Canonical models of modular curves and the Galois action on CM-points

We use [Milne0] again. Note first that canonical models for \mathrm{Sh}_{K(N)}(\mathrm{GL}_2, \mathbb{H}) over the field E(\mathrm{GL}_2, \mathbb{H})={\mathbb Q} exist according to Deligne [Deligne], see also [Milne1]. In particular, \mathrm{S}_{K(N)} are defined over {\mathbb Q}. (iii) follows sinse \mathrm{S} is the limit of the \mathrm{S}_{K(N)} and E_K just defines the splitting into irreducible components of a curve over {\mathbb Q}.

For g\in \mathrm{GL}_2(\hat{{\mathbb Z}}) we have g\cdot K(N)\cdot g^{-1}=K(N) and hence by [Milne0], Theorem 13.6, the maps

are defined over {\mathbb Q}. Since the action of g on S is the limit of its actions on the S_{K(N)} (i) follows.

(ii) follows from [Milne0], Theorem 13.6 when one considers g=1.

\Box

By definition \tau= a\sqrt{-n}+b for some a,b\in {\mathbb Q} and a positive square-free integer n. Then \tau= r \sqrt{-n} for r=\left( \begin{array}{ll} a\ b\\ 0\ 1 \end{array}\right), that is \tau \in \mathrm{S}_{\sqrt{-n}}.

In order to prove uniqueness consider the assumption that [\sqrt{-m},g_1]=[\sqrt{-n},g_2] for a square-free positive integer m and g_1,g_2\in \mathrm{GL}_2(\mathbb{A}_f). This is equivalent to [\sqrt{-m},1]=[\sqrt{-n},g], for g=g_2g_1^{-1}. By (?) the latter is equivalent to

that is \sqrt{-m}=q\sqrt{-n}. This is possible only if m=n.

The uniqueness of representation follows. \Box

Suppose

and hence, for any a\in \mathrm{GL}_2(\mathbb{A}_f),

We claim that r_\tau(\sigma)\in N_\tau .

Indeed, let \tau be the fixed point of q\in \mathrm{GL}_2^+({\mathbb Q}), q\notin Z({\mathbb Q}). Then by ? [\tau,1] is a fixed point of q and q^{-1} and so

Hence [\tau,r_\tau(\sigma)] is a fixed point of q and thus by ?(ii) r_\tau(\sigma)\in N_\tau .

Conversely, if r\in N_\tau then for each q\in G_\tau\setminus Z({\mathbb Q}) there is q'\in G_\tau,

Define the map \sigma_{r,\tau}: \mathrm{S}_\tau\to \mathrm{S}_\tau, a partial transformation of \mathrm{S}, by setting for each a\in \mathrm{GL}_2(\mathbb{A}_f)

Then \sigma_{r,\tau} trivially satisfies (?). Also (?) is satisfied because rq=q'r. This proves that \sigma_{r,\tau} acts as an automorphism on \mathrm{S}_\tau and r=r_\tau(\sigma_{r,\tau}).

Now note that by (?) [\tau, r]=[\tau,1], for r\in \mathrm{GL}_2(\mathbb{A}_f), holds if and only if r\in G_\tau. This proves (?). \Box

Compare the conclusion of ? with [Milne0], 12.8 describing the action on the special points \mathrm{S}_\tau of the canonical model by \sigma\in \mathrm{Gal}({\mathbb Q}(\tau)^\mathrm{ab}:{\mathbb Q}(\tau)).

In the notation of [Milne0], (60)-(62):

\tau=x, {\mathbb Q}(\tau)=E(x) and T(\mathbb{A}_f)=T_\tau as in our ?. Further

Consequently, r_\tau(\sigma) is the same as r_x(s) of [Milne0], 12.8, for s\in \mathbb{A}^\times_{{\mathbb Q}(\tau)} such that \sigma=\mathrm{art}(s), the Artin symbol. Thus the matrices r_x(s)\in T_\tau of (Proof) deliver all the Galois automorphism of \mathrm{S}_\tau(\mathrm{CM}) which fix {\mathbb Q}(\tau).

