In some proofs of our lecture we used following relation $$\rho(g)u=u*\delta_g$$

The convolution was defined as $(u*v)(h)= \sum_{g\in\Gamma} u(g) v(g^{-1}h)$ for $u,v\in l^2(\Gamma)$ and $h\in\Gamma$ and the right regular representation as $(\rho(g)v)(h)= v(hg)$ for $v\in l^2(\Gamma)$ and $h\in\Gamma$

The problem is that my calculation yields another relation:


Every summand is zero, except the one that satisfies $f=hg^{-1}$ (since $g=f^{-1}h \Rightarrow f^{-1}=gh^{-1} \Rightarrow f=hg^{-1}$). Hence it follows

$$(u*\delta_g)(h)= u(hg^{-1})=(\rho(g^{-1})u)(h)$$

Have I made a mistake somewhere? Thanks for your help.


1 Answer 1


The issue is in your formula for the right regular representation in terms of the convolution product: You state that $$\rho(g)u=u*\delta_g.$$ But the correct formula is $$\rho(g)u=u*\delta_{g^{-1}}.$$ With this, one obtains \begin{align*} (\rho(g)u)(h) &=(u*\delta_{g^{-1}})(h)\\ &=\sum_{f\in\Gamma}u(f)\delta_{g^{-1}}(f^{-1}h)\\ &=u(hg). \end{align*}

  • $\begingroup$ Thanks for your help I also thought that possibly the definition is wrong in the lecture. But usually I am wrong and not the script. $\endgroup$ Aug 8 at 9:15
  • $\begingroup$ @Schrödinger'scat Ahh I'm so sorry, I got it backwards. Please see my correction. You can now check that $\rho(g_1g_2)=\rho(g_1)\rho(g_2)$, whereas the previous formula gave $\rho(g_1g_2)=\rho(g_2)\rho(g_1)$ for all $g_1,g_2\in\Gamma$. $\endgroup$
    – Aweygan
    Aug 8 at 12:24

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.