Dirichlet series expansions of p-adic L-functions

Abstract

We study p-adic L-functions $L_p(s,\chi )$ for Dirichlet characters $\chi$. We show that $L_p(s,\chi )$ has a Dirichlet series expansion for each regularization parameter c that is prime to p and the conductor of $\chi$. The expansion is proved by transforming a known formula for p-adic L-functions and by controlling the limiting behavior. A finite number of Euler factors can be factored off in a natural manner from the p-adic Dirichlet series. We also provide an alternative proof of the expansion using p-adic measures and give an explicit formula for the values of the regularized Bernoulli distribution. The result is particularly simple for $c=2$, where we obtain a Dirichlet series expansion that is similar to the complex case.

1 Introduction

Let p be a prime, let $q=p$ if p is odd and $q=4$ if $p=2$, and let $\chi$ be a Dirichlet character of conductor f. A p-adic L-function $L_p(s,\chi )$ for a Dirichlet character $\chi$ is a p-adic meromorphic function and an analogue of the complex L-function. For powers of the Teichmüller character $\omega$ of conductor q, one obtains the p-adic zeta functions $\zeta _{p,i}=L_p(s,\omega ^{1-i})$, where $i=0,\, 1,\, \dots ,\, p-2$ ($i=0, 1$ if $p=2$). It is well known that $L_p(s,\chi )$ is identically zero for odd $\chi$. p-adic L-functions have a long history and the primary constructions going back to Kubota-Leopoldt [6] and Iwasawa [3] are via the interpolation of special values of complex L-functions.

It can also be shown that p-adic L-functions are in fact Iwasawa functions.

It is well known that for $Re(s) > 0$,

$$\begin{aligned} (1-2^{1-s}) \zeta (s) = \sum _{n=1}^{\infty } \frac{(-1)^{n+1}}{n^s} \end{aligned}$$

and, more generally, if $c\ge 2$ is an integer,

$$\begin{aligned} (1-\chi (c)c^{1-s})L(s, \chi ) = \sum _{n=1}^{\infty } \chi (n) \frac{a_{c,n}}{n^s}, \end{aligned}$$

where $a_{c,n} = 1-c$ if $n\equiv 0 \text { mod}\ c$ and $a_{c,n} = 1$ if $n\not \equiv 0 \text { mod}\ c$. In the following, we derive similar, but slightly different, expansions for p-adic L-functions.

An explicit formula for $L_p(s,\chi )$ is given in [9] (Theorem 5.11): let F be any multiple of q and f. Then $L_p(s,\chi )$ is a meromorphic function (analytic if $\chi \ne 1)$ on $\{ s \in {\mathbb {C}}_p\ |\ |s| < qp^{-1/(p-1)} \}$ such that

$$L_{p} (s,\chi ) = \frac{1}{F}\frac{1}{{s - 1}}\sum\limits_{\substack{\mathop {a=1}\\ {p\not\mid a}}}^F \chi (a)\langle a\rangle ^{{1 - s}} \sum\limits_{{j = 0}}^{\infty } {\left( {\begin{array}{*{20}c} {1 - s} \\ j \\ \end{array} } \right)} \left( {\frac{F}{a}} \right)^{j} B_{j} .{\text{ }}$$

(1)

In Sect. 2, we will use formula (1) to derive a Dirichlet series expansion of $L_p(s,\chi )$.

p-adic L-functions can be also be defined using distributions and measures. Let $\chi$ have conductor $f=dp^m$ with $(d,p)=1$. Choose an integer $c \ge 2$, where $(c,dp)=1$. Then there is a measure $E_{1,c}$ on $({\mathbb {Z}}/d {\mathbb {Z}})^{\times } \times \, {\mathbb {Z}}_p^{\times }$ (the regularized Bernoulli distribution) such that

$$\begin{aligned} -(1- \chi (c) \langle c \rangle ^{1-s}) L_p(s, \chi ) = \int _{({\mathbb {Z}}/d {\mathbb {Z}})^{\times } \times \, {\mathbb {Z}}_p^{\times }} \chi \omega ^{-1}(a)\langle a\rangle ^{-s} \ dE_{1,c} \end{aligned}$$

(2)

(see [9] Theorem 12.2). In Sect. 3, we give an explicit formula for the values of $E_{1,c}$ and derive the Dirichlet series expansion from (2).