Now let \mathsf{i}: {\mathbb C} \to {\mathbb C} be complex conjugation. \mathsf{i} induces a non-trivial automorphism of {\mathbb Q}(\tau) and an automorphism of \mathrm{S}_\tau(\mathrm{CM}).

Claim. \mathsf{i}

corresponds to the action by an element

The transformation \eta: \mathbb{H}\cup \mathbb{H}^-\to \mathbb{H}\cup \mathbb{H}^- induced on \mathbb{H}\cup \mathbb{H}^-\subset \mathrm{S} by \mathbf{i} is studied (in much more general setting) in [MilneCC]. Lemma 3.2 of that paper states in particular that (in our notation n:=\mathbf{d}_{-1}), for g\in \mathrm{GL}_2({\mathbb Q}),

In particular,

It implies, as proved further in [MilneCC], Conjecture B (a form of Langlands’ conjecture on complex conjugation):

for all g\in \mathrm{GL}_2(\mathbb{A}_f).

Now let \rho\in \mathrm{Gal}({\mathbb Q}(\tau)^\mathrm{ab}:{\mathbb Q}).

Then, either it fixes {\mathbb Q}(\tau) and has the form (Proof) or \mathsf{i}\circ\rho fixes {\mathbb Q}(\tau) and so \rho can be represented by the product of a matrix from T_{\tau} with the above matrix from N_{\tau}, which gives us again a matrix of the form \left( \begin{array}{ll} -c\ nd\\ -d\ \ c \end{array}\right) from N_{\tau}. This concludes the proof of (i) and (ii).

Since

the set of special points in C_g (the graph of g on \mathrm{S}) is invariant under \sigma. Hence the set of special points in the algebraic curve C_{g,N}\subset \mathbb{Y}(N)\times \mathbb{Y}(N) is invariant under \sigma. But the set of special points is Zariski dense in the special curve C_{g,N}, and hence C_{g,N} is invariant under \sigma. Finally, this implies C_g, the limit of the C_{g,N}, is invariant under \sigma.

We have established in (iii) that \mathbf{S}^\mathrm{Full} satisfies (?), that is s\mapsto g*s is definable in the structure. We conclude by ? that \mathbf{S}^\mathrm{Full} is an expansion of \mathbf{S}^\mathrm{Pure} by possibly extra relations, which implies the embedding of the automorphism groups. \Box

(i) By the preservation of connected components property (Remark), given \sigma there is a \mu such that for each \tau_i:= \sqrt{-m_i},

This proves the first statement of (i). For the second statement

use the fact that \mathrm{Gal}({\mathbb Q}^\mathrm{ab}:{\mathbb Q}) acts faithfully and transitively on the set of components \mathrm{S}^\mu of \mathrm{S} (see [Milne1], 3.4) so for each \mu there is a \sigma and \bar{r}_{\bar{\tau}}(\sigma) such that \bigwedge_i \det r_{\tau_i}(\sigma)= q_i \mu.

(ii) In matrix terms we have

where \epsilon_i is +1 if r_i\in T_{\sqrt{-m_i}} and -1, if r_i\in N_{\sqrt{-m_i}}\setminus T_{\sqrt{-m_i}}.

As a result we have the equalities

Assume that \epsilon_i=\epsilon_j. Then the equality between the algebraic expressions takes place for some a_i,b_i,a_j,b_j\in {\mathbb R}, not both a_i,b_i zero, as well. By the Hasse principle, there are s_i,t_i\in {\mathbb Q},\ \ i=1,\ldots, n such that not both s_i,t_i zero and

Consider, for i=1,\ldots,n, the matrices with rational entries,

which by definition belongs to G_{\sqrt{-m_i}}.

By dividing the algebraic expressions of (?) on respective expressions of (?) we get

and recalling the formula for \det r_i we have

Choose representatives for \bar{r}_{\bar{\tau}}(\sigma) modulo G_{\bar{\tau}} as

Suppose \epsilon_{i_1}=\ldots=\epsilon_{i_k}=1 and \epsilon_{i_{k+1}}=\ldots=\epsilon_{i_n}=-1.

