From Baum-Connes to the Novikov conjecture I What is the Baum-Connes conjecture? I What is the Novikov conjecture? I How does Baum-Connes ⇒ Novikov ? 1 / 30 Table of contents Brief description of these conjectures Details on the Baum-Connes conjecture Details on the Novikov conjecture Baum-Connes ⇒ Novikov 2 / 30 The Baum-Connes conjecture The Baum-Connes conjecture expresses a relationship K-theory of the C*-algebra of a group −→ K-homology of the classifying space of proper actions of the group. 3 / 30 The Baum-Connes conjecture The Baum-Connes conjecture expresses a relationship K-theory of the C*-algebra of a group −→ K-homology of the classifying space of proper actions of the group. G locally compact group Cr∗ (G ) reduced group C*-algebra of G E G classifying space for proper actions of G 3 / 30 The Baum-Connes conjecture The Baum-Connes conjecture expresses a relationship K-theory of the C*-algebra of a group −→ K-homology of the classifying space of proper actions of the group. G locally compact group Cr∗ (G ) reduced group C*-algebra of G E G classifying space for proper actions of G Analytic assembly map µ : KjG (E G ) −→ Kj (Cr∗ (G )) 3 / 30 The Baum-Connes conjecture The Baum-Connes conjecture expresses a relationship K-theory of the C*-algebra of a group −→ K-homology of the classifying space of proper actions of the group. G locally compact group Cr∗ (G ) reduced group C*-algebra of G E G classifying space for proper actions of G Analytic assembly map µ : KjG (E G ) −→ Kj (Cr∗ (G )) Conjecture (P. Baum, A. Connes; 1982) The assembly map µ is an isomorphism of abelian groups. 3 / 30 The Baum-Connes conjecture: µ : KjG (E G ) −→ Kj (Cr∗ (G )) is an isomorphism. I a way of calculating the operator K-theory of Cr∗ (G ) by means of group homology and representation theory 4 / 30 The Baum-Connes conjecture: µ : KjG (E G ) −→ Kj (Cr∗ (G )) is an isomorphism. I a way of calculating the operator K-theory of Cr∗ (G ) by means of group homology and representation theory I relates an analytical object Kj (Cr∗ (G )) with a geometrical/topological object KjG (E G ) 4 / 30 The Baum-Connes conjecture: µ : KjG (E G ) −→ Kj (Cr∗ (G )) is an isomorphism. I a way of calculating the operator K-theory of Cr∗ (G ) by means of group homology and representation theory I relates an analytical object Kj (Cr∗ (G )) with a geometrical/topological object KjG (E G ) I implies several other conjectures in diverse fields (topology, geometry, functional analysis, ..) 4 / 30 The Baum-Connes conjecture: µ : KjG (E G ) −→ Kj (Cr∗ (G )) is an isomorphism. I a way of calculating the operator K-theory of Cr∗ (G ) by means of group homology and representation theory I relates an analytical object Kj (Cr∗ (G )) with a geometrical/topological object KjG (E G ) I implies several other conjectures in diverse fields (topology, geometry, functional analysis, ..) I injectivity of µ has geometrical/topological implications: e.g. Novikov conj, Gromov-Lawson-Rosenberg conj 4 / 30 The Baum-Connes conjecture: µ : KjG (E G ) −→ Kj (Cr∗ (G )) is an isomorphism. I a way of calculating the operator K-theory of Cr∗ (G ) by means of group homology and representation theory I relates an analytical object Kj (Cr∗ (G )) with a geometrical/topological object KjG (E G ) I implies several other conjectures in diverse fields (topology, geometry, functional analysis, ..) I injectivity of µ has geometrical/topological implications: e.g. Novikov conj, Gromov-Lawson-Rosenberg conj I surjectivity of µ has analytical implications: e.g. Kaplansky-Kadison conj 4 / 30 Groups for which Baum-Connes conjecture has been verified Some examples of classes of groups I Compact groups I Abelian groups I Lie groups with finitely many conn. comps. I hyperbolic groups I a-T-menable groups (includes amenable groups, groups acting properly on trees, ...) I p-adic groups I adelic groups 5 / 30 Groups for which Baum-Connes conjecture has been verified Some examples of classes of groups I Compact groups I Abelian groups I Lie groups with finitely many conn. comps. I hyperbolic groups I a-T-menable groups (includes amenable groups, groups acting properly on trees, ...) I p-adic groups I adelic groups Counterexamples? Have been established for generalized forms of the conjecture: I Baum-Connes conjecture with coefficients. µ : KjG (E G , A) −→ Kj (A or G ). I coarse Baum-Connes conjecture: µ : K∗ (X ) −→ K∗ (C ∗ (X )). 5 / 30 The Novikov conjecture The Novikov conjecture is a statement about homotopy invariance of higher signatures associated to a discrete group and closed, connected, oriented, smooth manifolds. 6 / 30 The Novikov conjecture The Novikov conjecture is a statement about homotopy invariance of higher signatures associated to a discrete group and closed, connected, oriented, smooth manifolds. Let Γ be a discrete group and BΓ its classifying space. M a smooth, closed, connected, oriented manifold. LM ∈ H 4∗ (M; Q) the total L-class of M. 6 / 30 The Novikov conjecture The Novikov conjecture is a statement about homotopy invariance of higher signatures associated to a discrete group and closed, connected, oriented, smooth manifolds. Let Γ be a discrete group and BΓ its classifying space. M a smooth, closed, connected, oriented manifold. LM ∈ H 4∗ (M; Q) the total L-class of M. For x ∈ H ∗ (BΓ; Q) and f : M −→ BΓ continuous map, define σx (M, f ) := hLM ∪ f ∗ (x), [M]i ∈ Q ”higher signature” . 