The expansion is particularly simple for $c=2$, and this parameter can be used for $p \ne 2$ and Dirichlet characters with odd conductor. For this case we obtain similar results as in [1, 2], and [4]. In Sect. 4, we provide examples for different parameters c.

2 Expansions of p-adic L-functions

First, we derive an approximation of $L_p(s,\chi )$ that is close to the original definition of Kubota-Leopoldt (see [6]).

For $r \in {\mathbb {C}}_p^{\times }$ we write $\delta (r)$ for a term with p-adic absolute value $\le |r|$.

Proposition 2.1

Let p be a prime number, $\chi$ an even Dirichlet character of conductor f, and F a multiple of q and f. For $s \in {\mathbb {C}}_p$ with $|s| < qp^{-1/(p-1)}$, we have

$$L_{p} (s,\chi ) = \frac{1}{F}\frac{1}{{s - 1}}\sum\limits_{\substack{\mathop {a=1}\\{p\not\mid a}}}^F \chi (a)\langle a\rangle ^{{1 - s}} + \delta (F/qp).$$

(3)

Proof

We use formula (1) above and look at the series $\sum _{j=0}^{\infty } \left( {\begin{array}{c}1-s\\ j\end{array}}\right) \left( \frac{F}{a} \right) ^j B_j$. The first two terms are $1+(1-s) \frac{-F}{2a}$. We claim that the p-adic absolute value of the other terms ($j \ge 2$) is less than or equal to $| (s-1)F^2/qp |$. To this end, we note that $|1/j!|\le p^{(j-1)/(p-1)}$ and

$$\begin{aligned} \left| \left( {\begin{array}{c}1-s\\ j\end{array}}\right) \right| \le |1-s|\, p^{(j-1)/(p-1)}(qp^{-1/(p-1)})^{j-1}= |1-s|\, q^{j-1} \end{aligned}$$

since we assumed that $|s| < qp^{-1/(p-1)}$. Since $|F|\le \frac{1}{q}$, $|a|=1$, and $|B_j| \le p$, we obtain

$$\begin{aligned} \left| \left( {\begin{array}{c}1-s\\ j\end{array}}\right) \left( \frac{F}{a} \right) ^j B_j \right| \le |1-s| \, q^{j-1} q^{2-j} |F|^2\, p = |1-s| \, |F|^2\, q p. \end{aligned}$$

Then (1) implies

$$ L_{p} (s,\chi ) = \frac{1}{F}\frac{1}{{s - 1}}\sum\limits_{\substack{\mathop {a=1}\\ {p\not\mid a}}}^F \chi (a)\langle a\rangle ^{{1 - s}} + \frac{1}{2}\sum\limits_{\substack{\mathop{a = 1} \\ {p\not\mid a}}}^{F} \chi \omega ^{{ - 1}} (a)\langle a\rangle ^{{ - s}} + \delta \left( {F/qp} \right). $$

It remains to show that the second sum can be absorbed into $\delta (F/qp)$. We have

$$\begin{aligned} \sum\limits_{\substack{\mathop {a=1}\\ {p\not\mid a}}}^F \chi \omega ^{{ - 1}} (a)\left\langle a \right\rangle ^{{ - s}} & = \sum\limits_{\substack{\mathop {b=1}\\ {p\not\mid b}}}^F \chi \omega ^{{ - 1}} (F - b)\left\langle {F - b} \right\rangle ^{{ - s}} \\ & = - \sum\limits_{\substack{\mathop {b=1}\\{p\not\mid b}}}^F \chi \omega ^{{ - 1}} (b)\left\langle {b - F} \right\rangle ^{{ - s}} \\ & = - \sum\limits_{\substack{\mathop {b=1}\\ {p\not\mid b}}}^F \chi \omega ^{{ - 1}} (b)\left\langle b \right\rangle ^{{ - s}} + \delta (F/qp^{{ - 1/(p - 1)}} ). \end{aligned}$$

