# Zhiwei Yun, Lecture 2 :::{.remark} Today: what is rigidity? ::: ## Automorphic Data :::{.definition title="Automorphic data"} Given the following: - $k = \FF_q$ - $S \subseteq \abs{X}$ a finite set, e.g. $X= \PP^1$ and $S = \ts{0, \infty}$ or $S=\ts{0,1,\infty}$. - For each $x\in S$, a compact open $K_x \subseteq G(F_x)$ - For each $x\in S$, a character: \begin{tikzcd} {K_x} && \CC\units \\ \\ & {L_\chi \text{ finite}} \arrow[from=1-1, to=3-2] \arrow[from=3-2, to=1-3] \arrow["\chi"', from=1-1, to=1-3] \end{tikzcd} > [Link to Diagram](https://q.uiver.app/?q=WzAsMyxbMCwwLCJLX3giXSxbMiwwLCJcXENDXFx1bml0cyJdLFsxLDIsIkxfXFxjaGkgXFx0ZXh0eyBmaW5pdGV9Il0sWzAsMl0sWzIsMV0sWzAsMSwiXFxjaGkiLDJdXQ==) The pair $(K_S, \chi_S)$ is **automorphic data**. ::: :::{.remark} Note that $\chi_S = 1$ recovers $f\in \mca_K$ where $K = K_S \times \prod_{x\not\in S} G(\OO_x)$. ::: :::{.definition title="Typical data"} A map \[ f\in C^0(\dcoset{G(F)} {G(\AA)} {\prod_{x\not\in S} G(\OO_x) } \to \CC) \] is **$(K_S, \chi_X)\dash$typical** iff \[ f(gk_x) = \chi_x(k_x) f(g)\qquad \forall x\in S, k_x\in K_x, g\in G(\AA) .\] ::: :::{.remark} We want to make $\dim \mca_c(K_S,\chi_S) = 1$. In this case, the Hecke algebra $\mch_{K_y} \actson f\in \mca_c(K_S, \chi_S)$ by a character, making $f$ a Hecke eigenform. ::: ## Examples of naive ridigidy :::{.example title="?"} Let - $X=\PP^1$, - $G = \SL_2$, - $S=\ts{0, 1,\infty}$, - $K_x = I_x$ Iwahori for all $x\in S$, where \[ I_x = \ts{A\da \matt a b c d \in \Mat_{2}(\OO_X) \st c\in \mfm_x} .\] Then choosing characters $\chi_x: k\units\to \CC\units$ *generically* will imply $\dim \mca_c(K_S, \chi_S) = 1$. Here generic means that $\prod \chi_i^{\pm 1}\neq 1$. By global Langlands for $\SL_2$, any $f\in \mca_c(K_S, \chi_S)$ will yield a 2-dimensional local system on $\PP^1\smts{0,1,\infty}$ ramified at the 3 punctures. These will be solutions to hypergeometric differential equations. For $G=\PGL_2$ (where the example works similarly), for $\chi_0, \chi_1 = 1$ and $\chi_\infty$ quadratic, there is a cover \begin{tikzcd} \ts{E_t}: y^2 = x(x-1)(x-t) \ar[d] \\ \PP^1\smts{0,1,\infty} \end{tikzcd} Moreover $\ts{H^1(E_t)}$ will be a rank 2 local system on this base. ::: :::{.example title="?"} Let - $X = \PP^1$ - $S = \ts{0, \infty}$ - $K_0 = I_0$, $\chi_0 =1$ - $\tau$ a uniformizer at $\infty$ - $K_\infty = I_\infty^+ = \ts{\matt a b c d \st a\equiv d\equiv 1 \mod \mfm_x, b\in \OO_x, c\in \mfm_x}$, the pro $p$ part There is a map $K_\infty \to k$ where $\matt abcd \to b+{c\over \tau}\mod\tau$. Any character $k \mapsvia{\psi} \CC\units$ can be extended to $\chi_\infty: K_\infty \to k \mapsvia{\psi} \CC\units$, and $\dim \mca_c(K_S, \chi_S)$. This yields a **Kloosterman local system** on $\PP^1\smts{0,\infty}$, where \[ \Kl(a) = \sum_{x\in k\units} \psi\qty{x + a\over x} \] recovers the classical Kloosterman sum by taking trace of Frobenius. ::: ## Naive Rigidity :::{.definition title="Rigidity (Naive Definition)"} \[ (K_S, \chi_S) \text{ is rigid }\iff \dim \mca_c(K_S, \chi_S) = 1 .