6 / 30 The Novikov conjecture The Novikov conjecture is a statement about homotopy invariance of higher signatures associated to a discrete group and closed, connected, oriented, smooth manifolds. Let Γ be a discrete group and BΓ its classifying space. M a smooth, closed, connected, oriented manifold. LM ∈ H 4∗ (M; Q) the total L-class of M. For x ∈ H ∗ (BΓ; Q) and f : M −→ BΓ continuous map, define σx (M, f ) := hLM ∪ f ∗ (x), [M]i ∈ Q ”higher signature” . Conjecture (S. Novikov; 1965) All higher signatures determined by Γ are oriented homotopy invariants. 6 / 30 The Novikov conjecture The Novikov conjecture is a statement about homotopy invariance of higher signatures associated to a discrete group and closed, connected, oriented, smooth manifolds. Let Γ be a discrete group and BΓ its classifying space. M a smooth, closed, connected, oriented manifold. LM ∈ H 4∗ (M; Q) the total L-class of M. For x ∈ H ∗ (BΓ; Q) and f : M −→ BΓ continuous map, define σx (M, f ) := hLM ∪ f ∗ (x), [M]i ∈ Q ”higher signature” . Conjecture (S. Novikov; 1965) All higher signatures determined by Γ are oriented homotopy invariants. Meaning: for any x ∈ H ∗ (BΓ; Q), any orientation preserving homotopy equivalence h : N −→ M of closed oriented manifolds N and M, and f : M −→ BΓ, σx (M, f ) = σx (N, f ◦ h). 6 / 30 The Novikov conjecture: for x ∈ H ∗ (BΓ; Q) and reference map f : M −→ BΓ, the characteristic number hLM ∪ f ∗ (x), [M]i ∈ Q is an oriented homotopy invariant. 7 / 30 The Novikov conjecture: for x ∈ H ∗ (BΓ; Q) and reference map f : M −→ BΓ, the characteristic number hLM ∪ f ∗ (x), [M]i ∈ Q is an oriented homotopy invariant. I characteristic numbers are interesting and important for classification of manifolds (e.g. Euler characteristic) 7 / 30 The Novikov conjecture: for x ∈ H ∗ (BΓ; Q) and reference map f : M −→ BΓ, the characteristic number hLM ∪ f ∗ (x), [M]i ∈ Q is an oriented homotopy invariant. I characteristic numbers are interesting and important for classification of manifolds (e.g. Euler characteristic) I the conjecture expresses a relationship between a characteristic class of the manifold to the underlying homotopy theory 7 / 30 The Novikov conjecture: for x ∈ H ∗ (BΓ; Q) and reference map f : M −→ BΓ, the characteristic number hLM ∪ f ∗ (x), [M]i ∈ Q is an oriented homotopy invariant. I characteristic numbers are interesting and important for classification of manifolds (e.g. Euler characteristic) I the conjecture expresses a relationship between a characteristic class of the manifold to the underlying homotopy theory I Hirzebruch signature theorem: σ(M) = hLM, [M]i. The signature is an oriented homotopy invariant of M 7 / 30 The Novikov conjecture: for x ∈ H ∗ (BΓ; Q) and reference map f : M −→ BΓ, the characteristic number hLM ∪ f ∗ (x), [M]i ∈ Q is an oriented homotopy invariant. I characteristic numbers are interesting and important for classification of manifolds (e.g. Euler characteristic) I the conjecture expresses a relationship between a characteristic class of the manifold to the underlying homotopy theory I Hirzebruch signature theorem: σ(M) = hLM, [M]i. The signature is an oriented homotopy invariant of M I LM is characteristic class, a certain polynomial in rational Pontrjagin classes. These are themselves not homotopy invariants 7 / 30 The Novikov conjecture: for x ∈ H ∗ (BΓ; Q) and reference map f : M −→ BΓ, the characteristic number hLM ∪ f ∗ (x), [M]i ∈ Q is an oriented homotopy invariant. I characteristic numbers are interesting and important for classification of manifolds (e.g. Euler characteristic) I the conjecture expresses a relationship between a characteristic class of the manifold to the underlying homotopy theory I Hirzebruch signature theorem: σ(M) = hLM, [M]i. The signature is an oriented homotopy invariant of M I LM is characteristic class, a certain polynomial in rational Pontrjagin classes. These are themselves not homotopy invariants I There are no known counterexamples to the Novikov conjecture 7 / 30 Details on the Baum-Connes conjecture 8 / 30 Details on the Baum-Connes conjecture I The classifying space E G for proper actions 8 / 30 Details on the Baum-Connes conjecture I The classifying space E G for proper actions I KK and KK G 8 / 30 Details on the Baum-Connes conjecture I The classifying space E G for proper actions I KK and KK G I K-theory and K-homology 8 / 30 Details on the Baum-Connes conjecture I The classifying space E G for proper actions I KK and KK G I K-theory and K-homology I Reduced group C*-algebra Cr∗ (G ) 8 / 30 Details on the Baum-Connes conjecture I The classifying space E G for proper actions I KK and KK G I K-theory and K-homology I Reduced group C*-algebra Cr∗ (G ) I G -equivariant K-homology KjG 8 / 30 Details on the Baum-Connes conjecture I The classifying space E G for proper actions I KK and KK G I K-theory and K-homology I Reduced group C*-algebra Cr∗ (G ) I G -equivariant K-homology KjG I The analytic assembly map µ as the G -index IndexG 8 / 30 The classifying space E G G locally compact, Hausdorff, second countable group. 9 / 30 The classifying space E G G locally compact, Hausdorff, second countable group. X is called a G -space if equipped with a continuous action G × X −→ X 9 / 30 The classifying space E G G locally compact, Hausdorff, second countable group. X is called a G -space if equipped with a continuous action G × X −→ X Interested in proper actions (the G -action is locally induced from action of a compact subgroup). 9 / 30 The classifying space E G G locally compact, Hausdorff, second countable group. X is called a G -space if equipped with a continuous action G × X −→ X Interested in proper actions (the G -action is locally induced from action of a compact subgroup). E G := a universal example for proper actions of G , meaning E G is a proper G -space such that: if X is any proper G -space, there exists a unique (up to G -homotopy) G -map X −→ E G . E G can be constructed (a G-CW-complex) as an infinite join W ∗ W ∗ W ∗ · · · where W = ∪G /H is the disjoint union over compact subgroups H ⊂ G . 