The last step can be justified by noting that

$$\begin{aligned} \frac{\langle b-F\rangle ^{-s}}{\langle b\rangle ^{-s}} = \left( 1-\frac{F}{b}\right) ^{-s} = 1 +\sum _{j=1}^{\infty } \left( {\begin{array}{c}-s\\ j\end{array}}\right) \left( \frac{-F}{b}\right) ^j = 1 + \delta (F/qp^{-1/(p-1)}), \end{aligned}$$

since $|s|< qp^{-1/(p-1)}$ (this is the same estimate as earlier, without the presence of the Bernoulli number). This proves the proposition. $\square$

Remark 2.2

For $F=fp^n$ and $n \rightarrow \infty$, formula (3) gives the original definition of $L_p(s,\chi )$ by Kubota and Leopoldt (see [6]).

Remark 2.3

Suppose that $p\ne 2$. Then the error term in the above Proposition (as well as in the following Theorem 2.4) can be improved to $\delta (F/p^{2-(p-2)/(p-1)})$. First we note that $B_j=0$ for odd $j \ge 3$. By the von Staudt–Clausen Theorem (see [9] 5.10), we have for even $j \ge 2$: $|B_j| =p$ iff $(p-1) \mid j$, and otherwise $|B_j| \le 1$. Furthermore, $|1/j!| = p^{(j-S_j)/(p-1)}$, where $S_j$ is the sum of the digits of j, written to the base p (see [5]). Since $j \equiv S_j \text { mod}\ (p-1)$, $j \equiv 0 \text { mod}\ (p-1)$ is equivalent to $S_j \equiv 0 \text { mod}\ (p-1)$. We conclude that $|B_j |=p$ yields $S_j \ge p-1$ and $|1/j!| \le p^{(j-1)/(p-1) } p^{-(p-2)/(p-1)}$. This implies the above error term. We also see that this error term cannot be further improved.$\lozenge$

Now we give the Dirichlet expansion of $L_p(s,\chi )$. For $m \in {\mathbb {N}}$, we denote by $\{x\}_{m}$ the unique representative of $x \text { mod}\ m{\mathbb {Z}}$ between 0 and $m-1$.

Theorem 2.4

Let p be a prime number, $\chi$ be an even Dirichlet character of conductor f, and F a multiple of q and f. Let $c>1$ be an integer satisfying $(c,F)=1$. For $a \in {\mathbb {Z}}$, define

$$\begin{aligned} \epsilon _{a,c,F} = \frac{c-1}{2} - \{ -a F^{-1} \}_{c} \, \in \left\{ - \frac{c-1}{2} ,\, - \frac{c-1}{2} + 1,\, \dots ,\ \frac{c-1}{2}\right\} . \end{aligned}$$

Then we have for $s \in {\mathbb {C}}_p$ with $|s| < qp^{-1/(p-1)}$ the formula

$$\begin{aligned} -(1- \chi (c) \langle c \rangle ^{1-s}) L_p(s, \chi ) = \sum\limits_{\substack{\mathop {a=1} \\ {p\not\mid a}}}^F \chi \omega ^{-1}(a) \langle a \rangle ^{-s} \epsilon _{a,c,F} + \delta (F/qp) . \end{aligned}$$

Proof

Use (3) with cF in place of F, and subtract $\chi (c) \langle c \rangle ^{1-s}$ times (3) with F, to obtain

$$\begin{aligned} (1 - \chi (c)\langle c\rangle ^{{1 - s}} )L_{p} (s,\chi ) & = \frac{1}{{cF}}\frac{1}{{s - 1}}\sum\limits_{\substack{\mathop {a=1}\\{p\not\mid a}}}^{cF} \chi (a)\langle a\rangle ^{{1 - s}} \\ & - \frac{1}{F}\frac{1}{{s - 1}}\sum\limits_{\substack{\mathop {a=1}\\ {p\not\mid a}}}^F \chi (ac)\langle ac\rangle ^{{1 - s}} + \delta (F/qp). \end{aligned}$$

(4)

