Published March 2, 2023|4 minute read

This post motivates the complete zeta function ζA(z)=π−z/2 Γ(z/2) ζ(z)\zeta_{\mathbb{A}}(z) = \pi^{-z/2} \, \Gamma(z/2) \, \zeta(z), explaining how it arises in a natural way by including contributions from all possible completions of Q\mathbb{Q}. The argument is from Andreas Knauf’s excellent lecture notes; I just fill in some details.

By Ostrowski's theorem, the only possible absolute values on Q\mathbb{Q} are the usual absolute value ∣⋅∣=∣⋅∣∞|\cdot| = |\cdot|_\infty and the pp-adic absolute value ∣⋅∣p|\cdot|_p. The completion of Q\mathbb{Q} with respect to these absolute values give the real numbers R=Q∞\mathbb{R} = \mathbb{Q}_\infty and the pp-adic rationals Qp\mathbb{Q}_p, respectively.

One performs Fourier analysis over R\mathbb{R} by executing a change of basis to the eigenfunctions χu,∞:=exp(2πiux)\chi_{u, \infty} \defeq \mathrm{exp}(2\pi iux) of the translation group R/Z\mathbb{R}/\mathbb{Z}. Here the frequency uu takes values in Q∞\mathbb{Q}_\infty. An analogous basis exists for the pp-adics; we may take χu,p:=exp(2πi[ux]p)\chi_{u, p} \defeq \mathrm{exp}(2\pi i [ux]_p), where u∈Qpu \in \mathbb{Q}_p and [ux]p[ux]_p denotes the “fractional part.”

In either case, we may define the Fourier transform

Fvf(u):=∫Qvf(x) χu,v(x) dxv.\mathcal{F}_v f(u) \defeq \int_{\mathbb{Q}_v} f(x) \, \chi_{u, v}(x) \, \dd{x}_v.

It is natural to ask about the F\mathcal{F}-invariant functions. In the case of Q∞\mathbb{Q}_\infty, the answer is the familiar Gaussian exp(−πx2)\mathrm{exp}(-\pi x^2). For Qp\mathbb{Q}_p, a possible solution is the closed ball

γp(x):={1,∣x∣p≤1,0,∣x∣p>1\gamma_p(x) \defeq \begin{cases} 1, &|x|_p \leq 1, \\ 0, &|x|_p > 1 \end{cases}

of radius 1. Indeed, we may write out

Fpγp(u)=∫Qp1∣x∣p≤1exp(2πi[ux]p)dxp=∫Zpexp(2πi[ux]p)dxp.\begin{aligned} \mathcal{F}_p \gamma_p(u) &= \int_{\mathbb{Q}_p} \mathbf{1}_{|x|_p \leq 1} \mathrm{exp}(2\pi i [ux]_p) \dd{x}_p \\ &= \int_{\mathbb{Z}_p} \mathrm{exp}(2\pi i [ux]_p) \dd{x}_p. \end{aligned}

Here the measure dxp\dd{x}_p satisfies

dxp(a+pnZp)=p−n\dd{x}_p(a + p^n \mathbb{Z}_p) = p^{-n}

for integers aa and n≥0n \geq 0. If ∣u∣p≤1|u|_p \leq 1, then all ux∈Zpux \in \mathbb{Z}_p, so the integral reduces to

∫Zpdxp=dxp(Zp)=1=γp(u),\begin{aligned} \int_{\mathbb{Z}_p} \dd{x}_p &= \dd{x}_p(\mathbb{Z}_p) \\ &= 1 \\ &= \gamma_p(u), \end{aligned}

as claimed. If ∣u∣p=pk>1|u|_p = p^k > 1, then the integral becomes

∑a=0pk∫a+pkZpexp(2πi[ux]p)dxp=∑a=0pkexp(2πi[ua]p) dxp(a+pkZp)=p−k∑a=0pkexp(2πia/pk)=0=γp(u),\begin{aligned} &\sum_{a=0}^{p^k} \int_{a + p^k \mathbb{Z}_p} \mathrm{exp}(2\pi i [ux]_p) \dd{x}_p \\ &= \sum_{a=0}^{p^k} \mathrm{exp}(2\pi i [ua]_p) \, \dd{x}_p(a + p^k \mathbb{Z}_p) \\ &= p^{-k} \sum_{a=0}^{p^k} \mathrm{exp}(2\pi i a/p^k) \\ &= 0 \\ &= \gamma_p(u), \end{aligned}

again as claimed, where the second equality involves reshuffling the values of [ua]p[ua]_p.