9 / 30 Kasparov’s operator K-theory KK and KK G Homotopy invariant bifunctor KKj (−, −) : C*-algebras −→ abelian groups. j = 0, 1. 10 / 30 Kasparov’s operator K-theory KK and KK G Homotopy invariant bifunctor KKj (−, −) : C*-algebras −→ abelian groups. j = 0, 1. Contravariant in first argument, covariant in second. 10 / 30 Kasparov’s operator K-theory KK and KK G Homotopy invariant bifunctor KKj (−, −) : C*-algebras −→ abelian groups. j = 0, 1. Contravariant in first argument, covariant in second. KKj (A, B) = homotopy classes of triples (E , π, F ) where 10 / 30 Kasparov’s operator K-theory KK and KK G Homotopy invariant bifunctor KKj (−, −) : C*-algebras −→ abelian groups. j = 0, 1. Contravariant in first argument, covariant in second. KKj (A, B) = homotopy classes of triples (E , π, F ) where I E is count.gen. Hilbert B-module 10 / 30 Kasparov’s operator K-theory KK and KK G Homotopy invariant bifunctor KKj (−, −) : C*-algebras −→ abelian groups. j = 0, 1. Contravariant in first argument, covariant in second. KKj (A, B) = homotopy classes of triples (E , π, F ) where I E is count.gen. Hilbert B-module I π : A −→ LA (E ) *-homomorphism 10 / 30 Kasparov’s operator K-theory KK and KK G Homotopy invariant bifunctor KKj (−, −) : C*-algebras −→ abelian groups. j = 0, 1. Contravariant in first argument, covariant in second. KKj (A, B) = homotopy classes of triples (E , π, F ) where I E is count.gen. Hilbert B-module I π : A −→ LA (E ) *-homomorphism I F ∈ LA (E ) satisfies π(a)(F 2 − 1), [π(a), F ] ∈ KA (E ) 10 / 30 Kasparov’s operator K-theory KK and KK G Homotopy invariant bifunctor KKj (−, −) : C*-algebras −→ abelian groups. j = 0, 1. Contravariant in first argument, covariant in second. KKj (A, B) = homotopy classes of triples (E , π, F ) where I E is count.gen. Hilbert B-module I π : A −→ LA (E ) *-homomorphism I F ∈ LA (E ) satisfies π(a)(F 2 − 1), [π(a), F ] ∈ KA (E ) I grading: for j = 0, E has Z2 -grading, for j = 1 no grading 10 / 30 Kasparov’s operator K-theory KK and KK G Homotopy invariant bifunctor KKj (−, −) : C*-algebras −→ abelian groups. j = 0, 1. Contravariant in first argument, covariant in second. KKj (A, B) = homotopy classes of triples (E , π, F ) where I E is count.gen. Hilbert B-module I π : A −→ LA (E ) *-homomorphism I F ∈ LA (E ) satisfies π(a)(F 2 − 1), [π(a), F ] ∈ KA (E ) I grading: for j = 0, E has Z2 -grading, for j = 1 no grading I addition: by direct sum ⊕ 10 / 30 Kasparov’s operator K-theory KK and KK G Homotopy invariant bifunctor KKj (−, −) : C*-algebras −→ abelian groups. j = 0, 1. Contravariant in first argument, covariant in second. KKj (A, B) = homotopy classes of triples (E , π, F ) where I E is count.gen. Hilbert B-module I π : A −→ LA (E ) *-homomorphism I F ∈ LA (E ) satisfies π(a)(F 2 − 1), [π(a), F ] ∈ KA (E ) I grading: for j = 0, E has Z2 -grading, for j = 1 no grading I addition: by direct sum ⊕ I homotopy: a continuous path of triples 10 / 30 Kasparov’s operator K-theory KK and KK G Homotopy invariant bifunctor KKj (−, −) : C*-algebras −→ abelian groups. j = 0, 1. Contravariant in first argument, covariant in second. KKj (A, B) = homotopy classes of triples (E , π, F ) where I E is count.gen. Hilbert B-module I π : A −→ LA (E ) *-homomorphism I F ∈ LA (E ) satisfies π(a)(F 2 − 1), [π(a), F ] ∈ KA (E ) I grading: for j = 0, E has Z2 -grading, for j = 1 no grading I addition: by direct sum ⊕ I homotopy: a continuous path of triples KK G (−, −) : G-C*-algebras −→ abelian groups. 10 / 30 Kasparov’s operator K-theory KK and KK G Homotopy invariant bifunctor KKj (−, −) : C*-algebras −→ abelian groups. j = 0, 1. Contravariant in first argument, covariant in second. KKj (A, B) = homotopy classes of triples (E , π, F ) where I E is count.gen. Hilbert B-module I π : A −→ LA (E ) *-homomorphism I F ∈ LA (E ) satisfies π(a)(F 2 − 1), [π(a), F ] ∈ KA (E ) I grading: for j = 0, E has Z2 -grading, for j = 1 no grading I addition: by direct sum ⊕ I homotopy: a continuous path of triples KK G (−, −) : G-C*-algebras −→ abelian groups. KKjG (A, B) = homotopy classes of triples (E , π, F ) as above, in addition I E carries a unitary representation G −→ LA (E ) I π is G -equivariant I F is G -continuous (g 7→ g · F norm cont.) and G -equivariant 10 / 30 Kasparov’s operator K-theory KK and KK G Homotopy invariant bifunctor KKj (−, −) : C*-algebras −→ abelian groups. j = 0, 1. Contravariant in first argument, covariant in second. KKj (A, B) = homotopy classes of triples (E , π, F ) where I E is count.gen. Hilbert B-module I π : A −→ LA (E ) *-homomorphism I F ∈ LA (E ) satisfies π(a)(F 2 − 1), [π(a), F ] ∈ KA (E ) I grading: for j = 0, E has Z2 -grading, for j = 1 no grading I addition: by direct sum ⊕ I homotopy: a continuous path of triples KK G (−, −) : G-C*-algebras −→ abelian groups. KKjG (A, B) = homotopy classes of triples (E , π, F ) as above, in addition I E carries a unitary representation G −→ LA (E ) I π is G -equivariant I F is G -continuous (g 7→ g · F norm cont.) and G -equivariant {e} KKj (A, B) = KKj (A, B) with {e} trivial group. 10 / 30 K-theory and K-homology C*-algebra A. 11 / 30 K-theory and K-homology C*-algebra A. K-theory. Kj (A) = KKj (C, A). j = 0, 1. 11 / 30 K-theory and K-homology C*-algebra A. K-theory. Kj (A) = KKj (C, A). j = 0, 1. More conventional definition: K0 (A) built from equivalence classes of idempotents in matrix algebras ∪Mn (A) i.e. built from finitely generated projective A-modules. 