§6 · Agent 6 · GL₂/T: the Iwahori module · 2026-10-18T18:48:03Z

Proposition 6.5 · Sheaves fail for $\mathrm{GL}2/T$: $\mathrm{IC}{A_\pm}\notin\mathcal C_I^{\mathrm{Iwa}}$

proved ★ key step

Runs the Iwahori summand criterion to the end for GL2/T: the summands of the generators are exactly the ICs of the non-closed orbit closures, so the constant sheaves on the two closed I-orbits A± are not Iwahori-generated and C_I^Iwa is a proper thick subcategory.

Proposition 6.5.

Over \(k=\overline{\mathbb F}_q\) (\(p\) odd) or \(k=\mathbb C\), for \(X=\mathrm{GL}_2/T\): (a) for every \(w\in\widetilde W=\mathbb Z^2\rtimes S_2\), \(\mathrm{IC}_w\ast_I\mathrm{IC}_0\) is a finite direct sum of shifts of objects \(\mathrm{IC}_{i,n}\) (\(i\in\{0,1\}\), \(n\ge0\)), and each \(\mathrm{IC}_{i,n}\) occurs for some \(w\): \(M_I=\{\mathrm{IC}_{i,n}\}\); (b) \(\mathcal C_I^{\mathrm{Iwa}}=\mathrm{Thick}\{\mathrm{IC}_{i,n}\}=\mathrm{Thick}\big(\mathrm{For}\,\mathcal C_K\cup\mathrm{For}\,\mathcal C_{K'}\big)\), \(K'=\omega K\omega^{-1}=\mathrm{Stab}(v_1)\); (c) every \(\mathcal F\in\mathcal C_I^{\mathrm{Iwa}}\) has all Jordan–Hölder factors of all \({}^p\mathcal H^i(\mathcal F)\) among the \(\mathrm{IC}_{i,n}\); in particular \(\mathrm{IC}_{A_+}=k_{A_+}\) and \(\mathrm{IC}_{A_-}=k_{A_-}\) are not in \(\mathcal C_I^{\mathrm{Iwa}}\), so $$\mathcal C_I^{\mathrm{Iwa}}\subsetneq\mathcal C_I=\mathrm{Thick}\big(\mathcal C_I^{\mathrm{Iwa}}\cup\{\mathrm{IC}_{A_+},\mathrm{IC}_{A_-}\}\big).$$

Hypotheses of the summand criterion (I_I), checked for this \(X\): level presentation \(Y_N\) with \(2N+3\) orbits (Setup 6.1(d); finite level from Chen–Yi's placidness, §5 and Appendix A, assumed as in hypothesis.html I1I1 · Finite levels: LX has a level presentation in which every object of C_I is pulled back from a finite level Y_{N,M} and each level has finitely many pieces (Iwahori-generation pages, Proposition III).); connected stabilizers, so no non-constant local systems (Setup 6.1(c)); finite-length perverse heart at each level; decomposition of \(\mathrm{IC}_w\ast_I\mathrm{IC}_0\) (step 2 below).

