A P P -adic class formula for Anderson t t -modules Alexis Lucas Abstract . In 2012, Taelman proved a class formula for L L -series associated to Drinfeld 𝔽 q ​ [ θ ] \mathbb{F}_{q}[\theta] modules and considered it as a function field analog of the Birch and Swinnerton-Dyer conjecture. Since then, Taelman’s class formula has been generalized to the setting of Anderson t t -modules. Let P P be a monic prime of 𝔽 q ​ [ θ ] \mathbb{F}_{q}[\theta] , we define the P P -adic L L -series associated with Anderson t t -modules and prove a P P -adic class formula à la Taelman linking a P P -adic regulator, the class module and a local factor at P P . Next, we extend this result to the multi-variable setting à la Pellarin. Finally, we give some applications to Drinfeld modules defined over 𝔽 q ​ [ θ ] \mathbb{F}_{q}[\theta] itself. 2020 Mathematics Subject Classification 11G09, 11M38, 11R58 1. Introduction 1.1. Class formula à la Taelman In 2010, Taelman introduced the notion of L -series associated to Drinfeld \mathbb{F}_{q}[\theta] -modules and conjectured a class formula, see [ 20 , Conjecture 1] . He [ 21 ] proved this class formula in 2012, also considered as a function field analogue of the Birch and Swinnerton-Dyer conjecture. These results were generalised by Fang [ 13 ] and Demeslay [ 11 ] in the context of Anderson t -modules that are Drinfeld modules of higher dimension. Finally, Anglès, Tavares Ribeiro and Ngo Dac [ 4 ] proved the class formula for a general ring A in the context of admissible Anderson A -modules, including in particular all Drinfeld A -modules. The objective of the present article is to construct P -adic analogs of these L -series associated to Anderson t -modules. We call them P -adic L -series, and prove a P -adic class formula à la Taelman. We then extend these results to the setting of variables following the work of Anglès, Pellarin and Tavares Ribeiro [ 5 ] and Pellarin [ 19 ] . The key ingredient will be the notion of z -deformation introduced by Anglès, Tavares Ribeiro [ 7 ] as well as the introduction of evaluation of z not only at z=1 but at elements of \overline{\mathbb{F}}_{q} . We then study the vanishing of the P -adic L -series at z=1 . Finally, we investigate in detail the case of \mathbb{F}_{q}[\theta] -Drinfeld modules defined over \mathbb{F}_{q}[\theta] . 1.2. Main results Let us give more precise statements of our results. Let \mathbb{F}_{q} be a finite field with q elements and \theta an indeterminate over \mathbb{F}_{q} . Let us consider A=\mathbb{F}_{q}[\theta] and let K=\mathbb{F}_{q}(\theta) be the rational function field. Let L/K be a finite field extension of degree n . We denote by \mathscr{O}_{L} the integral closure of A in L . We consider the valuation v_{\infty} of K normalized such that v_{\infty}(\theta^{-1})=1 . Let K_{\infty} be the completion of K with respect to this valuation and we set L_{\infty}=L\otimes_{K}K_{\infty} . Let \tau:x\mapsto x^{q} be the Frobenius map. If M is an A -module having a finite number of elements, we denote by [M]_{A} the monic generator of the Fitting ideal of M . An Anderson t -module E of dimension d defined over \mathscr{O}_{L} is a non-constant \mathbb{F}_{q} -algebra homomorphism E:A\rightarrow M_{d}(\mathscr{O}_{L})\{\tau\} such that if a\in A and E_{a}={\sum}\limits_{i=0}^{r_{a}}E_{a,i}\tau^{i} then we require (E_{a,0}-aI_{d})^{d}=0 . We denote by \partial_{E}:A\rightarrow M_{d}(\mathscr{O}_{L}) the homomorphism of \mathbb{F}_{q} -algebras \partial_{E}(a)=E_{a,0} . If B is an \mathscr{O}_{L} -algebra, then we can define two A -module structures on B^{d} : the first is denoted by E(B) where A acts on B^{d} via E , and the second is denoted by \operatorname{Lie}_{E}(B) where A acts on B^{d} via \partial_{E} . There exists a unique series \exp_{E}\in M_{d}(L)\{\{\tau\}\} , called the exponential series, such that \exp_{E}\partial_{E}(\theta)=E_{\theta}\exp_{E} . Moreover, \exp_{E} converges on \operatorname{Lie}_{E}(L_{\infty}) . The key notion will be the notion of z -deformation introduced by Anglès and Tavares Ribeiro [ 7 ] . Let z be a new variable such that \tau(z)=z . Set \widetilde{A}=\mathbb{F}_{q}(z)A , \widetilde{K}=\mathbb{F}_{q}(z)K , \widetilde{\mathscr{O}_{L}}=\mathbb{F}_{q}(z)\mathscr{O}_{L} and \widetilde{K_{\infty}}=\mathbb{F}_{q}(z)((\theta^{-1})) . We set \widetilde{L_{\infty}}=L\otimes_{K}\widetilde{K}_{\infty} . We consider \widetilde{E} the z -twist of E , introduced in [ 7 ] , that is the homomorphism of \mathbb{F}_{q}(z) -algebras \widetilde{E}:\widetilde{A}\rightarrow M_{d}(\widetilde{\mathscr{O}_{L}})\{\tau\} given by \widetilde{E}_{a}={\sum}\limits_{i=0}^{r_{a}}E_{a,i}z^{i}\tau^{i},\text{ for all }a\in A. We also have an exponential series \exp_{\widetilde{E}} associated with \widetilde{E} that converges on \operatorname{Lie}_{\widetilde{E}}(\widetilde{L_{\infty}}) . Taelman showed that the module of z -units U(\widetilde{E};\widetilde{\mathscr{O}_{L}})=\{x\in\operatorname{Lie}_{\widetilde{E}}(L_{\infty})\ |\ \exp_{\widetilde{E}}(x)\in\widetilde{E}(\widetilde{\mathscr{O}_{L}})\} is an \widetilde{A} -lattice in \operatorname{Lie}_{\widetilde{E}}(\widetilde{L_{\infty}}) and that the Taelman unit module U(E;\mathscr{O}_{L})=\{x\in\operatorname{Lie}_{E}(L_{\infty})\ |\ \exp_{E}(x)\in E(\mathscr{O}_{L})\} is an A -lattice in \operatorname{Lie}_{E}(L_{\infty}) . Moreover, he proved that the class module H(E;\mathscr{O}_{L})=\dfrac{E(L_{\infty})}{E(\mathscr{O}_{L})+\exp_{E}(\operatorname{Lie}_{E}(\mathscr{O}_{L}))} is finite. The local factors associated with \widetilde{E} and E at a monic prime polynomial Q are respectively z_{Q}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})=\dfrac{\left[\operatorname{Lie}_{\widetilde{E}}(\widetilde{\mathscr{O}_{L}}/Q\widetilde{\mathscr{O}_{L}})\right]_{\widetilde{A}}}{[\widetilde{E}(\widetilde{\mathscr{O}_{L}}/Q\widetilde{\mathscr{O}_{L}})]_{\widetilde{A}}}\in\widetilde{K}^{\ast}\text{ and }z_{Q}(E/\mathscr{O}_{L})=\dfrac{\left[\operatorname{Lie}_{E}(\mathscr{O}_{L}/Q\mathscr{O}_{L})\right]_{A}}{[E(\mathscr{O}_{L}/Q\mathscr{O}_{L})]_{A}}\in K^{\ast}. Set m=dn and consider \mathscr{C} an A -basis of \operatorname{Lie}_{E}(\mathscr{O}_{L}) . Fix (u_{1}(z),\ldots,u_{m}(z)) an \widetilde{A} -basis of U(\widetilde{E};\widetilde{\mathscr{O}_{L}}) and (u_{1},\ldots,u_{m}) an A -basis of U(E;\mathscr{O}_{L}) . Demeslay proved in [ 11 ] that the following product converges in \mathbb{T}_{z}(K_{\infty}) , the completion of K_{\infty}[z] with respect to the Gauss norm: L(\widetilde{E}/\widetilde{\mathscr{O}_{L}})={\prod}\limits_{Q}z_{Q}(\widetilde{E}/\widetilde{\mathscr{O}_{L}}) where the product runs over all the monic irreducible polynomials Q of A . We call this product the L -series associated with \widetilde{E} and \widetilde{\mathscr{O}_{L}} . By evaluation at z=1 we obtain: L(E/\mathscr{O}_{L})=\operatorname{ev}_{z=1}L(\widetilde{E},\widetilde{\mathscr{O}}_{L})={\prod}\limits_{Q}z_{Q}(E/\mathscr{O}_{L})\in K_{\infty}^{\ast}. We call this product the L -series associated with E and \mathscr{O}_{L} . Demeslay [ 11 ] proved the following class formulas à la Taelman L(\widetilde{E}/\widetilde{\mathscr{O}_{L}})=\dfrac{\det_{\mathscr{C}}(u_{1}(z),\ldots,u_{m}(z))}{\operatorname{sgn}(\det_{\mathscr{C}}(u_{1}(z),\ldots,u_{m}(z)))} and L(E/\mathscr{O}_{L})=\dfrac{\det_{\mathscr{C}}(u_{1},\ldots,u_{m})}{\operatorname{sgn}(\det_{\mathscr{C}}(u_{1},\ldots,u_{m}))}\left[H(E;\mathscr{O}_{L})\right]_{A}. We consider the P -adic setting. Let P be an irreducible monic polynomial of A and v_{P} its associated valuation on K such that v_{P}(P)=1 . We consider K_{P}\simeq\mathbb{F}_{q^{\deg(P)}}((P)) (resp. A_{P} ) the completion of K (resp. A ) with respect to P . Consider the Tate algebra in the variable z , \mathbb{T}_{z}(K_{P}) , that is the completion of K_{P}[z] with respect to the Gauss norm. Our first main result is the construction and the convergence of the following P -adic L -series. The key argument will be the evaluation at z=\zeta\in\overline{\mathbb{F}}_{q} , see Subsections 3.4 and 4.3 . Theorem A (Theorem 4.10 ) . The following product converges in \mathbb{T}_{z}(K_{P}) L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})={\prod}\limits_{Q\neq P}z_{Q}(\widetilde{E}/\widetilde{\mathscr{O}_{L}}) where the product runs over all the monic irreducible polynomials Q of A different from P . We then define a P -adic logarithm \operatorname{Log}_{E,P} which converges on \{x\in\mathscr{O}_{L}^{d}\ |\ v_{P}(x)\geq 0\} and a P -adic regulator associated with the unit module as follows. Let (u_{1},\ldots,u_{m}) be an A -basis of the unit module. We then define the P -adic regulator of the unit module by: R_{P}(U(E;\mathscr{O}_{L}))=\dfrac{\det_{\mathscr{C}}\left(\operatorname{Log}_{E,P}(\exp_{E}(u_{1})),\ldots,\operatorname{Log}_{E,P}(\exp_{E}(u_{m}))\right)}{\operatorname{sgn}(\det_{\mathscr{C}}(u_{1},\ldots,u_{m}))}\in K_{P}. The construction does not depend on \mathscr{C} nor on the choice of the basis of the unit module. We do the same with the variable z . We then prove the following P -adic class formula à la Taelman. Theorem B (Theorem 4.21 ) . Let E be an Anderson t -module defined over \mathscr{O}_{L} . Then we have the P -adic class formula for \widetilde{E} : z_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})=R_{P}(U(\widetilde{E};\widetilde{\mathscr{O}_{L}})) and the class formula for E : z_{P}(E/\mathscr{O}_{L})L_{P}(E/\mathscr{O}_{L})=R_{P}(U(E;\mathscr{O}_{L}))\left[H(E;\mathscr{O}_{L}))\right]_{A}. A major difference from the \infty -adic case is that our P -adic L -series obtained will vanish at z=1 in certain cases that we are able to characterize. Set U(E;P\mathscr{O}_{L})=\{x\in\operatorname{Lie}_{E}(L_{\infty})\ |\ \exp_{E}(x)\in E(P\mathscr{O}_{L})\} and \mathscr{U}(E;P\mathscr{O}_{L})=\exp_{E}(U(E;P\mathscr{O}_{L})) which is provided with an A_{P} -structure module Theorem C (Proposition 5.4 and Conjecture Conjecture ) . We have the following assertions. (1) If the exponential map \exp_{E}:L_{\infty}^{d}\rightarrow L_{\infty}^{d} is not injective, then L_{P}(E/\mathscr{O}_{L})=0 . (2) If the A -rank of \exp_{E}(U(E;\mathscr{O}_{L})) and the A_{P} -rank of \mathscr{U}(E;P\mathscr{O}_{L}) coincide, then the two following assertions are equivalent: (a) L_{P}(E/\mathscr{O}_{L})\neq 0 , (b) \exp_{E}:L_{\infty}^{d}\rightarrow L_{\infty}^{d}\text{ is injective} . We extend the previous results to the multi-variable setting in the spirit of the work of Anglès, Pellarin and Tavares Ribeiro [ 5 ] , Demeslay [ 11 ] and Pellarin [ 19 ] . Consider s\geq 1 an integer and t_{1},\ldots,t_{s} new variables. Set k=\mathbb{F}_{q}(t_{1},\ldots,t_{s}) , A_{s}=k[\theta] , \mathscr{O}_{L,s}=k\mathscr{O}_{L} and K_{s,P}=\mathbb{F}_{q^{\deg(P)}}(t_{1},\ldots,t_{s})((P)) . Let s\geq 0 be a non-negative integer and E be an Anderson A_{s} -module defined over \mathscr{O}_{L,s} . We consider the Tate algebra with multi-variable \mathbb{T}_{s}(K_{P}) that is the completion of K_{P}[t_{1},\ldots,t_{s}] . We prove the following result. Theorem D (Theorem 5.8 and Theorem 5.3 ) . (1) The infinite product L_{P}({E}/{\mathscr{O}}_{L,s})={\prod}\limits_{Q\neq P}z_{Q}(E/\mathscr{O}_{L,s}) where Q runs through the monic primes of A different from P , converges in K_{s,P} and we have the class formula: z_{P}(E/\mathscr{O}_{L,s})L_{P}({E}/{\mathscr{O}}_{L,s})=R_{P}(U({E};{\mathscr{O}}_{L,s}))\left[H(E;\mathscr{O}_{L,s})\right]_{A_{s}}. (2) If E is defined over \mathbb{F}_{q}[t_{1},\ldots,t_{s}]\mathscr{O}_{L} , then L_{P}(E/\mathscr{O}_{L,s})\in\mathbb{T}_{s}(K_{P}) . The key point for proving the second assertion will be the successive evaluation at t_{i}=\zeta_{i}\in\overline{\mathbb{F}_{q}} so using techniques from Subsections 3.4 and 4.3 . Finally, in Section 6 , we give a detailed application of the previous theorems and obtain bounds on the vanishing order at z=1 of the P -adic L -series. Theorem E (Proposition 6.7 ) . Consider \phi an A -Drinfeld module defined over A of rank r . Then the vanishing order at z=1 of the P -adic L -series is less than or equal to r . The P -adic L -series is studied in detail in a work of Caruso, Gazda and the author in the context of A -Drinfeld modules defined over A , see [ 9 ] . 1.3. Some remarks Recently, Anglés [ 3 ] defined the P -adic L -series L_{P}(\widetilde{C}/\widetilde{\mathscr{O}_{L}}) associated with the Carlitz module defined over \mathscr{O}_{L} . He was able to prove that the P -adic L -series L_{P}(\widetilde{C}/\widetilde{\mathscr{O}_{L}}) is a meromorphic series without pole at z=1 , see [ 3 , Theorem 6.6] . In particular he defined the P -adic L -series as the limit at z=1 of this series which is an element of K_{P} and proved a P -adic class formula, see [ 3 , Theorem 6.7] . The present work generalizes this result for Anderson t -modules, including all Drinfeld modules, by proving that the P -adic L -series is an element of \mathbb{T}_{z}(K_{P}) so that we can evaluate at z=1 . Acknowledgments This work was carried out during my PhD thesis under the supervision of Tuan Ngo Dac and Floric Tavares Ribeiro. I would like to thank them warmly for the interesting discussions, the sharing of ideas and the time devoted to this work. I would also like to thank Bruno Anglès for the interesting discussions and remarks. 2. Notation and background 2.1. Notation We keep the notation in the Introduction and we introduce the following notation. \bullet \mathbb{C}_{\infty} : the completion of a fixed algebraic closure of K_{\infty} , \bullet \tau:\mathbb{C}_{\infty}\rightarrow\mathbb{C}_{\infty} the Frobenius endomorphism, \bullet M_{d}(R)=M_{d\times d}(R) , for a ring R the left R -module of d\times d matrices, \bullet I_{d} : the identity matrix of M_{d}(R) . Let us fix an integer d\geq 1 and B an \mathbb{F}_{q} -algebra. If M=(m_{i,j}) is a matrix with coefficients in \mathbb{C}_{\infty} and k\in\mathbb{N} , we set \tau^{k}(M)=M^{(k)} to be the matrix whose ij -entry is given by \tau^{k}(m_{i,j})^{(k)}=m_{i,j}^{q^{k}} . We denote by M_{d}(B)\{\tau\} the non-commutative ring of twisted polynomials in \tau with coefficients in M_{d}(B) equipped with the usual addition and the commutation rule \tau^{k}M=M^{(k)}\tau^{k} for all k\in\mathbb{N} and all M\in M_{d}(B) . Let M_{d}(B)\{\{\tau\}\} be the non-commutative ring of twisted power series in \tau with coefficients in M_{d}(B) . If k is a field containing \mathbb{F}_{q} , we set (kK)_{\infty}=k\hat{\otimes}_{\mathbb{F}_{q}}K_{\infty}=k((\frac{1}{\theta})). If x\in(kK)_{\infty}^{\times} , we can write x uniquely as x={\sum}\limits_{n\geq N}x_{n}\dfrac{1}{\theta^{n}} , x_{n}\in k and x_{N}\neq 0 . Then we call x_{N}\in K the sign denoted by \operatorname{sgn}(x) . We say that such an x\in(kK)_{\infty} is monic if \operatorname{sgn}(x)=1 . 2.2. Fitting ideals We recall here some definitions about Fitting ideals of modules over Dedekind rings. Let R be a Dedekind ring, and M be a finite and torsion R -module. By the structure theorem, there exists s\in\mathbb{N} and I_{1},\ldots,I_{s} ideals of R such that we have an isomorphism of R -modules M\simeq R/I_{1}\times\ldots\times R/I_{s}. We then define the Fitting ideal of M by \operatorname{Fitt}_{R}(M)=I_{1}\ldots I_{s}. We have the following properties that can be found in the appendix of [ 17 ] except the second one which appears in [ 12 , Corollary 20.5] . Proposition 2.1 . (1) We have \operatorname{Fitt}_{R}(M)\subseteq\operatorname{Ann}_{R}(M)=\{x\in R\ |\ x.m=0\ \forall m\in M\} . (2) If 0\rightarrow M_{1}\rightarrow M\rightarrow M_{2}\rightarrow 0 is exact, then \operatorname{Fitt}_{R}(M_{1})\operatorname{Fitt}_{R}(M_{2})=\operatorname{Fitt}_{R}(M) . 