\(\Theta\)-correspondences \((U(1),U(2))\). I: Unramified case (Q2568211): Difference between revisions
From MaRDI portal
Removed claim: reviewed by (P1447): Item:Q244304 |
Changed an Item |
||
Property / reviewed by | |||
Property / reviewed by: Anne-Marie Aubert / rank | |||
Normal rank |
Revision as of 23:53, 11 February 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | \(\Theta\)-correspondences \((U(1),U(2))\). I: Unramified case |
scientific article |
Statements
\(\Theta\)-correspondences \((U(1),U(2))\). I: Unramified case (English)
0 references
10 October 2005
0 references
Let \(F\) be a local \(p\)-adic field of odd residual characteristic with \(q\) being the order of the residue field. Let \({\mathfrak o}_F\) be the ring of integers of \(F\). Let \(D\) be the quaternion division algebra over \(F\), and let \(E\) be the unramified quadratic extension of \(F\) contained in \(D\). Let \(\varpi\), \(\pi\) be uniformizers of \(F\), \(D\), respectively, such that \(\pi^2=\varpi\). One may and will choose a unit element \(\varepsilon\) such that \(\varepsilon\pi=-\pi\varepsilon\), where \(\varepsilon\) is a generator of \(E\) and \(\{1,\varepsilon,\pi,\varepsilon\pi\}\) is a basis of \(D\) over \(F\). Let \(x\mapsto\bar x\) denote the involution on \(D\) defined by \[ \bar x:=x_1-x_2\varepsilon-x_3\pi-x_4\varepsilon\pi,\;\;\text{for any \(x= x_1+x_2\varepsilon+x_3\pi+x_4\varepsilon\pi\in D\), with \(x_i\in F\).} \] The restriction of this involution to \(E\) is the nontrivial element of the Galois group \({\text{Gal}}(E/F)\). Let \(D^1\) denote the elements of norm one in \(D\), that is, \(D^1:=\left\{x\in D\,:\,x\bar x=1\right\}\). Let \((W_1,(\;,\;)_1)\) (resp \((W_2,(\;,\;)_2)\)) be a Hermitian (resp. skew-Hermitian) space over \(E\), and let \(G_i\) denote the isometries group of \((\;,\;)_i\) for \(i=1,2\). Then \((G_1,G_2)\) is a reductive dual pair for the symplectic group \({\text{SP}}(W)\), where \(W=W_1\otimes_E W_2\) is equipped with the symplectic form \(\langle\;,\;\rangle:=\frac{1}{2}{\text{Tr}}_{E/F}(\overline{(\;,\;)_1}\otimes (\;,\;)_2)\) over \(F\). In the paper under review the author considers the special case where \(W_1=E\), with \((\;,\;)_1\) the norm form, and \(W_2=D\), viewed as a two-dimensional left vector space over \(E\), with \[ (x,y)_2:=\frac{1}{2}{\text{Tr}}_{D/E}(\varepsilon x\bar y),\;x,y\in D. \] Then \(W\) is isomorphic to \(W_2\) via \(a\otimes x\mapsto ax\), and there is an imbedding \(\sigma: E^1\hookrightarrow \Aut(W)\), \(\lambda\mapsto\sigma_\lambda\) such that, for any \(\lambda\in E^1\), \(\sigma_\lambda\) is an \(E\)-linear map of \(W\) onto \(W\), and for any \(\delta\in D^1\), we have \(\sigma_\lambda(\delta)\in D^1\) and \(\overline{\sigma_\lambda(w)}=\sigma_\lambda(\bar{w})\) for all \(w\in W\). Then we have \(G_1=U(1)=E^1\) (the group of norm one elements in \(E\)) and \(G_2=U(2)=D^1\rtimes E^1\), where the semidirect product is defined as follows: \[ (\delta_1,\lambda_1)\cdot(\delta_2,\lambda_2):= (\delta_1\sigma_{\lambda_1}(\delta_2),\lambda_1\lambda_2). \] Let \({\mathfrak o}_D\) be the ring of integers of \(D\), and let \({\mathfrak o}_D\) denote its maximal ideal. For any positive integer \(r\), put \[ E_r^1:=\left\{x\in E^1\,:\,x\equiv 1\pmod{\mathfrak o}_E^{r}\right\}, \] \[ D_r^1:=\left\{x\in D^1\,:\,x\equiv 1\pmod{\mathfrak o}_D^{r}\right\}. \] Let \(E^0\) (resp. \(D^0\)) denote traceless elements in \(E\) (resp. \(D\)). For any \(x\in D\), let \(v_D(x)\) denote its order. Let \(\alpha\in D^0\) be an element such that \(v_D(\alpha)= -n-1\), where \(n\) is a positive integer. Put \(r:=\left[\frac{n+1}{2}\right]\), where \([\;\;]\) is the greatest integer part function. Let \(\chi_\alpha: D_r^1\to\mathbb C^\times\) be the character of \(D_r^1\) defined by \[ \chi_\alpha(h):=\chi\left({\text{Tr}}(\alpha(h-1)\right),\;\;\text{for any \(h\in D_r^1\).