We get, for some q\in {\mathbb Q}_+, from above that

(iii) The condition \bar{r}=\langle r_1,\ldots,r_n\rangle\in \prod_i N_{\tau_i} implies that \bar{r} determines a transformation of \mathrm{S}_{\bar{\tau}} which satisfies (?) and (?). The conditions (?) implies that \bar{r} satisfies (Remark). Thus the transformation belongs to \mathrm{Aut}\, \mathbf{S}^\mathrm{Pure}_{\bar{\tau}}. \Box

By ?(ii) {\mathbb Q}(\mathrm{CM}_{\tau_i})={\mathbb Q}(\tau_i)^\mathrm{ab}. Hence {\mathbb Q}(\bigcup_{i=1}^n\mathrm{CM}_{\tau_i})=\mathrm{F}_{\bar{\tau}}.

Elements \sigma of \mathrm{Aut}\, \mathbf{S}^\mathrm{Full}_{\bar{\tau}} by construction act on \bigcup_i\mathrm{CM}_{\tau_i} as an automorphism group of the field, hence

By ? and definitions \bar{N}^{\mathrm{Gal}}_{\bar{\tau}} is isomorphic to the subgroup of the group of transformations \mathrm{Aut}\, \mathbf{S}^\mathrm{Pure}_{\bar{\tau}} induced by Galois automorphisms of {\mathbb Q}(\mathrm{CM}_{\bar{\tau}}) over {\mathbb Q}, which is exactly \mathrm{Aut}\, \mathbf{S}^\mathrm{Full}_{\bar{\tau}}. This proves the second isomorphism. \Box

First we claim that as groups of transformations, for each i,

via the canonical map

In order to prove it note that T^0_{\tau_i}\cong \mathbb{A}^0_{ {\mathbb Q}(\tau_i)} where \mathbb{A}^0_{ {\mathbb Q}(\tau_i)} is the group of finite {\mathbb Q}(\tau_i)-adeles of {\mathbb Q}-norm 1. Now the isomorphism

a corollary to Artin’s reciprocity law, completes the proof of our claim.

Now the definition of \bar{T}^{0,\mathrm{Gal}}_{\bar{\tau}} together with the second statement of ? and the claim above gives us

It is easy to check using the definition of \mathrm{F}_{\bar{\tau}} that the right-hand side of the latter is exactly \mathrm{Gal}(\mathrm{F}_{\bar{\tau}}:{\mathbb Q}^\mathrm{ab}).

Thus

For the proof of the second statement of the Lemma we use the proof of the adelic Mumford-Tate conjecture for the product of CM elliptic curves, [Cadoret--Moonen] and [Ullmo--Yafaev].

Let {\bf T}^*_{\bar{\tau}} denote the \mathbb{Q}-torus such that {\bf T}^*_{\bar{\tau}}(\mathbb{A}_f)=T^*_{\bar{\tau}}. We can form a CM-Shimura datum ({\bf T}*_{\bar{\tau}},\{h\}) associated with a product E_1\times\cdots\times E_n of elliptic curves, with E_i having CM by \mathbb{Q}(\tau_i). As in [Ullmo--Yafaev], Remark 2.8, for any neat compact open subgroup K of T^*_{\bar{\tau}} and L a finite extension of \mathbb{Q}(\bar{\tau}) such that [h,1]\in{\rm Sh}_{K}({\bf T}^*_{\bar{\tau}},\{h\})(L), we obtain a homomorphism

which describes the action of {\rm Gal}(\bar{L}/L) on the fiber of

over [h,1] (which is a K-torsor). In particular, \rho factors through the quotient {\rm Gal}(\mathrm{F}_{\bar{\tau}}:L).