2.3. Lattices and Ratio of co-volumes We use the following notation from [ 22 , Section 7.2.3] . We fix k a field containing \mathbb{F}_{q} and recall that (kK)_{\infty}=k\hat{\otimes}_{\mathbb{F}_{q}}K_{\infty}=k\left(\left(\frac{1}{\theta}\right)\right) . In what follows, we fix V to be a finite-dimensional (kK)_{\infty} -vector space endowed with the natural topology coming from (kK)_{\infty} . Definition 2.2 . A sub- k[\theta] -module M of V is a k[\theta] -lattice in V if M is discrete in V and if M generates V over (kK)_{\infty} . Proposition 2.3 . Let M be a sub- k[\theta] -module of V . The following are equivalent: (1) M is a k[\theta] -lattice in V . (2) There exists a (kK)_{\infty} -basis (v_{1},\ldots,v_{n}) of V such that M is the free k[\theta] -module of basis (v_{1},\ldots,v_{n}) . Proof. See [ 22 , Proposition 7.2.3] . ∎ Let M and M^{\prime} be two k[\theta] -lattices in V . Let \mathscr{B} and \mathscr{B}^{\prime} be k[\theta] -basis of M and M^{\prime} , respectively. The ratio of co-volumes of M in M^{\prime} is then defined as [M^{\prime}:M]_{k[\theta]}=\dfrac{\det_{\mathscr{B}^{\prime}}\mathscr{B}}{\operatorname{sgn}(\det_{\mathscr{B}^{\prime}}\mathscr{B})}\in(kK)_{\infty}^{\ast}. Note that this is independent of the choices of \mathscr{B} and \mathscr{B}^{\prime} . The definition immediately implies that if M_{0}, M_{1} and M_{2} are lattices in V , then [M_{0}:M_{1}]_{k[\theta]}[M_{1}:M_{2}]_{k[\theta]}=[M_{0}:M_{2}]_{k[\theta]}. We also observe that for two lattices M,M^{\prime} in V we have [M^{\prime}:M]_{k[\theta]}=[M:M^{\prime}]_{k[\theta]}^{-1}. 3. The \infty -adic case From now on, let L/K be a finite fields extension. Recall that we denote by: \mathscr{O}_{L} the integral closure of A in L , \mathscr{O}_{L}[z]\simeq\mathbb{F}_{q}[z]\otimes_{\mathbb{F}_{q}}\mathscr{O}_{L} , \widetilde{\mathscr{O}_{L}}=\mathbb{F}_{q}(z)\otimes_{\mathbb{F}_{q}}\mathscr{O}_{L} , \widetilde{L_{\infty}}=L\otimes_{\mathbb{F}_{q}}\widetilde{K_{\infty}} . In this section, we extend the notion of the Taelman unit module and class module by twisting with some elements \zeta\in\overline{\mathbb{F}_{q}} . 3.1. Anderson modules An Anderson t -module (or shortly a t -module) E of dimension d defined over \mathscr{O}_{L} is an \mathbb{F}_{q} -algebra homomorphism E:A\rightarrow M_{d}(\mathscr{O}_{L})\{\tau\} such that if a\in A and E_{a}={\sum}\limits_{i=0}^{r_{a}}E_{a,i}\tau^{i} , then we require (E_{a,0}-aI_{d})^{d}=0 and that \deg_{\tau}(E_{\theta})>0 . Let E:A\rightarrow M_{d}(\mathscr{O}_{L})\{\tau\} be a t -module of dimension d\geq 1 , completely determined by the value at \theta : E_{\theta}={\sum}\limits_{i=0}^{r}E_{\theta,i}\tau^{i} with E_{\theta,i}\in M_{d}(\mathscr{O}_{L}) and (E_{\theta,0}-\theta I_{d})^{d}=0 . For a\in A , if E_{a}={\sum}\limits_{i=0}^{r_{a}}E_{a,i}\tau^{i} , we define \partial_{E}(a)=E_{a,0} . Then the map \partial_{E}:A\rightarrow M_{d}(A) is a homomorphism of \mathbb{F}_{q} -algebras. We then consider the z -twist of E , introduced in [ 7 ] , (remember that \tau acts as the identity over \mathbb{F}_{q}(z) ) denoted by \widetilde{E} to be the homomorphism of \mathbb{F}_{q}(z) -algebras \widetilde{E}:\widetilde{A}\rightarrow M_{d}(\widetilde{\mathscr{O}_{L}})\{\tau\} given by: \widetilde{E}_{\theta}={\sum}\limits_{i=0}^{r}E_{\theta,i}z^{i}\tau^{i}. Recall the following notation taken from [ 4 ] . Let E be a t -module of dimension d over R an extension of \mathbb{F}_{q} and let B be an R -algebra. We can then define two A -module structures on B^{d} . The first is denoted E(B) where A acts on B^{d} via E : a.x=E_{a}(x)\in B^{d}\text{ for all }a\in A,x\in B^{d}. The second is \operatorname{Lie}_{E}(B) where A acts on B^{d} via \partial_{E} : a.x=\partial_{E}(a)x\text{ for all }a\in A,x\in B^{d}. We have the following results that can be found in [ 1 , Proposition 2.1.4] . Proposition 3.1 . There exists a unique element \operatorname{exp}_{E}\in M_{d}(L)\{\{\tau\}\} such that: (1) \operatorname{exp}_{E}\partial_{E}(a)=E_{a}\operatorname{exp}_{E}\ \text{ hold in }M_{d}(L)\{\{\tau\}\}\text{ for all }a\in A, (2) \operatorname{exp}_{E}\equiv I_{d}\text{ mod }M_{d}(L)\{\{\tau\}\}\tau. We call \exp_{E} the exponential map associated with the t -module E , and denote this element by \exp_{E}={\sum}\limits_{i=0}^{\infty}d_{n}\tau^{n}. Proposition 3.2 . There exists a unique element \operatorname{log}_{E}\in M_{d}(L)\{\{\tau\}\} such that: (1) \operatorname{log}_{E}E_{a}=\partial_{E}(a)\operatorname{log}_{E}\text{ hold in }M_{d}(L)\{\{\tau\}\}\text{ for all }a\in A , (2) \operatorname{log}_{E}\equiv I_{d}\text{ mod }M_{d}(L)\{\{\tau\}\}\tau. In addition, we have the equalities in M_{d}(L)\{\{\tau\}\} : \log_{E}\exp_{E}=\exp_{E}\log_{E}=I_{d}. We call \log_{E} the logarithm map associated to the t -module E , and we denote this element by \log_{E}={\sum}\limits_{n=0}^{\infty}l_{n}\tau^{n}. We also have exponential and logarithm series for the z -twist of the t -module \widetilde{E} which we denote by \exp_{\widetilde{E}} and \log_{\widetilde{E}} and given by: \exp_{\widetilde{E}}={\sum}\limits_{n\geq 0}d_{n}z^{n}\tau^{n}\text{ and }\log_{\widetilde{E}}={\sum}\limits_{n\geq 0}l_{n}z^{n}\tau^{n}. 3.2. Unit module and class module We consider an over-additive valuation v_{\infty} on the-finite dimensional K_{\infty} -vector space L_{\infty} (for example with respect to the choise of a basis of L/K) . The key point is the next result from [ 15 , Theorem 4.6.9] . Lemma 3.3 . We have \lim\limits_{i\rightarrow+\infty}\dfrac{v_{\infty}(d_{i})}{q^{i}}=+\infty. Corollary 3.4 . The exponential map \exp_{E} converges on \operatorname{Lie}_{E}(L_{\infty}) and induces a homomorphism of A -modules: \exp_{E}:\operatorname{Lie}_{E}(L_{\infty})\rightarrow E(L_{\infty}). We also have the convergence of \exp_{\widetilde{E}} on \operatorname{Lie}_{\widetilde{E}}(\widetilde{L_{\infty}}) (resp. \operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{\infty})) ) that induces a homomorphism of \widetilde{A} -modules (resp. A[z] ): \exp_{\widetilde{E}}:\operatorname{Lie}_{\widetilde{E}}(\widetilde{L_{\infty}})\rightarrow\widetilde{E}(\widetilde{L_{\infty}}). (resp. \exp_{\widetilde{E}}:\operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{\infty}))\rightarrow\widetilde{E}(\mathbb{T}_{z}(L_{\infty})) ). We can now define the Taelman unit module U(E;\mathscr{O}_{L})=\{x\in\operatorname{Lie}_{E}(L_{\infty})\ |\ \exp_{E}(x)\in E(\mathscr{O}_{L})\} provided with A -module structure, as well as the module of z -units: U(\widetilde{E};\widetilde{\mathscr{O}_{L}})=\left\{x\in\operatorname{Lie}_{\widetilde{E}}(\widetilde{L_{\infty}})\ |\ \exp_{\widetilde{E}}(x)\in\widetilde{E}(\widetilde{\mathscr{O}_{L}})\right\} provided with \widetilde{A} -module structure, and the module of z -units at the integral level: U(\widetilde{E};\mathscr{O}_{L}[z])=\left\{x\in\operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{\infty}))\ |\ \exp_{\widetilde{E}}(x)\in\widetilde{E}(\mathscr{O}_{L}[z])\right\} provided with A[z] -module structure. We also define the class module (introduced by Taelman in [ 21 ] ): H(E;\mathscr{O}_{L})=\dfrac{E(L_{\infty})}{E(\mathscr{O}_{L})+\exp_{{E}}(\operatorname{Lie}_{{E}}(L_{\infty}))} as well as the class module for the z -deformation H(\widetilde{E};\widetilde{\mathscr{O}_{L}})=\dfrac{\widetilde{E}(\widetilde{L_{\infty}})}{\widetilde{E}(\widetilde{\mathscr{O}_{L}})+\exp_{\widetilde{E}}(\operatorname{Lie}_{\widetilde{E}}(\widetilde{L_{\infty}}))} and finally the class module at the integral level H(\widetilde{E};\mathscr{O}_{L}[z])=\dfrac{\widetilde{E}(\mathbb{T}_{z}(L_{\infty}))}{\widetilde{E}(\mathscr{O}_{L}[z])+\exp_{\widetilde{E}}(\operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{\infty})))}. Consider one of the following cases: \bullet k_{0}=\mathbb{F}_{q} , \varphi=E and (k_{0}L)_{\infty}=L_{\infty} , \bullet k_{0}=\mathbb{F}_{q}[z] and \varphi=\widetilde{E} , \bullet k_{0}=\mathbb{F}_{q}(z) , \varphi=\widetilde{E} and (k_{0}L)_{\infty}=\widetilde{L_{\infty}} . We have the following result from [ 11 , Proposition 2.8] . Proposition 3.5 . (1) The class module H(\varphi;k_{0}\mathscr{O}_{L}) is a finite-dimensional k_{0} -vector space, so a finite and torsion k_{0}A -module. (2) If k_{0} is a field, then the module of units U(\varphi;k_{0}\mathscr{O}_{L}) is a k_{0}A -lattice in \operatorname{Lie}_{\varphi}((k_{0}L)_{\infty}) . We also have the following result in [ 7 , Proposition 2.3] . Proposition 3.6 . We have the following equality: U(\widetilde{E};\widetilde{\mathscr{O}_{L}})=\mathbb{F}_{q}(z)U(\widetilde{E};\mathscr{O}_{L}[z]). Consider the evaluation morphism: \operatorname{ev}_{z=1}:\operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{\infty}))\rightarrow\operatorname{Lie}_{E}(L_{\infty}). It induces an exact sequence of A -modules: 0\longrightarrow(z-1)\operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{\infty}))\longrightarrow\operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{\infty}))\overset{\operatorname{ev}_{z=1}}{\longrightarrow}\operatorname{Lie}_{E}(L_{\infty})\longrightarrow 0. For all x\in\operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{\infty})) we have \operatorname{ev}_{z=1}(\exp_{\widetilde{E}}(x))=\exp_{E}(\operatorname{ev}_{z=1}(x)). Moreover, if f(z)\in\widetilde{L_{\infty}} belongs to the \infty -adic convergence domain of the logarithm map \log_{\widetilde{E}} , then we have \operatorname{ev}_{z=1}(\log_{\widetilde{E}}(f(z)))=\log_{E}(\operatorname{ev}_{z=1}(f(z))). We recall the notion of Stark units introduced by B. Anglès and F. Tavares Ribeiro in [ 7 , section 2.5] . Definition 3.7 . The module of Stark units U_{\operatorname{St}}(E;\mathscr{O}_{L}) is defined by: U_{\operatorname{St}}(E;\mathscr{O}_{L})=\operatorname{ev}_{z=1}U(\widetilde{E};\mathscr{O}_{L}[z]). Given the compatibility between the exponential and the evaluation morphism, U_{\operatorname{St}}(E;\mathscr{O}_{L}) is a sub- A -module of U(E,\mathscr{O}_{L}). We have the following result from [ 7 , Theorem 1] . Theorem 3.8 . The A -module U_{\operatorname{St}}(E;\mathscr{O}_{L}) is an A -lattice in \operatorname{Lie}_{E}(L_{\infty}) . 3.3. The L series For a monic prime P of A , we define the local factor at P associated with E : z_{P}(E/\mathscr{O}_{L})=\dfrac{\left[\operatorname{Lie}_{E}(\mathscr{O}_{L}/P\mathscr{O}_{L})\right]_{A}}{\left[E(\mathscr{O}_{L}/P\mathscr{O}_{L})\right]_{A}}\in K and the local factor at P associated with \widetilde{E} : z_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})=\dfrac{\left[\operatorname{Lie}_{\widetilde{E}}(\widetilde{\mathscr{O}_{L}}/P\widetilde{\mathscr{O}_{L}})\right]_{\widetilde{A}}}{\left[\widetilde{E}(\widetilde{\mathscr{O}_{L}}/P\widetilde{\mathscr{O}_{L}})\right]_{\widetilde{A}}}\in\widetilde{K}. We then define the L -series associated with E and \mathscr{O}_{L} by the Eulerian product: L(E/\mathscr{O}_{L})={\prod}\limits_{P\in A}z_{P}(E/\mathscr{O}_{L}) where P runs through the monic primes of A , and the L -series associated with \widetilde{E} and \widetilde{\mathscr{O}_{L}} by the Eulerian product: L(\widetilde{E}/\widetilde{\mathscr{O}_{L}})={\prod}\limits_{P\in A}z_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}}). We have the convergence of the L -series and the class formula for z -deformation from [ 11 , Theorem 2.7] . Theorem 3.9 (Class formula for the z -deformation) . The product defining L(\widetilde{E}/\widetilde{\mathscr{O}_{L}}) converges in \widetilde{K_{\infty}}^{\ast} and we have the formula (1) L(\widetilde{E}/\widetilde{\mathscr{O}_{L}})=\left[\operatorname{Lie}_{\widetilde{E}}\left(\widetilde{\mathscr{O}_{L}}\right):U(\widetilde{E};\widetilde{\mathscr{O}_{L}})\right]_{\widetilde{A}}. Adapting the proof of [ 22 , Corollary 7.5.6] in the higher dimensional case we obtain that the polynomial \left[\operatorname{Fitt}_{\widetilde{A}}(\widetilde{E}(\widetilde{\mathscr{O}_{L}}/{P}\widetilde{\mathscr{O}_{L}}))\right]_{\widetilde{A}}\in A[z] is a unit in \mathbb{T}_{z}(K_{\infty}) . We then obtain: Corollary 3.10 . The L -series L(\widetilde{E}/\widetilde{\mathscr{O}_{L}}) converges in \mathbb{T}_{z}(K_{\infty})^{\times}. We can evaluate the L -series at z=1 : L(E/\mathscr{O}_{L})=\operatorname{ev}_{z=1}L(\widetilde{E}/\widetilde{\mathscr{O}_{L}})={\prod}\limits_{Q}\dfrac{\left[\operatorname{Lie}_{E}(\mathscr{O}_{L}/Q\mathscr{O}_{L})\right]_{A}}{\left[E(\mathscr{O}_{L}/Q\mathscr{O}_{L}))\right]_{A}}\in K_{\infty}^{\ast} where Q runs through the monic primes of A . We have the following class formula for t -modules obtained by Fang in [ 13 ] , generalizing Taelman’s class formula for Drinfeld modules. Theorem 3.11 (Class formula for Anderson t -modules) . The product defining L(E/\mathscr{O}_{L}) converges in K_{\infty}^{\ast} , and we have the equalities (2) L(E/\mathscr{O}_{L})=\left[\operatorname{Lie}_{E}(\mathscr{O}_{L}):U(E;\mathscr{O}_{L})\right]_{A}\left[H(E;\mathscr{O}_{L})\right]_{A}=\left[\operatorname{Lie}_{E}(\mathscr{O}_{L}):U_{\operatorname{st}}(E;\mathscr{O}_{L})\right]_{A}. 3.4. Evaluation at z=\zeta\in\overline{\mathbb{F}}_{q} . We want to extend the notion of Stark units by evaluating the variable z at z=\zeta for all \zeta\in\overline{\mathbb{F}}_{q} . Let \zeta be an element of \overline{\mathbb{F}}_{q} and consider \mathbb{F}_{q}(\zeta) the finite field obtained by adding \zeta to \mathbb{F}_{q} . Let us define the ring A_{\zeta}=\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}A . We define a Frobenius \tau_{\zeta}=\operatorname{id}\otimes\tau acting on A_{\zeta} . Let us define \widetilde{A_{\zeta}}=\mathbb{F}_{q}(z)\otimes_{\mathbb{F}_{q}}A_{\zeta} on which we extend the Frobenius \tau_{\zeta} by \tau_{\zeta}=\operatorname{id}\otimes\tau_{\zeta} , still denoted by \tau_{\zeta} (i.e., the Frobenius \tau_{\zeta} acts as the identity on \mathbb{F}_{q}(z) ). Denote by A_{\zeta}[z]=\mathbb{F}_{q}[z]\otimes_{\mathbb{F}_{q}}A_{\zeta} . Set \mathscr{O}_{L,\zeta}=\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}\mathscr{O}_{L} . It is also equiped with the following Frobenius \tau_{\zeta}=\operatorname{id}\otimes\tau . Similarly as the z -deformation, let us twist the t -module E into an Anderson A_{\zeta} -module E_{\zeta} defined over M_{d}(\mathscr{O}_{L,\zeta}) by (E_{\zeta})_{\theta}={\sum}\limits_{i=0}^{r}E_{\theta,i}\zeta^{i}\tau_{\zeta}^{i}\in M_{d}(\mathscr{O}_{L,\zeta})\{\tau_{\zeta}\} then extend to A_{\zeta} by \mathbb{F}_{q}(\zeta) -linearity. Set M_{w}=\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}L_{w} where w=\infty or w=P . Consider \mathbb{F}_{q}[z]\otimes_{\mathbb{F}_{q}}\mathscr{O}_{L,\zeta}=\mathscr{O}_{L,\zeta}[z] then set \widetilde{\mathscr{O}_{L,\zeta}}=\mathbb{F}_{q}(z)\otimes_{\mathbb{F}_{q}}\mathscr{O}_{L,\zeta} and \widetilde{M}_{w}=\mathscr{O}_{L,\zeta}[z]\otimes_{\mathscr{O}_{L}[z]}\mathbb{T}_{z}(L_{w})\simeq\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z}(L_{w}) and consider \mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}\widetilde{L_{w}} . We extend v_{\infty} to M_{\infty} as follows. Let’s fix (f_{1},\ldots,f_{m}) a \mathbb{F}_{q} -basis of \mathbb{F}_{q}(\zeta) . We set v_{\infty}\left({\sum}\limits_{i=1}^{m}f_{i}\otimes x_{i}\right)=\min_{i=1,\ldots,h}v_{\infty}(x_{i}) for x_{i}\in K_{\infty} . The topology over M_{\infty} does not depend on the choice of the basis (f_{1},\ldots,f_{m}) . We then consider v_{\infty} an over-additive valuation on the \mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}K_{\infty} -vector space of finite dimension M_{\infty} . We then extend similarly v_{\infty} to \mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}\widetilde{L_{\infty}} . Remark that we cannot just replace v_{\infty} by v_{P} on \mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}K_{P} with theses constructions, in fact we do not obtain a valuation over M_{v_{P}} . See Subsection 4.3 for more details. Finally, we deform E into E^{(\zeta)} an Anderson A_{\zeta} -module on M_{d}(\mathscr{O}_{L,\zeta}) by E^{(\zeta)}_{\theta}={\sum}\limits_{i=0}^{r}E_{\theta,i}\tau_{\zeta}^{i} and extend it to A_{\zeta} by \mathbb{F}_{q}(\zeta) -linearity. We finally extend it to an Anderson \widetilde{A_{\zeta}} -module \widetilde{E}^{(\zeta)} on M_{d}(\widetilde{\mathscr{O}_{L,\zeta}}) in the usual way. We have exponential maps associated with each of the Anderson modules. From the definitions we have the equalities \exp_{E^{(\zeta)}}={\sum}\limits_{n\geq 0}d_{n}\tau_{\zeta}^{n}\text{ and }\log_{E^{(\zeta)}}={\sum}\limits_{n\geq 0}l_{n}\tau_{\zeta}^{n}, and the map \exp_{E^{(\zeta)}} (resp. \exp_{\widetilde{E}^{(\zeta)}} ) converges on \operatorname{Lie}_{E^{(\zeta)}}(M_{\infty}) (resp. on \operatorname{Lie}_{\widetilde{E}^{(\zeta)}}(\widetilde{M}_{\infty}) ). Moreover, we have the following equalities in M_{d}(\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}L)\{\{\tau_{\zeta}\}\} : \exp_{E_{\zeta}}={\sum}\limits_{n\geq 0}d_{n}\zeta^{n}\tau_{\zeta}^{n}\text{ and }\log_{E_{\zeta}}={\sum}\limits_{n\geq 0}l_{n}\zeta^{n}\tau_{\zeta}^{n}. Consider the evaluation morphism at z=\zeta : \operatorname{ev}_{\zeta}=\operatorname{ev}_{z=\zeta}:\widetilde{M}_{\infty}\rightarrow M_{\infty} whose kernel is given by (z-\zeta)\widetilde{M}_{\infty} , then we consider the following evaluation morphism still denoted by \operatorname{ev}_{\zeta} : \operatorname{ev}_{\zeta}:\operatorname{Lie}_{\widetilde{E}^{(\zeta)}}(\widetilde{M}_{\infty})\rightarrow\operatorname{Lie}_{E^{(\zeta)}}(M_{\infty}). For x\in\operatorname{Lie}_{\widetilde{E}^{(\zeta)}}(\widetilde{M}_{\infty}) , we have in \operatorname{Lie}_{E^{(\zeta)}}(M_{\infty}) : \operatorname{ev}_{\zeta}(\exp_{\widetilde{E}^{(\zeta)}}(x))=\exp_{E_{\zeta}}(\operatorname{ev}_{\zeta}(x)). Let us consider the module of \zeta -units at the integral level: U(\widetilde{E}^{(\zeta)};\mathscr{O}_{L,\zeta}[z])=\left\{x\in\operatorname{Lie}_{\widetilde{E}^{(\zeta)}}(\widetilde{M}_{\infty})\ |\ \exp_{\widetilde{E}^{(\zeta)}}(x)\in\widetilde{E}(\mathscr{O}_{L,\zeta}[z])\right\} as well as the module of the \zeta -classes at the integral level: H(\widetilde{E}^{(\zeta)};\mathscr{O}_{L,\zeta}[z])=\dfrac{\widetilde{E}^{(\zeta)}(\widetilde{M}_{\infty})}{\widetilde{E}^{(\zeta)}(\mathscr{O}_{L,\zeta}[z])+\exp_{\widetilde{E}^{(\zeta)}}(\operatorname{Lie}_{\widetilde{E}^{(\zeta)}}(\widetilde{M}_{\infty}))}=A_{\zeta}[z]\otimes_{\mathbb{F}_{q}[z]}H(\widetilde{E},\mathscr{O}_{L}[z]) provided with a structure of A_{\zeta}[z] -modules. Next, consider the \zeta -unit module: U(E_{\zeta};\mathscr{O}_{L,\zeta})=\{x\in\operatorname{Lie}_{E_{\zeta}}(M_{\infty})\ |\ \exp_{E_{\zeta}}(x)\in E_{\zeta}(\mathscr{O}_{L,\zeta})\} and the \zeta -class module H(E_{\zeta},\mathscr{O}_{L,\zeta})=\dfrac{E_{\zeta}(M_{\infty})}{E_{\zeta}(\mathscr{O}_{L,\zeta})+\exp_{E_{\zeta}}(\operatorname{Lie}_{E_{\zeta}}(M_{\infty}))} provided with their A_{\zeta} -module structure via E_{\zeta}. Results to come in this section are adapted from [ 11 ] and [ 7 ] . Proposition 3.12 . (1) The exponential map \exp_{E_{\zeta}}:\operatorname{Lie}_{E_{\zeta}}(M_{\infty})\rightarrow E_{\zeta}(M_{\infty}) is locally an isometry. (2) The exponential map \exp_{\widetilde{E}^{(\zeta)}}:\operatorname{Lie}_{\widetilde{E^{(\zeta)}}}(\widetilde{M}_{\infty})\rightarrow\widetilde{E}^{(\zeta)}(\widetilde{M}_{\infty}) is locally an isometry. Proof. The proof is a direct corollary of Lemma 3.3 , we omit the proof. ∎ Proposition 3.13 . (1) The module of \zeta -classes H(E_{\zeta};\mathscr{O}_{L,\zeta}) is a \mathbb{F}_{q}(\zeta) -vector space of finite dimension, hence a torsion A_{\zeta} -module of finite type. (2) The module of \zeta -units U(E_{\zeta};\mathscr{O}_{L,\zeta}) is an A_{\zeta} -lattice in M_{\infty} . (3) The class module H(\widetilde{E}^{(\zeta)};\mathscr{O}_{L,\zeta}[z]) is a \mathbb{F}_{q}(\zeta)[z] -module of finite type. Proof. The proof follows the proof of [ 11 , Proposition 2.6] for the two first assertions, and the proof of [ 7 , Proposition 2] for the last one, by replacing A by A_{\zeta} , \mathscr{O}_{L} by \mathscr{O}_{L,\zeta} and E by E_{\zeta} . We omit the details. ∎ Just as Stark’s units consist of the evaluation at z=1 of the z -units, we define the evaluation at z=\zeta of the \zeta -units at the integral level: U_{\zeta}(E;\mathscr{O}_{L})=\operatorname{ev}_{\zeta}U(\widetilde{E}^{(\zeta)};\mathscr{O}_{L,\zeta}[z])\subseteq U(E_{\zeta};\mathscr{O}_{L,\zeta}) provided with an A_{\zeta} -module structure via E_{\zeta} . Theorem 3.14 . There exists an A_{\zeta} -module isomorphism: \dfrac{U(E_{\zeta};\mathscr{O}_{L,\zeta})}{U_{\zeta}(E;\mathscr{O}_{L})}\simeq H(\widetilde{E}^{(\zeta)};\mathscr{O}_{L,\zeta}[z])[z-\zeta] where H(\widetilde{E}^{(\zeta)};\mathscr{O}_{L,\zeta}[z])[z-\zeta] is the (z-\zeta) -torsion of the \zeta -class module at the integral level. In the following, we will denote by M=\mathscr{O}_{L,\zeta} and \widetilde{M}=\mathscr{O}_{L,\zeta}[z] . Proof. We follow the proof of [ 7 , Proposition 2.6] . Consider the map \displaystyle\alpha:M_{\infty}^{d} \displaystyle\rightarrow\widetilde{M}_{\infty}^{d} \displaystyle x \displaystyle\mapsto\dfrac{\exp_{\widetilde{E}^{(\zeta)}}(x)-\exp_{E_{\zeta}}(x)}{z-\zeta}. We divide the proof into several steps. Step 1: The map is well defined since \operatorname{ev}_{\zeta}(\exp_{\widetilde{E}^{(\zeta)}}(x))=\exp_{E_{\zeta}}(x) for x\in M_{\infty}^{d} , thus (z-\zeta) divide \exp_{\widetilde{E}^{(\zeta)}}(x)-\exp_{E_{\zeta}}(x) in \widetilde{M}_{\infty}^{d}. Step 2: We still denote \alpha to be the restriction: \alpha:U(E_{\zeta},M)\rightarrow H(\widetilde{E}^{(\zeta)};\widetilde{M}). Let us prove that it is a homomorphism of A_{\zeta} -modules. Let x\in U(E_{\zeta};M) be a unit and a\in A_{\zeta} . Then: \displaystyle(z-\zeta)\alpha(ax) \displaystyle=\exp_{\widetilde{E}^{(\zeta)}}(ax)-\exp_{E_{\zeta}}(ax) \displaystyle=\widetilde{E}^{(\zeta)}_{a}(\exp_{\widetilde{E}^{(\zeta)}}(x))-(E_{\zeta})_{a}(\exp_{E_{\zeta}}(x)) \displaystyle={\sum}\limits_{i=0}^{r_{a}}E_{a,i}z^{i}\tau_{\zeta}^{i}(\exp_{\widetilde{E}^{(\zeta)}}(x))-{\sum}\limits_{i=0}^{r_{a}}E_{a,i}\zeta^{i}\tau_{\zeta}^{i}(\exp_{E_{\zeta}}(x)) \displaystyle={\sum}\limits_{i=0}^{r_{a}}E_{a,i}z^{i}\tau_{\zeta}^{i}(\exp_{\widetilde{E}^{(\zeta)}}(x)-\exp_{E_{\zeta}}(x))+{\sum}\limits_{i=1}^{r_{a}}E_{a,i}(z^{i}-\zeta^{i})\tau_{\zeta}^{i}(\exp_{E_{\zeta}}(x)). Thus \alpha(ax)=\widetilde{E}_{a}^{(\zeta)}(\alpha(x))+\underbrace{{\sum}\limits_{i=0}^{h}a_{i}\dfrac{z^{i}-\zeta^{i}}{z-\zeta}\tau_{\zeta}^{i}(\underbrace{\exp_{E_{\zeta}}(x)}_{\in M^{d}})}_{\in\widetilde{M}^{d}}. We have proved that \alpha(ax)=\widetilde{E}_{a}^{(\zeta)}(\alpha(x))\text{ mod }\left(\widetilde{M}^{d}+\exp_{\widetilde{E}^{(\zeta)}}(\operatorname{Lie}_{\widetilde{E}^{(\zeta)}}(\widetilde{M}_{s,\infty}))\right), so \alpha(ax)=\widetilde{E}_{a}^{(\zeta)}(\alpha(x)) in H(\widetilde{E}^{(\zeta)},\widetilde{M}) . Step 3: We claim that the image of U(E_{\zeta},M) is in the (z-\zeta) -torsion of the \zeta -class module at the integral level. In fact, let x\in U(E_{\zeta};M) be a unit. We have: (z-\zeta)\alpha(x)=\exp_{\widetilde{E}^{(\zeta)}}(x)-\exp_{E_{\zeta}}(x)=0\text{ mod }\left(E_{\zeta}(M)+\exp_{\widetilde{E}^{(\zeta)}}(\operatorname{Lie}_{\widetilde{E}^{(\zeta)}}(\widetilde{M}_{\infty}))\right). Step 4: Let us prove the surjectivity of \alpha on H(\widetilde{E}^{(\zeta)};\widetilde{M})[z-\zeta] . Let x\in\widetilde{E}^{(\zeta)}(\widetilde{M}_{\infty}) be such that (z-\zeta)x=\exp_{\widetilde{E}^{(\zeta)}}(u)+v with u\in\operatorname{Lie}_{\widetilde{E}^{(\zeta)}}(\widetilde{M}_{\infty}) and v\in\widetilde{E}^{(\zeta)}(\widetilde{M}). We write u=u_{1}+(z-\zeta)u_{2} , u_{1}\in M_{\infty}^{d} , u_{2}\in\widetilde{M}_{\infty}^{d} and v=v_{1}+(z-\zeta)v_{2} , v_{1}\in M^{d} , v_{2}\in\widetilde{M}^{d} . We have: (z-\zeta)x=\exp_{\widetilde{E}^{(\zeta)}}(u_{1})+v_{1}+(z-\zeta)(v_{2}+\exp_{\widetilde{E}^{(\zeta)}}(u_{2})). By evaluating at z=\zeta yields \exp_{E_{\zeta}}(u_{1})+v_{1}=0 . Thus u_{1}\in U(E_{\zeta};M). Moreover, we have: \alpha(u_{1})=x-(\underbrace{\exp_{\widetilde{E}^{(\zeta)}}(u_{2})}_{\in\exp_{\widetilde{E}^{(\zeta)}}(\widetilde{M}_{\infty}^{d})}+\underbrace{v_{2}}_{\in\widetilde{M}^{d}}) thus \alpha(u_{1})=x\text{ mod }\left(\widetilde{M}^{d}+\exp_{\widetilde{E}^{(\zeta)}}(\widetilde{M}_{\infty}^{d})\right). Step 5: We now consider the kernel of \alpha:U(E_{\zeta};M)\rightarrow H(\widetilde{E}^{(\zeta)},\widetilde{M}) denoted by \kappa . We want to prove that \kappa=U_{\zeta}(E,\mathscr{O}_{L}). We proceed by double inclusion. \supseteq Let x\in U_{\zeta}(E,\mathscr{O}_{L}) be a unit and write x=\operatorname{ev}_{\zeta}(u) with u\in U(\widetilde{E}^{(\zeta)};\widetilde{M}). We have \operatorname{ev}_{\zeta}(x-u)=0 thus we can find v\in\widetilde{M}_{\infty}^{d} such that x=u+(z-\zeta)v. We have \alpha(x)=\dfrac{\exp_{\widetilde{E}^{(\zeta)}}(u)-\exp_{E_{\zeta}}(x)}{z-\zeta}+\exp_{\widetilde{E}^{(\zeta)}}(v). But \exp_{E_{\zeta}}(x)=\operatorname{ev}_{\zeta}\exp_{\widetilde{E}^{(\zeta)}}(u)\in M^{d} so \alpha(x)=0\text{ mod }\left(\widetilde{M}^{d}+\exp_{\widetilde{E}^{(\zeta)}}(\widetilde{M}_{\infty}^{d})\right). \subseteq Let x\in U(E_{\zeta};M) be such that \alpha(x)\in\widetilde{M}^{d}+\exp_{\widetilde{E}^{(\zeta)}}(\widetilde{M}_{\infty}^{d}) . Let us express \alpha(x)=u+\exp_{\widetilde{E}^{(\zeta)}}(v) . We have (z-\zeta)\alpha(x)=\exp_{\widetilde{E}^{(\zeta)}}(x)+\exp_{E_{\zeta}}(x)=(z-\zeta)u+\exp_{\widetilde{E}^{(\zeta)}}((z-\zeta)v). Thus \exp_{\widetilde{E}^{(\zeta)}}(x-(z-\zeta)v)=(z-\zeta)u+\exp_{E_{\zeta}}(x)\in\widetilde{M}^{d} so x-(z-\zeta)v\in U(\widetilde{E}^{(\zeta)};\widetilde{M}) . Finally we obtain \operatorname{ev}_{\zeta}(x-(z-\zeta)v)=x\in U_{\zeta}(E;\mathscr{O}_{L}). ∎ Corollary 3.15 . The unit module U_{\zeta}(E;\mathscr{O}_{L}) is an A_{\zeta} -lattice in M_{\infty}^{d} . Moreover, we have the following equalities \left[H(\widetilde{E}^{(\zeta)};\widetilde{M})[z-\zeta]\right]_{A_{\zeta}}=\left[H(E_{\zeta};M)\right]_{A_{\zeta}}=\left[\dfrac{U(E_{\zeta};M)}{U_{\zeta}(E;\mathscr{O}_{L})}\right]_{A_{\zeta}}. Let us start by proving the following result. Lemma 3.16 . We have an exact sequence of A_{\zeta} -modules: 0\longrightarrow(z-\zeta)H(\widetilde{E}^{(\zeta)};\widetilde{M})\longrightarrow H(\widetilde{E}^{(\zeta)};\widetilde{M})\overset{\operatorname{ev}_{\zeta}}{\longrightarrow}H(E_{\zeta};M)\longrightarrow 0. Proof of Lemma 3.16 . We apply the snake lemma to the following commutative diagram (where the lines are exact sequences of A_{\zeta} -modules) and the i_{j} represent natural injections: (z-\zeta)(\exp_{\widetilde{E}^{(\zeta)}}(\widetilde{M}_{\infty}^{d})+\widetilde{M}^{d}) \exp_{\widetilde{E}^{(\zeta)}}(\widetilde{M}_{\infty}^{d})+\widetilde{M}^{d} \exp_{E_{\zeta}}(M_{\infty}^{d})+M^{d} (z-\zeta)\widetilde{M}_{\infty}^{d} \widetilde{M}_{\infty}^{d} M_{\infty}^{d} \operatorname{ev}_{z=\zeta} \operatorname{ev}_{z=\zeta} i_{1} i_{2} i_{3} ∎ Proof of Corollary 3.15 . We deduce by Lemma 3.16 an exact sequence of \mathbb{F}_{q}(\zeta) -vector spaces of finite dimension (and of finitely-generated A_{\zeta} -modules): 0\longrightarrow H(\widetilde{E}^{(\zeta)};\widetilde{M})[z-\zeta]\longrightarrow H(\widetilde{E}^{(\zeta)};\widetilde{M})\overset{.(z-\zeta)}{\longrightarrow}H(\widetilde{E}^{(\zeta)},\widetilde{M})\overset{\operatorname{ev}_{\zeta}}{\longrightarrow}H(E_{\zeta};M)\longrightarrow 0. By Proposition 2.1 we obtain: \left[H(\widetilde{E}^{(\zeta)};\widetilde{M})[z-\zeta]\right]_{A_{\zeta}}=\left[H(E_{\zeta};M)\right]_{A_{\zeta}}=\left[\dfrac{U(E_{\zeta};M)}{U_{\zeta}(E;\mathscr{O}_{L})}\right]_{A_{\zeta}}, the last equality coming from Theorem 3.14 . Since H(\widetilde{E}^{(\zeta)};\widetilde{M})[z-\zeta] is a \mathbb{F}_{q}(\zeta) -vector space of finite dimension and U(E_{\zeta};M) is an A_{\zeta} -lattice in M_{\infty}^{d} , the result follows. ∎ 4. The P -adic case We keep the notation of Section 3 . Recall that L is a finite extension of K of degree n , \mathscr{O}_{L} denotes the integral closure of A in L and E is an Anderson t -module defined over \mathscr{O}_{L} of dimension d . The goal of this section is to define and study some P -adic L -series associated with Anderson t -modules by removing the local factor at P of the classical L -series. 4.1. Introduction and notation Recall that the local factor at Q associated with E is defined by z_{Q}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})=\frac{\left[\operatorname{Lie}_{\widetilde{E}}(\widetilde{\mathscr{O}_{L}}/Q\widetilde{\mathscr{O}_{L}})\right]_{\widetilde{A}}}{\left[\widetilde{E}(\widetilde{\mathscr{O}_{L}}/Q\widetilde{\mathscr{O}_{L}})\right]_{\widetilde{A}}}\in\widetilde{K} (resp. z_{Q}(E/\mathscr{O}_{L})=\frac{\left[\operatorname{Lie}_{E}(\mathscr{O}_{L}/Q\mathscr{O}_{L})\right]_{A}}{\left[E(\mathscr{O}_{L}/Q\mathscr{O}_{L})\right]_{A}}\in K ). The goal of this section is to study the following infinite product of local factors z_{Q}(E/\mathscr{O}_{L}) (resp. z_{Q}(\widetilde{E}/\widetilde{\mathscr{O}_{L}}) ) that we call the P -adic L -series (resp. the z -twisted P -adic L -series): L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})={\prod}\limits_{Q\neq P}z_{Q}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})\text{ (resp. }L_{P}({E}/{\mathscr{O}}_{L})={\prod}\limits_{Q\neq P}z_{Q}({E}/{\mathscr{O}}_{L})) where Q runs through the monic primes of A different from P . More precisely, let us denote by v_{P} a finite place of K associated to an irreducible monic polynomial P of A . Let \mathbb{F}_{P}=\mathbb{F}_{q^{\deg(P)}} be the residue field associated to v_{P} and let K_{P}=\mathbb{F}_{P}((P)) (resp. A_{P}=\mathbb{F}_{P}[[P]] ) be the completion of K (resp. A )for v_{P} . Let \mathbb{C}_{P} be the completion of an algebraic closure of K_{P} and v_{P} the valuation on \mathbb{C}_{P} normalized such that v_{P}(P)=1 . Set \widetilde{K_{P}}=\mathbb{F}_{q}(z)\hat{\otimes}_{\mathbb{F}_{q}(z)}K_{P}=\mathbb{F}_{P}(z)((P)) on which we extend the valuation v_{P} : v_{P}({\sum}\limits_{n\geq N}\alpha_{n}(z)P^{n})=N,N\in\mathbb{Z},\alpha_{n}(z)\in\mathbb{F}_{P}(z),\alpha_{N}(z)\neq 0. Let |\cdot|_{P} be the absolute value on \mathbb{C}_{P} defined by |x|_{P}=q^{-v_{P}(x)}. Let \mathscr{B}=(f_{1},\ldots,f_{n}) be an A -basis of \mathscr{O}_{L} (that is also a K -basis of L ). We set L_{P}=L\otimes_{K}K_{P} and \widetilde{L_{P}}=L\otimes_{K}\widetilde{K_{P}} . In what follows, the reader will be careful not to confuse the notation L_{P}(E/\mathscr{O}_{L}) for the P -adic L -series and L_{P} for the tensor product L\otimes_{K}K_{P} . Then L_{P} is a K_{P} -vector space with \mathscr{B} as a basis, and \widetilde{L_{P}} is a \widetilde{K_{P}} -vector space with \mathscr{B} as a basis. In particular on L_{P} all the norms of K_{P} -vector space of finite dimension are equivalent. Let us work with the following. Consider the sup norm |.