11 / 30 K-theory and K-homology C*-algebra A. K-theory. Kj (A) = KKj (C, A). j = 0, 1. More conventional definition: K0 (A) built from equivalence classes of idempotents in matrix algebras ∪Mn (A) i.e. built from finitely generated projective A-modules. Proper G -space X . G -equivariant K-homology. KjG (X ) := KKjG (C0 (X ), C). 11 / 30 K-theory and K-homology C*-algebra A. K-theory. Kj (A) = KKj (C, A). j = 0, 1. More conventional definition: K0 (A) built from equivalence classes of idempotents in matrix algebras ∪Mn (A) i.e. built from finitely generated projective A-modules. Proper G -space X . G -equivariant K-homology. KjG (X ) := KKjG (C0 (X ), C). A generalized index map. The G -index construction IndexG : KjG (X ) −→ Kj (Cr∗ (G )), in other words 11 / 30 K-theory and K-homology C*-algebra A. K-theory. Kj (A) = KKj (C, A). j = 0, 1. More conventional definition: K0 (A) built from equivalence classes of idempotents in matrix algebras ∪Mn (A) i.e. built from finitely generated projective A-modules. Proper G -space X . G -equivariant K-homology. KjG (X ) := KKjG (C0 (X ), C). A generalized index map. The G -index construction IndexG : KjG (X ) −→ Kj (Cr∗ (G )), in other words IndexG : KKjG (C0 (X ), C) −−−−→ KKj (C, Cr∗ (G )) (E , π, F ) G-module −−−−→ (E, F) −−−−−−−−−−→ Cr∗ (G )-module completed to 11 / 30 The reduced group C*-algebra Cr∗ (G ) Locally compact G with a fixed Haar measure. 12 / 30 The reduced group C*-algebra Cr∗ (G ) Locally compact G with a fixed Haar measure. Hilbert space L2 (G ). Left regular representation λ : G −→ B(L2 (G )), λs (ξ)(t) = ξ(s −1 t). 12 / 30 The reduced group C*-algebra Cr∗ (G ) Locally compact G with a fixed Haar measure. Hilbert space L2 (G ). Left regular representation λs (ξ)(t) = ξ(s −1 t). R Consider L1 (G ) with convolution product, (f × g )(t) = G f (s)g (s −1 t) ds, and involution f ∗ (t) = ∆(t)−1 f (t −1 ). This makes L1 (G ) a Banach algebra. λ : G −→ B(L2 (G )), 12 / 30 The reduced group C*-algebra Cr∗ (G ) Locally compact G with a fixed Haar measure. Hilbert space L2 (G ). Left regular representation λs (ξ)(t) = ξ(s −1 t). R Consider L1 (G ) with convolution product, (f × g )(t) = G f (s)g (s −1 t) ds, and involution f ∗ (t) = ∆(t)−1 f (t −1 ). This makes L1 (G ) a Banach algebra. λ : G −→ B(L2 (G )), The left regular representation induces a *-homomorphism λ : L1 (G ) −→ B(L2 (G )), f 7→ λf , Z λf (ξ)(t) = f (s)(λs ξ)(t) ds. G 12 / 30 The reduced group C*-algebra Cr∗ (G ) Locally compact G with a fixed Haar measure. Hilbert space L2 (G ). Left regular representation λs (ξ)(t) = ξ(s −1 t). R Consider L1 (G ) with convolution product, (f × g )(t) = G f (s)g (s −1 t) ds, and involution f ∗ (t) = ∆(t)−1 f (t −1 ). This makes L1 (G ) a Banach algebra. λ : G −→ B(L2 (G )), The left regular representation induces a *-homomorphism λ : L1 (G ) −→ B(L2 (G )), f 7→ λf , Z λf (ξ)(t) = f (s)(λs ξ)(t) ds. G The reduced group C*-algebra of G is defined Cr∗ (G ) := λ(L1 (G )) ⊆ B(L2 (G )). 12 / 30 The G -equivariant K-homology KjG (X ) Let X be a G -compact, proper G -space. 13 / 30 The G -equivariant K-homology KjG (X ) Let X be a G -compact, proper G -space. G -equivariant K-homology of X KjG (X ) := KKjG (C0 (X ), C) = ”G-equivariant abstract elliptic operators” 13 / 30 The G -equivariant K-homology KjG (X ) Let X be a G -compact, proper G -space. G -equivariant K-homology of X KjG (X ) := KKjG (C0 (X ), C) = ”G-equivariant abstract elliptic operators” KjG (X ) consists of homotopy classes of triples (H, π, F ) where 13 / 30 The G -equivariant K-homology KjG (X ) Let X be a G -compact, proper G -space. G -equivariant K-homology of X KjG (X ) := KKjG (C0 (X ), C) = ”G-equivariant abstract elliptic operators” KjG (X ) consists of homotopy classes of triples (H, π, F ) where I H is Hilbert space with unitary representation G −→ B(H) I G -covariant representation π : C0 (X ) −→ B(H) I F is G -equivariant operator, F ∈ B(H) I the operators π(f )(F 2 − 1) and [π(f ), F ] are compact I j = 0 : H has Z2 -grading and representations are grading preserving, F is grading reversing. j = 1, no grading. Homotopy: norm continuous path (Ft )t∈[0,1] connecting the triples (H, π, F0 ) and (H, π, F1 ) Addition operation ⊕: form the direct sum of the triples 13 / 30 The G -equivariant K-homology KjG (X ) X is assumed G -compact proper G -space. homotopy classes of G -equivariant G Kj (X ) = abstract elliptic operators on X These are abelian groups with addition ⊕. 14 / 30 The G -equivariant K-homology KjG (X ) X is assumed G -compact proper G -space. homotopy classes of G -equivariant G Kj (X ) = abstract elliptic operators on X These are abelian groups with addition ⊕. In general, when Z is a proper G -space not G -compact, KjG (Z ) = lim −→ G -compact X ⊂ Z KjG (X ) 14 / 30 The assembly map µ µ : KjG (X ) −→ Kj (Cr∗ (G )) = KKj (C, Cr∗ (G )) 15 / 30 The assembly map µ µ : KjG (X ) −→ Kj (Cr∗ (G )) = KKj (C, Cr∗ (G )) (H, π, F ) 7→ (H, F) 15 / 30 The assembly map µ µ : KjG (X ) −→ Kj (Cr∗ (G )) = KKj (C, Cr∗ (G )) (H, π, F ) Given an element (H, π, F ) ∈ KjG (X ) 7→ (H, F) we construct an element (H, F) ∈ Kj (Cr∗ (G )). 