Let $0<a_0 < F$ with $(a_0,\, p)=1$. Since we assumed $(c,\, F)=1$ and $p \mid F$, there is a unique number of the form $a_0 c$ with $0<a_0c < cF$ and $(a_0 c,\, p)=1$ in each congruence class modulo F relatively prime to p. The first sum in (4) can be written as

$$\begin{aligned} \frac{1}{cF} \frac{1}{s-1} \sum\limits_{\substack{\mathop {a_0=1}\\ {p\not\mid a_0}}}^F \chi (a_0 c) \langle a_0 c \rangle ^{1-s} \left( \sum\limits_{\substack{\mathop {a=1} \\ \mathop{a \equiv a_{0} c{\text{ }}\bmod \;F}}}^{cF} \left\langle 1+ \frac{a-a_0 c}{a_0 c} \right\rangle ^{1-s} \right) \\ \quad = \frac{1}{cF} \frac{1}{s-1} \sum\limits_{\substack{\mathop {a_0=1} \\ {p\not\mid a_0}}}^F \chi (a_0 c) \langle a_0 c \rangle ^{1-s} \left( \sum\limits_{\substack{\mathop{a=1} \\ \mathop {a \equiv a_{0} c{\text{ }}\bmod \;F}}}^{cF} \left( 1+ (1-s) \frac{a-a_0 c}{a_0 c} \right) \right) + \delta (F/q) . \end{aligned}$$

Note that $| \frac{a-a_0 c}{a_0 c} | \le |F|$, so this is the same type of estimate used in the proof of Proposition 2.1. Subtracting the second sum in (4) yields

$$\begin{gathered} (1 - \chi (c)\langle c\rangle ^{{1 - s}} )L_{p} (s,\chi ) \hfill \\ = \frac{{ - 1}}{{cF}}\sum\limits_{\substack{\mathop {a_0=1}\\ {p\not\mid a_0}}}^F \chi (a_{0} c)\langle a_{0} c\rangle ^{{1 - s}} \left( \sum\limits_{\substack{\mathop{a=1} \\ {a \equiv a_{0} c{\text{ }}\bmod \;F}}}^{cF} \frac{a-a_0 c}{a_0 c} \right) + \delta (F/qp) \hfill \\ = \frac{{ - 1}}{c} \sum\limits_{\substack{\mathop {a_0=1}\\{p\not\mid a_0}}}^F \chi \omega ^{{ - 1}} (a_{0} c)\langle a_{0} c\rangle ^{{ - s}} \left( \sum\limits_{\substack{\mathop{a=1} \\ \mathop {a \equiv a_{0} c{\text{ }}\bmod \;F}}}^{cF} \frac{a-a_0 c}{F} \right) + \delta (F/qp) . \hfill \\ \end{gathered}$$

We compute the inner sum. Let $b=\{a_0 c\}_F$. Then $a_0 c = b + \{ -F^{-1} b \}_c\, F$, since the latter sum is congruent to b modulo F and congruent to 0 modulo c. If a satisfies $a \equiv a_0 c \text { mod}\ F$ and $0<a<cF$, then $a=b+jF$ with $0\le j <c$. Hence

$$\begin{aligned} \sum\limits_{\substack{\mathop{a=1} \\ \mathop {a \equiv a_{0} c{\text{ }}\bmod \;F}}}^{cF} \frac{a-a_0 c}{F} = \sum _{j=0}^{c-1} (j - \{ - F^{-1} b \}_c) = c\ \epsilon _{b,c,F}\ . \end{aligned}$$

Since $b \equiv a_0 c \text { mod}\ F$, we have $\chi \omega ^{-1}(b) \langle b \rangle ^{-s} = \chi \omega ^{-1}(a_0 c) \langle a_0 c \rangle ^{-s} + \delta (F/q)$ by the same estimate as earlier, so

$$- (1 - \chi (c)\langle c\rangle ^{{1 - s}} )L_{p} (s,\chi ) = \sum\limits_{\substack{\mathop {b=1}\\ {p\not\mid b}}}^F \chi \omega ^{{ - 1}} (b)\langle b\rangle ^{{ - s}} \epsilon _{{b,c,F}} + \delta (F/qp).{\text{ }}$$