As explained on the following page of [Ullmo--Yafaev], \rho yields the {\rm Gal}(\bar{L}:L) representation on the Tate-module of E_1\times\cdots\times E_n. Take L to satisfy {\mathbb Q}(\bar{\tau})\subseteq L\subset\mathrm{F}_{\bar{\tau}}. By [Cadoret--Moonen], Theorem A (ii), provided h_\mathbb{R} is “maximal” (see Definition 2.1), the latter has open image in T^*_{\bar{\tau}} and, therefore,

(the equality due to the fact that T^0_{\bar{\tau}} is the kernel of the determinant map T^*_{\bar{\tau}}\to\mathbb{A}^\times_f) is open in T^0_{\bar{\tau}}.

To see that h_\mathbb{R} is maximal, let \sigma_i and \bar{\sigma}_i denote distinct embeddings of \mathbb{Q}(\tau_i). The cocharacter group of {\bf T}^*_{\bar{\tau}} is equal to the set of elements

such that a_i+b_i is independent of i. It has a basis consisting of the \sigma_i-\bar{\sigma}_i, for i=1,\ldots,n, and \mu, where \mu=\sum_i\sigma_i is the cocharacter associated with h. The {\rm Gal}(\bar{\mathbb{Q}}:\mathbb{Q})-orbit of \mu consists of the sums of n elements in which, for each i, the sum includes either \sigma_i or \bar{\sigma}_i. Since

the result follows from the the paragraph following Definition 2.3 of [Cadoret--Moonen]. \Box

\det/{\mathbb Q}_+(r_1,\ldots,r_n) takes its values in \mathbb{A}_f^\times/{\mathbb Q}_+^\times\cong \hat{{\mathbb Z}}^\times. It is well-defined by the definitions of \bar{N}^{\mathrm{Gal}}_{\bar{\tau}} and \bar{N}^*_{\bar{\tau}} , and is surjective on both lines by ?(i). The kernel of \det/{\mathbb Q}_+ consists of tuples \bar{r} of matrices r_i=\left( \begin{array}{ll} a_i\ \ b_im_i\\ -b_i\ a_i \end{array}\right) such that \det r_i=a_i^2+m_ib_i^2=q_i\in {\mathbb Q}^+, where a_i,b_i\in \mathbb{A}_f. By the Hasse principle there exists s_i,t_i\in {\mathbb Q} such that s_i^2+m_it_i^2=q_i and thus there exists g_i\in G_{\tau_i} such that \det r_ig_i^{-1}=1. Thus \bar{r}\in \bar{T}^0_{\bar{\tau}}.

It follows that the kernel is \bar{T}^{0,\mathrm{Gal}}_{\bar{\tau}} in the top line and \bar{T}^0_{\bar{\tau}} in the bottom line. \Box

The first statement is immediate from ? - ?. The second statement follows from the first together with ? and ?. \Box

By ? and ? \bar{N}^\mathrm{Gal}_{\bar{\tau}} acts on the compositum \mathrm{F}_{\bar{\tau}} of the fields {\mathbb Q}(\tau_i)^\mathrm{ab} as its Galois group over {\mathbb Q}. At the same time coordinate-wise \bar{r}=\langle r_1,\ldots,r_n\rangle\in N^*_{\bar{\tau}} acts on the i-th coordinate by [\tau_i,a]\mapsto [\tau_i, r_i a], which by ? corresponds to the Galois action on {\mathbb Q}(\tau_i)^\mathrm{ab}. Moreover, r_i(\sigma) \in T_{\tau_i} if and only if \sigma fixes \tau_i in {\mathbb Q}(\tau_i)^\mathrm{ab}. If r_i(\sigma) \notin T_{\tau_i} then r_i(\sigma) acts on T_{\tau_i} by conjugation as \mathbf{d}_{-1}.

Let \delta\in \{ 1,-1\}^n be an arbitrary sequence with values \pm 1. The condition of independence of the \tau_i guaranties the existence of \sigma_\delta\in \mathrm{Gal}(\mathrm{F}_{\bar{\tau}}:{\mathbb Q}) such that

Thus

It implies that

\Box

Since

the isomorphisms (i) follow from the isomorphism statements in ? and ?. For the same reason (ii) follows from ?. (iii) follows from ? and ?.