15 / 30 The assembly map µ µ : KjG (X ) −→ Kj (Cr∗ (G )) = KKj (C, Cr∗ (G )) (H, π, F ) Given an element (H, π, F ) ∈ KjG (X ) 7→ (H, F) we construct an element (H, F) ∈ Kj (Cr∗ (G )). Let H0 = π(Cc (X ))H. This is a right module for the convolution algebra Cc (G ) ⊂ Cr∗ (G ), Z ξ · f := f (s) s −1 · ξ ds, ξ ∈ H0 , f ∈ Cc (G ), G 15 / 30 The assembly map µ µ : KjG (X ) −→ Kj (Cr∗ (G )) = KKj (C, Cr∗ (G )) (H, π, F ) Given an element (H, π, F ) ∈ KjG (X ) 7→ (H, F) we construct an element (H, F) ∈ Kj (Cr∗ (G )). Let H0 = π(Cc (X ))H. This is a right module for the convolution algebra Cc (G ) ⊂ Cr∗ (G ), Z ξ · f := f (s) s −1 · ξ ds, ξ ∈ H0 , f ∈ Cc (G ), G and carries the Cc (G )-valued inner product hv1 , v2 i(g ) = hv1 , g · v2 i. 15 / 30 The assembly map µ µ : KjG (X ) −→ Kj (Cr∗ (G )) = KKj (C, Cr∗ (G )) (H, π, F ) Given an element (H, π, F ) ∈ KjG (X ) 7→ (H, F) we construct an element (H, F) ∈ Kj (Cr∗ (G )). Let H0 = π(Cc (X ))H. This is a right module for the convolution algebra Cc (G ) ⊂ Cr∗ (G ), Z ξ · f := f (s) s −1 · ξ ds, ξ ∈ H0 , f ∈ Cc (G ), G and carries the Cc (G )-valued inner product hv1 , v2 i(g ) = hv1 , g · v2 i. Define H = completion of H0 in the norm associated to this inner product, pass from the dense subspace Cc (G ) to its completion Cr∗ (G ). The operator F : H −→ H passes to an operator F : H −→ H. 15 / 30 The assembly map µ This gives us a map µ : KjG (X ) −→ Kj (Cr∗ (G )) µ([(H, π, F )]) = [(H, F)] 16 / 30 The assembly map µ This gives us a map µ : KjG (X ) −→ Kj (Cr∗ (G )) µ([(H, π, F )]) = [(H, F)] which is the G -index construction µ((H, π, F )) = IndexG (F ). 16 / 30 The assembly map µ This gives us a map µ : KjG (X ) −→ Kj (Cr∗ (G )) µ([(H, π, F )]) = [(H, F)] which is the G -index construction µ((H, π, F )) = IndexG (F ). Construction passes to direct limits, and we get the analytic assembly map µ : KjG (E G ) −→ Kj (Cr∗ (G )). 16 / 30 Details on the Novikov conjecture 17 / 30 Details on the Novikov conjecture I Signature of a manifold 17 / 30 Details on the Novikov conjecture I Signature of a manifold I Signature theorem and the signature operator 17 / 30 Details on the Novikov conjecture I Signature of a manifold I Signature theorem and the signature operator I Higher signatures and homological formulation of Novikov conjecture 17 / 30 Details on the Novikov conjecture I Signature of a manifold I Signature theorem and the signature operator I Higher signatures and homological formulation of Novikov conjecture I Dual chern character 17 / 30 Details on the Novikov conjecture I Signature of a manifold I Signature theorem and the signature operator I Higher signatures and homological formulation of Novikov conjecture I Dual chern character I Homotopy invariance of analytic signature 17 / 30 Details on the Novikov conjecture I Signature of a manifold I Signature theorem and the signature operator I Higher signatures and homological formulation of Novikov conjecture I Dual chern character I Homotopy invariance of analytic signature I Strong Novikov conjecture 17 / 30 Signature of manifold Let M be a closed, connected, oriented smooth manifold of dimension n = 4k. 18 / 30 Signature of manifold Let M be a closed, connected, oriented smooth manifold of dimension n = 4k. The intersection form (cup product) Z 2k 2k α∧β H (M) × H (M) −→ R, ([α], [β]) 7→ M is a symmetric bilinear form. 18 / 30 Signature of manifold Let M be a closed, connected, oriented smooth manifold of dimension n = 4k. The intersection form (cup product) Z 2k 2k α∧β H (M) × H (M) −→ R, ([α], [β]) 7→ M is a symmetric bilinear form. Define the signature of the manifold σ(M) = signature of the intersection form 18 / 30 Signature of manifold Let M be a closed, connected, oriented smooth manifold of dimension n = 4k. The intersection form (cup product) Z 2k 2k α∧β H (M) × H (M) −→ R, ([α], [β]) 7→ M is a symmetric bilinear form. Define the signature of the manifold σ(M) = signature of the intersection form Hirzebruch Signature Theorem gives a cohomological formula for σ(M), in terms of pairing between L-class and fundamental class of the manifold. 18 / 30 Signature of manifold Let M be a closed, connected, oriented smooth manifold of dimension n = 4k. The intersection form (cup product) Z 2k 2k α∧β H (M) × H (M) −→ R, ([α], [β]) 7→ M is a symmetric bilinear form. Define the signature of the manifold σ(M) = signature of the intersection form Hirzebruch Signature Theorem gives a cohomological formula for σ(M), in terms of pairing between L-class and fundamental class of the manifold. The fundamental class [M] ∈ Hn (M; Z) = Z is the generator element. It pairs with top-degree differential forms Z hω, [M]i = ω. M 18 / 30 Signature of manifold The total L-class LM ∈ L k≥0 H 4k (M; Q) LM = 1 + L1 (p1 ) + L2 (p1 , p2 ) + L3 (p1 , p2 , p3 ) + . . . consists of homogeneous polynomials Lk in the rational Pontrjagin classes pi = pi (M; Q) ∈ H 4i (M; Q) 19 / 30 Signature of manifold The total L-class LM ∈ L k≥0 H 4k (M; Q) LM = 1 + L1 (p1 ) + L2 (p1 , p2 ) + L3 (p1 , p2 , p3 ) + . . . consists of homogeneous polynomials Lk in the rational Pontrjagin classes pi = pi (M; Q) ∈ H 4i (M; Q) Theorem (Hirzebruch) σ(M) = hLM, [M]i. 19 / 30 Signature of manifold The total L-class LM ∈ L k≥0 H 4k (M; Q) LM = 1 + L1 (p1 ) + L2 (p1 , p2 ) + L3 (p1 , p2 , p3 ) + . . . consists of homogeneous polynomials Lk in the rational Pontrjagin classes pi = pi (M; Q) ∈ H 4i (M; Q) Theorem (Hirzebruch) σ(M) = hLM, [M]i. One way of approaching this formula is I Define the signature operator ∂, a classical elliptic differential operator I Identify the signature of M with the index of ∂ via Hodge theory and harmonic forms I Compute index(∂) by Atiyah-Singer index thm 19 / 30 Signature operator of M Let Ω = ⊕Ωk be sections of exterior algebra of differential forms, Ωk = C ∞ (Λk T ∗ M). d : Ωi −→ Ωi+1 exterior differentiation with formal adjoint d ∗ : Ωi+1 −→ Ωi with R respect to pairing (α, β) = M α ∧ ∗β, Hodge star ∗ : Ωi −→ Ωn−i d + d ∗ : Ω −→ Ω is an elliptic first-order operator and (d + d ∗ )2 = ∆ Laplacian. 20 / 30 Signature operator of M Let Ω = ⊕Ωk be sections of exterior algebra of differential forms, Ωk = C ∞ (Λk T ∗ M). d : Ωi −→ Ωi+1 exterior differentiation with formal adjoint d ∗ : Ωi+1 −→ Ωi with R respect to pairing (α, β) = M α ∧ ∗β, Hodge star ∗ : Ωi −→ Ωn−i d + d ∗ : Ω −→ Ω is an elliptic first-order operator and (d + d ∗ )2 = ∆ Laplacian. Denote by Ω± the ±1-eigenspaces for ∗, then d + d ∗ interchanges Ω+ and Ω− . 