(πG(xG),1)x_{\scalebox{0.5}{$L$}}=(\pi_{G}(x_{\scalebox{0.5}{$G$}}),1) and yL=(πG(yG),1)y_{\scalebox{0.5}{$L$}}=(\pi_{G}(y_{\scalebox{0.5}{$G$}}),1). Put zL=(1,zG)z_{\scalebox{0.5}{$L$}}=(1,z_{\scalebox{0.5}{$G$}}). Then, [xL,yL]=zL[x_{\scalebox{0.5}{$L$}},y_{\scalebox{0.5}{$L$}}]=z_{\scalebox{0.5}{$L$}}. By Lemma 7.13, fA,B=λL𝔟(λG𝔟)1f_{A,B}=\lambda_{L}^{\mathfrak{b}}\circ(\lambda_{G}^{\mathfrak{b}})^{-1} is a bijection 𝒮G(A,B)𝒮L(A,B)\mathcal{S}_{G}(A,B)\to\mathcal{S}_{L}(A,B), and we have fA,B(H𝔟(𝟏))=M𝔟(𝟏)f_{A,B}(H^{\mathfrak{b}}(\mathbf{1}))=M^{\mathfrak{b}}(\mathbf{1}), where 𝟏\mathbf{1} is the identity element of (Z/A)2(Z/A)^{2}. Moreover,

M𝔟(𝟏)\displaystyle M^{\mathfrak{b}}(\mathbf{1}) =(xL+yL)+2yL+2zL\displaystyle=\mathbb{Z}(x_{\scalebox{0.5}{$L$}}+y_{\scalebox{0.5}{$L$}})+2\mathbb{Z}y_{\scalebox{0.5}{$L$}}+2\mathbb{Z}z_{\scalebox{0.5}{$L$}}
={kxL+lyL+mzL|(k,l)(1,1)+(0,2),m2},\displaystyle=\{kx_{\scalebox{0.5}{$L$}}+ly_{\scalebox{0.5}{$L$}}+mz_{\scalebox{0.5}{$L$}}\ |\ (k,l)\in\mathbb{Z}(1,1)+\mathbb{Z}(0,2),\ m\in 2\mathbb{Z}\},
H𝔟(𝟏)\displaystyle H^{\mathfrak{b}}(\mathbf{1}) ={(xGyG)kyG2lzG2m|k,l,m}.\displaystyle=\{(x_{\scalebox{0.5}{$G$}}y_{\scalebox{0.5}{$G$}})^{k}y_{\scalebox{0.5}{$G$}}^{2l}z_{\scalebox{0.5}{$G$}}^{2m}\ |\ k,l,m\in\mathbb{Z}\}.
Remark 7.15.

We keep the notation of the above example. Proposition 3.1 of [3] claims that, for each nn, a one-to-one correspondence between 𝒮n(G)\mathcal{S}_{n}(G) and 𝒮n(L)\mathcal{S}_{n}(L) is induced by the bijection f:GLf^{\prime}:G\to L defined by

f(xGkyGlzGm)=kxL+lyL+mzLfor k,l,m.f^{\prime}(x_{\scalebox{0.5}{$G$}}^{k}y_{\scalebox{0.5}{$G$}}^{l}z_{\scalebox{0.5}{$G$}}^{m})=kx_{\scalebox{0.5}{$L$}}+ly_{\scalebox{0.5}{$L$}}+mz_{\scalebox{0.5}{$L$}}\quad\text{for $k,l,m\in\mathbb{Z}$}.

However, it is not true. Indeed, M𝔟(𝟏)M^{\mathfrak{b}}(\mathbf{1}) corresponds to H=f1(M𝔟(𝟏))={xGkyGlzGm|(k,l)(1,1)+(0,2),m2}H^{\prime}=f^{\prime-1}(M^{\mathfrak{b}}(\mathbf{1}))=\{x_{\scalebox{0.5}{$G$}}^{k}y_{\scalebox{0.5}{$G$}}^{l}z_{\scalebox{0.5}{$G$}}^{m}\ |\ (k,l)\in\mathbb{Z}(1,1)+\mathbb{Z}(0,2),\ m\in 2\mathbb{Z}\}, however, HH^{\prime} is not a group because xGyGHx_{\scalebox{0.5}{$G$}}y_{\scalebox{0.5}{$G$}}\in H^{\prime}, and (xGyG)2=xG2yG2zG1H(x_{\scalebox{0.5}{$G$}}y_{\scalebox{0.5}{$G$}})^{2}=x_{\scalebox{0.5}{$G$}}^{2}y_{\scalebox{0.5}{$G$}}^{2}z_{\scalebox{0.5}{$G$}}^{-1}\not\in H^{\prime}.

