Let $\Bbb{F}_q$ be a finite field with $q$ elements, let $\Bbb{F}_{q^2}$ be its quadratic extension, and consider the finite "upper" half space ${\frak{H}}_q := \Bbb{F}_{q^2} - \Bbb{F}_q$. Apeing a construction found in one of S. Lang's immortal works one can show for $g \in \text{GL}_2(\Bbb{F}_q)$ and $f: {\frak{H}}_q \longrightarrow \Bbb{C}$ that $\sigma_\chi$ given by

\begin{equation} \Big( \sigma_\chi(g) f \Big)(z) \ =\ f \Bigg( {az + b \over {cz +d} }\Bigg) \, \chi\big(cz + d \big)\end{equation}

defines a $q(q-1)$ dimensional representation of $\text{GL}_2(\Bbb{F}_q)$ on the vector space $\Bbb{C}\big[ {\frak{H}}_q \big]$ consisting of all complex-valued function on ${\frak{H}}_q$ where

\begin{equation} g^{-1} \ = \ \begin{pmatrix} a & b \\ c & d\end{pmatrix} \end{equation}

and where $\chi: \Bbb{F}_{q^2}^* \longrightarrow \Bbb{C}^*$ is a fixed (indecomposable) multiplicative character.

Is it possilbe to identify the Weil representation of $\text{GL}_2(\Bbb{F}_q)$ corresponding to $\chi$ as a direct summand of $\Bbb{C}\big[ {\frak{H}}_q \big]$ and if so, with which multiplicity does it occur ?




It feels like it should be. In a sense it's really almost there! But alas, it's not.

Set $G=GL_2(\mathbb{F}_q)$. Fix $\epsilon\in\mathfrak{H}_q$, such that $\epsilon^2\in\mathbb{F}_q^\times$. We define the usual action of $G$ on $\mathfrak{H}_q$ by:

$$\left( \begin{array}{ccc} a & b \\ c & d \end{array} \right)(z)=\frac{az+b}{cz+d}$$

and also the "automorphic factor":

$$j_\chi(\left( \begin{array}{ccc} a & b \\ c & d \end{array} \right), z)=\chi(cz+d)$$

Exercise 0.0 (you're probably familiar with this): Show that the above is an action, i.e. $gg'(z)=g(g'(z))$

Exercise 0.1 Show that $j_\chi$ is a cocycle in the sense that: $j_\chi(gg',z)=j_\chi(g,g'(z))j_\chi(g',z)$

Let $T$ be the stabiliser of $\epsilon\in\mathfrak{H}_q$.

Exercise 1: Show that $T$ is the subgroup consisting of the matrices $$\left( \begin{array}{ccc} a & b\epsilon^2 \\ b & a \end{array} \right),\ \ (a,b)\not=0$$

Exercise 2: Show that $$d:\left( \begin{array}{ccc} a & b\epsilon^2 \\ b & a \end{array} \right)\mapsto a+\epsilon b$$ is an isomorphism $T\cong\mathbb{F}_{q^2}^\times$.

Define $\varphi:\mathbb{C}[\mathfrak{H}_q]\rightarrow\mathbb{C}[G]$ as:


Endow $\mathbb{C}[G]$ with an action of $G$ on the left, by: $$(gf)(h):=f(g^{-1}h).$$

Exercise 3: Show that $\varphi$ is an intertwining operator, i.e: $g\varphi(f)=\varphi(\sigma_\chi(g)f)$.

Exercise 3$\frac{1}{2}$: Show that $\varphi$ is injective.

Define $\text{Ind}_T^G\ \chi\circ d$ to be the subspace of $\mathbb{C}[G]$ of functions: $$\{ F\in\mathbb{C}[G]\ |\ F(gt)=F(g)\chi(d(t)), \text{for all } g\in G, t\in T \}$$

Exercise 4: Show that $\varphi(\mathbb{C}[\mathfrak{H}_q])=\text{Ind}_T^G\ \chi\circ d$.

So in order to answer the question, we must compute $(W(\chi), \text{Ind}_T^G\ \chi\circ d)$. By Frobenius reciprocity, this is equal to $(Res_T\ W(\chi), \chi\circ d)_T$

Given the character of $Res_T\ W(\chi)$, this is a very easy computation. It is well known that: $$Tr_{W(\chi)}(\left( \begin{array}{ccc} a & 0 \\ 0 & a \end{array} \right))=(q-1)\chi(a),\ \ a\not=0$$ and $$Tr_{W(\chi)}(\left( \begin{array}{ccc} a & b\epsilon^2 \\ b & a \end{array} \right))=-\chi(a+\epsilon b)-\chi(a-\epsilon b),\ \ b\not=0$$

So, finally:

Exercise 5: Show that $(W(\chi), \sigma_\chi)=0$.

If you've come this far, it is now also possible to prove:

Exercise 6: Let $\theta\in\widehat{\mathbb{F}_{q^2}^\times}$ be a generator. Show that $(W(\chi\theta^{i(q-1)}), \sigma_\chi)=1$ for $1\le i\le q$.

And we get: $$\sigma_\chi=\bigoplus_{1\le i\le q} W(\chi\theta^{i(q-1)})$$

  • $\begingroup$ Dear Dror, Thank you ! Comprehensive :) In the last line, what does the notation $\chi^{\theta}$ mean ? Am I correct in understanding that $\sigma_{\chi}$ is a direct sum of Weil representations (each with multiplicity one) --- just that $W(\chi)$ is not one of them ? This comes as a surprise to me because I always thought that Weil representatiosn do not arise through inductions --- which exercise 4 asserts. best, Ines $\endgroup$ Apr 11 '15 at 18:52
  • $\begingroup$ @Ines: You're welcome :P The latex is being a bit weird. It's not a superscript, it's just a multiplication of characters, $\chi\cdot\theta^{i(q-1)}$. Indeed the above Weil representations are components of $\sigma_\chi$. This shouldn't be too much of a surprise, since every representation of a finite group is a component of an induction, as the regular representation is such. Moreover, I would say that induction from $T$ is not only the formally correct choice to get $W(\chi)$, but also the intuitive one. But, yeah, intuition comes after working with these things for a long while. $\endgroup$ Apr 11 '15 at 19:36
  • $\begingroup$ Sure, like everything else they occur as summands of an induction --- but individually they are not induced representations. In view of your exercise 6 there are no $\text{GL}_(\Bbb{F}_q)$-invariants in this upper-half plane model --- so are there no finite analogues of modular forms? Ines $\endgroup$ Apr 12 '15 at 15:48
  • $\begingroup$ I think you are confusing some ideas, and doing the exercises should help with understanding what is going on. Exercise 4 does not claim that the Weil representations are inductions. As for modular forms, what would a function in $\sigma_\chi$ look like if it was $GL_2(\mathbb{F}_q)$-invariant? And in the global case, which are the $SL_2(\mathbb{Z})$-invariant modular forms? These are good elementary questions. $\endgroup$ Apr 12 '15 at 16:13

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.