20 / 30 Signature operator of M Let Ω = ⊕Ωk be sections of exterior algebra of differential forms, Ωk = C ∞ (Λk T ∗ M). d : Ωi −→ Ωi+1 exterior differentiation with formal adjoint d ∗ : Ωi+1 −→ Ωi with R respect to pairing (α, β) = M α ∧ ∗β, Hodge star ∗ : Ωi −→ Ωn−i d + d ∗ : Ω −→ Ω is an elliptic first-order operator and (d + d ∗ )2 = ∆ Laplacian. Denote by Ω± the ±1-eigenspaces for ∗, then d + d ∗ interchanges Ω+ and Ω− . Define the signature operator ∂ as the restriction of d + d ∗ as an operator from Ω+ to Ω− , i.e. ∂ : Ω+ −→ Ω− 20 / 30 Signature operator of M Let Ω = ⊕Ωk be sections of exterior algebra of differential forms, Ωk = C ∞ (Λk T ∗ M). d : Ωi −→ Ωi+1 exterior differentiation with formal adjoint d ∗ : Ωi+1 −→ Ωi with R respect to pairing (α, β) = M α ∧ ∗β, Hodge star ∗ : Ωi −→ Ωn−i d + d ∗ : Ω −→ Ω is an elliptic first-order operator and (d + d ∗ )2 = ∆ Laplacian. Denote by Ω± the ±1-eigenspaces for ∗, then d + d ∗ interchanges Ω+ and Ω− . Define the signature operator ∂ as the restriction of d + d ∗ as an operator from Ω+ to Ω− , i.e. ∂ : Ω+ −→ Ω− Recall index(∂) = dim(ker ∂) − dim(ker ∂ ∗ ). 20 / 30 Signature operator of M Let Ω = ⊕Ωk be sections of exterior algebra of differential forms, Ωk = C ∞ (Λk T ∗ M). d : Ωi −→ Ωi+1 exterior differentiation with formal adjoint d ∗ : Ωi+1 −→ Ωi with R respect to pairing (α, β) = M α ∧ ∗β, Hodge star ∗ : Ωi −→ Ωn−i d + d ∗ : Ω −→ Ω is an elliptic first-order operator and (d + d ∗ )2 = ∆ Laplacian. Denote by Ω± the ±1-eigenspaces for ∗, then d + d ∗ interchanges Ω+ and Ω− . Define the signature operator ∂ as the restriction of d + d ∗ as an operator from Ω+ to Ω− , i.e. ∂ : Ω+ −→ Ω− Recall index(∂) = dim(ker ∂) − dim(ker ∂ ∗ ). ker ∂ = ker ∂ ∗ ∂ = (ker ∆) ∩ Ω+ . Put H+ := ker ∆ ∩ Ω+ . Likewise ker ∂ ∗ = H− := ker ∆ ∩ Ω− . Then re-phrase index(∂) = dim(H+ ) − dim(H− ). 20 / 30 Signature operator of M Let Ω = ⊕Ωk be sections of exterior algebra of differential forms, Ωk = C ∞ (Λk T ∗ M). d : Ωi −→ Ωi+1 exterior differentiation with formal adjoint d ∗ : Ωi+1 −→ Ωi with R respect to pairing (α, β) = M α ∧ ∗β, Hodge star ∗ : Ωi −→ Ωn−i d + d ∗ : Ω −→ Ω is an elliptic first-order operator and (d + d ∗ )2 = ∆ Laplacian. Denote by Ω± the ±1-eigenspaces for ∗, then d + d ∗ interchanges Ω+ and Ω− . Define the signature operator ∂ as the restriction of d + d ∗ as an operator from Ω+ to Ω− , i.e. ∂ : Ω+ −→ Ω− Recall index(∂) = dim(ker ∂) − dim(ker ∂ ∗ ). ker ∂ = ker ∂ ∗ ∂ = (ker ∆) ∩ Ω+ . Put H+ := ker ∆ ∩ Ω+ . Likewise ker ∂ ∗ = H− := ker ∆ ∩ Ω− . Then re-phrase index(∂) = dim(H+ ) − dim(H− ). j H+ There are isomorphisms degree forms, degree 2k, 4k−j ∼ = H− , which give cancellations except for the middle 20 / 30 Signature operator of M Let Ω = ⊕Ωk be sections of exterior algebra of differential forms, Ωk = C ∞ (Λk T ∗ M). d : Ωi −→ Ωi+1 exterior differentiation with formal adjoint d ∗ : Ωi+1 −→ Ωi with R respect to pairing (α, β) = M α ∧ ∗β, Hodge star ∗ : Ωi −→ Ωn−i d + d ∗ : Ω −→ Ω is an elliptic first-order operator and (d + d ∗ )2 = ∆ Laplacian. Denote by Ω± the ±1-eigenspaces for ∗, then d + d ∗ interchanges Ω+ and Ω− . Define the signature operator ∂ as the restriction of d + d ∗ as an operator from Ω+ to Ω− , i.e. ∂ : Ω+ −→ Ω− Recall index(∂) = dim(ker ∂) − dim(ker ∂ ∗ ). ker ∂ = ker ∂ ∗ ∂ = (ker ∆) ∩ Ω+ . Put H+ := ker ∆ ∩ Ω+ . Likewise ker ∂ ∗ = H− := ker ∆ ∩ Ω− . Then re-phrase index(∂) = dim(H+ ) − dim(H− ). 4k−j ∼ There are isomorphisms = H− , which give cancellations except for the middle degree forms, degree 2k, thus j H+ 2k 2k index(∂) = dim(H+ ) − dim(H− ). 20 / 30 Signature operator of M Consider the quadratic form of the intersection form Q : α 7→ R M α ∧ ∗α. 21 / 30 Signature operator of M R Consider the quadratic form of the intersection form Q : α 7→ M α ∧ ∗α. This form is 2k and negative definite on H 2k , and so the signature of M can positive definite on H+ − be computed as 21 / 30 Signature operator of M R Consider the quadratic form of the intersection form Q : α 7→ M α ∧ ∗α. This form is 2k and negative definite on H 2k , and so the signature of M can positive definite on H+ − be computed as σ(M) = signature of Q = #{+1 eigenvalues of Q} − #{−1 eigenvalues of Q} 2k 2k = dim(H+ ) − dim(H− ) = index(∂). 21 / 30 Signature operator of M Atiyah-Singer index formula 22 / 30 Signature operator of M Atiyah-Singer index formula gives Z LM. index(∂) = M 22 / 30 Signature operator of M Atiyah-Singer index formula gives Z LM. index(∂) = M Hence Z LM = hLM, [M]i. σ(M) = index(∂) = M 22 / 30 Homological formulation of the Novikov conjecture Let Γ = π1 (M) be a discrete group, and BΓ its classifying space. 23 / 30 Homological formulation of the Novikov conjecture Let Γ = π1 (M) be a discrete group, and BΓ its classifying space. e −→ M. Let τ : M −→ BΓ denote the classifying map for the universal cover M 23 / 30 Homological formulation of the Novikov conjecture Let Γ = π1 (M) be a discrete group, and BΓ its classifying space. e −→ M. Let τ : M −→ BΓ denote the classifying map for the universal cover M ∗ Let x ∈ H (BΓ; Q). 23 / 30 Homological formulation of the Novikov conjecture Let Γ = π1 (M) be a discrete group, and BΓ its classifying space. e −→ M. Let τ : M −→ BΓ denote the classifying map for the universal cover M ∗ Let x ∈ H (BΓ; Q). Novikov conjecture: σx (M) = hLM ∪ τ ∗ x, [M]i ∈ Q is an oriented homotopy invariant. 