Let $G$ be a connected, reductive group over a $p$-adic field. Let $A_0$ be a maximal split torus of $G$ and $P = MU$ a parabolic subgroup with Levi $M$ containing $A_0$, and opposite parabolic $\overline{P} = M \overline{U}$. I'm trying to understand where the integration formula

$$\int\limits_G f(g) dg = \gamma(P)^{-1} \int\limits_P \int\limits_{\overline{U}} f(p \bar{u}) d\bar{u} d_lp$$

comes from, where $\gamma(P) = \int\limits_{\overline{U}} \delta_P(m_p(\bar{u})) \, d\bar{u}$. This is claimed in writeup (page 6) by Waldspurger on Harish-Chandra's unpublished notes on the Plancherel formula for $p$-adic groups, in the second equality of (2).

enter image description here

I know that the product map $U \times M \times \overline{U} \rightarrow G$ gives an open immersion into $G$. So the right hand side of this equality seems like it is integration over some open set in $G$. When $P$ is minimal, I believe this open set is dense. Can the second equality in (2) be derived from the first equality?


The integral $$\int\limits_{U \times M \times \overline{U}} f(um\bar u) \delta_P(m)^{-1} d\bar u dm du = \int\limits_{P \times \overline{U}} f(p \bar{u}) d \bar u d_lp $$ is integration with respect to the product measure $d_lp d \bar{u}$ on the open dense set $P \times \overline{U} \cong P \overline{U}$ of $G$. One has to check that the Haar measure on $G$ restricts to a Haar measure on the group $P \times \overline{U}$, even though $(p, \bar u) \mapsto p \bar u$ is not a group homomorphism.

Therefore $\int\limits_{P \times \overline{U}} f(p \bar u) d \bar ud_lp$ is a scalar multiple of $\int\limits_G f(g) dg$, and to find the scalar, it suffices to compute the integral over $f = \operatorname{Char} K$. The measure $dg$ is chosen so that $\int\limits_G f(g)dg = 1$.

We extend $\delta_P$ to a continuous function on $G$ by making it trivial on $K$, so that $$\gamma(P) = \int\limits_{\overline{U}} \delta_P( \bar{u}) d \bar u$$ For each $\bar{u} \in \overline{U}$, write $\bar{u} = p_u k_u$ for $p_u \in P$ and $k_u \in K$ (nonuniquely), so that $\delta_P(\bar{u}) = \delta_P(p_u)$. Now for $f = \operatorname{Char} K$,

$$\int\limits_{\overline{U}} \int\limits_P f(p \bar{u})d\bar{u} d_lp = \int\limits_{\overline{U}} \int\limits_P f(p p_u k_u) d_lp d \bar{u}$$

$$ = \int\limits_{\overline{U}} \int\limits_P f(pk_u) \delta_P(p_u) d_lp d \bar{u} = \int\limits_{\overline{U}} \int\limits_{P \cap K} \delta_P(\bar{u}) d_lp d \bar{u} = \gamma(P)$$

since $d_lp(P \cap K) = 1$. This gives the second equality in (2) and also guarantees the convergence of the integral defining $\gamma(P)$, since the Radon measure of the open compact set $K \cap P \overline{U}$ is finite and positive.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.