\] ::: :::{.warnings} If $\pi_1 G\neq 1$, then $\pi_0 \Bun_G\geq 2$, yielding multiple components. It's also not clear if this type of dimension bound will hold after a base change $k\to k'$. ::: ## Base Change :::{.remark} For $k'/k$ finite, write $X'\da X\tensor_k k'$ for the base change. Let $S\to S'$ be the preimage of $S$ in $S'$, and consider $k'_x \da K_x\tensor_k k'$ How can we base change a character? We need a norm map to fill in the following diagram: \begin{tikzcd} {K_x\tensor_k k'} && {K_x} && \CC\units \arrow["{\chi_x}", from=1-3, to=1-5] \arrow["{\Norm(\wait)}", dashed, from=1-1, to=1-3] \end{tikzcd} > [Link to Diagram](https://q.uiver.app/?q=WzAsMyxbMCwwLCJLX3hcXHRlbnNvcl9rIGsnIl0sWzIsMCwiS194Il0sWzQsMCwiXFxDQ1xcdW5pdHMiXSxbMSwyLCJcXGNoaV94Il0sWzAsMSwiXFxOb3JtKFxcd2FpdCkiLDAseyJzdHlsZSI6eyJib2R5Ijp7Im5hbWUiOiJkYXNoZWQifX19XV0=) ::: :::{.example title="Base-changing a character"} \begin{tikzcd} {I_x} && k\units && \CC\units \\ \\ {I_x' = I_x(k')} && {(k')\units} && k\units && \CC\units \arrow["\chi", from=1-3, to=1-5] \arrow[from=1-1, to=1-3] \arrow["\chi", from=3-5, to=3-7] \arrow["{\Norm_{k'/k}(\wait)}", from=3-3, to=3-5] \arrow[from=3-1, to=3-3] \end{tikzcd} > [Link to Diagram](https://q.uiver.app/?q=WzAsNyxbMCwwLCJJX3giXSxbMiwwLCJrXFx1bml0cyJdLFs0LDAsIlxcQ0NcXHVuaXRzIl0sWzAsMiwiSV94JyA9IElfeChrJykiXSxbMiwyLCIoaycpXFx1bml0cyJdLFs0LDIsImtcXHVuaXRzIl0sWzYsMiwiXFxDQ1xcdW5pdHMiXSxbMSwyLCJcXGNoaSJdLFswLDFdLFs1LDYsIlxcY2hpIl0sWzQsNSwiXFxOb3JtX3trJy9rfShcXHdhaXQpIl0sWzMsNF1d) Here $I_x(k') = \ts{a,b,c,d\in \OO_x\hat{\tensor} k' \cong k'\fps{t}}$. ::: :::{.example title="?"} \begin{tikzcd} {I_x} && k\units && \CC\units \\ \\ {I_x' = I_x(k')} && {(k')\units} && k\units && \CC\units \\ {\matt a b c d } && {b + {c\over \tau}\mod \tau} \arrow["\chi", from=1-3, to=1-5] \arrow[from=1-1, to=1-3] \arrow["\chi", from=3-5, to=3-7] \arrow["{\Trace_{k'/k}(\wait)}", from=3-3, to=3-5] \arrow[from=3-1, to=3-3] \arrow[maps to, from=4-1, to=4-3] \end{tikzcd} > [Link to Diagram](https://q.uiver.app/?q=WzAsOSxbMCwwLCJJX3giXSxbMiwwLCJrXFx1bml0cyJdLFs0LDAsIlxcQ0NcXHVuaXRzIl0sWzAsMiwiSV94JyA9IElfeChrJykiXSxbMiwyLCIoaycpXFx1bml0cyJdLFs0LDIsImtcXHVuaXRzIl0sWzYsMiwiXFxDQ1xcdW5pdHMiXSxbMCwzLCJcXG1hdHQgYSBiIGMgZCAiXSxbMiwzLCJiICsge2NcXG92ZXIgXFx0YXV9XFxtb2QgXFx0YXUiXSxbMSwyLCJcXGNoaSJdLFswLDFdLFs1LDYsIlxcY2hpIl0sWzQsNSwiXFxUcmFjZV97aycva30oXFx3YWl0KSJdLFszLDRdLFs3LDgsIiIsMCx7InN0eWxlIjp7InRhaWwiOnsibmFtZSI6Im1hcHMgdG8ifX19XV0=) ::: :::{.remark} We now geometrize this process to send characters to character sheaves, i.e. rank one local systems on $K_x$. We have a way of taking $(K_S, \chi_S)$ to $(K'_S, \chi_S')$ for extensions $K\to K'$, so we can form $\mca(k'; K'_S, \chi'_S)$. ::: :::{.definition title="Weakly rigid"} Automorphic data $(K_S, \chi_S)$ is **weakly rigid** iff $\dim \mca_c(k'; K'_S, \chi_S')$ is uniformly bounded for all extensions $k\to k'$. ::: ## Relevant Points :::{.remark} Recall that there is a bijection \[ \dcoset{G(F)}{G(\AA)}{K} \mapstofrom \Bun_G(K)(k) ,\] so functions $f\in \mca_c(K_S, \chi_S)$ are functions on $\Bun_G(K_S^+)(k)$ where for $x\in S$, $K^+_x \normal K_x$ with $\ro{\chi_x}{K_x^+} = 1$ and $K_x/K_x^+$ are the $k\dash$points of a finite dimensional group $L_x$. ::: :::{.example title="?"