This completes the proof. $\square$

We can take the limit of $F=fp^n$ as $n \rightarrow \infty$ and obtain:

Corollary 2.5

Let p be a prime number, $\chi$ an even Dirichlet character of conductor f, and $c>1$ an integer satisfying $(c,pf)=1$. Then we have for $s \in {\mathbb {C}}_p$ with $|s| < qp^{-1/(p-1)}$,

$$\begin{aligned} -(1- \chi (c) \langle c \rangle ^{1-s}) L_p(s, \chi ) = \lim _{n \rightarrow \infty } \sum\limits_{\substack{\mathop {a=1} \\ {p\not\mid a}}}^{fp^n} \chi \omega ^{-1}(a) \frac{ \epsilon _{a,c,fp^n} }{ \langle a \rangle ^{s}}. \end{aligned}$$

The next Theorem shows that a finite number of Euler factors can be factored off in a similar way as in [8], where a weak Euler product was obtained. The main statement is that the remaining Dirichlet series has the expected form, similar to the complex case.

Theorem 2.6

Let p be a prime number and let $\chi$ be an even Dirichlet character of conductor f. Let S be any finite (or empty) set of primes not containing p and set $S^+ = S \cup \{p\}$. Let F be a multiple of q, f and all primes in S. Let $c>1$ be an integer satisfying $(c,F)=1$. Then we have for $s \in {\mathbb {C}}_p$ with $|s| < qp^{-1/(p-1)}$ the formula

$$- (1 - \chi (c)\langle c\rangle ^{{1 - s}} ) \cdot \prod\limits_{{l \in S}} {(1 - \chi \omega ^{{ - 1}} (l)\langle l\rangle ^{{ - s}} )} \; \cdot \;L_{p} (s,\chi ) = \sum\limits_{\substack{\mathop {a=1}\\{ \left( {a,S^{ + } } \right) = 1 }}}^F\chi \omega ^{{ - 1}} (a)\frac{{ \epsilon_{{a,c,F}} }}{{\langle a\rangle ^{s} }}\; + \;\delta (F/qp).$$

Proof

We prove the statement by induction on |S|. By Theorem 2.4, the formula is true for $S=\varnothing$. Now assume the formula is true for S, and $l \ne p$ is a prime with $l \notin S$ and $(c,l)=1$. It suffices to prove the following formula:

$$\begin{aligned} \begin{aligned} (1-\chi \omega ^{-1}(l) \langle l \rangle ^{-s} ) \sum\limits_{\substack{\mathop {a=1} \\ \left( {a,S^{ + } } \right) = 1 }}^F \chi \omega ^{-1}(a) \langle a \rangle ^{-s} \epsilon _{a,c,F}\ = \\ \sum\limits_{\substack{\mathop {a=1} \\ \left( {a,S^{ + } \cup \left\{ l \right\}} \right) = 1 }}^{lF} \chi \omega ^{-1}(a) \langle a \rangle ^{-s} \epsilon _{a,c,lF}\ + \ \delta (F/qp) . \end{aligned} \end{aligned}$$

(5)

Note that $|1-\chi \omega ^{-1}(l) \langle l \rangle ^{-s} | \le 1$ and $|lF|=|F|$, so we can keep the error term. We can use lF in place of F and write the left side of (5) as

$$\begin{aligned} \sum\limits_{\substack{\mathop {a=1} \\ \left( {a,S^{ + } } \right) = 1 }}^{lF} \chi \omega ^{-1}(a) \langle a \rangle ^{-s} \epsilon _{a,c,lF} \ - \sum\limits_{\substack{ \mathop {a=1} \\ \left( {a,S^{ + } } \right) = 1 }}^{F} \chi \omega ^{-1}(la) \langle la \rangle ^{-s} \epsilon _{a,c,F} + \delta (F/qp) . \end{aligned}$$

(6)

Now we have

$$\begin{aligned} \epsilon _{la,c,lF} = \frac{c-1}{2} - \{ -la (lF)^{-1} \}_{c} = \frac{c-1}{2} - \{ -a F^{-1} \}_{c} = \epsilon _{a,c,F}. \end{aligned}$$