(iv) follows from the description of components in ? in terms of determinants along with the statement ?(i).

\Box

By definition

We need to show that for \sigma\in \mathrm{Aut}\, \mathbf{S}^\mathrm{Pure} the condition

for all \tau\in \mathbb{H} quadratic and all g\in \mathrm{GL}_2 , implies that \sigma is the identity (recall the notation in ?).

We assume first that in (?) \sigma acts trivially on \mathrm{S}_\approx. g=1 and that \tau=\sqrt{-n}.

By definition

It follows r\in \Delta and hence b=0 and a=1, that is r=\mathbf{d}_{-1} or the identity. We assume, towards a contradiction, that \sigma acts non-trivially on S_{\tau} and thus r=\mathbf{d}_{-1}.

Now for the more general case we have with the same r

for any q\in G_{\tau} and thus (?) implies

But \det q\in {\mathbb Q}_+ and \det g^{-1}\cdot q\mathbf{d}_{-1}\cdot g=\det q\mathbf{d}_{-1} is a negative rational number. In \Delta this is only satisfied for g^{-1}\cdot q\mathbf{d}_{-1}\cdot g=\mathbf{d}_{-1}.

Thus q\mathbf{d}_{-1}=\left(\begin{array}{ll} -a\ - nb\\ -b\ \ \ \ a \end{array}\right)\in \mathrm{GL}_2({\mathbb Q}). This is an involution only when b=0 and a=\pm 1. Now the extra condition g^{-1}\cdot q\mathbf{d}_{-1}\cdot g=\mathbf{d}_{-1} for all g gives the final contradiction. \Box

First note that any coset a\Delta\subset \mathrm{GL}_2(\mathbb{A}_f), a\in \mathrm{GL}_2(\mathbb{A}_f), contains exactly one element of determinant in {\mathbb Q}_+. The uniquiness of the representation (?) follows.

The action (?) is well-defined because of the uniquiness of a and the fact that \det g\in {\mathbb Q}_+.

Finally, note that the action (?) is isomorphic to the action of g on the component \mathrm{S}^1 of \mathrm{S}, which according to ?(b) represents the action of g on \mathrm{S}_\approx. \Box

(b) proves one direction. In the opposite direction use the fact that by definition

where \det r_1=\det r_2 \in \Delta. By ? there is \sigma\in \mathrm{Aut}\, \mathbf{S}^\mathrm{Pure} such that r_i=r_{\tau_i}(\sigma) modulo {\mathbb Q}_+. Using the Hasse principle as in the proof of ?(ii) we can make this an equality. This proves the claim.

By definition, see (?), we need to prove that any automorphism of \mathbf{S}^\mathrm{Pure} induces an automorphism of \tilde{\mathbb{H}}^\mathrm{Pure} and, conversely, any automorphism of \tilde{\mathbb{H}}^\mathrm{Pure} can be lifted to an automorphism of \mathbf{S}^\mathrm{Pure}. The former is proved in ?. So we need to show that any \sigma\in \mathrm{Aut}\,\tilde{\mathbb{H}}^\mathrm{Pure} can be lifted to an automorphism \rho of \mathbf{S}^\mathrm{Pure}.

Claim. Any \sigma\in \mathrm{Aut}\,\tilde{\mathbb{H}}^\mathrm{Pure}(\mathrm{CM}) can be lifted to an automorphism \rho of \mathbf{S}^\mathrm{Pure}(\mathrm{CM}).

Consider special points s_1,\ldots,s_n,\ldots\in \mathrm{S}_\approx with representatives [\tau_1, a_1],\ldots,[\tau_1, a_n],\ldots\in \mathrm{S},

Since \sigma preserves R, we get

and hence

By ? there is a \rho\in \mathrm{Aut}\, \mathbf{S}^\mathrm{Pure}(\mathrm{CM}) such that r_i=r_{\tau_i}(\rho), i=1,...,n..., and hence

Claim proved.