|_{P} with respect to this basis. In other words, if x={\sum}\limits_{i=1}^{n}f_{i}\otimes x_{i} with the x_{i}\in K_{P} , then we set |x|_{P}=\max_{i=1,\ldots,n}|x_{i}|_{P}. We obtain over L_{P} a norm of K_{P} -algebra. We then consider the over-additive valuation of K_{P} -vector spaces of finite dimension on L_{P} defined by: v_{P}(x)=-\log_{q}|x|_{P}=\min_{i=1,\ldots,d}v_{P}(x_{i}). For all d\geq 1 , we extend these definitions to L_{P}^{d} : if x=(x_{1},\ldots,x_{d})\in L_{P}^{d} , then we set |x|=\max_{i=1,\ldots,d}|x_{i}| or equivalently v_{P}(x)=-\log_{q}|x|=\min_{i=1,\ldots,d}v_{P}(x_{i}). In particular for all x\in\mathscr{O}_{L}^{d} we get v_{P}(x)\geq 0 . Consider \mathbb{T}_{z}(K_{P})=\left\{f(z)={\sum}\limits_{n\geq 0}a_{n}z^{n}\ |\ a_{n}\in K_{P}\text{ and }\lim\limits_{n\rightarrow+\infty}v_{P}(a_{n})=+\infty\right\}\subset\widetilde{K_{P}} and \mathbb{T}_{z}(L_{P})=\left\{f(z)={\sum}\limits_{n\geq 0}a_{n}z^{n}\ |\ a_{n}\in L_{P}\text{ and }\lim\limits_{n\rightarrow+\infty}v_{P}(a_{n})=+\infty\right\}=L\otimes_{K}\mathbb{T}_{z}(K_{P}). We define a K_{P} -vector space structure over \operatorname{Lie}_{E}(L_{P}) . We take inspiration from the \infty -adic case in [ 13 , Lemma 1.7] and [ 11 , section 2.3] . Proposition 4.1 . We can extend the homomorphism \partial_{E}:A\rightarrow M_{d}(\mathscr{O}_{L}) into a homomorphism from K_{P} to M_{d}(L_{P}) in the following way: \displaystyle\partial_{E}:K_{P} \displaystyle\rightarrow M_{d}(L_{P}), \displaystyle{\sum}\limits_{i\geq-N}\alpha_{i}P^{i} \displaystyle\mapsto{\sum}\limits_{i\geq-N}\alpha_{i}\partial_{E}(P)^{i}. Moreover, with respect to this action, L_{P}^{d} is a K_{P} -vector space of dimension m=d[L:K] denoted by \operatorname{Lie}_{E}(L_{P}) . Proof. Consider n\geq 0 . There exists a unique integer t_{n} such that q^{d}t_{n}\leq n<(t_{n}+1)q^{d}. We have: \partial_{E}(P)^{n}=\partial_{E}(P)^{q^{d}t_{n}}\partial_{E}(P)^{n-q^{d}t_{n}}=P^{q^{d}t_{n}}\underbrace{\partial_{E}(P)^{n-q^{d}t_{n}}}_{\in M_{d}(\mathscr{O}_{L})}. We obtain that \lim\limits_{n\rightarrow+\infty}v_{P}(\partial_{E}(P)^{n})=+\infty thus the map \partial_{E} is well-defined. Denote by W_{P}=\mathbb{F}_{q^{\deg(P)}}((P^{q^{d}}))\subseteq K_{P}. Then for all x\in W_{P} we have \partial_{E}(x)=xI_{d} ,thus we have an isomorphism of W_{P} -vector spaces: \operatorname{Lie}_{E}(L_{P})\simeq L_{P}^{d}. Hence \operatorname{Lie}_{E}(L_{P}) is a K_{P} -vector space of finite dimension. We have: \dim_{W_{P}}(\operatorname{Lie}_{E}(L_{P}))=\dim_{K_{P}}(\operatorname{Lie}_{E}(L_{P}))\dim_{W_{P}}(K_{P}). But from the isomorphism of W_{P} -vector spaces \operatorname{Lie}_{E}(L_{P})\simeq L_{P}^{d} we have: \dim_{W_{P}}(\operatorname{Lie}_{E}(L_{P}))=\dim_{W_{P}}(L_{P}^{d})=\dim_{K_{P}}(L_{P}^{d})\dim_{W_{P}}(K_{P}) thus \dim_{K_{P}}(\operatorname{Lie}_{E}(L_{P}))=\dim_{K_{P}}(L_{P}^{d})=d[L:K]. ∎ We also have that \operatorname{Lie}_{E}(\mathscr{O}_{L}) is an A -lattice in \operatorname{Lie}_{E}(L_{P}) . Finally, everything is still valid by adding the variable z , in other words \operatorname{Lie}_{E}(\widetilde{L_{P}}) is a \widetilde{K_{P}} -vector space of dimension d[L:K] and \operatorname{Lie}_{\widetilde{E}}(\widetilde{\mathscr{O}_{L}}) is an \widetilde{A} -lattice in \operatorname{Lie}_{E}(\widetilde{L_{P}}) . In particular, we have: \partial_{\widetilde{E}}:\mathbb{T}_{z}(K_{P}))\rightarrow M_{d}(\mathbb{T}_{z}(L_{P})). Remark that the topologies of L_{P}^{d} and \operatorname{Lie}_{E}(L_{P}) are equivalent. Consider the unique t -module F over \mathscr{O}_{L} satisfying PF_{a}=E_{a}P for all a\in A . If E_{a}={\sum}\limits_{i=0}^{r_{a}}E_{a,i}\tau^{i} , then F_{a}={\sum}\limits_{i=0}^{r_{a}}E_{a,i}P^{q^{i}-1}\tau^{i} . In particular for all a\in A we have: \partial_{E}(a)=\partial_{F}(a) . From [ 4 , Section 3.2] we have the following equalities in M_{d}(L)\{\{\tau\}\} : \log_{F}=P^{-1}\log_{E}P={\sum}\limits_{n\geq 0}l_{n}P^{q^{n}-1}\tau^{n}, and \exp_{F}=P^{-1}\exp_{E}P={\sum}\limits_{n\geq 0}d_{n}P^{q^{n}-1}\tau^{n}. We now study the link between the local factors of E and F . Lemma 4.2 . Let Q\in A be an monic prime. If Q\neq P , then we have the following equalities: z_{Q}({F}/{\mathscr{O}}_{L})=z_{Q}({E}/{\mathscr{O}}_{L}) and z_{Q}(\widetilde{F}/\widetilde{\mathscr{O}_{L}})=z_{Q}(\widetilde{E}/\widetilde{\mathscr{O}_{L}}) . Otherwise z_{P}(F/{\mathscr{O}_{L}})=1 and z_{P}(\widetilde{F}/\widetilde{\mathscr{O}_{L}})=1 . Proof. See [ 4 , Lemma 3.7] . ∎ We then obtain the following result. Corollary 4.3 . We have the following equality in K[[z]] : L(\widetilde{F}/\widetilde{\mathscr{O}_{L}})=L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}}). 4.2. P -adic exponential and P -adic logarithm We define (D_{n})_{n\geq 0} and (L_{n})_{n\geq 0} as the following sequences of elements of A : \left\{\begin{aligned} &D_{0}=1,\\ &D_{n}=\prod\limits_{k=0}^{n-1}(\theta^{q^{n-k}}-\theta)^{q^{k}},\end{aligned}\right.\text{ and }\left\{\begin{aligned} &L_{0}=1,\\ &L_{n}=\prod\limits_{k=1}^{n}(\theta-\theta^{q^{k}}).\end{aligned}\right. We first estimate the P -adic valuation of D_{n} and L_{n} for all n\geq 0 . Lemma 4.4 . We have the following equalities for n\geq 1 : (1) v_{P}(D_{n})=q^{n}\dfrac{q^{-\deg(P)\left(\left\lfloor\frac{n}{\deg(P)}\right\rfloor+1\right)}-q^{-\deg(P)}}{q^{-\deg(P)}-1}=q^{n}\dfrac{q^{-\deg(P)\left\lfloor\frac{n}{\deg(P)}\right\rfloor}-1}{1-q^{\deg(P)}}, (2) v_{P}(L_{n})=\left\lfloor\dfrac{n}{\deg(P)}\right\rfloor. Proof. See [ 2 , Section 2] . ∎ We recall that \partial_{E}(a)\in M_{d}(A) is the constant coefficient of E_{a}\in M_{d}(A)\{\tau\} for all a\in A , see Section 2 . Set s\in\mathbb{N} the smallest integer such that (\partial_{E}(\theta)-\theta I_{d})^{q^{s}}=0. There exists because \partial_{E}(\theta)-\theta I_{d} is nilpotent. Then for all a\in A we have \partial_{E}(a^{q^{s}})=a^{q^{s}}I_{d} . Recall that \exp_{E}={\sum}\limits_{n\geq 0}d_{n}\tau^{n}\in M_{d}(L)\{\{\tau\}\} and \log_{E}={\sum}\limits_{n\geq 0}l_{n}\tau^{n}\in M_{d}(L)\{\{\tau\}\} . Following [ 15 , Theorem 4.6.9] , using functional equation of the logarithm map (resp. the exponential map) \log_{E}E_{\theta^{q^{s}}}=\partial_{E}(\theta^{q^{s}})\log_{E} , an immediate induction tells us that l_{n} has the form l_{n}=\dfrac{a_{n}}{L_{n}^{q^{s}}} with a_{n}\in M_{d}(\mathscr{O}_{L}) . Reasoning in a similar way for the exponential map and by Lemma 4.4 we obtain the following result. Proposition 4.5 . We have the following inequalities for all n\geq 0 : (1) v_{P}(l_{n})\geq-q^{s}\left\lfloor\dfrac{n}{\deg(P)}\right\rfloor, (2) v_{P}(d_{n})\geq-q^{s+n}\dfrac{q^{-\deg(P)\left\lfloor\frac{n}{\deg(P)}\right\rfloor}-1}{1-q^{\deg(P)}}. So far, we have considered the exponential and logarithm series as functions of L_{\infty}^{d} , but now we want to look at them as functions of L_{P}^{d} , which we denote by \exp_{E,P} and \log_{E,P} . Note that formally (i.e., in M_{d}(L)\{\{\tau\}\}) , these are always the same series. We do the same for z -twist. Let us denote by \operatorname{ev}_{z=1,P}:\mathbb{T}_{z}(L_{P})^{d}\rightarrow L_{P}^{d} the P -adic evaluation morphism at z=1 , whose kernel is given by (z-1)\mathbb{T}_{z}(L_{P})^{d} . We can first study the P -adic convergence domain of the P -adic logarithms maps associated with \widetilde{F} and \widetilde{E} . We consider the following sets: \bullet \Omega_{z}=\left\{x\in\mathbb{T}_{z}(L_{P})^{d}\ |\ v_{P}(x)\geq 0\right\} and \Omega_{z}^{+}=\left\{x\in\mathbb{T}_{z}(L_{P})^{d}\ |\ v_{P}(x)>0\right\} , \bullet \Omega=\left\{x\in L_{P}^{d}\ |\ v_{P}(x)\geq 0\right\} and \Omega^{+}=\left\{x\in L_{P}^{d}\ |\ v_{P}(x)>0\right\}, \bullet \mathcal{D}_{z}=\left\{x\in\mathbb{T}_{z}(L_{P})^{d}\ |\ v_{P}(x)>-1+\dfrac{q^{s}}{q^{\deg(P)}-1}\right\} , \bullet \mathcal{D}_{z}^{+}=\left\{x\in\mathbb{T}_{z}(L_{P})^{d}\ |\ v_{P}(x)>\dfrac{q^{s}}{q^{\deg(P)}-1}\right\}, \bullet \mathcal{D}=\left\{u\in(L_{P})^{d}\ |\ v_{P}(u)>-1+\dfrac{q^{s}}{q^{\deg(P)}-1}\right\} , \bullet \mathcal{D}^{+}=\left\{u\in(L_{P})^{d}\ |\ v_{P}(u)>\dfrac{q^{s}}{q^{\deg(P)}-1}\right\}. Proposition 4.6 . (1) We have the P -adic convergences: \log_{\widetilde{E},P}:\Omega_{z}^{+}\rightarrow\operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{P})) and \log_{{E},P}:\Omega^{+}\rightarrow\operatorname{Lie}_{{E}}((L_{P}). Moreover, \log_{\widetilde{E},P}:\mathcal{D}_{z}^{+}\rightarrow\mathcal{D}_{z}^{+} is an isometry and \log_{{E},P}:\mathcal{D}^{+}\rightarrow\mathcal{D}^{+} is an isometry. (2) The first assertion remains true by replacing E by F and deleting the ” + ”. As a particular case, we have the convergence \log_{\widetilde{F},P}:\widetilde{E}(\mathscr{O}_{L}[z])\rightarrow\operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{P})). Proof. We give the proof only for \log_{\widetilde{E},P} , the arguments are similar in the other cases. Consider f(z)\in\mathbb{T}_{z}(L_{P})^{d}. We have (first formally): \log_{\widetilde{E},P}f(z)={\sum}\limits_{n\geq 0}L_{n}z^{n}\tau^{n}(f(z)). For all n\geq 0 we have: v_{P}\left(L_{n}\tau^{n}(f(z))\right)\geq v_{P}(L_{n})+v_{P}\left(\tau^{n}(f(z))\right)\geq-q^{s}\left\lfloor\dfrac{n}{\deg(P)}\right\rfloor+q^{n}v_{P}(f(z)) and this last quantity tends to \infty when n tends to \infty if v(f(z))>0 . Moreover, if v_{P}(f(z))>\dfrac{q^{s}}{q^{\deg(P)}-1} , then we have for all n\geq 1 : v_{P}\left(L_{n}\tau^{n}(f(z))\right)-v_{P}(f(z))\geq(q^{n}-1)v_{P}(f(z))+v(L_{n})>\dfrac{q^{s}}{q^{\deg(P)}-1}(q^{n}-1)-q^{s}\left\lfloor\dfrac{n}{\deg(P)}\right\rfloor. Write n=b\deg(P)+i\geq 1 with b\in\mathbb{N} and 0\leq i<\deg(P) . Then: \dfrac{q^{s}}{q^{\deg(P)}-1}(q^{n}-1)-q^{s}\left\lfloor\dfrac{n}{\deg(P)}\right\rfloor=q^{s}\left(\dfrac{q^{b\deg(P)+i}-1}{q^{\deg(P)}-1}-b\right). But we have: \dfrac{q^{b\deg(P)+i}-1}{q^{\deg(P)}-1}-b\geq\dfrac{q^{b\deg(P)}-1}{q^{\deg(P)}-1}-b={1+q^{\deg(P)}+\ldots+(q^{\deg(P)})^{b-1}}-b\geq 0. ∎ We have results for the P -adic convergences of the exponentials series using similar arguments. Proposition 4.7 . (1) We have the P -adic convergences: \exp_{\widetilde{E},P}:\mathcal{D}_{z}^{+}\rightarrow\mathbb{T}_{z}(L_{P})^{d} and \exp_{{E},P}:\mathcal{D}^{+}\rightarrow L_{P}^{d}. Moreover, \exp_{\widetilde{E},P}:\mathcal{D}_{z}^{+}\rightarrow\mathcal{D}_{z}^{+} is an isometry and \exp_{{E},P}:\mathcal{D}^{+}\rightarrow\mathcal{D}^{+} is an isometry. (2) The first assertion remains true by replacing E by F and deleting ” + ”. In particular for all x\in\mathcal{D}_{z} we have the following P -adic equality: (3) \exp_{F,P}(\operatorname{ev}_{z=1,P}(x))=\operatorname{ev}_{z=1,P}(\exp_{\widetilde{F}}(x)). Similarly for all x\in\Omega_{z} we have the following P -adic equality: (4) \log_{F,P}(\operatorname{ev}_{z=1,P}(x))=\operatorname{ev}_{z=1,P}(\log_{\widetilde{F}}(x)). Similarly in their convergence domain, all of the P -adic exponential and logarithm maps verify the functional identities of the exponential and the logarithm maps: \displaystyle\forall(a,x)\in A\times\Omega_{z},\ \log_{\widetilde{F},P}\partial_{\widetilde{F}}(a)x=\widetilde{F}_{a}\log_{\widetilde{F},P}x, \displaystyle\forall(a,x)\in A\times\mathcal{D}_{z},\ \exp_{\widetilde{F},P}\partial_{\widetilde{F}}(a)x=\widetilde{E}_{a}\log_{\widetilde{F},P}x. Moreover, for all x\in\mathcal{D}_{z}^{+} we have \exp_{\widetilde{E},P}(\log_{\widetilde{E},P}(x))=\log_{\widetilde{E},P}(\exp_{\widetilde{E},P}(x))=x. The same goes without the variable z , and the same goes for \widetilde{E} (resp. E ) over \Omega_{z}^{+} and \mathcal{D}_{z}^{+} (resp. over \Omega^{+} and \mathcal{D}^{+} ). 4.3. Evaluation at z=\zeta\in\overline{\mathbb{F}}_{q} : the P -adic setting Consider \mathbb{F}_{P} to be the residual field associated with P . Set \mathbb{F}=\mathbb{F}_{q}(\zeta)\cap\mathbb{F}_{P} and G=\operatorname{Gal}(\mathbb{F}/\mathbb{F}_{q}) . Let us first remark that the valuation w defined in 3.4 is not a valuation over \mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}K_{P} . We have an isomorphism of \mathbb{F} -vector spaces: \mathbb{F}\otimes_{\mathbb{F}_{q}}\mathbb{F}\simeq{\prod}\limits_{g\in G}\mathbb{F}\simeq{\prod}\limits_{g\in G}\left(\mathbb{F}\otimes_{\mathbb{F}}\mathbb{F}\right) given by \eta:x\otimes y\mapsto\left(g(x)y,g\in G\right). In particular, through this isomorphism, the Frobenius \tau_{\zeta} is identified with (\operatorname{id}\otimes\tau,\ldots,\operatorname{id}\otimes\tau) . First, we extend the scalars from \mathbb{F} to \mathbb{F}_{q}(\zeta) . We obtain a (canonical) isomorphism (of \mathbb{F}_{q}(\zeta) -vector spaces) \eta^{\prime}:\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}\mathbb{F}\rightarrow\prod\limits_{g\in G}\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}}\mathbb{F} , given by the following. Let (f_{1},\ldots,f_{l}) be an \mathbb{F} -basis of \mathbb{F}_{q}(\zeta) and a_{1},\ldots,a_{l}\in\mathbb{F} . We set \eta^{\prime}\left({\sum}\limits_{i=1}^{l}a_{i}f_{i}\otimes_{\mathbb{F}_{q}}x_{i}\right)=\left({\sum}\limits_{i=1}^{l}g(a_{i})f_{i}\otimes_{\mathbb{F}}x_{i},g\in G\right). Note that the isomorphism is canonical, but not the topologies that will appear. We then naturally extend the scalars (on the right) from \mathbb{F} to L_{P} . We obtain an isomorphism (of \mathbb{F}_{q}(\zeta) -vector spaces on the left and L_{P} -modules on the right) induced by \eta , also denoted \eta : \eta:\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}L_{P}\simeq{\prod}\limits_{g\in G}\left(\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}}L_{P}\right). In particular we obtain L_{P} -vector spaces of dimension [\mathbb{F}_{q}(\zeta):\mathbb{F}] on each component of the product, so an L_{P} -vector space of dimension [\mathbb{F}_{q}(\zeta):\mathbb{F}_{q}] . For g\in G , we set H_{g}=\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}}K_{P}\simeq\mathbb{F}_{q}(\zeta)((P)) (where the action on the left of \mathbb{F} is determined by g\in G ). For x={\sum}\limits_{i=1}^{l}f_{i}\otimes x_{i}\in H_{g} , we consider the usual valuation on H_{g} : v_{g}(x)=\min_{i=1,\ldots,m}v_{P}(x_{i}). For all g\in G we provide L\otimes_{K}H_{g} with the topology v_{g} induced by its structure of H_{g} -vector space of finite dimension with respect to the choice of the basis \mathscr{B} of L/K . In particular, if we set \operatorname{pr}_{g} the projection on the g -component of the product, then we obtain v_{g}\left(\operatorname{pr}_{g}\left(\eta\left((\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}}\mathscr{O}_{L})\right)\right)\right)\geq 0 . Let v_{P} be the over-additive valuation on the product {\prod}_{g\in G}(L\otimes_{K}H_{g}) : v_{P}(x_{g},g\in G)=\min_{g\in G}(v_{g}(x_{g})) verifying v_{P}(\eta(1\otimes P))=1 . Remark that the Frobenius \tau_{\zeta} is equal to (\operatorname{id}\otimes\tau,\ldots,\operatorname{id}\otimes\tau) on {\prod}\limits_{g\in G}\left(\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}}L_{P}\right) . Remark 4.8 . Following exactly the same ideas, by extending the scalars from \mathbb{F} to \mathbb{T}_{z}(L_{P}) or \widetilde{L_{P}} we obtain the isomorphisms of \mathbb{F}_{q}(\zeta) -vector spaces on the left and \widetilde{L_{P}} (resp. \mathbb{T}_{z}(L_{P}) ) on the right: \eta_{z}:\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z}(L_{P})\simeq{\prod}\limits_{g\in G}\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}}\mathbb{T}_{z}(L_{P}) and \mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}\widetilde{L_{P}}\simeq{\prod}\limits_{g\in G}\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}}\widetilde{L_{P}}. We are now interested in the case of the higher dimension d . We extend v_{P} onto (\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}L_{P})^{d} (the same goes with z ) (topology of finite-dimensional vector spaces, for example with respect to the canonical basis). Then set \Omega_{\zeta,d}=\{x\in(\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}L_{P})^{d}\ |\ v_{P}(x)\geq 0\}\supseteq(\mathbb{F}_{q}(\zeta)\otimes\mathscr{O}_{L})^{d} and \Omega_{\zeta,d,z}=\{x\in(\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z}(L_{P}))^{d}\ |\ v_{P}(x)\geq 0\}\supseteq(\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}\mathscr{O}_{L}[z])^{d}. Proposition 4.9 . We have the following convergences: \log_{F^{(\zeta)},P}:\Omega_{\zeta,d}\rightarrow(\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}L_{P})^{d} and \log_{F^{(\zeta)},P}:\Omega_{\zeta,d,z}\rightarrow(\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z}(L_{P}))^{d}. Proof. It follows from Proposition 4.5 and the definitions of the objects. We omit the proof. ∎ 4.4. The P -adic L -series Recall that m=d[L:K] where d in the dimension of the t -module E and that F is the t -module given by F=P^{-1}EP . Let \mathscr{C}=(g_{1},\ldots,g_{m}) be an A -basis of \operatorname{Lie}_{E}(\mathscr{O}_{L}) , it is also a \widetilde{K_{\infty}} -basis of \operatorname{Lie}_{\widetilde{E}}(L_{\infty}) , a \widetilde{K_{P}} -basis of \operatorname{Lie}_{\widetilde{E}}(L_{P}) and a \mathbb{T}_{z}(K_{P}) -basis of \operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{P})) . The same goes by replacing E by F since \partial_{E}=\partial_{F} . Let us remark, from Corollary 4.6 , that for any z -unit y(z)\in U(\widetilde{F},\mathscr{O}_{L}[z]) we have \exp_{\widetilde{F}}(y(z))\in\widetilde{E}(\mathscr{O}_{L}[z])\subseteq\Omega_{z} and therefore \log_{\widetilde{F},P}(\exp_{\widetilde{F}}(y(z)))\in\operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{P})). Moreover, for a family (x_{1}(z),\ldots,x_{m}(z)) of elements of \operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{P})) we have \operatorname{Mat}_{\mathscr{C}}(x_{1}(z),\ldots,x_{m}(z))\in M_{m}(\mathbb{T}_{z}(K_{P})) thus \operatorname{det}_{\mathscr{C}}(x_{1}(z),\ldots,x_{m}(z))\in\mathbb{T}_{z}(K_{P}). Next, formally in (L[[z]])^{d} we have the following equality for all f(z)\in(L[[z]])^{d} : \log_{\widetilde{F},P}(\exp_{\widetilde{F}}(f(z)))=f(z). Let \left(v_{1}(z),\ldots,v_{m}(z)\right)\subset U(\widetilde{F};\mathscr{O}_{L}[z]) be an \widetilde{A} -basis of U(\widetilde{F};\widetilde{\mathscr{O}_{L}}). Remark that the family (1\otimes v_{1}(z),\ldots,1\otimes v_{m}(z))\subseteq\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z}(L_{\infty})^{d} is also an \widetilde{A_{\zeta}} -basis of U(\widetilde{F}^{(\zeta)};\widetilde{M}) . Set w(z)=\det_{\mathscr{C}}\left(v_{1}(z),\ldots,v_{m}(z)\right)\in\mathbb{T}_{z}(K_{\infty}) and w_{P}(z)=\det_{\mathscr{C}}\left(\log_{\widetilde{F},P}(\exp_{\widetilde{F}}(v_{1}(z))),\ldots,\log_{\widetilde{F},P}(\exp_{\widetilde{F}}(v_{m}(z)))\right)\in\mathbb{T}_{z}(K_{P}). By the above discussions and the class formula, we have the following equality in K[[z]] : L_{P}(\widetilde{F}/\widetilde{\mathscr{O}_{L}})=\dfrac{w_{P}(z)}{\operatorname{sgn}(w(z))}. Since w_{P}(z)\in\mathbb{T}_{z}(K_{P}) , to study the P -adic convergence we want to prove that \operatorname{sgn}(w(z)) divides w_{P}(z) in \mathbb{T}_{z}(K_{P}) . Remark that the possible P -adic poles are the zeros of \operatorname{sgn}(w(z))\in\mathbb{F}_{q}[z] hence elements of \overline{\mathbb{F}}_{q} . We will prove that the meromorphic series \dfrac{w_{P}(z)}{\operatorname{sgn}(w(z))} does not have a pole in \overline{\mathbb{F}}_{q} . Theorem 4.10 . The meromorphic series \dfrac{w_{P}(z)}{\operatorname{sgn}(w(z))} does not have a pole in \overline{\mathbb{F}}_{q} . In other words, we have the convergence \dfrac{w_{P}(z)}{\operatorname{sgn}(w(z))}\in\mathbb{T}_{z}(K_{P}) . Proof. Let \zeta\in\overline{\mathbb{F}}_{q} be a root of \operatorname{sgn}(w(z)) . Recall that \mathscr{C}_{\zeta}=(1\otimes g_{1},\ldots,1\otimes g_{m})\subseteq\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}\mathscr{O}_{L} if \mathscr{C}=(g_{1},\ldots,g_{m}) . Then \operatorname{Lie}_{F_{\zeta}}(\mathbb{F}_{q}(\zeta)\otimes_{\mathbb{F}_{q}}\mathscr{O}_{L}) is an A_{\zeta} -lattice in M_{\infty} and admits \mathscr{C}_{\zeta} as an A_{\zeta} -basis. Consider (w_{1},\ldots,w_{m}) an A_{\zeta} -basis of U_{\zeta}(F;\mathscr{O}_{L})=\operatorname{ev}_{\zeta}U(F^{(\zeta)};\widetilde{M}) and (w_{1}(z),\ldots,w_{m}(z))\subseteq U(F^{(\zeta)};\widetilde{M}) be such that \operatorname{ev}_{\zeta}w_{i}(z)=w_{i} for i=1,\ldots,m . Set W^{\prime}(z)=\det_{\mathscr{C}_{\zeta}}(w_{1}(z),\ldots,w_{m}(z))\in\widetilde{M}_{\infty}\backslash(z-\zeta)\widetilde{M}_{\infty} and W_{P}^{\prime}(z)=\det_{\mathscr{C}_{\zeta}}\left(\log_{\widetilde{F}^{(\zeta)},P}(\exp_{\widetilde{F}^{(\zeta)}}(w_{1}(z))),\ldots,\log_{\widetilde{F}^{(\zeta)},P}(\exp_{\widetilde{F}^{(\zeta)}}(w_{m}(z)))\right)\in\widetilde{M}_{v}. Recall that the family (1\otimes v_{1}(z),\ldots,1\otimes v_{m}(z)) is an \widetilde{A}_{\zeta} -basis of U(\widetilde{F}^{(\zeta)};\widetilde{M}) . Let us set W(z)=\det_{\mathscr{C}_{\zeta}}(1\otimes v_{1}(z),\ldots,1\otimes v_{m}(z))=1\otimes w(z)\in\widetilde{M}_{\infty}, W_{P}(z)=1\otimes w_{P}(z)\in\widetilde{M}_{v}, and \Delta=\det_{(1\otimes v_{1}(z),\ldots,1\otimes v_{m}(z))}(w_{1}(z),\ldots,w_{m}(z))\in\widetilde{A}_{\zeta}. From the equality W^{\prime}(z)=\Delta W(z) we obtain 1\otimes\underbrace{L(\widetilde{F}/\widetilde{\mathscr{O}_{L}})}_{\in\mathbb{T}_{z}(K_{\infty})}=1\otimes\dfrac{w(z)}{\operatorname{sgn}(w(z))}=\dfrac{1\otimes w(z)}{1\otimes\operatorname{sgn}(w(z))}=\dfrac{W(z)}{1\otimes\operatorname{sgn}(w(z))}=\dfrac{W^{\prime}(z)}{\Delta(1\otimes\operatorname{sgn}(w(z))}. Since 1\otimes L(\widetilde{F}/\widetilde{\mathscr{O}_{L}}) does not have a pole in \overline{\mathbb{F}}_{q} and W^{\prime}(z) is not divisible by z-\zeta , we obtain that \Delta(1\otimes\operatorname{sgn}(w(z)) is not divisible by z-\zeta . From the equality \dfrac{W_{P}(z)}{1\otimes\operatorname{sgn}(w(z))}=\dfrac{W_{P}^{\prime}(z)}{\Delta(1\otimes\operatorname{sgn}(w(z))} we can evaluate at z=\zeta so \zeta is not a pole of \dfrac{w_{P}(z)}{\operatorname{sgn}(w(z))} . Finally, the P -adic L -series is a meromorphic series without any pole in \overline{\mathbb{F}}_{q} : it is an element of \mathbb{T}_{z}(K_{P}) . ∎ Corollary 4.11 ( P -adic L -series for P^{-1}\widetilde{E}P ) . Consider the t -module F=P^{-1}EP . Consider (v_{1}(z),\ldots,v_{r}(z))\subseteq U(\widetilde{F},\mathscr{O}_{L}[z]) an \widetilde{A} -basis of U(\widetilde{F};\widetilde{\mathscr{O}_{L}}) , \mathscr{C} an \widetilde{A} -basis of \operatorname{Lie}_{\widetilde{F}}(\widetilde{\mathscr{O}_{L}}) . Then the following product converges in \mathbb{T}_{z}(K_{P}) L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})={\prod}\limits_{Q\neq P}\dfrac{\left[\operatorname{Lie}_{\widetilde{E}}(\widetilde{\mathscr{O}_{L}}/Q\widetilde{\mathscr{O}_{L}})\right]_{\widetilde{A}}}{\left[E(\widetilde{\mathscr{O}_{L}}/Q\widetilde{\mathscr{O}_{L}})\right]_{\widetilde{A}}} where the product runs over all the monic irreducible polynomials Q of A different from P . Further, we have the equality: L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})=\dfrac{\det_{\mathscr{C}}\left(\log_{\widetilde{F},P}\exp_{\widetilde{F}}v_{1}(z),\ldots,\log_{\widetilde{F},P}\exp_{\widetilde{F}}v_{m}(z)\right)}{\operatorname{sgn}\left(\det_{\mathscr{C}}\left(v_{1}(z),\ldots,v_{m}(z)\right)\right)}. We can then define the P -adic L -series associated with E and \mathscr{O}_{L} : L_{P}(E/\mathscr{O}_{L})=\operatorname{ev}_{z=1,P}L_{P}(\widetilde{E},\widetilde{\mathscr{O}_{L}})={\prod}\limits_{Q\neq P}\dfrac{\left[\operatorname{Lie}_{E}(\mathscr{O}_{L}/Q\mathscr{O}_{L})\right]_{A}}{\left[E(\mathscr{O}_{L}/Q\mathscr{O}_{L})\right]_{A}}\in K_{P}. We also have the following equalities from [ 4 , Proposition 3.3] : L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})={\prod}\limits_{\mathfrak{P}\nmid P}\dfrac{\left[\operatorname{Lie}_{\widetilde{E}}(\widetilde{\mathscr{O}_{L}}/\mathfrak{P}\widetilde{\mathscr{O}_{L}})\right]_{\widetilde{A}}}{\left[E(\widetilde{\mathscr{O}_{L}}/\mathfrak{P}\widetilde{\mathscr{O}_{L}})\right]_{\widetilde{A}}}\in\mathbb{T}_{z}(K_{P}) where the product runs over all the primes of \mathscr{O}_{L} not dividing P , and L_{P}(E/\mathscr{O}_{L})={\prod}\limits_{\mathfrak{P}\nmid P}\dfrac{\left[\operatorname{Lie}_{E}(\mathscr{O}_{L}/\mathfrak{P}\mathscr{O}_{L})\right]_{A}}{\left[E(\mathscr{O}_{L}/\mathfrak{P}\mathscr{O}_{L})\right]_{A}}\in K_{P}. 4.5. A P -adic class formula associated with the t -module P^{-1}EP The next step is to introduce a P -adic regulator and obtain a P -adic class formula. We begin with P -twisted t -modules. Recall that \mathscr{C} is a fixed {A} -basis of \operatorname{Lie}_{{E}}({\mathscr{O}_{L}})=\operatorname{Lie}_{F}(\mathscr{O}_{L}) . Definition 4.12 . Consider V\subseteq U(F;\mathscr{O}_{L}) a sub- A -lattice and let (v_{1},\ldots,v_{m}) be an A -basis of V . Then we define the P -adic regulator associated with V by R_{P}(V)=\dfrac{\det_{\mathscr{C}}\left(\log_{F,P}(\exp_{F}(v_{1})),\ldots,\log_{F,P}(\exp_{F}(v_{m}))\right)}{\operatorname{sgn}(\det_{\mathscr{C}}(v_{1},\ldots,v_{m}))}\in K_{P} which is independent of the choice of the basis of V and of \operatorname{Lie}_{F}(\mathscr{O}_{L}) . Theorem 4.13 . [ P -adic class formula for P^{-1}EP ] We have the following equality in K_{P} : L_{P}(E/\mathscr{O}_{L})=R_{P}(U(F;\mathscr{O}_{L}))\left[H(F;\mathscr{O}_{L})\right]_{A}=R_{P}(U_{\operatorname{st}}(F;\mathscr{O}_{L})). Proof. We use notation as in the proof of Theorem 4.10 with \zeta=1 . In particular (v_{1},\ldots,v_{m}) is an A -basis of U(F;\mathscr{O}_{L}) . Consider (u_{1},\ldots,u_{m}) an A -basis of U_{\operatorname{st}}(F;\mathscr{O}_{L}) . Denote by b_{i}=\exp_{F}(u_{i})\in F(\mathscr{O}_{L}) for i=1,\ldots,m and by a_{i}=\exp_{F}(v_{i})\in F(\mathscr{O}_{L}) for i=1,\ldots,m . We have L(\widetilde{F}/\widetilde{\mathscr{O}_{L}})=\dfrac{w^{\prime}(z)}{f(z)\Delta}. As the L -series L(F/\mathscr{O}_{L}) (at infinity) is equal to \dfrac{\det_{\mathscr{C}}(u_{1},\ldots,u_{m})}{\operatorname{sgn}(\det_{\mathscr{C}}(u_{1},\ldots,u_{m}))}=\dfrac{\operatorname{ev}_{z=1}(w^{\prime}(z))}{\operatorname{sgn}(\det_{\mathscr{C}}(u_{1},\ldots,u_{m}))}=\operatorname{ev}_{z=1}\dfrac{w^{\prime}(z)}{f(z)\Delta} we first have (5) \operatorname{ev}_{z=1}(f(z)\Delta)=\operatorname{sgn}(\det_{\mathscr{C}}(u_{1},\ldots,u_{m}))\in\mathbb{F}_{q}^{\ast}. Let us consider P_{1}=\operatorname{Mat}_{(v_{1},\ldots,v_{m})}(u_{1},\ldots,u_{m})\in M_{m}(A) . By [ 7 , Theorem 1] we have \dfrac{\det(P_{1})}{\operatorname{sgn}(\det(P_{1}))}=\left[H(F;\mathscr{O}_{L})\right]_{A} . Moreover, we have: P_{1}=\operatorname{Mat}_{(\log_{F,P}(a_{1}),\ldots,\log_{F,P}(a_{m}))}(\log_{F,P}(b_{1}),\ldots,\log_{F,P}(b_{m})). Then we have in K_{P} : (6) \det(P_{1}){\det_{\mathscr{C}}(\log_{F,P}(a_{1}),\ldots,\log_{F,P}(a_{m})}={\det_{\mathscr{C}}(\log_{F,P}(b_{1}),\ldots,\log_{F,P}(b_{m})}). From the following equality in K_{\infty} : \det(P_{1})\det_{\mathscr{C}}(v_{1},\ldots,v_{m})=\det_{\mathscr{C}}(u_{1},\ldots,u_{m}) we deduce by comparing signs: (7) \operatorname{sgn}(\det(P_{1}))\operatorname{sgn}(\det_{\mathscr{C}}(v_{1},\ldots,v_{m}))=\operatorname{sgn}(\det_{\mathscr{C}}(u_{1},\ldots,u_{m})). We finally have: \displaystyle L_{P}(E/\mathscr{O}_{L}) \displaystyle=\operatorname{ev}_{z=1,P}\left(\dfrac{w^{\prime}_{P}(z)}{f(z)\Delta}\right) \displaystyle=\dfrac{\det_{\mathscr{C}}(\log_{F,P}(b_{1}),\ldots,\log_{F,P}(b_{m}))}{\operatorname{sgn}(\det_{\mathscr{C}}(u_{1},\ldots,u_{m}))}\text{ by Equality \eqref{eq:sgn} }, \displaystyle=\det(P_{1})\dfrac{\det_{\mathscr{C}}(\log_{F,P}(a_{1}),\ldots,\log_{F,P}(a_{m}))}{\operatorname{sgn}(\det_{\mathscr{C}}(u_{1},\ldots,u_{m}))}\text{ by Equality }\eqref{r1}, \displaystyle=\dfrac{\det(P_{1})}{\operatorname{sgn}(\det(P_{1}))}\dfrac{\det_{\mathscr{C}}(\log_{F,P}(a_{1}),\ldots,\log_{F,P}(a_{m}))}{\operatorname{sgn}(\det_{\mathscr{C}}(v_{1},\ldots,v_{m}))}\text{ by Equality }\eqref{r2}, \displaystyle=R_{P}(U(F;\mathscr{O}_{L}))\left[H(F;\mathscr{O}_{L})\right]_{A} and the second line equals R_{P}(U_{\operatorname{st}}(E;\mathscr{O}_{L})) . ∎ 4.6. A P -adic class formula So far we’ve worked mainly with the t -module F , and now we want to link everything to the t -module E . Set h(z)=\left[\widetilde{E}(\widetilde{\mathscr{O}_{L}}/P\widetilde{\mathscr{O}_{L}})\right]_{\widetilde{A}}\in A[z] and h(1)=\left[E(\mathscr{O}_{L}/P\mathscr{O}_{L})\right]_{A}\in A\backslash\{0\} . Consider s\in\mathbb{N} such that \partial_{\widetilde{E}}(h(z)^{q^{s}})=h(z)^{q^{s}}I_{d} (e.g., s such that q^{s}\geq d ) and denote by g(z)=h(z)^{q^{s}}\in A[z] . By Proposition 2.1 , we have for all b(z)\in\mathscr{O}_{L}[z]^{d} : \widetilde{E}_{g(z)}(b(z))\in P\mathscr{O}_{L}[z]^{d} and for all b\in\mathscr{O}_{L}^{d} : E_{g(1)}(b)\in P\mathscr{O}_{L}^{d}. Then by Corollary 4.6 and the above discussion, we can define the following maps: \displaystyle\operatorname{Log}_{\widetilde{E},P}:\Omega_{z} \displaystyle\rightarrow\dfrac{1}{g(z)}\mathbb{T}_{z}(L_{P})^{d} \displaystyle x \displaystyle\mapsto\dfrac{1}{g(z)}\log_{\widetilde{E},P}(\widetilde{E}_{g(z)}(x)) and \begin{aligned} \operatorname{Log}_{{E},P}:\Omega&\rightarrow L_{P}^{d}\\ x&\mapsto\dfrac{1}{g(1)}\log_{{E},P}({E}_{g(1)}(x))\end{aligned}. Moreover, if x\in\Omega_{z}^{+} , we have \log_{\widetilde{E},P}(\widetilde{E}_{g(z)}(x))=g(z)\log_{\widetilde{E},P}(x) in \mathbb{T}_{z}(L_{P})^{d} . We obtain the following equality in \mathbb{T}_{z}(L_{P})^{d} for such x : \log_{\widetilde{E},P}(x)=\operatorname{Log}_{\widetilde{E},P}(x) thus the map \operatorname{Log}_{\widetilde{E},P} extends the map \log_{\widetilde{E},P} from \Omega_{z}^{+} to \Omega_{z} . The same applies without z . Lemma 4.14 . We have the following properties. (1) For all a\in{A}[z] and x\in\Omega_{z} we have the following equality in \dfrac{1}{g(z)}\mathbb{T}_{z}(L_{P})^{d} : \partial_{\widetilde{E}}(a)\operatorname{Log}_{\widetilde{E},P}(x)=\operatorname{Log}_{\widetilde{E},P}(\widetilde{E}_{a}(x)). (2) For all x\in\Omega_{z} we have the equality in L_{P}^{d} : \operatorname{ev}_{z=1,P}(\operatorname{Log}_{\widetilde{E},P}(x))=\operatorname{Log}_{E,P}(\operatorname{ev}_{z=1,P}(x)). (3) For all a\in A and x\in\Omega we have the following equality in L_{P}^{d} : \partial_{E}(a)\operatorname{Log}_{E,P}(x)=\operatorname{Log}_{E,P}(E_{a}(x)). Proof. (1) We have the following equalities in \mathbb{T}_{z}(L_{P})^{d} for all x\in\Omega_{z} and a\in A[z] : \displaystyle g(z)\operatorname{Log}_{\widetilde{E},P}(\widetilde{E}_{a}(x))=\log_{\widetilde{E},P}(\underbrace{\widetilde{E}_{g(z)}(\widetilde{E}_{a}(x))}_{v_{P}>0}) \displaystyle=\partial_{\widetilde{E}}(a)\log_{\widetilde{E},P}(\widetilde{E}_{g(z)}(x)) \displaystyle=\partial_{\widetilde{E}}(a)g(z)\operatorname{Log}_{\widetilde{E},P}(x). (2) We have the following equality in \Omega^{+} for all x\in\Omega_{z} : \operatorname{ev}_{z=1,P}(\widetilde{E}_{g(z)}(x))=E_{g(1)}(\operatorname{ev}_{z=1,P}(x)). Then we have the following equalities: \displaystyle\operatorname{ev}_{z=1,P}(g(z)\operatorname{Log}_{\widetilde{E},P}(x))=\operatorname{ev}_{z=1,P}(\log_{\widetilde{E},P}\underbrace{(\widetilde{E}_{g(z)}(x)))}_{v_{P}>0} \displaystyle=\log_{E,P}(E_{g(1)}(\operatorname{ev}_{z=1,P}(x))) \displaystyle=g(1)\operatorname{Log}_{E,P}(\operatorname{ev}_{z=1,P}(x)). (3) The proof is similar as for the first assertion. ∎ Proposition 4.15 . The logarithm map \log_{\widetilde{E},P} is injective on \Omega_{z}^{+} . We begin with the following lemma. Lemma 4.16 . For all a\in A\backslash\{0\} and for all x\in\mathbb{T}_{z}(L_{P})^{d}\backslash\{0\} we have \widetilde{E}_{a}(x)\neq 0 . Proof. Fix a\in A\backslash\{0\} and x\in\mathbb{T}_{z}(K_{P})^{d} such that E_{a}(x)=0 . We can suppose without loss of generality that \partial_{E}(a)=aI_{d} , even if it means applying \widetilde{E}_{a^{q^{s}-1}} to \widetilde{E}_{a}(x) . We can also assume that z does not divide x in \mathbb{T}_{z}(L_{P})^{d} . Denote by \widetilde{E}_{a}=a+{\sum}\limits_{i=1}^{r_{a}}\widetilde{E}_{a,i}\tau^{i} with \widetilde{E}_{a,i}\in zM_{d}(\mathscr{O}_{L}[z]) for all i=1,\ldots,r_{a} , and x={\sum}\limits_{n\geq 0}y_{n}z^{n} with a_{n}\in L_{P}^{d} and y_{0}\neq 0 . We