Thus (6) is equal to

$$\begin{aligned} \sum\limits_{\substack{\mathop {a=1}\\ \left( {a,S^{ + } } \right) = 1 }}^{lF} \chi \omega ^{- 1}(a) \langle a \rangle ^{-s} \epsilon _{a,c,lF} \ - \sum\limits_{\substack{\mathop {a=1}\\ \left( {a,S^{ + } } \right) = 1 }}^F \chi \omega ^{-1}(la) \langle la \rangle ^{-s} \epsilon _{la,c,lF} + \ \delta (F/qp) \\ & = \sum\limits_{\substack{\mathop{a = 1} \\ \left( {a,S^{ + } } \right) = 1 \\ l\not\mid a} }^{{lF}} {} \chi \omega ^{-1}(a) \langle a \rangle ^{-s} \epsilon _{a,c,lF}\ + \ \delta (F/qp), \end{aligned}$$

which shows equation (5).

$\square$

Remark 2.7

What happens if S contains more and more primes? It is well known that the Euler product does not converge p-adically (see [2]), since the factors $(1-\chi \omega ^{-1}(l) \langle l \rangle ^{-s} )$ have absolute value $\le 1$ and do not converge to 1 as $l \rightarrow \infty$. Furthermore, there are infinitely many primes l with $\chi \omega ^{-1} (l) = 1$ and $(1 - \langle l \rangle ^{-s})^{-1}$ has a pole at $s=0$. We have for $l \ne p$ and $|s| < qp^{-1/(p-1)}$,

$$\begin{aligned} 1 - \langle l \rangle ^{-s} = - \sum _{j=1}^{\infty } \left( {\begin{array}{c}-s\\ j\end{array}}\right) ( \langle l \rangle -1 )^j . \end{aligned}$$

The p-adic absolute value of each term of the above series is less than

$$\begin{aligned} (qp^{-1/(p-1)})^j p^{(j-1)/(p-1)} q^{-j} = p^{-1/(p-1)} < 1 . \end{aligned}$$

Hence the product $\prod _{l \in S} (1-\chi \omega ^{-1}(l) \langle l \rangle ^{-s} )$ approaches 0 as S expands to include all primes.

3 Regularized Bernoulli distributions

Let p be a prime number and let d be a positive integer with $(d,p)=1$. Define $X_n = ({\mathbb {Z}}/dp^n {\mathbb {Z}})$ and $X = \varprojlim X_n \cong {\mathbb {Z}}/d {\mathbb {Z}}\times {\mathbb {Z}}_p$. Let $k \ge 1$ be an integer. Then the Bernoulli distribution $E_k$ on X is defined by

$$\begin{aligned} E_k( a + dp^ n X) = (dp^n)^{k-1} \frac{1}{k} B_k \left( \frac{ \{ a \}_{d p^n} }{d p^n} \right) , \end{aligned}$$

where $B_k(x)$ is the k-th Bernoulli polynomial and $B_k=B_k(0)$ are the Bernoulli numbers (see [5, 7]). For $k=1$, one has $B_1(x)= x - \frac{1}{2}$. Choose $c \in {\mathbb {Z}}$ with $c \ne 1$ and $(c,dp)=1$. Then the regularization $E_{k,c}$ of $E_k$ is defined by

$$\begin{aligned} E_{k,c} ( a + dp^n X ) = E_k ( a + dp^n X) - c^k E_k \left( \left\{ \frac{a}{c} \right\} _{d p^n} + d p^n X \right) . \end{aligned}$$

One shows that the regularized Bernoulli distributions $E_{k,c}$ are measures (see [7]). In the following, we consider only $k=1$; the cases $k \ge 2$ are similar.

