On splendid derived and stable equivalences between blocks and finite groups (Q5945141)

Let \(\mathcal O\) be a complete discrete valuation ring with a residue field \(k\) of prime characteristic \(p\). In this paper all \(\mathcal O\)-algebras are assumed to be free and finitely generated \(\mathcal O\)-modules and all modules over an \(\mathcal O\)-algebra are finitely generated unitary modules. By definition an \(\mathcal O\)-algebra \(A\) is symmetric if \(A\cong A^*=\Hom_{\mathcal O}(A,{\mathcal O})\) (the \(\mathcal O\)-dual of \(A\)) as an \(A\)-\(A\)-bimodule. Let \(A\), \(B\) be symmetric \(\mathcal O\)-algebras. As is well-known, if \(M\) is an \(A\)-\(B\) bimodule and \(N\) is a \(B\)-\(A\)-bimodule that induce a Morita equivalence between \(A\) and \(B\) (i.e., \(M\otimes_BN\cong A\) and \(N\otimes_AM\cong B\) as bimodules), then \(N\cong M^*\) as \(B\)-\(A\)-bimodules. J. Rickard proved that \(A\) and \(B\) are derived equivalent if and only if there exists a bounded complex \(X\) of \(A\)-\(B\)-bimodules such that each component of \(X\) is projective in \(A\)-mod and in mod-\(B\) and the total bimodule complexes \(X\otimes_BX^*\) and \(X^*\otimes_AX\) are chain homotopic equivalent to \(A\) and \(B\) viewed as bimodule complexes concentrated in degree \(0\), resp. (such a complex \(X\) is called a Rickard tilting complex). Also if the \(A\)-\(B\)-bimodule \(M\) is projective in \(A\)-mod and in mod-\(B\) and if \(M\otimes_BM^*\cong A\oplus X\) where \(X\) is a projective \(A\)-\(A\)-bimodule and \(M^*\otimes_AM\cong B\oplus Y\) where \(Y\) is a projective \(B\)-\(B\)-bimodule as bimodules, then \(M\), as defined by M. Broué, is said to induce a stable equivalence of Morita type between \(A\) and \(B\).NEWLINENEWLINENEWLINEThe results of this paper deal with these topics in the context of Finite Group Block Theory. Let \(G\) be a finite group, let \(b\) be a block idempotent of \({\mathcal O}G\) and let \(P\) be a defect group of \(b\). Let \(\Delta P=\{(p,p)\mid p\in P\}\), so that \({\mathcal O}G\) and \({\mathcal O}Gb\) are left \({\mathcal O}\Delta P\)-modules. A source idempotent in \(({\mathcal O}Gb)^{\Delta P}\) is a primitive idempotent \(i\in({\mathcal O}Gb)^{\Delta P}\) such that \(Br_{\Delta P}(i)\neq 0\) where \(Br_{\Delta P}\colon{\mathcal O}G\to k(C_G(P))\) is the Brauer homomorphism.NEWLINENEWLINENEWLINELet \(H\) also be a finite group and let \(c\) be a block idempotent of \({\mathcal O}H\) with \(P\) also as a defect group. In subsequent work, J. Rickard defined a splendid tilting complex for the symmetric \(\mathcal O\)-algebras \({\mathcal O}Gb\) and \({\mathcal O}Hc\) to be a Rickard tilting complex \(X\) such that each indecomposable component of each bimodule in \(X\) is isomorphic to a summand of \({\mathcal O}Gb\otimes_{{\mathcal O}Q}{\mathcal O}Hc\) for some subgroup \(Q\) of \(P\). He demonstrated that if \(X\) is such a complex and \(b\) and \(c\) are principal blocks with the ``same \(p\)-local structure'', then \(X\) is ``compatible with the \(p\)-local structure''.NEWLINENEWLINENEWLINELet \(i\in({\mathcal O}Gb)^{\Delta P}\) and \(j\in({\mathcal O}Hc)^{\Delta P}\) be source