Acknowledgements

The author extends his sincere gratitude to the reviewers for their meticulous evaluation and numerous valuable suggestions. In particular, he is grateful to them for providing references [2, 3, 5], which are highly relevant to this study, and for their insightful suggestions for future research. In addition, he wishes to thank them for providing a direct proof of Lemma 7.9 that does not rely on Chernikov’s result. He also would like to thank the editors for their helpful suggestions and advice.

References

  • [1] A. N. Andrianov: Rationality theorems for Hecke series and zeta functions of the groups GLnGL_{n} and SPnSP_{n} over local fields, Izv. Math. 3(3) (1969), 439–476.
  • [2] M. N. Berman, I. Glazer, and M. M. Schein: Pro-isomorphic zeta functions of nilpotent groups and Lie rings under base extension, Trans. Am. Math. Soc. 375 (2022), 1051–1100.
  • [3] M. N. Berman, B. Klopsch, U. Onn: On pro-isomorphic zeta functions of DD^{*}-groups of even Hirsch length, arXiv:1511.06360v5.
  • [4] J. Dulinski: Jacobi-Hecke algebras and a rationality theorem for a formal Hecke series, Manuscripta Math. 99(2) (1999), 255–285.
  • [5] F. J. Grunewald, D. Segal, and G. C. Smith: Subgroups of finite index in nilpotent groups, Invent. Math. 93 (1988), 185–223.
  • [6] E. Hecke: Über Modulfunktionen und die Dirichletschen Reihen mit Eulerscher Produktentwicklung, I, II, Math. Ann. 114(1) (1937), 1–28, 316–351.
  • [7] T. Hina and T. Sugano: On the local Hecke series of some classical groups over pp-adic fields, J. Math. Soc. Japan 35(1) (1983), 133–152.
  • [8] F. Hyodo: A formal power series of a Hecke ring associated with the Heisenberg Lie algebra over p{\mathbb{Z}}_{p}, Int. J. Number Theory 11(8) (2015), 2305–2323.
  • [9] F. Hyodo: A note on a Hecke ring associated with the Heisenberg Lie algebra, Math. J. Okayama Univ. 64 (2022), 215–225.
  • [10] F. Hyodo: A formal power series over a noncommutative Hecke ring and the rationality of the Hecke series for GSp4GSp_{4}, Preprint, 2022, arXiv:2208.09622v1.
  • [11] N. Iwahori, H. Matsumoto: On some Bruhat decomposition and the structure of the Hecke rings of pp-adic Chevalley groups, Inst. Hautes Études Sci. Publ. Math. 25(1) (1965), 5–48.
  • [12] I. Satake: Theory of spherical functions on reductive algebraic groups over pp-adic fields, Inst. Hautes Études Sci. Publ. Math 18 (1963), 5–69.
  • [13] M. P. F. du Sautoy and L. Woodward: Zeta functions of groups and rings, Lecture Notes in Mathematics, vol. 1925, Springer-Verlag, Berlin, 2008.
  • [14] D. Segal: Polycyclic groups, Cambridge University Press, Cambridge, 1983.
  • [15] G. Shimura: Sur les intégrales attachées aux formes automorphes, J. Math. Soc. Japan 11(4) (1959), 291–311.
  • [16] G. Shimura: On modular correspondences for Sp(n,)Sp(n,\mathbb{Z}) and their congruence relations, Proc. Nat. Acad. Sci. U.S.A. 49(6) (1963), 824–828.
  • [17] G. Shimura: Introduction to the arithmetic theory of automorphic functions, Princeton University Press, Princeton, 1971.
  • [18] T. Tamagawa: On the ζ\zeta-functions of a division algebra, Ann. of Math. 77(2) (1963), 387–405.
  • [19] R.B. Warfield: Nilpotent Groups, Lecture Notes in Mathematics, vol. 513, Springer-Verlag, Berlin, 1976.
  • [20] E. Witt: Treue Darstellung Liescher Ringe, J. Reine Angew. Math. 1937(177) (1937), 152–160.

Department of Health Informatics, Faculty of Health and Welfare Services Administration, Kawasaki University of Medical Welfare, Kurashiki, 701-0193, Japan

Email address: fumitake.hyodo@mw.kawasaki-m.ac.jp