Theorem 3.1

Let p be a prime, $c,\, d \in {\mathbb {N}}$, and $c \ge 2$ such that $(c,dp)=1$. Let X be as above, and let $E_{1,c}$ be the regularized Bernoulli distribution on X. For $a \in \{0,1,\dots , dp^n-1\}$, we have

$$\begin{aligned} E_{1,c} ( a + dp^n X) = \frac{c-1}{2} - \{ -a (dp^n)^{-1} \}_{c} = \epsilon _{a,c,dp^n}. \end{aligned}$$

Proof

By definition,

$$\begin{aligned} E_{1,c}(a+dp^n X) = E_1(a + dp^n X) - c E_1(c^{-1}a + dp^n X) = \frac{a}{dp^n} - \frac{1}{2} - c \left( \frac{ \{c^{-1} a \}_{dp^n}}{dp^n} \right) + \frac{c}{2}. \end{aligned}$$

We give the standard representative of $c^{-1} a \text { mod}\ dp^n$:

$$\begin{aligned} \{c^{-1} a \}_{dp^n} = \frac{ \{ -a (dp^n)^{-1} \}_{c}\ dp^n + a}{c} \end{aligned}$$

Note that the numerator is divisible by c, since $\{ -a (dp^n)^{-1} \}_{c}\ dp^n \equiv -a \text { mod}\ c$. Hence the quotient is an integer between 0 and $dp^n-1$. Furthermore, the numerator is congruent to a modulo $dp^n$, and so the quotient has the desired property. We obtain

$$\begin{aligned} E_{1,c}(a+dp^n X) = \frac{a}{dp^n} + \frac{c-1}{2} - \frac{ \{ -a (dp^n)^{-1} \}_{c}\ dp^n + a}{dp^n} = \frac{c-1}{2} - \{ -a (dp^n)^{-1} \}_{c} \end{aligned}$$

which is the assertion. $\square$

Now the Dirichlet series expansion in Corollary 2.5 follows from Theorem 3.1 and the integral formula (2).

4 Expansions for different regularization parameters

We look at the coefficients $\epsilon _{a,c,dp^n}$ for different parameters c and the resulting Dirichlet series expansions. The following observation follows directly from the definition.

Remark 4.1

The sequence of values $E_{1,c} ( a + dp^n X) = \epsilon _{a,c,dp^n}$ for $a=$ 0,1, 2, $\dots$, $dp^n-1$ is periodic with period c. The sequence begins with $\frac{c-1}{2}$ and continues with a permutation of $\frac{c-3}{2}, \dots , - \frac{c-1}{2}$. If we restrict to values of n such that $dp^n$ lies in a fixed congruence class modulo c, then the values do not change as $n \rightarrow \infty$. $\lozenge$

The measure $E_{1,c}$ and the Dirichlet series expansion are particularly simple for $c=2$. Note that we assumed that d and p are odd in this case. If a is even, then $\{ -a (dp^n)^{-1} \}_{2} = 0$ and

$$\begin{aligned} E_{1,2}(a+dp^n X) = \epsilon _{a,2,dp^n} = \frac{1}{2}. \end{aligned}$$

If a is odd, then $-a(dp^n)^{-1}$ is odd, $\{ -a (dp^n)^{-1} \}_{2} = 1$ and

$$\begin{aligned} E_{1,2}(a+dp^n X) = \epsilon _{a,2,dp^n} = - \frac{1}{2}. \end{aligned}$$

Hence $E_{1,2}$ is up to the factor $\frac{1}{2}$ equal to the following simple measure:

Definition 4.2

Let $p \ne 2$ be a prime, and let $X \cong {\mathbb {Z}}/d{\mathbb {Z}}\times {\mathbb {Z}}_p$ be as above. Then

$$\begin{aligned} \mu (a + dp^n X) = (-1)^{\{a\}_{dp^n}} \end{aligned}$$

defines a measure on X. We call $\mu$ the alternating measure, since the measure of all clopen balls is $\pm 1$.$\lozenge$

The corresponding integral is also called the fermionic p-adic integral (see [4]).

Now we obtain the following Dirichlet series expansion from Corollary 2.5.