idempotents for \(b\) and \(c\), resp. In a previous work, the author of this article defined a splended tilting complex \(X\) with respect to \((i,j)\) by requiring \(X\) to be a splendid tilting complex such that every indecomposable bimodule component of any bimodule in \(X\) is isomorphic to a direct summand of \({\mathcal O}Gi\otimes_{{\mathcal O}Q}j{\mathcal O}H\) for some subgroup \(Q\) of \(P\). With such a complex, he extended the results of J. Rickard to not-necessarily principal blocks \(b\) and \(c\).NEWLINENEWLINENEWLINESimilarly if the bimodule \(M\) is an \({\mathcal O}Gb\)-\({\mathcal O}Hc\)-bimodule that induces a Morita equivalence (a stable equivalence of Morita type) between \({\mathcal O}Gb\) and \({\mathcal O}Hc\) and \(M\) is isomorphic to a bimodule summand of \({\mathcal O}Gb\otimes_{{\mathcal O}P}{\mathcal O}Hc\) (\({\mathcal O}Gi\otimes_{{\mathcal O}P}j{\mathcal O}H\), resp.), then adjoin the qualifiers ``splendid'' and ``with respect to \((i,j)\)'' as above.NEWLINENEWLINENEWLINEThe main results of this paper are: Theorem 2.1. Let \(G\), \(H\) and \(L\) be finite groups and let \(b\), \(c\) and \(d\) be block idempotents of \({\mathcal O}G\), \({\mathcal O}H\) and \({\mathcal O}L\), resp. having a common defect group \(P\). Let \(i\in({\mathcal O}Gb)^{\Delta P}\), \(j\in({\mathcal O}Hc)^{\Delta P}\) and \(\ell\in({\mathcal O}Ld)^{\Delta P}\) be source idempotents. For any subgroup \(Q\) of \(P\), denote by \(e_Q\), \(f_Q\) and \(g_Q\) the unique blocks of \(kC_G(Q)\), \(k(C_H(Q))\) and \(kC_L(Q)\) determined by \(i\), \(j\) and \(\ell\), resp. Assume that for any two subgroups \(Q\), \(R\) of \(P\), we have \(E_G((Q,e_Q),(R,e_R))=F_H((Q,f_Q),(R,f_R))=E_L((Q,g_Q),(R,g_R))\). Let \(X\) be a splendid tilting complex with respect to \((i,j)\) of \({\mathcal O}Gb\)-\({\mathcal O}Hc\)-bimodules and let \(Y\) be a splendid tilting complex with respect to \((j,\ell)\) of \({\mathcal O}Hc\)-\({\mathcal O}Ld\)-bimodules. Then the following hold: (i) The total complex \(X\otimes_{{\mathcal O}Hc}Y\) is a splendid tilting complex with respect to \((i,\ell)\) of \({\mathcal O}Gb\)-\({\mathcal O}Ld\)-bimodules. (ii) For any subgroup \(Q\) of \(P\) there is an isomorphism of \(kC_G(Q)e_Q\)-\(kC_L(Q)g_Q\)-bimodule complexes \(e_Q(X\otimes_{{\mathcal O}Hc}Y)(\Delta Q)g_Q\cong e_QX(\Delta Q)f_Q\otimes_{kC_H(Q)f_Q}f_QY(\Delta Q)g_Q\).NEWLINENEWLINENEWLINETheorem 3.1. Let \(G\), \(H\) be finite groups, let \(b\), \(c\) be blocks of \({\mathcal O}G\) and \({\mathcal O}H\), resp. with a common defect group \(P\). Let \(i\in({\mathcal O}Gb)^{\Delta P}\) and \(j\in({\mathcal O}Hc)^{\Delta P}\) be source idempotents. For any subgroup \(Q\) of \(P\), denote by \(e_Q\), \(f_Q\) the unique blocks of \(kC_G(Q)\), \(kC_H(Q)\) determined by \(i\), \(j\) resp. Let \(M\) be an indecomposable direct summand of the \({\mathcal O}Gb\)-\({\mathcal O}Hc\)-bimodule \({\mathcal O}Gi\otimes_{{\mathcal O}P}j{\mathcal O}H\). The following are equivalent: (i) The bimodule \(M\) and its \(\mathcal O\)-dual \(M^*\) induce a stable equivalence of Morita type between \({\mathcal O}Gb\) and \({\mathcal O}Hc\). (ii) For any nontrivial subgroup \(Q\) of \(P\) the bimodule \(e_QM(\Delta Q)f_Q\) and its \(\mathcal O\)-dual induce a Morita equivalence between \(kC_G(Q)e_Q\) and \(kC_H(Q)f_Q\) and for any two subgroups \(Q\), \(R\) of \(P\), we have \(E_Q((Q,e_Q), (R,e_R))=E_H((Q,f_Q),(R,f_R))\). In that case, there is a bimodule isomorphism \(M_j\cong{\mathcal O}Gi\oplus U\) for some projective \({\mathcal O}Gb\)-\({\mathcal O}P\)-bimodule \(U\) which induces for any nontrivial subgroup \(Q\) of \(P\) algebra isomorphisms \((i{\mathcal O}Gi)^{\Delta Q}/(i{\mathcal O}Gi)^{\Delta Q}_1\cong(j{\mathcal O}Hj)^{\Delta Q}/(j{\mathcal O}Hj)^{\Delta Q}_1\) and \((i{\mathcal O}Gi)(\Delta Q)\cong(j{\mathcal O}Hj)(\Delta Q)\) which in turn induce fusion compatible bijections between the sets of local points of \(Q\) in \(i{\mathcal O}Gi\) and \(j{\mathcal O}Hj\), resp.NEWLINENEWLINENEWLINETheorem 4.1. (Puig and Scott) Under the hypotheses of Theorem 3.1, the following are equivalent: (i) The bimodule \(M\) and its \(\mathcal O\)-dual \(M^*\) induce a Morita equivalence between \({\mathcal O}Gb\) and \({\mathcal O}Hc\). (ii) There is an algebra isomorphism \(\varphi\colon i{\mathcal O}Gi\to j{\mathcal O}Hj\) mapping \(ui\) to \(uj\) for any \(u\in P\) such that \(M\cong{\mathcal O}Gi\otimes_{i{\mathcal O}Gi}{_\varphi(j{\mathcal O}H)}\).NEWLINENEWLINENEWLINEIf these statements hold, there is an isomorphism of \({\mathcal O}G\)-\({\mathcal O}P\)-bimodules \({\mathcal O}Gi\cong Mj\); moreover the correspondence sending \(M\) to the algebra isomorphism \(\varphi\) induces a bijection between the isomorphism classes of indecomposable direct summands of \({\mathcal O}Gi\otimes_{{\mathcal O}P}j{\mathcal O}H\) inducing a Morita equivalence between \({\mathcal O}Gb\), \({\mathcal O}Hc\) and the set of \(((j{\mathcal O}Hj)^P)^\times\)-conjugacy classes of algebra isomorphisms \(i{\mathcal O}Gi\cong j{\mathcal O}Hj\) mapping \(ui\) to \(uj\) for all \(u\in P\).NEWLINENEWLINENEWLINEFor the final main result of this paper, let \(G\) be a finite group with block \(b\) and having nontrivial cyclic defect group \(P\). Set \(Z=\Omega_1(P)\) and \(H=N_G(Z)\). Since \(C_G(P)\leq H\), there is a unique block \(c\) of \({\mathcal O}H\) having defect group \(P\) and such that \(Br_{\Delta P}(b)=Br_{\Delta P}(c)\). Let \(i\in({\mathcal O}Gb)^{\Delta P}\) and \(j\in({\mathcal O}Hc)^{\Delta P}\) be source idempotents. Let \(e\) be the unique block of \(kC_G(P)=kC_H(P)\) such that \(Br_{\Delta P}(i)e=Br_{\Delta P}(i)\). Set \(E=N_G(P,e)/C_G(P)\) and let \(P\rtimes E\) denote the naturally defined semi-direct product. Here \(E\) is a cyclic \(p'\)-group and \(P\rtimes E\) is a Frobenius group with kernel \(P\). The unit element \(1\) of \({\mathcal O}(P\rtimes E)\) is both its unique block and source idempotent.NEWLINENEWLINENEWLINETheorem 5.8. The complex constructed by R. Rouquier in this situation is a splendid tilting complex with respect to \((i,1)\) of \({\mathcal O}Gb\)-\({\mathcal O}(P\rtimes E)\)-bimodules.NEWLINENEWLINENEWLINEThe proofs of these four main results are presented in Sections 2-5. A brief review of fundamental tools and structures in this area is presented in Sections 6 and 7.