Now define the function

ζv(s):=∫Qvγv(x) ∣x∣vsdxv∗,\zeta_v(s) \defeq \int_{\mathbb{Q}_v} \gamma_v(x) \, |x|_v^s \dd{x}^*_v,

where we now use the multiplicatively-invariant normalized Haar measure

dx∞∗:=dx∞∣x∣∞anddxp∗:=pp−1dxp∣x∣p,\dd{x}_\infty^* \defeq \frac{\dd{x}_\infty}{\abs{x}_\infty} \qquad \mathrm{and} \qquad \dd{x}_p^* \defeq \frac{p}{p-1} \frac{\dd{x}_p}{\abs{x}_p},

since the main object of integration ∣x∣ps|x|_p^s transforms multiplicatively (i.e. ∣x∣ps∣y∣ps=∣xy∣ps|x|_p^s |y|_p^s = |xy|_p^s). For the case of Q∞\mathbb{Q}_\infty, we straightforwardly obtain the result

∫Re−πx2∣x∣sdx∞∗=2∫0∞e−πx2xs−1dx=π−s/2Γ(s/2).\begin{aligned} \int_{\mathbb{R}} e^{-\pi x^2} |x|^s \dd{x}_\infty^* &= 2 \int_0^\infty e^{-\pi x^2} x^{s-1} \dd{x} \\ &= \pi^{-s/2} \Gamma(s/2). \end{aligned}For the case of Qp\mathbb{Q}_p, we obtain∫Qp1∣x∣p≤1∣x∣psdxp∗=pp−1∫Zp∣x∣ps−1dxp=pp−1∑n=0∞∫∣x∣p=p−n∣x∣ps−1dxp=pp−1∑n=0∞(p1−s)ndxp(pnZp−pn+1Zp)=pp−1∑n=0∞(p1−s)n p−1pn+1=11−p−s,\begin{aligned} &\int_{\mathbb{Q}_p} \mathbb{1}_{|x|_p \leq 1} |x|_p^s \dd{x}_p^* \\ &= \frac{p}{p-1} \int_{\mathbb{Z}_p} |x|_p^{s-1} \dd{x}_p \\ &= \frac{p}{p-1} \sum_{n=0}^\infty \int_{|x|_p = p^{-n}} |x|_p^{s-1} \dd{x}_p \\ &= \frac{p}{p-1} \sum_{n=0}^\infty (p^{1-s})^n \dd{x}_p(p^n \mathbb{Z}_p - p^{n+1} \mathbb{Z}_p) \\ &= \frac{p}{p-1} \sum_{n=0}^\infty (p^{1-s})^n \, \frac{p-1}{p^{n+1}} \\ &= \frac{1}{1-p^{-s}}, \end{aligned}

where we have evaluated

dxp(pnZp−pn+1Zp)=∑a=1p−1dxp(apn+pn+1Zp)=p−1pn+1.\begin{aligned} &\dd{x}_p(p^n \mathbb{Z}_p - p^{n+1} \mathbb{Z}_p) \\ &= \sum_{a=1}^{p-1} \dd{x}_p(a p^n + p^{n+1} \mathbb{Z}_p) \\ &= \frac{p-1}{p^{n+1}}. \end{aligned}

It is then natural to combine all the ζv\zeta_v to obtain an adelic zeta function

ζA(s):=∏vζv(s)=π−s/2 Γ(s/2) ζ(s),\zeta_{\mathbb{A}}(s) \defeq \prod_v \zeta_v(s) = \pi^{-s/2} \, \Gamma(s/2) \, \zeta(s),

which obeys the simple functional equation ζA(s)=ζA(1−s)\zeta_{\mathbb{A}}(s) = \zeta_{\mathbb{A}}(1-s).