23 / 30 Homological formulation of the Novikov conjecture Let Γ = π1 (M) be a discrete group, and BΓ its classifying space. e −→ M. Let τ : M −→ BΓ denote the classifying map for the universal cover M ∗ Let x ∈ H (BΓ; Q). Novikov conjecture: σx (M) = hLM ∪ τ ∗ x, [M]i ∈ Q is an oriented homotopy invariant. Then observe: hLM ∪ τ ∗ x, [M]i = hτ ∗ x, LM ∩ [M]i = hx, τ∗ (LM ∩ [M])i. 23 / 30 Homological formulation of the Novikov conjecture Let Γ = π1 (M) be a discrete group, and BΓ its classifying space. e −→ M. Let τ : M −→ BΓ denote the classifying map for the universal cover M ∗ Let x ∈ H (BΓ; Q). Novikov conjecture: σx (M) = hLM ∪ τ ∗ x, [M]i ∈ Q is an oriented homotopy invariant. Then observe: hLM ∪ τ ∗ x, [M]i = hτ ∗ x, LM ∩ [M]i = hx, τ∗ (LM ∩ [M])i. So an equivalent formulation is: τ∗ (LM ∩ [M]) ∈ H∗ (BΓ; Q) is an oriented homotopy invariant. 23 / 30 e e in K Γ (M) The element [∂] in K∗ (M) and [∂] ∗ Consider signature operator ∂ of M. 24 / 30 e e in K Γ (M) The element [∂] in K∗ (M) and [∂] ∗ Consider signature operator ∂ of M. It gives canonically an element [∂] := (L2 (Λ∗ M), ∂(1 + ∂ 2 )−1/2 ) ∈ K0 (M). 24 / 30 e e in K Γ (M) The element [∂] in K∗ (M) and [∂] ∗ Consider signature operator ∂ of M. It gives canonically an element [∂] := (L2 (Λ∗ M), ∂(1 + ∂ 2 )−1/2 ) ∈ K0 (M). e −→ M. The signature operator ∂ lifts Γ = π1 (M) acts on the universal cover M e e This gives the Γ-equivariant (abstract) canonically to a Γ-equivariant operator ∂ on M. elliptic operator e ∈ K Γ (M). e [∂] 0 24 / 30 e e in K Γ (M) The element [∂] in K∗ (M) and [∂] ∗ Consider signature operator ∂ of M. It gives canonically an element [∂] := (L2 (Λ∗ M), ∂(1 + ∂ 2 )−1/2 ) ∈ K0 (M). e −→ M. The signature operator ∂ lifts Γ = π1 (M) acts on the universal cover M e e This gives the Γ-equivariant (abstract) canonically to a Γ-equivariant operator ∂ on M. elliptic operator e ∈ K Γ (M). e [∂] 0 The Γ-index e −→ Kj (C ∗ Γ) IndexΓ : KjΓ (M) r is constructed as before; a triple (H, π, F ) carrying a representation of Γ is completed to a triple built on a Hilbert C*-module for Cr∗ Γ. 24 / 30 e −→ M with classifying map τ : M −→ BΓ. Universal principal Universal cover M Γ-bundle E Γ −→ BΓ. e −−−−→ M y EΓ y M −−−−→ BΓ τ 25 / 30 e −→ M with classifying map τ : M −→ BΓ. Universal principal Universal cover M Γ-bundle E Γ −→ BΓ. e −−−−→ M y EΓ y M −−−−→ BΓ τ By universal property of E Γ, there is a map σ : E Γ −→ E Γ. e [∂] e IndexΓ (∂) µ σ∗ e −−−−→ K Γ (E Γ) −−− K∗Γ (M) −→ K∗Γ (E Γ) −−−−→ K∗ (Cr∗ Γ) ∗ ∼ y= K∗ (M) −−−−→ K∗ (BΓ) τ∗ [∂] 25 / 30 Dual Chern character homomorphism Chern character isomorphism ch∗ : K ∗ (X ) ⊗ Q −→ H ∗ (X ; Q). 26 / 30 Dual Chern character homomorphism Chern character isomorphism ch∗ : K ∗ (X ) ⊗ Q −→ H ∗ (X ; Q). Dually there is ch∗ : K∗ (X ) ⊗ Q −→ H∗ (X ; Q) hch∗ (x), ch∗ (y )i = hx, y i, x ∈ K ∗ (X ) ⊗ Q, y ∈ K∗ (X ) ⊗ Q. 26 / 30 Dual Chern character homomorphism Chern character isomorphism ch∗ : K ∗ (X ) ⊗ Q −→ H ∗ (X ; Q). Dually there is ch∗ : K∗ (X ) ⊗ Q −→ H∗ (X ; Q) hch∗ (x), ch∗ (y )i = hx, y i, x ∈ K ∗ (X ) ⊗ Q, y ∈ K∗ (X ) ⊗ Q. One can deduce from the signature/index theorem: ch∗ ([∂]) = LM ∩ [M]. 26 / 30 Homotopy invariance of analytic signature e −→ M denote the universal cover and Γ = π1 (M). Then the signature operator Let M e ∂ of M lifts to a Γ-equivariant elliptic operator ∂e on M. 27 / 30 Homotopy invariance of analytic signature e −→ M denote the universal cover and Γ = π1 (M). Then the signature operator Let M e ∂ of M lifts to a Γ-equivariant elliptic operator ∂e on M. Theorem (Kasparov. Mischchenko. Lusztig. Kaminker-Miller) Let M and N be closed connected oriented manifolds and f : M −→ N an orientation preserving homotopy equivalence. Let Γ = π1 (M) = π1 (N). Denote the signature operators by ∂M and ∂N . 27 / 30 Homotopy invariance of analytic signature e −→ M denote the universal cover and Γ = π1 (M). Then the signature operator Let M e ∂ of M lifts to a Γ-equivariant elliptic operator ∂e on M. Theorem (Kasparov. Mischchenko. Lusztig. Kaminker-Miller) Let M and N be closed connected oriented manifolds and f : M −→ N an orientation preserving homotopy equivalence. Let Γ = π1 (M) = π1 (N). Denote the signature operators by ∂M and ∂N . Then IndexΓ (∂eM ) = IndexΓ (∂eN ) ∈ Kj (Cr∗ Γ). 27 / 30 Homotopy invariance of analytic signature e −→ M denote the universal cover and Γ = π1 (M). Then the signature operator Let M e ∂ of M lifts to a Γ-equivariant elliptic operator ∂e on M. Theorem (Kasparov. Mischchenko. Lusztig. Kaminker-Miller) Let M and N be closed connected oriented manifolds and f : M −→ N an orientation preserving homotopy equivalence. Let Γ = π1 (M) = π1 (N). Denote the signature operators by ∂M and ∂N . Then IndexΓ (∂eM ) = IndexΓ (∂eN ) ∈ Kj (Cr∗ Γ). I Using general theory of Fredholm complexes; co-chain complexes of Hilbert C*-modules 27 / 30 Homotopy invariance of analytic signature e −→ M denote the universal cover and Γ = π1 (M). Then the signature operator Let M e ∂ of M lifts to a Γ-equivariant elliptic operator ∂e on M. Theorem (Kasparov. Mischchenko. Lusztig. Kaminker-Miller) Let M and N be closed connected oriented manifolds and f : M −→ N an orientation preserving homotopy equivalence. Let Γ = π1 (M) = π1 (N). Denote the signature operators by ∂M and ∂N . Then IndexΓ (∂eM ) = IndexΓ (∂eN ) ∈ Kj (Cr∗ Γ). I I Using general theory of Fredholm complexes; co-chain complexes of Hilbert C*-modules Associate a generalized signature operator to a complex; and its analytic index in K-theory 27 / 30 Homotopy invariance of analytic signature e −→ M denote the universal cover and Γ = π1 (M). Then the signature operator Let M e ∂ of M lifts to a Γ-equivariant elliptic operator ∂e on M. Theorem (Kasparov. Mischchenko. Lusztig. Kaminker-Miller) Let M and N be closed connected oriented manifolds and f : M −→ N an orientation preserving homotopy equivalence. Let