} \] The group \(U(2)\) acts on \(\chi_\alpha\) via \((g\cdot\chi_\alpha):=\chi_\alpha(g^{-1}hg)\). Let \({\text{Stab}}(\chi_\alpha)\) denote the stabilizer of \(\chi_\alpha\) in \(U(2)\). Then \[ {\text{Stab}}(\chi_\alpha)=\begin{cases} L^1D_r^1\rtimes E^1_{\left[\frac{r+1}{2}\right]}& \text{ if \(n\) even,}\cr E^1D_{r-1}^1\rtimes E^1&\text{ if \(n\) odd,} \cr \end{cases} \] where \(L=F(\alpha)\) and \(L^1\) is the group of norm one elements in \(L\). \(\bullet\) Assume first that \(n\) is even. For any extension \(\varphi\) of \(\chi_\alpha\) to \(L^1D_r^1\), and any character \(\xi\) of \(E^1_{\left[\frac{r+1}{2}\right]}\), let \(\varphi_{\alpha,\xi}\) be the character of \({\text{Stab}}(\chi_\alpha)\) defined as \(\varphi_{\alpha,\xi}(x,\lambda):=\varphi(x)\cdot\xi(\lambda)\). Then the induced representation \[ \rho_{\alpha,\varphi,\xi}:={\text{Ind}}_{{\text{Stab}}(\chi_\alpha)}^{U(2)}(\varphi_{\alpha,\xi}) \] is an irreducible representation of \(U(2)\). \(\bullet\) Assume now that \(n\) and \(r\) are odd. Again for any extension \(\psi\) of \(\chi_\alpha\) to \(E^1D_{r-1}^1\), and any character \(\xi\) of \(E^1\), let \(\psi_{\alpha,\xi}\) be the character of \({\text{Stab}}(\chi_\alpha)\) defined as \(\psi_{\alpha,\xi}(x,\lambda):=\psi(x)\cdot\xi(\lambda)\). Then the induced representation \[ \rho_{\alpha,\psi,\xi}:={\text{Ind}}_{{\text{Stab}}(\chi_\alpha)}^{U(2)} (\psi_{\alpha,\xi}) \] is an irreducible representation of \(U(2)\). \(\bullet\) Assume finally that \(n\) is odd and \(r\) is even. Then for any extension \(\psi\) of \(\chi_\alpha\) to \(E^1D_r^1\), and any character \(\xi\) of \(E^1\), let \(\psi_{\alpha,\xi}\) be the character of \(E^1D_r^1\rtimes E^1\) defined as \(\psi_{\alpha,\xi}(x,\lambda):=\psi(x)\cdot\xi(\lambda)\), and let \(\rho'_{\alpha,\psi,\xi}\) denote the unique irreducible \(q\)-dimensional representation occurring in the induced representation \[ {\text{Ind}}_{E^1D_r^1\rtimes E^1}^{{\text{Stab}}(\chi_\alpha)} (\psi_{\alpha,\xi}). \] Then the induced representation \[ \rho_{\alpha,\psi,\xi}:={\text{Ind}}_{{\text{Stab}}(\chi_\alpha)}^{U(2)} (\rho'_{\alpha,\psi,\xi}) \] is an irreducible representation of \(U(2)\). Any irreducible representation of \(U(2)\) is either a character or is one of the representations described above. The author proves that (1) all smooth characters of \(U(1)\) (resp. \(U(2)\)) occur in the Weil representation; (2) any smooth irreducible representation \(\rho_{\alpha,\varphi,1}\) of \(U(2)\) such that \(\alpha=a\varepsilon\pi^{-2r-2}\), with \(a\in\mathfrak o_F^\times\) and \(r\geq 1\) an odd integer, occurs in the Weil representation; 3 any smooth irreducible representation \(\rho_{-\alpha,\bar\varphi,\varphi}\) of \(U(2)\) such that \(\alpha=a\varepsilon\pi^{-2r-2}\), with \(a\in\mathfrak o_F^\times\), \(r\geq 2\) an even integer, and \(\bar\varphi(\lambda):=\varphi(\bar\lambda)\) for all \(\lambda\in E^1\), occurs in the Weil representation; 4 none of the other representations of \(U(2)\) occur in the Weil representation. For Part, see ibid. 111, No. 2, 287--317 (2005; Zbl 1084.11019).
0 references
eigenfunction
0 references
lattice model
0 references
unitary group
0 references
Howe correspondence
0 references