  1. Generators. \(\mathcal H_{c,I}=\mathrm{Thick}\{\mathrm{IC}_w\}\): at a finite level of \(\mathrm{Fl}\) the cells \(IwI/I\cong\mathbb A^{\ell(w)}\) have connected stabilizers, so every object lies in \(\mathrm{Thick}\{\Delta_w\}\), and the cone of \(\Delta_w\to\mathrm{IC}_w\) is supported on lower cells, so \(\Delta_w\in\mathrm{Thick}\{\mathrm{IC}_y:y\le w\}\) by induction. As \(-\ast_I\mathrm{IC}_0\) is exact, \(\mathcal C_I^{\mathrm{Iwa}}=\mathrm{Thick}\{\mathrm{IC}_w\ast_I\mathrm{IC}_0\}\). (One-sided action; \(G=\mathrm{GL}_2\) acts on the left only.)
  2. Bott–Samelson. For \(w=s_{i_1}\cdots s_{i_r}\omega^j\) reduced, the Bott–Samelson map onto \(\overline{IwI}/I\) is proper and birational, so by the decomposition theorem (BBD 5.4.5–5.4.6, resp. 6.2.5) \(\mathrm{IC}_w\) is a direct summand of \(\mathrm{IC}_{s_{i_1}}\ast\cdots\ast\mathrm{IC}_{s_{i_r}}\ast\mathrm{IC}_{\omega^j}\), which in turn is \(\bigoplus_y\mathrm{IC}_y[\text{shifts}]\). Hence \(\mathrm{IC}_w\ast\mathrm{IC}_0\) is a direct summand of \(B_{\underline w}:=\mathrm{IC}_{s_{i_1}}\ast\cdots\ast\mathrm{IC}_{\omega^j}\ast\mathrm{IC}_{0,0}\).
  3. Closure. By Lemma 6.4 and induction on \(r\), \(B_{\underline w}\in\mathrm{add}\{\mathrm{IC}_{i,n}[m]\}\). At a finite level the category has finite-dimensional Hom spaces and split idempotents, hence is Krull–Schmidt, and each \(\mathrm{IC}_{i,n}[m]\) is indecomposable; so every indecomposable summand of \(\mathrm{IC}_w\ast\mathrm{IC}_0\) is some \(\mathrm{IC}_{i,n}[m]\). Occurrence: by Lemma 6.4(a) the alternating word \(s_{i}\cdots s_1s_0\) of length \(n\) gives \(B=\mathrm{IC}_{i',n}\oplus(\text{lower})\), and \(B=\bigoplus_y\mathrm{IC}_y\ast\mathrm{IC}_0[\cdot]\), so \(\mathrm{IC}_{i',n}\) is a summand of some \(\mathrm{IC}_y\ast\mathrm{IC}_0\); apply \(\omega\) for the other index. This proves (a).
  4. (b). \(\subseteq\) by steps 1–3; \(\supseteq\) since each \(\mathrm{IC}_{i,n}\) is a retract of a generator. \(\mathcal C_K\) is generated by the \(\mathrm{IC}(Y_n)=\mathrm{IC}_{0,n}\) (finitely many \(K\)-orbits per \(Y_N\), connected stabilizers, dévissage), and \(\omega_*\) carries \(\mathcal C_K\) to \(\mathcal C_{K'}\).
  5. (c). Let \(\mathcal S\subseteq\mathcal C_I\) be the full subcategory of objects all of whose perverse cohomology objects have Jordan–Hölder factors in \(\{\mathrm{IC}_{i,n}\}\). It is closed under shifts and retracts (\({}^p\mathcal H^i\) of a retract is a retract), and under cones: in \({}^p\mathcal H^i(A)\to{}^p\mathcal H^i(B)\to{}^p\mathcal H^i(C)\to{}^p\mathcal H^{i+1}(A)\), \({}^p\mathcal H^i(C)\) is an extension of a subobject of \({}^p\mathcal H^{i+1}(A)\) by a quotient of \({}^p\mathcal H^i(B)\). So \(\mathcal S\) is thick and contains \(\mathcal C_I^{\mathrm{Iwa}}\) by (b). \(\mathrm{IC}_{A_\pm}\) is a simple perverse object different from all \(\mathrm{IC}_{i,n}\) (different support), hence not in \(\mathcal S\). The perverse t-structure is the one of each finite level; smooth pullback shifted by the relative dimension and closed pushforward \(Y_N\hookrightarrow Y_{N+1}\) are t-exact and preserve ICs, so \(\mathcal S\) is well defined on the colimit (checklist 1). Finally every object of \(\mathcal C_I\) lies in the thick hull of all simple objects by dévissage on the finitely many orbits of a level, and the simple objects are \(\mathrm{IC}_{i,n},\mathrm{IC}_{A_\pm}\) (Setup 6.1(c)).
  6. Fields. No weights are used beyond the decomposition theorem: steps 1–5 hold verbatim over \(\mathbb C\) (Betti, BBD 6.2.5) and over \(\overline{\mathbb F}_q\), \(p\) odd (\(p\ne2\) is used for the involution model of Setup 6.1).

Builds on

  • Setup 6.1 · The $I$-orbits on $L(\mathrm{GL}_2/T)$ as relative positions of an edge and a geodesic
  • Lemma 6.3 · $\mathrm{IC}(Y_n)=\nu_{n*}k$ with stalk rank $n+1-d(v_0,\gamma)$
  • Lemma 6.4 · Convolution rules for $\mathrm{IC}s\ast_I\mathrm{IC}{i,n}$ and $\mathrm{IC}_\omega$

Built on by

Original record, as the agent wrote it at 18:48 UTC

Sheaves fail for $\mathrm{GL}2/T$: $\mathrm{IC}{A_\pm}\notin\mathcal C_I^{\mathrm{Iwa}}$

Hypotheses of the summand criterion (I_I), checked for this \(X\): level presentation \(Y_N\) with \(2N+3\) orbits (Setup 6.1(d); finite level from Chen–Yi's placidness, §5 and Appendix A, assumed as in hypothesis.html I1); connected stabilizers, so no non-constant local systems (Setup 6.1(c)); finite-length perverse heart at each level; decomposition of \(\mathrm{IC}_w\ast_I\mathrm{IC}_0\) (step 2 below).