have E_{a}(x)=ay_{0}\text{ mod }z\mathbb{T}_{z}(L_{P})^{d}\neq 0\text{ mod }z\mathbb{T}_{z}(L_{P})^{d} . Hence E_{a}(x)\neq 0 . ∎ Proof of Proposition 4.15 . Let x be in \Omega_{z}^{+} such that \log_{\widetilde{E},P}(x)=0 . Then for all s\in\mathbb{N} we have \partial_{\widetilde{E}}(P^{q^{s}})\log_{\widetilde{E},P}(x)=0=\log_{\widetilde{E},P}(\widetilde{E}_{P^{q^{s}}}(x)). Since v_{P}(x)>0 , we consider s\in\mathbb{N} large enough such that {\widetilde{E}}_{P^{q^{s}}}(x) belongs to \mathcal{D}_{z}^{+} . For such an integer s we obtain: \widetilde{E}_{P^{q^{s}}}(x)=0 which implies that x=0 by Lemma 4.16 . ∎ Proposition 4.17 . The kernel of \operatorname{Log}_{E,P}:\mathscr{O}_{L}^{d}\rightarrow L_{P}^{d} consists exactly of the torsion points of E(\mathscr{O}_{L}) , denoted by E(\mathscr{O}_{L})_{\operatorname{Tors}} . Proof. Consider first x\in\mathscr{O}_{L}^{d} such that \operatorname{Log}_{E,P}(x)=0 . Then we have \log_{E,P}(E_{g(1)}(x))=0. Thus for all n\geq 0 we have: \log_{E,P}(E_{P^{q^{n}}g(1)}(x))=\partial_{E}(P^{q^{n}})\log_{E,P}(E_{g(1)}(x))=0. Since v_{P}(E_{g(1)}(x))>0 , we can consider n large enough such that E_{P^{q^{n}}g(1)}(x)\in\mathcal{D}^{+} . Then we find by applying the exponentiel map \exp_{E,P} to \log_{E,P}(E_{P^{q^{n}}g(1)}(x)) that 0=E_{P^{q^{n}g(1)}}(x) so x is a torsion point. Conversely, suppose that there is a non-zero polynomial a\in A such that E_{a}(x)=0 . We also have E_{a^{q^{s}}}(x)=0 . Then we have a^{q^{s}}\log_{E,P}(E_{g(1)}(x))=\log_{E,P}(E_{a^{q^{s}}g(1)}(x))=\log_{E,P}(E_{g(1)}(E_{a^{q^{s}}}(x)))=0. Since a is non-zero, we obtain \log_{E,P}(E_{g(1)}(x))=0 . ∎ Set \mathscr{O}_{{L},P}=\mathscr{O}_{L}\otimes_{A}{A}_{P} and \widetilde{\mathscr{O}_{{L},P}}=\widetilde{\mathscr{O}_{L}}\otimes_{A}A_{P} . By [ 7 , Lemma 3.21] , we can extend E by continuity to a homomorphism of \mathbb{F}_{q} -algebras E:A_{P}\rightarrow M_{d}(\mathscr{O}_{{L},P})\{\{\tau\}\}. We can also extend \widetilde{E} by continuity to a homomorphism of \mathbb{F}_{q}(z) -algebras \widetilde{E}:\widetilde{A}_{P}\rightarrow M_{d}(\widetilde{\mathscr{O}_{{L},P}})\{\{\tau\}\}. In particular, the A -module E(P\mathscr{O}_{L,P}) inherits a structure of A_{P} -module. Set U^{\prime}=g(1)U(E;\mathscr{O}_{L}) and U^{\prime}_{z}=g(z)U(\widetilde{E};\widetilde{\mathscr{O}_{L}}) (the multiplication is of course in \operatorname{Lie}_{E}(L_{\infty}) (resp. \operatorname{Lie}_{\widetilde{E}}(\widetilde{L_{\infty}}) ). Lemma 4.18 . We have the following properties. (1) We have that U^{\prime} and U_{z}^{\prime} are sub-lattices of U(E;\mathscr{O}_{L}) and U(\widetilde{E};\widetilde{\mathscr{O}_{L}}) respectively. (2) We have that U^{\prime} and U_{z}^{\prime} are sub-lattices of U(F;\mathscr{O}_{L}) and U(\widetilde{F};\widetilde{\mathscr{O}_{L}}) respectively. Proof. (1) The first point is clear. (2) We have to prove the inclusions U^{\prime}\subseteq U(F;\mathscr{O}_{L}) and U_{z}^{\prime}\subseteq U(\widetilde{F};\widetilde{\mathscr{O}_{L}}). We do it for F , the same arguments apply to \widetilde{F} . Let x\in\operatorname{Lie}_{E}(L_{\infty}) be such that \exp_{E}(x)\in\mathscr{O}_{L}^{d} , then: \exp_{F}(g(1)x)=P^{-1}\exp_{E}(P\partial_{E}(g(1))x)=P^{-1}E_{g(1)}\left({\exp_{E}(Px)}\right)\in P^{-1}.P\mathscr{O}_{L}^{d}=\mathscr{O}_{L}^{d}. ∎ By Lemma 4.18 , U^{\prime} (resp. U_{z}^{\prime} ) is a common sub- A -lattice (resp. sub- \widetilde{A} -lattice) for U(E;\mathscr{O}_{L}) and U(F;\mathscr{O}_{L}) (resp. U(\widetilde{E};\widetilde{\mathscr{O}_{L}}) and U(\widetilde{F};\widetilde{\mathscr{O}_{L}}) ). We then have: \left[U(F;\mathscr{O}_{L}):U(E;\mathscr{O}_{L})\right]_{A}=\left[U(F;\mathscr{O}_{L}):U^{\prime}\right]_{A}\left[U^{\prime}:U(E;\mathscr{O}_{L})\right]_{A}=\dfrac{\left[U(F;\mathscr{O}_{L}):U^{\prime}\right]_{A}}{\left[U(E;\mathscr{O}_{L}):U^{\prime}\right]_{A}}\in K^{\ast} and \left[U(\widetilde{F};\widetilde{\mathscr{O}_{L}}):U(\widetilde{E};\widetilde{\mathscr{O}_{L}})\right]_{\widetilde{A}}=\left[U(\widetilde{F};\widetilde{\mathscr{O}_{L}}):U_{z}^{\prime}\right]_{\widetilde{A}}\left[U_{z}^{\prime}:U(\widetilde{E};\widetilde{\mathscr{O}_{L}})\right]_{\widetilde{A}}=\dfrac{\left[U(\widetilde{F};\widetilde{\mathscr{O}_{L}}):U_{z}^{\prime}\right]_{\widetilde{A}}}{\left[U(\widetilde{E};\widetilde{\mathscr{O}_{L}}):U_{z}^{\prime}\right]_{\widetilde{A}}}\in\widetilde{K}^{\ast}. Let us define P -adic regulators associated to U^{\prime} as follows. Let (w_{1},\ldots,w_{m}) be an A -basis of U^{\prime} . We set: R_{P,E}(U^{\prime})=\dfrac{\det_{\mathscr{C}}\left(\operatorname{Log}_{E,P}(\exp_{E}(w_{1})),\ldots,\operatorname{Log}_{E,P}(\exp_{E}(w_{m}))\right)}{\operatorname{sgn}(\det_{\mathscr{C}}(w_{1},\ldots,w_{m}))}\in K_{P} and R_{P,F}(U^{\prime})=\dfrac{\det_{\mathscr{C}}\left(\log_{F,P}(\exp_{F}(w_{1})),\ldots,\log_{F,P}(\exp_{F}(w_{m}))\right)}{\operatorname{sgn}(\det_{\mathscr{C}}(w_{1},\ldots,w_{m}))}\in K_{P}. These regulators do not depend of the choice of the basis (w_{1},\ldots,w_{m}) . We can also define a P -adic regulator R_{P}(U_{\operatorname{st}}(E;\mathscr{O}_{L})) for U_{\operatorname{St}}(E;\mathscr{O}_{L}) and we have the following equality from the proof of Theorem 4.13 : (8) R_{P}(U(E;\mathscr{O}_{L}))\left[H(E;\mathscr{O}_{L})\right]_{A}=R_{P}(U_{\operatorname{St}}(E;\mathscr{O}_{L})). Similarly, we define the P -adic regulators R_{P,\widetilde{E}}(U_{z}^{\prime}) and R_{P,\widetilde{F}}(U_{z}^{\prime}) associated with U_{z}^{\prime} that are elements in \mathbb{T}_{z}(K_{P}) from Theorem 4.10 . Lemma 4.19 . We have the following equalities: (9) R_{P,\widetilde{E}}(U_{z}^{\prime})=R_{P,\widetilde{F}}(U_{z}^{\prime})\text{ and }R_{P,{E}}(U^{\prime})=R_{P,{F}}(U^{\prime}). Proof. We prove the first equality. We want to prove that for all u(z)\in U(\widetilde{E};\mathscr{O}_{L}[z]) we have the following equality in \mathbb{T}_{z}(L_{P})^{d} : \log_{\widetilde{F},P}(\exp_{\widetilde{F}}(u(z)))=\operatorname{Log}_{\widetilde{E},P}(\exp_{\widetilde{E}}(u(z))). Formally (i.e., in L[[z]]^{d}) it is true, and the two quantities belong to \mathbb{T}_{z}(L_{P})^{d} . Then the first equality of the lemma is clear. To prove the second equality, consider (u_{1},\ldots,u_{m}) an A -basis of U_{\operatorname{St}}(E;\mathscr{O}_{L}) .From the first equality of the lemma, we have the following equality in \mathbb{T}_{z}(K_{P}) : \det_{\mathscr{C}}(\log_{\widetilde{F},P}(\exp_{\widetilde{F}}(g(z)u_{i}(z))),i=1,\ldots,m)=\det_{\mathscr{C}}(\operatorname{Log}_{\widetilde{E},P}(\exp_{\widetilde{E}}(g(z)u_{i}(z))),i=1,\ldots,m). By evaluating at z=1 we obtain \det_{\mathscr{C}}(\log_{{F},P}(\exp_{{F}}(g(1)u_{i})),i=1,\ldots,m)=\det_{\mathscr{C}}(\operatorname{Log}_{{E},P}(\exp_{{E}}(g(1)u_{i})),i=1,\ldots,m). Consider now (v_{1},\ldots,v_{m}) an A -basis of U(E,\mathscr{O}_{L}) , then (g(1)v_{1},\ldots,g(1)v_{m}) is an A -basis of U^{\prime} . Write Q=\operatorname{Mat}_{(v_{1},\ldots,v_{m})}(u_{1},\ldots,u_{m})\in\operatorname{Gl}_{m}(A) . We then have \displaystyle\det_{\mathscr{C}}(\log_{F,P} \displaystyle(\exp_{F}(g(1)v_{i})),i=1,\ldots,m) \displaystyle=\det(Q)^{-1}\det_{\mathscr{C}}(\log_{F,P}(\exp_{F}(g(1)u_{i})),i=1,\ldots,m) \displaystyle=\det(Q)^{-1}\det_{\mathscr{C}}(\operatorname{Log}_{{E},P}(\exp_{{E}}(g(1)u_{i})),i=1,\ldots,m) \displaystyle=\det_{\mathscr{C}}(\operatorname{Log}_{E,P}(\exp_{E}(g(1)v_{i})),i=1,\ldots,m). We proved that the equality is true for one basis of U^{\prime} and since the P -adic regulators does not depend of the choice of the basis, the equality is true. ∎ Lemma 4.20 . (1) We have the following equalities: R_{P,\widetilde{F}}(U_{z}^{\prime})=R_{P}(U(\widetilde{F};\widetilde{\mathscr{O}_{L}}))\left[U(\widetilde{F};\widetilde{\mathscr{O}_{L}}):U_{z}^{\prime}\right]_{\widetilde{A}} and R_{P,\widetilde{E}}(U_{z}^{\prime})=R_{P}(U(\widetilde{E};\widetilde{\mathscr{O}_{L}}))\left[U(\widetilde{E};\widetilde{\mathscr{O}_{L}}):U_{z}^{\prime}\right]_{\widetilde{A}} (2) We have the following equalities: R_{P,{F}}(U^{\prime})=R_{P}(U({F};{\mathscr{O}_{L}}))\left[U({F};{\mathscr{O}_{L}}):U^{\prime}\right]_{{A}} and R_{P,{E}}(U^{\prime})=R_{P}(U({E};\widetilde{\mathscr{O}_{L}}))\left[U({E};{\mathscr{O}_{L}}):U^{\prime}\right]_{{A}} Proof. We only need to prove one of the equalities; all the others can be proven in a similar way. Let us prove the third one. By the structure theorem for finitely generated modules over a principal ideal domain, let us pick an A -basis (v_{1},\ldots,v_{m}) of U(F;\mathscr{O}_{L}) and a_{1},\ldots,a_{m}\in A such that (a_{1}v_{1},\ldots,a_{m}v_{m}) is an A -basis of U^{\prime} . Then \displaystyle R_{P,F}(U^{\prime}) \displaystyle=\dfrac{\det_{\mathscr{C}}(\log_{F,P}(\exp_{F}(a_{1}w_{1})),\ldots,\log_{F,P}(\exp_{F}(a_{m}w_{m})))}{\operatorname{sgn}\left(\det_{\mathscr{C}}(\log_{F,P}(\exp_{F}(a_{1}w_{1})),\ldots,\log_{F,P}(\exp_{F}(a_{m}w_{m})))\right)} \displaystyle=\dfrac{a_{1}\ldots a_{m}}{\operatorname{sgn}(a_{1}\ldots a_{m})}R_{P}(U(F;\mathscr{O}_{L})) \displaystyle=\left[U({F};{\mathscr{O}_{L}}):U^{\prime}\right]_{{A}}R_{P}(U(F;\mathscr{O}_{L})). ∎ The link between objects associated to F and those associated to E is contained in the local factor at P , and given by the following equalities from [ 4 , Lemma 3.4] : (10) z_{P}(E/\mathscr{O}_{L})=\left[U(F;\mathscr{O}_{L}):U(E;\mathscr{O}_{L})\right]_{A}\dfrac{\left[H(E;\mathscr{O}_{L})\right]_{A}}{\left[H(F;\mathscr{O}_{L})\right]_{A}}. and (11) z_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})=\left[U(\widetilde{F};\widetilde{\mathscr{O}_{L}}):U(\widetilde{E};\widetilde{\mathscr{O}_{L}})\right]_{\widetilde{A}}. We can state one of the main results of this work. Theorem 4.21 ( P -adic class formula) . We have the P -adic class formula for \widetilde{E} : z_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})=R_{P}(U(\widetilde{E};\widetilde{\mathscr{O}_{L}})) and the class formula for E : z_{P}(E/\mathscr{O}_{L})L_{P}(E/\mathscr{O}_{L})=R_{P}(U(E;\mathscr{O}_{L}))\left[H(E;\mathscr{O}_{L}))\right]_{A}=R_{P}(U_{\operatorname{st}}(E;\mathscr{O}_{L})). Proof. Let us start with the following equality from Corollary 4.11 : L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})=R_{P}(U(\widetilde{F};\widetilde{\mathscr{O}_{L}})). Then: \displaystyle z_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}}) \displaystyle=\dfrac{\left[U(\widetilde{F};\widetilde{\mathscr{O}_{L}}):U_{z}^{\prime}\right]_{\widetilde{A}}}{\left[U(\widetilde{E};\widetilde{\mathscr{O}_{L}}):U_{z}^{\prime}\right]_{\widetilde{A}}}R_{P}(U(\widetilde{F};\widetilde{\mathscr{O}_{L}}))\text{ by Equality }\eqref{facteurz}, \displaystyle=\dfrac{R_{P,\widetilde{F}}(U_{z}^{\prime})}{\left[U(\widetilde{E};\widetilde{\mathscr{O}_{L}}):U_{z}^{\prime}\right]_{\widetilde{A}}}\text{ by Lemma }\ref{reg}, \displaystyle=\dfrac{R_{P,\widetilde{E}}(U_{z}^{\prime})}{\left[U(\widetilde{E};\widetilde{\mathscr{O}_{L}}):U_{z}^{\prime}\right]_{\widetilde{A}}}\text{ by Equality }\eqref{egreg}, \displaystyle=R_{P}(U(\widetilde{E};\widetilde{\mathscr{O}_{L}}))\text{ by Lemma }\ref{reg}. Recall Theorem 4.13 : L_{P}(E/\mathscr{O}_{L})=R_{P}(U(F;\mathscr{O}_{L}))\left[H(F;\mathscr{O}_{L})\right]_{A}. Then: \displaystyle z_{P}(E/\mathscr{O}_{L}) \displaystyle L_{P}(E/\mathscr{O}_{L}) \displaystyle=\dfrac{\left[U(F;\mathscr{O}_{L}):U^{\prime})\right]_{A}}{\left[U(E;\mathscr{O}_{L}):U^{\prime}\right]_{A}}\dfrac{\left[H(E;\mathscr{O}_{L})\right]_{A}}{\left[H(F;\mathscr{O}_{L})\right]_{A}}R_{P}(U(F;\mathscr{O}_{L}))\left[H(F;\mathscr{O}_{L})\right]_{A}\text{ by Equality }\eqref{facteur}, \displaystyle=\dfrac{R_{P,F}(U^{\prime})}{\left[U(E;\mathscr{O}_{L}):U^{\prime}\right]_{A}}\left[H(E;\mathscr{O}_{L})\right]_{A}\text{ by Lemma }\ref{reg}, \displaystyle=\dfrac{R_{P,E}(U^{\prime})}{\left[U(E;\mathscr{O}_{L}):U^{\prime}\right]_{A}}\left[H(E;\mathscr{O}_{L})\right]_{A}\text{ by Equality }\eqref{egreg}, \displaystyle=R_{P}(U(E;\mathscr{O}_{L}))\left[H(E;\mathscr{O}_{L})\right]_{A}\text{ by Lemma }\ref{reg}. ∎ 4.7. Vanishing of the P -adic L -series We keep the notation as in Theorem 4.10 . In particular, (v_{1},\ldots,v_{m}) is an A -basis of U_{\operatorname{st}}(E;\mathscr{O}_{L}) and (u_{1},\ldots,u_{m}) is an A -basis of U(E;\mathscr{O}_{L}) . Proposition 4.22 . Suppose that there exists a non-zero element x\in U_{\operatorname{St}}(E;\mathscr{O}_{L}) such that \exp_{E}(x)=0 . Then L_{P}(E/\mathscr{O}_{L})=0. Proof. Write x={\sum}\limits_{i=1}^{m}a_{i}v_{i} with a_{i}\in A and suppose without loss of generality that a_{1}\neq 0 . Then x=x(1) with x(z)={\sum}\limits_{i=1}^{m}a_{i}v_{i}(z)\in U(\widetilde{E},\mathscr{O}_{L}[z]) . We have: \displaystyle\det_{\mathscr{C}}(\operatorname{Log}_{\widetilde{E},P}(\exp_{\widetilde{E}}(v_{i}(z))),i=1,\ldots,m) \displaystyle= \displaystyle\dfrac{1}{a_{1}}\det_{\mathscr{C}}(\operatorname{Log}_{\widetilde{E},P}(\exp_{\widetilde{E}}(x(z))),\operatorname{Log}_{\widetilde{E},P}(\exp_{\widetilde{E}}(v_{i}(z))),i=2,\ldots,m). Since \exp_{E}(x)=0 , we have: \operatorname{ev}_{z=1,P}(\operatorname{Log}_{\widetilde{E},P}(\exp_{\widetilde{E}}x(z)))=\operatorname{Log}_{E,P}(\exp_{E}(x))=0. We then conclude that R_{P}(U_{\operatorname{St}}(E;\mathscr{O}_{L}))=0 , so L_{P}(E/\mathscr{O}_{L})=0 from the P -adic class formula 4.21 . ∎ Theorem 4.23 . If the exponential map \exp_{E}:L_{\infty}^{d}\rightarrow L_{\infty}^{d} is not injective, then we have L_{P}(E/\mathscr{O}_{L})=0. Proof. Let x\in L_{\infty}^{d} be non-zero such that \exp_{E}(x)=0 . There exists a\in A\backslash\{0\} such that ax\in U_{\operatorname{st}}(E;\mathscr{O}_{L}) and we have \exp_{E}(ax)= . By Proposition 4.22 we have L_{P}(E/\mathscr{O}_{L})=0 . ∎ We believe that the converse statement holds. Conjecture . The P -adic L -series is non-zero if and only if the exponential map \exp_{E}:L_{\infty}^{d}\rightarrow L_{\infty}^{d} is injective. By [ 7 , Corollary 3.24] , it is true when d=1 (i.e., in the Drinfeld module case) and L=K . Remark that in the case \exp_{E}:L_{\infty}^{d}\rightarrow L_{\infty}^{d} is injective, which we will call the totally real case, then \mathscr{U}(E;\mathscr{O}_{L})=\exp_{E}(U(E;\mathscr{O}_{L}))\subseteq E(\mathscr{O}_{L}) is a free A -module of rank m , and the family (\operatorname{Log}_{E,P}(\exp_{E}(u_{i})),i=1,\ldots,m) is A -free. We would like to have that this family is A_{P} -free to obtain the non-vanishing of the P -adic L -series. Set: U(E;P\mathscr{O}_{L})=\{x\in\operatorname{Lie}_{E}(L_{\infty})\ |\ \exp_{E}(x)\in E(P\mathscr{O}_{L})\} and \mathscr{U}(E;P\mathscr{O}_{L})=\exp_{E}(U(E;P\mathscr{O}_{L})) . Then we can state an equivalent of the Leopoldt’s conjecture in [ 16 ] , introduced recently by Anglès in [ 3 , Section 6.3] for the Carlitz module. Conjecture (Conjecture A) . The A_{P} -rank of \mathscr{U}(E;P\mathscr{O}_{L}) is equal to the A -rank of \mathscr{U}(E;\mathscr{O}_{L}) . This conjecture is clear in the case d=1 and L=K . For further discussion of this conjecture, the reader may wish to see the paper by Anglès, Bosser and Taelman [ 6 ] where this conjecture is proved in the case of the Carlitz module defined on the P th cyclotomic extension. In the totally real case, the non-vanishing of the P -adic L -series L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}}) at z=1 is equivalent to the previous Leopoldt conjecture. This result can be seen as an analog to the following result from [ 10 ] . Theorem 4.24 . Let F be a totally real extension of \mathbb{Q} . Then the p -adic zeta function \zeta_{F,p}(s) has a simple pole at s=1 if and only if the (usual) Leopoldt conjecture is true for (F,p) . Definition 4.25 . We call order of vanishing of the P -adic L -series and denote by \operatorname{ord}_{z=1}L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}}) , the greatest integer n such that (z-1)^{n} divides L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}}) . Fox example if the exponential map \exp_{E}:L_{\infty}^{d}\rightarrow L_{\infty}^{d} is not injective, then for all P we have \operatorname{ord}_{z=1}L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})\geq 1 and the previous conjecture tells us that \operatorname{ord}_{z=1}L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})=0 if and only if \exp_{E} in injective. Here is a list of conjectures. Conjecture (Conjecture B) . : The vanishing order of the P -adic L -series at z=1 is independent of P . Caruso and Gazda [ 8 ] have already conjectured this in the context of Anderson motives. Caruso, Gazda and the author proved this conjecture in the case L=K and d=1 , see [ 9 , Theorem 2.17] . Conjecture (Conjecture C) . We have \operatorname{ord}_{z=1}L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L}})\leq[L:K]r_{\Omega_{E}}d where r_{\Omega_{E}} is the rank of the period lattices \Omega_{E} associated with E . We prove conjecture C in section 6 in the case d=1 and L=K . 