Note that the last argument in the proof of the claim also shows that \bigwedge_i R(s_1,s_i,t_1,t_i) implies the existence of \sigma: s_i\mapsto t_i.

Now let \sigma\in \mathrm{Aut}\, \tilde{\mathbb{H}}^\mathrm{Pure} and \sigma^\mathrm{CM} be its restriction to \tilde{\mathbb{H}}(\mathrm{CM}). By the claim there is \rho^\mathrm{CM}\in \mathrm{Aut}\, \mathbf{S}^\mathrm{Pure}(\mathrm{CM}) whose restriction to \tilde{\mathbb{H}}(\mathrm{CM}) is \sigma^\mathrm{CM}.

Since special points are Zariski dense in the components \mathrm{S}^\mu_{K(N)}, \rho^\mathrm{CM} determines the unique bijective map \mathrm{S}^\mu\to \mathrm{S}^{\rho^\mathrm{CM}(\mu)} on the set of components of \mathrm{S}. More precisely,

where \lambda_{\rho}=\det r_\tau(\rho^\mathrm{CM})\in \hat{{\mathbb Z}}^\times is obtained from the property (?) of automorphisms acting on CM-points.

Consider an arbitrary s\in \mathrm{S}_\approx. It follows from the definition of \approx that for each component \mathrm{S}^\mu there is a unique x\in \mathrm{S}^\mu such that x_\approx=s. Call this element x(s,\mathrm{S}^\mu). By construction any element of \mathrm{S} has the form x(s,\mathrm{S}^\mu) for some s\in \mathrm{S}_\approx and a component \mathrm{S}^\mu of \mathrm{S}.

Consider the element \mathbf{t}\in \mathrm{GL}_2^+({\mathbb Q}) and the set \mathbf{t}^\Delta definable in \tilde{\mathbb{H}}^\mathrm{Pure} by definition. Elements \mathbf{t}^\mu of \mathbf{t}^\Delta can be identified with curves C^\mu_{\mathbf{t}/\approx} on \tilde{\mathbb{H}}\times \tilde{\mathbb{H}} (see ?A(b)) and there is a one-to-one correspondence between elements \mathbf{t}^\mu of \mathbf{t}^\Delta and components \mathrm{S}^\mu of \mathrm{S} given by

(see ?A).

Clearly,

for \sigma and \rho^\mathrm{CM} as above. This gives us an extension of the action of \sigma\in \mathrm{Aut}\, \tilde{\mathbb{H}}^\mathrm{Pure} to components of \mathrm{S},

Define the map

This is the lift of \sigma\in \mathrm{Aut}\, \tilde{\mathbb{H}}^\mathrm{Pure} to a transformation \rho of \mathrm{S} and by definition the restriction of \rho to the CM-substructure is \rho^\mathrm{CM} above.

By construction \rho respects the splitting of \mathrm{S} into components.

In order to prove that \rho\in \mathrm{Aut}\, \mathbf{S}^\mathrm{Pure} it remains to prove that \rho respects the action by elements of \mathrm{GL}_2(\mathbb{A}_f).

Consider h\in \Delta'({\mathbb Q}_+)\cdot \mathrm{SL}_2(\mathbb{A}_f). By definition ?A(b)) via the action on \mathrm{S}^\mu h induces on \tilde{\mathbb{H}} the transformation h^\mu. This means that

and thus

which proves that any such h is invariant under \rho.

Next consider the action of an element d of \Delta(\mathbb{A}_f) on x(s,\mathrm{S}^\mu). Since any a\in \mathbb{A}_f can be represented as a=q\cdot \nu, for q\in {\mathbb Q}_+ and \nu\in \hat{{\mathbb Z}}^\times, d can be represented as d=\mathbf{d}_q\cdot \mathbf{d}_\nu. Thus, the action by d is invariant under \rho if and only if the action by \mathbf{d}_\nu is.

We have by construction

Thus

which proves that the actions by \mathbf{d}_\nu and by d on \mathrm{S} are invariant under \rho.