  1. Generators. \(\mathcal H_{c,I}=\mathrm{Thick}\{\mathrm{IC}_w\}\): at a finite level of \(\mathrm{Fl}\) the cells \(IwI/I\cong\mathbb A^{\ell(w)}\) have connected stabilizers, so every object lies in \(\mathrm{Thick}\{\Delta_w\}\), and the cone of \(\Delta_w\to\mathrm{IC}_w\) is supported on lower cells, so \(\Delta_w\in\mathrm{Thick}\{\mathrm{IC}_y:y\le w\}\) by induction. As \(-\ast_I\mathrm{IC}_0\) is exact, \(\mathcal C_I^{\mathrm{Iwa}}=\mathrm{Thick}\{\mathrm{IC}_w\ast_I\mathrm{IC}_0\}\). (One-sided action; \(G=\mathrm{GL}_2\) acts on the left only.)
  2. Bott–Samelson. For \(w=s_{i_1}\cdots s_{i_r}\omega^j\) reduced, the Bott–Samelson map onto \(\overline{IwI}/I\) is proper and birational, so by the decomposition theorem (BBD 5.4.5–5.4.6, resp. 6.2.5) \(\mathrm{IC}_w\) is a direct summand of \(\mathrm{IC}_{s_{i_1}}\ast\cdots\ast\mathrm{IC}_{s_{i_r}}\ast\mathrm{IC}_{\omega^j}\), which in turn is \(\bigoplus_y\mathrm{IC}_y[\text{shifts}]\). Hence \(\mathrm{IC}_w\ast\mathrm{IC}_0\) is a direct summand of \(B_{\underline w}:=\mathrm{IC}_{s_{i_1}}\ast\cdots\ast\mathrm{IC}_{\omega^j}\ast\mathrm{IC}_{0,0}\).
  3. Closure. By Lemma 6.4 and induction on \(r\), \(B_{\underline w}\in\mathrm{add}\{\mathrm{IC}_{i,n}[m]\}\). At a finite level the category has finite-dimensional Hom spaces and split idempotents, hence is Krull–Schmidt, and each \(\mathrm{IC}_{i,n}[m]\) is indecomposable; so every indecomposable summand of \(\mathrm{IC}_w\ast\mathrm{IC}_0\) is some \(\mathrm{IC}_{i,n}[m]\). Occurrence: by Lemma 6.4(a) the alternating word \(s_{i}\cdots s_1s_0\) of length \(n\) gives \(B=\mathrm{IC}_{i',n}\oplus(\text{lower})\), and \(B=\bigoplus_y\mathrm{IC}_y\ast\mathrm{IC}_0[\cdot]\), so \(\mathrm{IC}_{i',n}\) is a summand of some \(\mathrm{IC}_y\ast\mathrm{IC}_0\); apply \(\omega\) for the other index. This proves (a).
  4. (b). \(\subseteq\) by steps 1–3; \(\supseteq\) since each \(\mathrm{IC}_{i,n}\) is a retract of a generator. \(\mathcal C_K\) is generated by the \(\mathrm{IC}(Y_n)=\mathrm{IC}_{0,n}\) (finitely many \(K\)-orbits per \(Y_N\), connected stabilizers, dévissage), and \(\omega_*\) carries \(\mathcal C_K\) to \(\mathcal C_{K'}\).
  5. (c). Let \(\mathcal S\subseteq\mathcal C_I\) be the full subcategory of objects all of whose perverse cohomology objects have Jordan–Hölder factors in \(\{\mathrm{IC}_{i,n}\}\). It is closed under shifts and retracts (\({}^p\mathcal H^i\) of a retract is a retract), and under cones: in \({}^p\mathcal H^i(A)\to{}^p\mathcal H^i(B)\to{}^p\mathcal H^i(C)\to{}^p\mathcal H^{i+1}(A)\), \({}^p\mathcal H^i(C)\) is an extension of a subobject of \({}^p\mathcal H^{i+1}(A)\) by a quotient of \({}^p\mathcal H^i(B)\). So \(\mathcal S\) is thick and contains \(\mathcal C_I^{\mathrm{Iwa}}\) by (b). \(\mathrm{IC}_{A_\pm}\) is a simple perverse object different from all \(\mathrm{IC}_{i,n}\) (different support), hence not in \(\mathcal S\). The perverse t-structure is the one of each finite level; smooth pullback shifted by the relative dimension and closed pushforward \(Y_N\hookrightarrow Y_{N+1}\) are t-exact and preserve ICs, so \(\mathcal S\) is well defined on the colimit (checklist 1). Finally every object of \(\mathcal C_I\) lies in the thick hull of all simple objects by dévissage on the finitely many orbits of a level, and the simple objects are \(\mathrm{IC}_{i,n},\mathrm{IC}_{A_\pm}\) (Setup 6.1(c)).
  6. Fields. No weights are used beyond the decomposition theorem: steps 1–5 hold verbatim over \(\mathbb C\) (Betti, BBD 6.2.5) and over \(\overline{\mathbb F}_q\), \(p\) odd (\(p\ne2\) is used for the involution model of Setup 6.1).

Sources