5. The multi-variable setting We keep the notation from Sections 2 , 3 , 4 and from the Introduction. In particular L/K is a finite field extension of degree n and \mathscr{O}_{L} denotes the integral closure of A in L . The aim of this section is to extend the previous constructions to the case where the constant field is no longer \mathbb{F}_{q} but \mathbb{F}_{q}(t_{1},\ldots,t_{s}) where the t_{i} are new variables. One of the interests of these constructions is that in many cases, we can reduce the study of certain t -modules E:\mathbb{F}_{q}[\theta]\rightarrow M_{d}(\mathscr{O}_{L})\{\tau\} to the study of Drinfeld modules \phi:\mathbb{F}_{q}(t_{1},\ldots,t_{s})[\theta]\rightarrow\mathscr{O}_{L}(t_{1},\ldots,t_{s})\{\tau\} simpler to understand. For an application to the study of the tensor power of the Carlitz module C^{\otimes n} reduced to the study of the Carlitz module C , see the work of Anglès, Pellarin and Tavares Ribeiro in [ 5 ] . 5.1. Setup The goal of this section is to extend the developed theory to the multi-variable setting by replacing \mathbb{F}_{q} by k=\mathbb{F}_{q}(t_{1},\ldots,t_{s}) . Recall that the Frobenius map acts as the identity on k . We keep the notation in the Introduction and we introduce the following notation. \bullet A_{s}[z]\simeq k[z]\otimes_{k}A_{s} , \bullet \widetilde{A}_{s}=k(z)\otimes_{k}A_{s} , \bullet w : a place of K ( w=v_{P} a finite place or w=v_{\infty} the infinite place), \bullet \pi_{w} : a uniformiser of w ( \pi=P if w=v_{P} and \pi=\frac{1}{\theta} if w=v_{\infty} ), \bullet K_{w}=\mathbb{F}_{w}((\pi_{w})) denoted by K_{w}=K_{\infty} if w=v_{\infty} and K_{w}=K_{P} if w=v_{P} , \bullet \mathbb{F}_{w} : the residue field associated with w i.e., \mathbb{F}_{w}=\mathbb{F}_{P} if w=v_{P} and \mathbb{F}_{w}=\mathbb{F}_{q} if w=v_{\infty} , \bullet L_{w}=L\otimes_{K}K_{w} i.e., L_{w}=L_{P} if w=v_{P} and L_{w}=L_{\infty} else, \bullet k_{w}=\mathbb{F}_{w}(t_{1},\ldots,t_{s}) , \bullet K_{s,w}=kw((\pi_{w})) denoted by K_{s,P} if w=v_{P} and K_{s,\infty} if w=v_{\infty} , \bullet \widetilde{K_{s,w}}=k_{w}(z)((\pi_{w})), \bullet L_{s}=kL , \bullet L_{s,w}=L\otimes_{K}K_{s,w} denoted by L_{s,P} if w=v_{P} , \bullet \widetilde{L_{s,w}}=L\otimes_{K}\widetilde{K_{s,w}} denoted by \widetilde{L_{s,P}} if w=v_{P} , \bullet \mathscr{O}_{L,s}[z]\simeq k[z]\otimes_{k}\mathscr{O}_{L,s} , \bullet \widetilde{\mathscr{O}_{L,s}}=k(z)\otimes_{k}\mathscr{O}_{L,s} . We recall that every x\in\widetilde{K_{s,\infty}}^{\ast} (resp. \in K_{s,\infty} ) can be written uniquely as x=\sum\limits_{n\geq N}x_{n}\frac{1}{\theta^{n}} with N\in\mathbb{Z} , x_{n}\in k(z) (resp. x_{n}\in K ) and x_{N}\neq 0 . We call x_{N}\in k(z) (resp. k ) the sign of x denoted by \operatorname{sgn}(x) . We define the Tate algebra in variables \underline{t}=(t_{1},\ldots,t_{s}) : \mathbb{T}_{s}(K_{w})=\left\{{\sum}\limits_{{n}\in\mathbb{N}^{s}}a_{{n}}\underline{t}^{{n}}\in K_{w}[[\underline{t}]]\ |\ a_{n}\in K_{w},\lim\limits_{n\rightarrow+\infty}w(a_{n})=+\infty\right\} where \underline{t}^{n}=t_{1}^{n_{1}}\ldots t_{s}^{n_{s}} if n=(n_{1},\ldots,n_{s})\in\mathbb{N}^{s} . This is the completion of K[t_{1},\ldots,t_{s}] with respect to the Gauss norm associated with w . We set: \mathbb{T}_{s}(L_{w})=L\otimes_{K}\mathbb{T}_{s}(K_{w}). An Anderson t -module E of dimension d over \mathscr{O}_{L,s} is a non-constant k -algebra homomorphism E:A_{s}\rightarrow M_{d}({\mathscr{O}_{L}}_{,s}) , a\mapsto E_{a}={\sum}\limits_{i=0}^{r_{a}}E_{a,i}\tau^{i}\in M_{d}(\mathscr{O}_{L,s})\{\tau\} such that (E_{\theta,0}^{d}-\theta I_{d})^{d}=0 . We can consider \widetilde{E} , the z -twist of E , as in Section 3 . Following notation from Section 2 , we denote by [M]_{A_{s}} the monic generator of \operatorname{Fitt}_{A_{s}}(M) where M is a torsion A_{s} -module of finite type, e.g., M=E(\mathscr{O}_{L,s}/P\mathscr{O}_{L,s}) and M=\operatorname{Lie}_{E}(\mathscr{O}_{L,s}/P\mathscr{O}_{L,s}) . As in Proposition 3.1 there exists a unique element \exp_{E}\in M_{d}(L_{s})\{\{\tau\}\} called the exponential map associated with E and converging over L_{s,\infty}^{d} . Similarly, there exists a logarithm map \log_{E}\in M_{d}(L_{s})\{\{\tau\}\} as in Proposition 3.2 . 5.2. The \infty -case We can now define the module of units and class module in the multi-variable setting: U(E;\mathscr{O}_{L,s})=\{x\in\operatorname{Lie}_{E}(L_{s,\infty})\ |\ \exp_{E}(x)\in E(\mathscr{O}_{L,s})\} and the class module H(E;\mathscr{O}_{L,s})=\dfrac{E(L_{s,\infty})}{E(\mathscr{O}_{L,s})+\exp_{{E}}(\operatorname{Lie}_{{E}}(L_{s,\infty}))} both provided with A_{s} -module structure. We define the module of z -units: U(\widetilde{E};\widetilde{\mathscr{O}_{L,s}})=\left\{x\in\operatorname{Lie}_{\widetilde{E}}(\widetilde{L_{s,\infty}})\ |\ \exp_{\widetilde{E}}(x)\in\widetilde{E}(\widetilde{\mathscr{O}_{L,s}})\right\} and the class module for the z -deformation: H(\widetilde{E};\widetilde{\mathscr{O}_{L,s}})=\dfrac{\widetilde{E}(\widetilde{L_{s,\infty}})}{\widetilde{E}(\widetilde{\mathscr{O}_{L,s}})+\exp_{\widetilde{E}}(\operatorname{Lie}_{\widetilde{E}}(\widetilde{L_{s,\infty}}))} both provided with \widetilde{A_{s}} -module structure. We define the module of z -units at the integral level: U(\widetilde{E};\mathscr{O}_{L,s}[z])=\left\{x\in\operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{s,\infty}))\ |\ \exp_{\widetilde{E}}(x)\in\widetilde{E}(\mathscr{O}_{L,s}[z])\right\} and finally the class module at the integral level H(\widetilde{E};\mathscr{O}_{L,s}[z])=\dfrac{\widetilde{E}(\mathbb{T}_{z}(L_{s,\infty}))}{\widetilde{E}(\mathscr{O}_{L}[z])+\exp_{\widetilde{E}}(\operatorname{Lie}_{\widetilde{E}}(\mathbb{T}_{z}(L_{s,\infty})))} both provided with A_{s}[z] -module structure. We have the following result from [ 11 , Proposition 2.8] . Proposition 5.1 . The unit module U(E;\mathscr{O}_{L,s}) is an A_{s} -lattice in \operatorname{Lie}_{E}(L_{s,\infty}) and the module of z -units U(\widetilde{E},\widetilde{\mathscr{O}_{L,s}}) is an \widetilde{A_{s}} -lattice in \operatorname{Lie}_{\widetilde{E}}(\widetilde{L_{s,\infty}}) . Denote by z_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L,s}})=\dfrac{\left[\operatorname{Lie}_{\widetilde{E}}(\widetilde{\mathscr{O}_{L,s}}/{P}\widetilde{\mathscr{O}_{L,s}})\right]_{\widetilde{A}_{s}}}{\left[\widetilde{E}(\widetilde{\mathscr{O}_{L,s}}/{P}\widetilde{\mathscr{O}_{L,s}})\right]_{\widetilde{A}_{s}}} the local factor associated with \widetilde{E} at P and z_{P}({E}/{\mathscr{O}_{L,s}})=\dfrac{\left[\operatorname{Lie}_{{E}}({\mathscr{O}_{L,s}}/{P}{\mathscr{O}_{L,s}})\right]_{{A}_{s}}}{\left[{E}({\mathscr{O}_{L,s}}/{P}{\mathscr{O}_{L,s}})\right]_{{A}_{s}}} the local factor associated with E at P . We have the following class formula for t -modules defined over \mathscr{O}_{L,s} , see [ 11 , Theorem 2.9] . Theorem 5.2 . The following product L(\widetilde{E}/\widetilde{\mathscr{O}_{L,s}})={\prod}\limits_{P}z_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L,s}}) where P runs through the monic primes of A , converges in \widetilde{K_{s,\infty}} and we have the class formula: L(\widetilde{E}/\widetilde{\mathscr{O}_{L,s}})=\left[\operatorname{Lie}_{\widetilde{E}}\left(\widetilde{\mathscr{O}_{L,s}}\right):U(\widetilde{E};\widetilde{\mathscr{O}_{L,s}})\right]_{\widetilde{A_{s}}}. 5.3. The P -adic case We define the P -adic L -series in the multi-variable setting. 5.3.1. The P -adic class formula All results from Section 4 remain valid by replacing \mathbb{F}_{q} by k . In particular, we have the following P -adic class formula. Theorem 5.3 ( P -adic class formula) . We have the following assertions. (1) The infinite product L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L,s}})={\prod}\limits_{Q\neq P}z_{Q}(\widetilde{E}/\widetilde{\mathscr{O}_{L,s}}) where Q runs through the monic primes of A different from P , converges in \mathbb{T}_{z}(K_{s,P}) and we have the class formula: z_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L,s}})L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L,s}})=R_{P}(U(\widetilde{E};\widetilde{\mathscr{O}_{L,s}})). (2) The infinite product L_{P}({E}/{\mathscr{O}}_{L,s})={\prod}\limits_{Q\neq P}z_{Q}({E}/{\mathscr{O}_{L,s}}) where {Q} runs through the monic primes of A different from P , converges in {K}_{s,P} and we have the class formula: z_{P}(E/\mathscr{O}_{L,s})L_{P}({E}/{\mathscr{O}}_{L,s})=R_{P}(U({E};{\mathscr{O}}_{L,s}))\left[H(E;\mathscr{O}_{L,s})\right]_{A_{s}}. Proof. The proof follows the same lines as the proof of 4.10 by replacing \mathbb{F}_{q} by k . We omit the details.∎ Denote by U(E;P\mathscr{O}_{L,s})=\{x\in\operatorname{Lie}_{E}(L_{s,\infty})\ |\ \exp_{E}(x)\in E(P\mathscr{O}_{L,s})\} and consider the A_{s} -module \mathscr{U}(E;\mathscr{O}_{L,s})=\exp_{E}(U(E;\mathscr{O}_{L,s})) . Consider also the A_{P,s} -module \mathscr{U}(E;P\mathscr{O}_{L,s})=\exp_{E}(U(E;P\mathscr{O}_{L,s})) . Then the proof of Theorem 4.23 is still valid in the multi-variable setting by replacing \mathbb{F}_{q} by k . Proposition 5.4 . We have the following assertions. (1) If the exponential map \exp_{E}:L_{s,\infty}^{d}\rightarrow L_{s,\infty}^{d} is not injective, then L_{P}(E/\mathscr{O}_{L,s})=0 . (2) Assume that the A_{s} -rank of \mathscr{U}(E;\mathscr{O}_{L,s}) and the A_{P,s} -rank of \mathscr{U}(E;P\mathscr{O}_{L,s}) are equal. Then L_{P}(E/\mathscr{O}_{L})\neq 0 if and only if the exponential map \exp_{E}:L_{s,\infty}^{d}\rightarrow L_{s,\infty}^{d} is injective. 5.3.2. The integral level In the work of [ 5 ] , given a t -module E:A_{s}\rightarrow M_{d}(A_{s})\{\tau\} , they want to evaluate the variables (t_{1},\ldots,t_{s}) at some \zeta\in\overline{\mathbb{F}}_{q}^{s} . In this case, they need that all of the coefficients E_{\theta,i} of E_{\theta} , for i=0,\ldots,r , can be evaluated at \zeta . This is possible if all the E_{i,\theta} belong to M_{d}(\mathbb{F}_{q}[t_{1},\ldots,t_{s}]\mathscr{O}_{L}) . This is what we call the integral level. We suppose now that: E_{\theta}\in M_{d}(\mathscr{O}_{L}[t_{1},\ldots,t_{s}])\{\tau\} i.e., we want to work at the integral level. Theorem 5.5 . The L -series L(\widetilde{E}/\widetilde{\mathscr{O}_{L,s}}) converges in \mathbb{T}_{s,z}(K_{\infty}) and we have the class formula: L(\widetilde{E}/\widetilde{\mathscr{O}_{L,s}})=\dfrac{\det_{\mathscr{C}}\left(u_{1}(z),\ldots,u_{m}(z)\right)}{\operatorname{sgn}(\det_{\mathscr{C}}\left(u_{1}(z),\ldots,u_{m}(z)\right))} where (u_{1}(z),\ldots,u_{m}(z))\in U(\widetilde{E};\mathscr{O}_{L}[t_{1},\ldots,t_{s},z]) is an \widetilde{A_{s}} -basis of the z -unit module. Proof. The proof of [ 22 , Corollary 7.5.6] is still valid in the multi-variable setting at the integral level. We omit the details. ∎ The objective of this section is to prove that the P -adic L -series L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L,s}}) converges in \mathbb{T}_{s,z}(K_{P}) . Set \Omega_{s,z}=\{x\in\mathbb{T}_{z,s}(L_{P})^{d}\ |\ v_{P}(x)\geq 0\} and \Omega_{s,z}^{+}=\{x\in\mathbb{T}_{z,s}(L_{P})^{d}\ |\ v_{P}(x)>0\} . Following the proof of Proposition 4.5 we have the two following convergences: \log_{\widetilde{E},P}:\Omega_{s,z}^{+}\rightarrow\mathbb{T}_{s,z}(L_{P})^{d} and \log_{\widetilde{F},P}:\Omega_{s,z}\rightarrow\mathbb{T}_{s,z}(L_{P})^{d}. We deduce that the P -adic L -series L_{P}(\widetilde{E}/\widetilde{\mathscr{O}_{L,s}}) is written in the form \dfrac{w}{f} with w\in\mathbb{T}_{z,s}(K_{P}) and f\in\mathbb{F}_{q}[t_{1},\ldots,t_{s},z] . We then consider \zeta=(\zeta_{1},\ldots,\zeta_{s})\in\overline{\mathbb{F}_{q}}^{s} and we want to prove that we can evaluate the P -adic L -series at t_{i}=\zeta_{i} for all i=1,\ldots,s and at z=\zeta\in\overline{\mathbb{F}_{q}} (simultaneously). We use arguments very similar to those used for the convergence of the P -adic L -series, so we omit some of the details. We set \mathscr{K}(s)=\mathbb{F}_{q}(\zeta_{1})\otimes_{\mathbb{F}_{q}}\ldots\otimes_{\mathbb{F}_{q}}\mathbb{F}_{q}(\zeta_{s}) . We then consider the following notation for j=0,\ldots,s : \bullet k_{j}=\mathbb{F}_{q}(t_{j+1},\ldots,t_{s}) , e.g., k_{0}=k=\mathbb{F}_{q}(t_{1},\ldots,t_{s}) and k_{s}=\mathbb{F}_{q} , \bullet k_{j}A=k_{j}\otimes_{\mathbb{F}_{q}}A\simeq\mathbb{F}_{q}(t_{j+1},\ldots,t_{s})[\theta] , \bullet k_{j}K=k_{j}\otimes_{\mathbb{F}_{q}}K\simeq\mathbb{F}_{q}(t_{j+1},\ldots,t_{s},\theta) and \widetilde{k_{j}K}=\mathbb{F}_{q}(z)\otimes_{\mathbb{F}_{q}}k_{j}K\simeq\mathbb{F}_{q}(z,t_{j+1},\ldots,t_{s},\theta) , \bullet k_{j}\mathscr{O}_{L}=k_{j}\otimes_{\mathbb{F}_{q}}\mathscr{O}_{L} , \bullet A_{s,j}=\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}k_{j}A, \bullet \widetilde{A_{s,j}}=\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}k_{j}\widetilde{A} , \bullet \mathscr{O}_{L,s,j}=\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}k_{j}\mathscr{O}_{L} , \bullet \widetilde{\mathscr{O}_{L,s,j}}=\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}k_{j}\widetilde{\mathscr{O}_{L}}, \bullet For a place w of K extended to k_{j}K , \widetilde{K(j)_{w}} is the completion of \widetilde{k_{j}K} with respect to w . \bullet \widetilde{L(j)_{w}}=L\otimes_{K}K(j)_{w} , \bullet \widetilde{M_{s,j,w}}=\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{t_{j}}\left(L(j)_{w}\right) , \bullet \widetilde{L_{j,w}}=L\otimes_{K}\widetilde{K_{j,w}} . \bullet For a place w of K , \mathbb{T}_{z,j}(L_{w})=\mathbb{T}_{z,t_{j+1},\ldots,t_{s}}(L_{w}), e.g;, \mathbb{T}_{z,0}(L_{w})=\mathbb{T}_{z,t_{1},\ldots,t_{s}}(L_{w}) and \mathbb{T}_{z,s}(L_{w})=\mathbb{T}_{z}(L_{w}) . For all j=0,\ldots,s , we extend the Frobenius \tau into \tau_{s} on A_{s,j} by \tau_{s}=\operatorname{id}\otimes\tau where \operatorname{id} is the identity on \mathscr{K}(s) . We do the same for \widetilde{A_{s,j}} , for \mathscr{O}_{L,s,j} and for \widetilde{\mathscr{O}_{L,s,j}} . For j=1,\ldots,s we define E^{(j)} the homomorphism of \mathscr{K}(s)\otimes_{\mathbb{F}_{q}}k_{j} -algebras E^{(j)}:A_{s,j}\rightarrow M_{d}(\mathscr{O}_{L,s,j})\{\tau_{s}\} , that we call Anderson A_{s,j} -module defined over \mathscr{O}_{L,s,j} , by: E^{(j)}_{\theta}={\sum}\limits_{i=0}^{r}\operatorname{ev}_{t_{1}=\zeta_{1},\ldots,t_{j}=\zeta_{j}}(a_{i})\tau_{s}^{i} if E_{\theta}={\sum}\limits_{i=0}^{r}a_{i}\tau_{s}^{i}\in M_{d}(\mathscr{O}_{L,s})\{\tau\} . We also set E^{(0)}=E where we identify a_{i} with 1\otimes a_{i}\in\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathscr{O}_{L}[t_{1},\ldots,t_{s}] and replace \tau with \tau_{s} . Similarly, we define \widetilde{E^{(j)}} the z -twist of E^{(j)} , that is the homomohprism of \mathscr{K}(z)\otimes_{\mathbb{F}_{q}}k_{j}(z) -algebras \widetilde{E^{(j)}}:\widetilde{A_{s,j}}\rightarrow M_{d}(\widetilde{\mathscr{O}_{L,s,j}})\{\tau_{s}\} , that we call Anderson \widetilde{A_{s,j}} -module defined over \widetilde{\mathscr{O}_{L,s,j}} , by: \widetilde{E^{(j)}}_{\theta}={\sum}\limits_{i=0}^{r}\operatorname{ev}_{t_{1}=\zeta_{1},..,t_{j}=\zeta_{j}}(a_{i})z^{i}\tau_{s}^{i} Finally, we consider F=P^{-1}EP and construct \widetilde{F^{(j)}} and {F}^{(j)} in the same way. Lemma 5.6 . Consider j\in\{1,\ldots,s\} . (1) For all a\in A_{s,j} we have the following equalities in M_{d}(\mathscr{O}_{L,s,j})\{\tau_{s}\} : E^{(j)}_{a}=\operatorname{ev}_{t_{1}=\zeta_{1},\ldots,t_{j}=\zeta_{j}}E_{a}=\operatorname{ev}_{t_{j}=\zeta_{j}}E_{a}^{(j-1)}. (2) We have the following equalities in M_{d}\left(\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathbb{F}_{q}[t_{j+1},\ldots,t_{s}]L\right)\{\{\tau_{s}\}\} : \exp_{E^{(j)}}=\operatorname{ev}_{t_{1}=\zeta_{1},\ldots,t_{j}=\zeta_{j}}\exp_{E}=\operatorname{ev}_{t_{j}=\zeta_{j}}\exp_{E^{(j-1)}}. (3) We have the following equalities for all x in \left(\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{t_{j}}\left((\mathbb{F}_{q}(t_{j+1},\ldots,t_{s})L)_{\infty}\right)\right)^{d} : \exp_{E^{(j)}}(x)=\operatorname{ev}_{t_{1}=\zeta_{1},\ldots,t_{j}=\zeta_{j}}\exp_{E}(x)=\operatorname{ev}_{t_{j}=\zeta_{j}}\exp_{E^{(j-1)}}(x). Proof. It follows from definitions of the onjects, we omit the proof. ∎ We then define for all j=1,\ldots,s : U(j)=\operatorname{ev}_{t_{j}=\zeta_{j}}U\left(\widetilde{E}^{(j-1)};\widetilde{\mathscr{O}_{L,s,j}}[t_{j}]\right)\subseteq U\left(\widetilde{E}^{(j)};\widetilde{\mathscr{O}_{L,s,j}}\right). Following the same arguments we used to prove Theorem 3.14 , we have the following result. Theorem 5.7 . (1) For all j=1,\ldots,s , we have an \widetilde{A_{s,j}} -module isomorphism: \dfrac{U(j)}{U\left(\widetilde{E}^{(j)};\widetilde{\mathscr{O}_{L,s,j}}\right)}\simeq H\left(\widetilde{E}^{(j-1)};\widetilde{\mathscr{O}_{L,s,j}}[t_{j}]\right)[t_{j}-\zeta_{j}] given by f_{j}(x)=\dfrac{\exp_{\widetilde{E}^{(j-1)}}x-\exp_{\widetilde{E}^{(j)}}x}{t_{j}-\zeta_{j}} where H\left(\widetilde{E}^{(j-1)};\widetilde{\mathscr{O}_{L,s,j}}[t_{j}]\right)[t_{j}-\zeta_{j}] is the (t_{j}-\zeta_{j}) -torsion of the class module H\left(\widetilde{E}^{(j-1)};\widetilde{\mathscr{O}_{L,s,j}}[t_{j}]\right)=\dfrac{\widetilde{E^{(j-1)}}\left(\widetilde{M_{s,j,\infty}}\right)}{\widetilde{E^{(j-1)}}\left(\widetilde{\mathscr{O}_{L,s,j}}[t_{j}]\right)+\exp_{\widetilde{E^{(j-1)}}}\left(\widetilde{E^{(j-1)}}(\widetilde{M_{s,j,\infty}})\right)}. (2) The module U(j) is a sub- \widetilde{A_{s,j}} lattice of U\left(\widetilde{E}^{(j)};\widetilde{\mathscr{O}_{L,s,j}}\right) . We are now able to prove the main theorem of this section. Theorem 5.8 . The P -adic L -series does not have a pole in \overline{\mathbb{F}}_{q}^{s} . In other words we have: L_{P}({\widetilde{E}}/{{\mathscr{O}}}_{L,s})\in\mathbb{T}_{z,s}(K_{P}). Proof. We closely follow the proof of Theorem 4.10 . We identify \mathscr{C}=(g_{1},\ldots,g_{m}) with (1\otimes g_{1},\ldots,1\otimes g_{m})\subseteq\mathscr{K}(s)\otimes\operatorname{Lie}_{E}(\mathscr{O}_{L}) . Consider (u_{i,1})_{i=1,\ldots,m}\subseteq U(\widetilde{F};\mathscr{O}_{L}[t_{1},\ldots,t_{s},z]) an \widetilde{A}_{s} -basis of U(\widetilde{F};\widetilde{\mathscr{O}_{L,s}}) . Set w_{1}=\det_{\mathscr{C}}(u_{1,1},\ldots,u_{m,1})\in\mathbb{T}_{z,s}(K_{\infty}) with sign f_{1}\in\mathbb{F}_{q}[z,t_{1},\ldots,t_{s}] and set w_{1,P}=\det_{\mathscr{C}}(\log_{\widetilde{F},P}(\exp_{\widetilde{F}}(u_{1,1})),\ldots,\log_{\widetilde{F},P}(\exp_{\widetilde{F}}(u_{m,1})))\in\mathbb{T}_{z,s}(K_{P}). We want to prove that the quotient \dfrac{w_{1,P}}{f_{1}} is an element of \mathbb{T}_{s,z}(K_{P}) . We will prove that we can evaluate the last quantity at every \zeta=(\zeta_{1},\ldots,\zeta_{s})\in\overline{\mathbb{F}}_{q}^{s} and at z=\zeta\in\overline{\mathbb{F}_{q}} . We will prove by induction that for all k=1,\ldots,s , there exists (v_{i,k+1},i=1,\ldots,m) an \widetilde{A_{s,k}} -basis of U(k) and x_{k+1}\in\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\widetilde{A_{s,k}} such that \operatorname{ev}_{t_{1}=\zeta_{1},\ldots,t_{k}=\zeta_{k}}(1\otimes L(\widetilde{F}/\widetilde{\mathscr{O}_{L,s}})=\dfrac{\det_{\mathscr{C}}(v_{i,{k+1}},i=1,\ldots,m)}{x_{k+1}} and \displaystyle\operatorname{ev}_{t_{1}=\zeta_{1},\ldots,t_{k}=\zeta_{k}}(1\otimes\dfrac{w_{1,P}}{f_{1}}) \displaystyle=\dfrac{\det_{\mathscr{C}}(\log_{\widetilde{F}^{(k)},P}\exp_{\widetilde{F}^{(k)}}v_{i,{k+1}},i=1,\ldots,m)}{x_{k+1}} \displaystyle\in\dfrac{1}{x_{k+1}}\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z}(t_{k+1},\ldots,t_{s})(L_{P}). Step 1 : evaluation at t_{1}=\zeta_{1} . By Theorem 5.5 we have L(\widetilde{F}/\widetilde{\mathscr{O}_{L,s}})=\dfrac{w_{1}}{f_{1}}\in\mathbb{T}_{z,s}(K_{\infty}). Consider (v_{i,2})_{i=1,\ldots,m} an \widetilde{A_{s,1}} basis of U(1) that can be assumed to be at the entire level, i.e., (v_{i,2})\subseteq\left(\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z,t_{2},\ldots,t_{s}}(L_{\infty})\right)^{d} and let (u_{i,2})\subseteq\left(\mathscr{K}(s)\otimes\mathbb{T}_{z,s}(L_{\infty})\right)^{d} be above (i.e., \operatorname{ev}_{t_{1}=\zeta_{1}}u_{i,2}=v_{i,2}) for all i ). Set w_{2}=\det_{\mathscr{C}}(u_{1,2},\ldots,u_{m,2})\in\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z,s}(K_{\infty}) that is not divisible by t_{1}-\zeta_{1} and set w_{2,P}=\det_{\mathscr{C}}(\log_{\widetilde{F},P}(\exp_{\widetilde{F}}(u_{1,2})),\ldots,\log_{\widetilde{F},P}(\exp_{\widetilde{F}}(u_{m,2})))\in\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z,s}(K_{P}) the P -adic analog of w_{2} . Set \delta_{2}=\det_{(1\otimes u_{1,1},\ldots,1\otimes_{u_{m,1}})}(u_{1,2},\ldots,u_{m,2})\in\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\widetilde{A}_{s} . We have: 1\otimes_{\mathbb{F}_{q}}w_{1}=\dfrac{w_{2}}{\delta_{2}} and 1\otimes w_{1,P}=\dfrac{w_{2,P}}{\delta_{2}}. We deduce the following equality from the class formula: 1\otimes L(\widetilde{F}/\widetilde{\mathscr{O}_{L,s}})=\dfrac{w_{2}}{\delta_{2}(1\otimes f_{1})}. Since \zeta_{1} is not a pole of the L -series and t_{1}-\zeta_{1} does not divide w_{2} , we deduce that we can evaluate \delta_{2}(1\otimes f_{1}) at t_{1}=\zeta_{1} in a non-zero element x_{2} of \mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\widetilde{A}_{s-1} , in other words \operatorname{ev}_{t_{1}=\zeta_{1}}1\otimes L(\widetilde{F}/\widetilde{\mathscr{O}_{L,s}})=\dfrac{\det_{\mathscr{C}}(v_{i,2})}{x_{2}}\in\dfrac{1}{x_{2}}\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z,t_{2},\ldots,t_{s}}(L_{\infty}). Next, from the equality 1\otimes\dfrac{w_{1,P}}{f_{1}}=\dfrac{w_{2,P}}{\delta_{2}(1\otimes f_{1})} we deduce that we can evaluate the P -adic L -series at t_{1}=\zeta_{1} : \displaystyle\operatorname{ev}_{t_{1}=\zeta_{1}}1\otimes\dfrac{w_{1,P}}{f_{1}} \displaystyle=\dfrac{\det_{\mathscr{C}}\left(\log_{\widetilde{F}^{(1)},P}(\exp_{\widetilde{F}^{(1)}}(v_{i,2})),i=1,\ldots,m\right)}{x_{2}} \displaystyle\in\dfrac{1}{x_{2}}\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z,t_{2},\ldots,t_{s}}(L_{P}). Step k : Assume the result to be true up to rank s-1\geq k-1\geq 1 and we prove it at rank k . Consider (v_{i,{k+1}})_{i=1,\ldots,m} a \widetilde{A_{s,k}} basis of U(k) that can be assumed to be to the entire level, i.e., v_{i,{k+1}}\subseteq\left(\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z,t_{k+1},\ldots,t_{s}}(L_{\infty})\right)^{d} and let (u_{i,k+1})\subseteq\left(\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z,t_{k},\ldots,t_{s}}(L_{\infty})\right)^{d} be above (i.e., \operatorname{ev}_{t_{k}=\zeta_{k}}u_{i,k+1}=v_{i,k+1} ). Set w_{k+1}=\det_{\mathscr{C}}(u_{1,{k+1}},\ldots,u_{m,k+1})\in\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z,t_{k},\ldots,t_{s}}(K_{\infty}) that is not divisible by t_{k}-\zeta_{k} and set w_{k+1,P}=\det_{\mathscr{C}}(\log_{\widetilde{F}^{(k-1)}}(\exp_{\widetilde{F}^{(k-1)}}(u_{i,k+1})),i=1,\ldots,m)\in\mathscr{K}_{s}\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z,t_{k},\ldots,t_{s}}(K_{P}) the P -adic analog of w_{k+1} . Set \delta_{k+1}=\det_{(v_{1,k}),\ldots,v_{m,k}}(u_{1,k+1},\ldots,u_{m,k+1})\in\mathscr{K}(s)\otimes\widetilde{A_{s,k-1}}. We have: \det_{\mathscr{C}}(v_{1,k},\ldots,v_{m,k})=\dfrac{\det_{\mathscr{C}}(u_{1,k+1},\ldots,u_{m,k+1})}{\delta_{k+1}} and \det_{\mathscr{C}}\left(\log_{\widetilde{F}^{(k-1),P}}(\exp_{\widetilde{F}^{(k-1)}}(v_{1,k})),\ldots,\log_{\widetilde{F}^{(k-1),P}}(\exp_{\widetilde{F}^{(k-1)}}(v_{m,k}))\right)=\frac{w_{k+1,P}}{\delta_{k+1}}. Then we have the following equalities: \operatorname{ev}_{t_{1}=\zeta_{1},\ldots,t_{k}=\zeta_{k}}\left(1\otimes L(\widetilde{F}/\widetilde{\mathscr{O}_{L,s}})\right)=\dfrac{w_{k+1}}{x_{k}\delta_{k+1}} and \operatorname{ev}_{t_{1}=\zeta_{1},\ldots,t_{k}=\zeta_{k}}\left(1\otimes\dfrac{w_{1,P}}{f_{1}}\right)=\dfrac{w_{k+1,P}}{x_{k}\delta_{k+1}}. Since we can evaluate at t_{k}=\zeta_{k} the L -series and t_{k}-\zeta_{k} does not divide w_{k+1} , we can evaluate x_{k}\delta_{k+1} at t_{k}=\zeta_{k} into a non-zero element x_{k+1}\in\mathscr{K}(s)\otimes\widetilde{A_{s,k}} . We have: \operatorname{ev}_{t_{k}=\zeta_{k},\ldots,t_{z}=\zeta_{1}}(1\otimes L(\widetilde{F}/\widetilde{\mathscr{O}_{L,s}}))=\operatorname{ev}_{t_{k}=\zeta_{k}}\dfrac{w_{k+1}}{x_{k}\delta_{k+1}}=\dfrac{\det_{\mathscr{C}}(v_{i,{k+1}},i=1,\ldots,m)}{x_{k+1}} and \displaystyle\operatorname{ev}_{t_{k}=\zeta_{k},\ldots,t_{1}=\zeta_{1}}1\otimes\dfrac{w_{1,P}}{f_{1}} \displaystyle=\dfrac{\det_{\mathscr{C}}\left(\log_{\widetilde{F}^{(k)},P}(\exp_{\widetilde{F}^{(k)}}(v_{i,k+1}))\right)}{x_{k+1}} \displaystyle\in\dfrac{1}{x_{k+1}}\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathbb{T}_{z,t_{k+1},\ldots,t_{s}}(K_{P}). Last step : evaluation at z . We write z=t_{s+1} and use similar arguments to conclude. ∎ Remark 5.9 . If at some step j\in\{1,\ldots,s\} we have for all i=1,\ldots,r : \operatorname{ev}_{t_{1}=\zeta_{1},\ldots,t_{j}=\zeta_{j}}A_{i}=0, then we have for all k\geq j : E_{\theta}^{(k)}=\theta I_{d}+N_{k} with N_{k}\in M_{d}(\mathscr{K}(s)\otimes_{\mathbb{F}_{q}}\mathbb{F}_{q}[t_{k+1},\ldots,t_{s}]\mathscr{O}_{L}) a nilpotent matrix. Then we have: \partial_{E^{(k)}}(\theta^{q^{d}})=E^{(k)}_{\theta^{q^{d}}}=\theta^{q^{d}}I_{d}. Hence, if \exp_{E^{(k)}}={\sum}\limits_{n\geq 0}d_{n,k}\tau_{s}^{k} , then from the functional equation of the exponential map we obtain for all n\geq 0 : (\theta^{q^{d}})^{q^{n}}d_{n,k}=\theta^{q^{d}}d_{n,k}. Thus d_{0,k}=I_{d} and d_{n,k}=0 for all n\geq 1 so finally \exp_{E^{(k)}}=I_{d}\tau^{0} for all k\geq j . Then we have U(\widetilde{E}^{(j)};\widetilde{\mathscr{O}_{L,s,j}})=\operatorname{Lie}_{\widetilde{E}^{(j)}}(\widetilde{\mathscr{O}_{L,s,j}}) for all j\geq k , and the previous proof is still valid. 6. Applications In this section we investigate the case where L=K and d=1 , i.e., the case of Drinfeld modules defined over the ring A itself. 6.1. Preliminaries We consider a Drinfeld module \phi:A\rightarrow A\{\tau\} of rank r\geq 1 and we denote by \exp_{\phi}={\sum}\limits_{n\geq 0}d_{n}\tau^{n}\in K\{\{\tau\}\} its associated exponential map. We call a period of \phi any element \lambda\in\mathbb{C}_{\infty} such that \exp_{\phi}(\lambda)=0 . We denote by \Lambda_{\phi} its period lattice which is a free A -module of rank r . Let NP(\phi) be the Newton polygon associated with \exp_{\phi} which is defined as the lower convex hull of the points P_{n}=(q^{n}-1,v_{\infty}(d_{n})) ). Remark that the zeros of \exp_{\phi} are all simple since \dfrac{d}{dx}\exp_{\phi}x=1 . We have the following property about the edges of NP(\phi) that can be found in [ 18 , Theorem 2.5.2] . Proposition 6.1 . Consider \lambda a non-zero period of \phi of valuation x , and let N be the number of periods of valuation equal to x . Then NP(\phi) has a single edge with slope \lambda and length N . By Lemma 3.3 , we know that \lim\limits_{n\rightarrow+\infty}v_{\infty}(d_{n})=+\infty , then we define N_{0} as the smallest n such that v_{\infty}(d_{n}) is minimal, and N_{1} as the largest n such that v_{\infty}(d_{n}) is minimal. Another way of looking at N_{0} and N_{1} is that the edge of slope equals to 0 of NP(\phi) has endpoints P_{N_{0}} and P_{N_{1}} . We will use the concept of successive minimum basis from [ 14 , section 3] . Definition 6.2 . An ordered A -basis (\lambda_{1},\ldots,\lambda_{r} ) of the A -lattice \Lambda_{\phi} in \mathbb{C}_{\infty} is a successive minimum basis ( shortly an SMB) if for each 1\leq i\leq r , the vector \lambda_{i} has minimal valuation v_{\infty}(\lambda_{i}) among all w\in\Lambda_{\phi} not in the span {\sum}\limits_{1\leq js . Denote also by x_{i}=v_{\infty}(\lambda_{i}) for all i=1,\ldots,r . Then we consider n_{1},\ldots,n_{t}\in\{1,\ldots,s\} be such that x_{n_{i}}\in\mathbb{N} for i=1,\ldots,t , and we denote by S_{\phi}=\{n_{1},\ldots,n_{t}\} . Proposition 6.4 . We have the following equality: N_{0}-N_{1}=t. Proof. First we calculate N_{0}. To do this, we need to count the total length of the strictly negative slopes (which is also equal to q^{N_{0}}-1 by definition), in other words the number of non-zero periods of strictly positive valuation. Set \lambda={\sum}\limits_{i=1}^{r}a_{i}\lambda_{i}\in\Lambda_{\phi} . We have the equivalence by Proposition 6.3 : v_{\infty}(\lambda)>0\Leftrightarrow\left\{\begin{aligned} &a_{i}=0\text{ if }i>s,\\ &\deg(a_{i})\leq\left\lfloor x_{i}\right\rfloor\text{ if }i\leq s\text{ and }i\notin S_{\phi},\\ &\deg(a_{i})0 . Let N_{0},m_{1},\ldots,m_{l},N_{1} be the integers n such that v_{\infty}(d_{n}) is minimal. Let \beta_{n}\in\mathbb{F}_{q}^{\ast} be the sign of d_{n} , we obtain: \operatorname{sgn}(u_{\psi}(z))=z^{N_{0}}(\beta_{N_{0}}+\ldots+\beta_{N_{1}}z^{N_{1}-N_{0}})\in\mathbb{F}_{q}[z] that has at most r non-zero roots by Proposition 6.4 . Thus, u_{\psi}(z) is divisible at most by (z-1)^{r} . ∎ Note that we have proved more precisely: \operatorname{ord}_{z=1}L_{P}(\widetilde{\phi}/\widetilde{A})\leq\#\{i=1\ldots,r\ |\ v_{\infty}(\lambda_{i})\in\mathbb{Z}\}. Proposition 6.8 . The previous inequality is not an equality in general. Proof. Consider the Drinfeld module given by \phi_{\theta}=\theta+\theta\tau^{2} with q=3 . One can prove that the Newton polygon of the associated exponential map is the polygon beginning at the point (0,0) and has successive slopes of length (q^{2k+2}-q^{2k}) and equal to k+1 . Thus, the number of periods of an SMB having valuation \in\mathbb{Z} is equal to 2 . By [ 9 , Proposition 2.21] we have u_{\phi}(1)=1 . One can prove that \phi does not have A -torsion. Then by Proposition 6.5 we obtain \operatorname{ord}_{z=1}L_{P}(\widetilde{\phi}/\widetilde{A})=0 . ∎ Remark 6.9 . For any r\geq 1 , we can construct explicit Drinfeld modules of rank r whose vanishing order of the associated P -adic L -series equals r . In fact, denote by (-1)^{r}(z-1)^{r}=1+{\sum}\limits_{i=1}^{r}\alpha_{i}z^{i} , with \alpha_{i}\in\mathbb{F}_{q} and consider the Drinfeld module given by \phi_{\theta}=\theta+{\sum}\limits_{i=1}^{r}\alpha_{i}\theta^{q^{i}}\tau^{i} . By [ 9 , Proposition 2.21] , we have: u_{\phi}(z)=1+{\sum}\limits_{i=1}^{r}\alpha_{i}z^{i}=(-1)^{r}(z-1)^{r} . Then the vanishing order at z=1 is greater than or equal to r , so equals r . To conclude the text, we would like to ask the following question from a personal communication with Xavier Caruso and Quentin Gazda. Problem 6.10 . Do we have the following equality \operatorname{ord}_{z=1}L_{P}(\widetilde{\phi}/\widetilde{A})=\#\left\{i\in\{1,\ldots,r\}\ |\ \lambda_{i}\in\bigcup\limits_{n\geq 0}\mathbb{F}_{q^{p^{n}}}((\frac{1}{\theta}))\right\}?