Since any element of \mathrm{GL}_2(\mathbb{A}_f) can be represented in the form d\cdot h where d\in \Delta(\mathbb{A}_f) and h\in \mathrm{SL}_2(\mathbb{A}_f) we proved that \rho satisfies (?) and thus \rho\in \mathrm{Aut}\, \mathbf{S}^\mathrm{Pure}. \rho is the lifting of \sigma\in \mathrm{Aut}\, \tilde{\mathbb{H}}^\mathrm{Pure} to \mathrm{Aut}\, \mathbf{S}^\mathrm{Pure}.

This completes the proof of the theorem. \Box

See ?, A(b).

periods Cadoret–Moonen A.Cadoret and B.Moonen, Integral and adelic aspects of the Mumford-Tate conjecture Deligne1979 P.Deligne, Variétés de Shimura: interprétation modulaire, et techniques de construction de modèles canoniques, Automorphic forms, representations and L-functions (Part 2), Proc. Sympos. Pure Math., XXXIII, AMS, Providence, R.I., 1979, pp. 247–289 imagH W.Hodges, A Shorter Model Theory, CUP, 1997

Milne1 J.Milne, Canonical models of Shimura curves, Notes 2003 on Milne’s web-site Milne0 J.Milne, Introduction to Shimura varieties, In Harmonic Analysis, the trace formula and Shimura varieties, Clay Math. Proc., 2005, pp.265–378

Deligne P.Deligne, Variétés de Shimura: interprétation modulaire, et techniques de construction de modèles canoniques, pp. 247 – 289. In Automorphic forms, representations and L-functions (Proc. Sympos. Pure Math., Oregon State Univ., Corvallis, Ore., 1977), Part 2, Proc. Sympos. Pure Math., XXXIII. Amer. Math. Soc., Providence, R.I. MilneCC J.Milne and K-y.Shih, The action of complex conjugation on Shimura varieties, Annals of Maths, v113 (1981), pp.569–599

DRap P. Deligne, M. Rapoport, Les Schémas de Modules de Courbes Elliptiques, In: Deligne P., Kuijk W. (eds) Modular Functions of One Variable II. Lecture Notes in Mathematics, vol 349, pp. 143–316, Springer, Berlin, Heidelberg

Langlands R.Langlands, Automorphic representations, Shimura varieties, and motives, Ein Märchen Proc. Sympos. Pure Math., Vol. 33, Part 2, Amer. Math. Soc, Providence, R. I., 1979, pp. 205-246

DH C.Daw and A.Harris, Categoricity of modular and Shimura curves, Journal of the Institute of Mathematics of Jussieu, v. 16, 5, 2017, pp.1075 –1101 Ullmo–Yafaev E.Ullmo and A.Yafaev, Generalised Tate, Mumford-Tate and Shafarevich conjectures, Zspecial B.Zilber, Model theory of special subvarieties and Schanuel-type conjectures Annals of Pure and Applied Logic, Volume 167, Issue 10, October 2016, pp. 1000-1028

CZ B.Zilber and C.Daw, Modular curves and their pseudo-analytic cover, arxiv 2021 Xavier F.Bars, A.Kontogeorgis, and X.Xarles, Bielliptic and hyperelliptic modular curves X(N) and the group Aut X(N), Acta Arithmetica 161(3), 2012 Elkies N.D. Elkies, The Klein quartic in number theory, pp. 51–102 in The Eightfold Way: The Beauty of Klein’s Quartic Curve, S.Levy, ed.; Cambridge Univ. Press, 1999 Lang S. Lang, Algebra, Addison-Wesley, 1965 SchW O. Schreier and B. L. van der Waerden, Die Automorphismen der projektiven Gruppen, Abh. Math. Sem. Univ. Hamburg,6, 303 - 322 (1928) Segal D.Segal, Defining \mathbb{A} in \mathrm{G}(\mathbb{A}), arXiv:2007.11440 W A.Weil, Basic Number Theory, Springer-Verlag, 1967