} $I_x^+\normal I_x = K_x \to \GG_m(k)$. ::: :::{.example title="?"} $I_\infty^{++} = K_\infty^+ \normal I_\infty^+ = K_\infty \to k\sumpower{2}$ where $\matt abcd \mapsto \qty{b, {c\over \tau}} \mod \tau$. ::: :::{.remark} There is a right action \[ C^0(\Bun_G(K_S^+)(k) \to \CC?) \actsonl \prod_{x\in S} L_x(k) ,\] and the eigenfunctions with eigenvalues $(\chi_x)_{x\in S}$ are in $\mca_c(K_S, \chi_S)$. As a set, $\Bun_G(K_S^+)$ has commuting left and right actions, where quotienting by the right action yields a principal homogeneous space. The left action is by $\Aut(\mce)$ for $\mce\in \Bun_G(K_S)$, and permutes points in the fiber in $\tilde \mce \in \Bun_G(K_S^+)$. So there is an evaluation map which is well-defined up to conjugacy \[ \ev_\mce: \Aut(\mce) \to \prod_{x\in S} L_x(k) .\] ::: :::{.definition title="Relevant points"} A $k\dash$point $\mce\in \Bun_G(K_s)(k)$ is **$(K_S, \chi_S)\dash$relevant** iff \[ \ro{ \ev_\mce^*\qty{ \prod_{x\in S} \chi_x } }{\Aut(\mce)^0(k) } = 1 .\] Similarly one can define relevant $k'\dash$points for $k'/k$ a finite extension. ::: :::{.fact} \[ \dim \mca_c(k'; K_S, \chi_S) \leq \size \Rel(K_S), \quad \Rel(K_S) \da \ts{(K_S', \chi_S')\dash\text{relevant }k' \text{ points of } \Bun_G(K_S) } .\] Note that taking connected components in the definition is needed to make this stable under base change. ::: :::{.corollary title="?"} $(K_S, \chi_S)$ is weakly rigid $\iff \size \Rel(K_S) < \infty$. ::: :::{.example title="?"} Let - $G=\SL_2$ - $K_x = I_x$ - $x\in S \da \ts{0, 1,\infty}$ Note that \[ \Bun_G(K_S)(k) = \ts{ V\in \mathsf{VectBundle}^{\rank = 2}, \iota: \Extalg^2 V\cong \OO_X, \ts{\ell_x \subseteq V_x}_{x\in S} } ,\] where the $\ell$ are lines. So these are bundles with extra structure at fixed places, and are parameterized by 5-tuples $\mce = (V, \iota, \ell_0, \ell_1, \ell_\infty)$. For all $x\in S$ we have \[ I_x &\to \GG_m = L_x \\ \matt a b c d &\mapsto a\mod \tau ,\] and the evaluation map is $\ev: \Aut(\mce) \to \prod_{x\in S} \GG_m$. Each $\gamma\in \Aut(\mce)$ is a map $V\to V$ where $\gamma_0 \actson V_0$ preserving $\ell_0$. Is it the case that \[ \ro{\prod_{x\in S} \chi_x}{\Aut(\mce)^0} \underset{?}{=} 1 .\] For $V = \OO^2$ and $\ell_x \subseteq k^2$ in generic position, $\Aut(\mce) = \ts{\pm 1}$ so they are relevant. Other points are irrelevant: if $V = L \oplus L'$ with $\ell_x\in L_x$ or $L_x'$, $\Aut(\mce)$ will contain a copy of $\GG_m$ that acts by scaling each $L$ which will map nontrivially to $\prod_{x\in S} \GG_m$. Since $\prod \chi_i^{\pm 1}\neq 1$, we get $\ro{\ev_\mce^* \prod \chi_x}{\GG_m} \neq 1$. :::