Corollary 4.3

Let $p \ne 2$ be a prime number, and let $\chi$ be an even Dirichlet character of odd conductor f. Then we have for $s \in {\mathbb {C}}_p$ with $|s| < p^{(p-2)/(p-1)}$,

$$\begin{aligned} (1- \chi (2) \langle 2 \rangle ^{1-s}) L_p(s, \chi ) = \lim _{n \rightarrow \infty } \frac{1}{2} \sum\limits_{\substack{\mathop {a=1}\\{p\not\mid a}}}^{fp^{n}} (-1)^{a+1} \chi \omega ^{-1}(a) \frac{1}{\langle a \rangle ^{s}} . \end{aligned}$$

For $\chi =\omega ^{1-i}$ and odd $i=1,\, \dots ,\, p-2$, we obtain the branches of the p-adic zeta function:

$$\begin{aligned} \zeta _{p,i}(s) = L_p(s,\, \omega ^{1-i}) = \frac{1}{1-\omega (2)^{1-i} \langle 2 \rangle ^{1- s}} \cdot \lim _{n \rightarrow \infty } \frac{1}{2}\sum\limits_{\substack{\mathop {a=1} \\ {p\not\mid a}}}^{p^{n}} (-1)^{a+1}\omega (a)^{-i}\frac{1}{ \langle a \rangle ^{s}} \end{aligned}$$

Remark 4.4

Dirichlet series expansions of p-adic L-functions were studied by D. Delbourgo in [1] and [2]. He considers Dirichlet characters $\chi$ satisfying $(p,\, 2 f \phi (f))=1$ and their Teichmüller twists. We obtain the same expansion for $c=2$ and $\chi = \omega ^{1-i}$. However, we require $(c,\, fp)=1$ and use other methods for the proof.

Similar expansions for a slightly different p-adic L-function using a fermionic p-adic integral (i.e., $c=2$) were also obtained by M.-S. Kim and S. Hu (see [4]).

Example 4.5

We look at the case $c=3$. The sequence of values $\epsilon _{a,3,dp^n}$ is periodic with period 3. If $dp^n \equiv 1 \text { mod}\ 3$, then the sequence is $1,\ -1,\ 0,\, \dots$ . If $dp^n \equiv 2 \text { mod}\ 3$, then we obtain the sequence $1,\ 0,\ -1,\, \dots$ .

Corollary 4.6

Let p be a prime number, and let $\chi$ be an even Dirichlet character of conductor $f=dp^m$ such that $(3,\, dp)=1$. If $d \equiv 1 \text { mod}\ 3$, then define a sequence $\epsilon _0=1$, $\epsilon _1=-1$, $\epsilon _2=0$, $\dots$ with period 3. Otherwise, set $\epsilon _0=1$, $\epsilon _1=0$, $\epsilon _2=-1$ and extend it with period 3. Then we have for $s \in {\mathbb {C}}_p$ with $|s| < qp^{-1/(p-1)}$,

$$\begin{aligned} -(1- \chi (3) \langle 3 \rangle ^{1-s}) L_p(s, \chi ) = \lim _{n \rightarrow \infty } \sum\limits_{\substack{\mathop {a=1} \\ {p\not\mid a}}}^{ dp^{{2n}} } \chi \omega ^{-1}(a)\frac{\epsilon _a}{ \langle a \rangle ^{s} }. \end{aligned}$$

Example 4.7

For $c=5$, we get a periodic sequence with period 5 and we have $\epsilon _{a,5,dp^n}=2$ for $a \equiv 0 \text { mod}\ 5$. The next four coefficients are a permutation of the values $-2$, $-1$, 0 and 1, depending on the class of $dp^n \text { mod}\ 5$.

Example 4.8

Let $c=7$. Then $\epsilon _{0,7,dp^n} = 3$. Now suppose, for example, that $dp^n \equiv 3 \text { mod}\ 7$. Then $(dp^n)^{-1} \equiv 5 \text { mod}\ 7$. This yields the values

$$\begin{aligned} \epsilon _{1,7,dp^n} = 1,\ \epsilon _{2,7,dp^n} = -1,\ \epsilon _{3,7,dp^n} = -3,\ \epsilon _{4,7,dp^n} = 2,\ \epsilon _{5,7,dp^n} = 0,\ \epsilon _{6,7,dp^n} = -2, \end{aligned}$$

and these are extended with period 7.