\(p\)-adic Representation Theory (PadicRep)

3.2 Distributions on groups

Let \(G\) be a totally disconnected, locally compact, Hausdorff topological group.

3.2.1 Translations of functions

Definition 54 Translations
#

For \(\varphi : G \to \mathbb {C}\) and \(g \in G\) we define

\[ (R_g\varphi )(x)=\varphi (xg), \qquad (L_g\varphi )(x)=\varphi (g^{-1}x). \]

Right and left translations preserve local constancy and compact support.

Proof

3.2.2 Compactly supported distributions and convolution

Definition 57 Compactly supported distributions
#

A compactly supported distribution is a distribution whose support is compact.

Given \(D\) and \(\varphi \), the function \(x \mapsto D(R_x\varphi )\) is locally constant and compactly supported, and hence defines an element of \(C_c^{\infty }(G)\).

Definition 59 Convolution
#

The convolution of \(F\) and \(D\) is defined by

\[ (F * D)(\varphi ) = F\bigl(x \mapsto D(R_x\varphi )\bigr). \]
Lemma 60 Associativity
#

Convolution of compactly supported distributions is associative.

Proof

3.2.3 Translation actions on distributions

3.2.4 Delta distributions and a normalized Haar distribution

Definition 62 Delta distributions

The delta distribution at \(g\) evaluates test functions at \(g\).

Definition 63 Indicator functions

The indicator function of a compact clopen subset is a compactly supported locally constant function.

Definition 64 A normalized Haar distribution

We define a compactly supported distribution \(e_H\) attached to a compact open subgroup \(H\) as the average of a test function over \(H\) with respect to the normalized Haar measure on \(H\).

3.2.5 Convolution identities for delta and normalized Haar distributions

Lemma 65 Idempotence of the normalized Haar distribution

For a compact open subgroup \(H\), the normalized Haar distribution is idempotent:

\[ e_H * e_H = e_H. \]
Proof

For \(g \in G\) and test function \(\varphi \):

\[ (\delta _g * e_H)(\varphi ) = e_H(R_g\varphi ), \qquad (e_H * \delta _g)(\varphi ) = e_H(L_{g^{-1}}\varphi ). \]

If \(g \in H\), then \(\delta _g * e_H = e_H * \delta _g\).

Proof

3.2.6 Invariance of the normalized Haar distribution

For a compact open subgroup \(H\), \(e_H\) is left \(H\)-invariant, right \(H\)-invariant, and hence bi-\(H\)-invariant.

Proof

3.2.7 Finite decomposition for bi-invariant distributions

Theorem 68 Decomposition into convolutions with delta distributions

If \(F\) is compactly supported and bi-\(H\)-invariant, then there exist finitely many elements \(g_i \in G\) and coefficients \(a_i \in \mathbb {C}\) such that

\[ F(\varphi ) = \sum _i a_i \, ((e_H * \delta _{g_i})(\varphi )) \]

for all test functions \(\varphi \).

Proof

3.2.8 Local constancy (smoothness)

We say a distribution is left, right, or bi-invariant under an open subgroup if it is fixed by the corresponding translation action.

Let \(D\) be a compactly supported distribution on \(G\). The following statements are equivalent:

  1. \(D\) is left-invariant under some compact open subgroup of \(G\);

  2. \(D\) is right-invariant under some compact open subgroup of \(G\);

  3. \(D\) is bi-invariant under some compact open group of \(G\).

A compactly supported distribution satisfying the above condition is said to be locally constant.

Proof