Γ = π1 (M) = π1 (N). Denote the signature operators by ∂M and ∂N . Then IndexΓ (∂eM ) = IndexΓ (∂eN ) ∈ Kj (Cr∗ Γ). I I I Using general theory of Fredholm complexes; co-chain complexes of Hilbert C*-modules Associate a generalized signature operator to a complex; and its analytic index in K-theory Homotopy equivalent Fredholm complexes are shown to have same signature, i.e. same analytic index of their signature operators 27 / 30 Homotopy invariance of analytic signature e −→ M denote the universal cover and Γ = π1 (M). Then the signature operator Let M e ∂ of M lifts to a Γ-equivariant elliptic operator ∂e on M. Theorem (Kasparov. Mischchenko. Lusztig. Kaminker-Miller) Let M and N be closed connected oriented manifolds and f : M −→ N an orientation preserving homotopy equivalence. Let Γ = π1 (M) = π1 (N). Denote the signature operators by ∂M and ∂N . Then IndexΓ (∂eM ) = IndexΓ (∂eN ) ∈ Kj (Cr∗ Γ). I I I I Using general theory of Fredholm complexes; co-chain complexes of Hilbert C*-modules Associate a generalized signature operator to a complex; and its analytic index in K-theory Homotopy equivalent Fredholm complexes are shown to have same signature, i.e. same analytic index of their signature operators e −→ M and lifted signature operator ∂e is the signature operator Our scenario: M of a Fredholm complex Ω∗ (M, ψ Γ ) where ψ Γ is a bundle with fibers Cr∗ Γ. The 27 / 30 Strong Novikov conjecture By universal property of E Γ there is a Γ-map σ : E Γ −→ E Γ. 28 / 30 Strong Novikov conjecture By universal property of E Γ there is a Γ-map σ : E Γ −→ E Γ. Conjecture (Strong Novikov conjecture) The composition ∼ = σ µ Kj (BΓ) −−−−→ KjΓ (E Γ) −−−∗−→ KjΓ (E Γ) −−−−→ Kj (Cr∗ Γ) is rationally injective. 28 / 30 Strong Novikov conjecture By universal property of E Γ there is a Γ-map σ : E Γ −→ E Γ. Conjecture (Strong Novikov conjecture) The composition ∼ = σ µ Kj (BΓ) −−−−→ KjΓ (E Γ) −−−∗−→ KjΓ (E Γ) −−−−→ Kj (Cr∗ Γ) is rationally injective. Comments I Kj (BΓ) ∼ = KjΓ (E Γ) follows since Γ acts freely on E Γ (use induction hom, or descent hom) 28 / 30 Strong Novikov conjecture By universal property of E Γ there is a Γ-map σ : E Γ −→ E Γ. Conjecture (Strong Novikov conjecture) The composition ∼ = σ µ Kj (BΓ) −−−−→ KjΓ (E Γ) −−−∗−→ KjΓ (E Γ) −−−−→ Kj (Cr∗ Γ) is rationally injective. Comments I I Kj (BΓ) ∼ = KjΓ (E Γ) follows since Γ acts freely on E Γ (use induction hom, or descent hom) σ∗ is rational injective, follows from equivariant Chern character of Baum and Connes to group homology 28 / 30 Strong Novikov conjecture By universal property of E Γ there is a Γ-map σ : E Γ −→ E Γ. Conjecture (Strong Novikov conjecture) The composition ∼ = σ µ Kj (BΓ) −−−−→ KjΓ (E Γ) −−−∗−→ KjΓ (E Γ) −−−−→ Kj (Cr∗ Γ) is rationally injective. Comments I I Kj (BΓ) ∼ = KjΓ (E Γ) follows since Γ acts freely on E Γ (use induction hom, or descent hom) σ∗ is rational injective, follows from equivariant Chern character of Baum and Connes to group homology Thus: if µ is rationally injective (e.g. if Baum-Connes conj holds for Γ), then Strong Novikov conj holds for Γ 28 / 30 Baum-Connes ⇒ Novikov Let Γ be a discrete group satisfying the Baum-Connes conjecture, i.e. µ : KjΓ (E Γ) −→ Kj (Cr∗ Γ) is an isomorphism. 29 / 30 Baum-Connes ⇒ Novikov Let Γ be a discrete group satisfying the Baum-Connes conjecture, i.e. µ : KjΓ (E Γ) −→ Kj (Cr∗ Γ) is an isomorphism. Then Γ satisfies the Novikov conjecture, i.e. for any closed, connected, oriented, smooth manifold M with π1 (M) = Γ, then τ∗ (LM ∩ [M]) ∈ H∗ (BΓ; Q) is an oriented homotopy invariant. 29 / 30 Baum-Connes ⇒ Novikov Let Γ be a discrete group satisfying the Baum-Connes conjecture, i.e. µ : KjΓ (E Γ) −→ Kj (Cr∗ Γ) is an isomorphism. Then Γ satisfies the Novikov conjecture, i.e. for any closed, connected, oriented, smooth manifold M with π1 (M) = Γ, then τ∗ (LM ∩ [M]) ∈ H∗ (BΓ; Q) is an oriented homotopy invariant. Approach: Baum-Connes: ⇒ Strong Novikov: ⇒ Novikov: µ : KjΓ (E Γ) −→ Kj (Cr∗ Γ) is isomorphism µσ∗ : Kj (BΓ) −→ Kj (Cr∗ Γ) is rationally injective τ∗ (LM ∩ [M]) ∈ H∗ (BΓ; Q) is an oriented homotopy invariant 29 / 30 Sketch of Baum-Connes ⇒ Novikov LM ∩ [M] [∂] e IndexΓ (∂) ch H∗ (M; Q) ←−−∗−− K∗ (M) −−−−→ K∗ (Cr∗ Γ) τ τ y y∗ y∗ H∗ (BΓ; Q) ←−−−− K∗ (BΓ) −−−−→ K∗ (Cr∗ Γ) µσ∗ ch∗ τ∗ (LM ∩ [M]) I τ∗ [∂] e IndexΓ (∂) Start by considering [∂] ∈ K∗ (M) 30 / 30 Sketch of Baum-Connes ⇒ Novikov LM ∩ [M] [∂] e IndexΓ (∂) ch H∗ (M; Q) ←−−∗−− K∗ (M) −−−−→ K∗ (Cr∗ Γ) τ τ y y∗ y∗ H∗ (BΓ; Q) ←−−−− K∗ (BΓ) −−−−→ K∗ (Cr∗ Γ) µσ∗ ch∗ τ∗ (LM ∩ [M]) I I τ∗ [∂] e IndexΓ (∂) Start by considering [∂] ∈ K∗ (M) e ∈ K∗ (C ∗ Γ) in bottom right is oriented homotopy invariant by previous IndexΓ (∂) r Thm 30 / 30 Sketch of Baum-Connes ⇒ Novikov LM ∩ [M] [∂] e IndexΓ (∂) ch H∗ (M; Q) ←−−∗−− K∗ (M) −−−−→ K∗ (Cr∗ Γ) τ τ y y∗ y∗ H∗ (BΓ; Q) ←−−−− K∗ (BΓ) −−−−→ K∗ (Cr∗ Γ) µσ∗ ch∗ τ∗ (LM ∩ [M]) I I I τ∗ [∂] e IndexΓ (∂) Start by considering [∂] ∈ K∗ (M) e ∈ K∗ (C ∗ Γ) in bottom right is oriented homotopy invariant by previous IndexΓ (∂) r Thm As µσ∗ is rationally injective (SNC), the element τ∗ [∂] ∈ K∗ (BΓ) is an oriented homotopy invariant modulo torsion 30 / 30 Sketch of Baum-Connes ⇒ Novikov LM ∩ [M] [∂] e IndexΓ (∂) ch H∗ (M; Q) ←−−∗−− K∗ (M) −−−−→ K∗ (Cr∗ Γ) τ τ y y∗ y∗ H∗ (BΓ; Q) ←−−−− K∗ (BΓ) −−−−→ K∗ (Cr∗ Γ) µσ∗ ch∗ τ∗ (LM ∩ [M]) I I τ∗ [∂] e IndexΓ (∂) Start by considering [∂] ∈ K∗ (M) e ∈ K∗ (C ∗ Γ) in bottom right is oriented homotopy invariant by previous IndexΓ (∂) r Thm I As µσ∗ is rationally injective (SNC), the element τ∗ [∂] ∈ K∗ (BΓ) is an oriented homotopy invariant modulo torsion I Then ch∗ (τ∗ [∂]) = τ∗ (LM ∩ [M]) ∈ H∗ (BΓ; Q) in bottom left must be an oriented homotopy invariant 30 / 30