WHAT IS NONCOMMUTATIVE GEOMETRY Dr. FATIMA M. AZMI
Transcription
WHAT IS NONCOMMUTATIVE GEOMETRY Dr. FATIMA M. AZMI
WHAT IS NONCOMMUTATIVE GEOMETRY BY Dr. FATIMA M. AZMI In the early 80's, Alain Connes came up with the idea of generalizing the notion of space and this led to the birth of noncommutative geometry which came as set of tools and methods to deal with problems that was beyond the reach of classical methods. Example - M a compact Hausdor® space. - ¡ a discrete group acting on M . The quotient topology on the orbit space M=¡ may not seperate orbits in M=¡. To over come this problem, Alain Connes' key observation is that one can attach a non commutative algebra through non commutative quotient space that captures most of the information. His motivation came from one of the most important theory in functional analysis, which shows the duality between commutative algebra and geometry. In the late 40's two Russian mathematician Gelfand and Naimark proved this important theorem in functional analysis. 1 Gelfand - Naimark Theorem : 1- Let X be a compact Haudor® space, then C(X) is a commutative C ¤ algebra with norm jjf jj = supx2X jf (x)j < f 2 C(X) and f ¤(x) = f¹(x). 2- Given any commutative C ¤ algebra A one can construct a unique compact Hausdor® space X such that A can be identi¯ed with the algebra C(X). f compact Hausdor® space Xg () f comm. C ¤ algebra C(X)g Conclusion: Studying commuative C ¤ algebra amounts to studying compact Hausdor® spaces and vice vers. Connes' Proposal: Studying non commutative C* algebra amounts to studying non commutative spaces. The noncommutative approach for studying the orbit space M=¡ is as follows: 1. replace M by the C ¤-algebra C(M ) 2. replace the orbit space M=¡, not by a smaller algebra C(M=¡) but by a larger algebra, the crossed product algebra C(M ) £ ¡ = ff = X g2¡ fg g; wherefg 2 C(M )andg 2 ¡g It is a noncommutative algebra, and after completion in some fashion it provides a powerfull tool for the study of M=¡. 2 2-NONCOMMUTATIVE MANIFOLDS: - M a smooth Riemannian manifold - C 1(M ) a commutative * algebra which is not a C ¤ algebra and it lives on a manifold Other useful geometrical objects on M de¯ned in terms of the algebra C 1(M ) are for example: - Â(M ) = the collection of all smooth vector ¯elds on M . Note that Â(M ) is a C 1(M ) module. Moreover, any vector ¯eld X 2 Â(M ) gives rise to a di®erential operator acting on smooth functions X : C 1(M ) ! C 1(M ); f ! Xf and it satis¯es the Leibniz rule; X(f g) = X(f )g + f X(g); 8f; g 2 C 1(M ) Observation: f vector ¯elds over M g () fderivations of the algebra C 1(M ) - di®erential forms, general tensor ¯elds, curvature, connection and covariant derivative all these can be de¯ned purely algebraically from C 1(M). Thus they all can be studied in this algebraic context. The algebra C 1(M ) indeed characterizes the manifold M completely as shown by the following theorem. 3 Theorem : Two smooth manifolds M and N are di®eomorphic () the algebra of functions C 1(M ) and C 1(N) are isomorphic. Conclusion: All the di®erential geometric properties of the manifold are encoded in the algebra C 1(M ). Let A be a noncommutative algebra which is closely related to A == C 1(M ). Then one tries to mimick the various de¯nition, based on A. This is a fruitful idea, as many of the ordinary de¯nition still make sense in ths noncommutative algebraic context. The de¯nition of vector ¯elds as derivations of A works quite well, as do the de¯nition of di®erentail forms with exterior derivative, also tensor ¯elds, and so on. One gets this way noncommutative vector ¯elds, di®erential forms, tensor ¯elds, etc. 4 3- NONCOMMUATIVE VECTOR BUNDLES AND K THEORY - X is a compact Hausdor® space. - E is a complex vector bundle over X. - ¼ : E ! X is a bundle map - P = ¡(E) = fs : X ! E; ¼ s = idX g Then P is a ¯nitely generated projective C(X) module. Also conversly one can do the reverse. The Serret- Swan Theorem: fvector bundles on Xg Ã! f ¯nitely generated proj. C(X) moduleg Conclusion: Instead of studying vector bundles over compact Hausdor® space, one can study ¯nitely generated projective modules over commutative C ¤ algebra. Moreover one usally thinks of ¯nite projective modules over noncommutative algebra as noncommutative vector bundles. Thus, the idea in noncommutative geometry is to treat certain classes of noncommutative algebra as noncommutative spaces and try to extend tools of geometry, topology and analysis to this setting. It should be emphasized, however that as a rule this extension is never straight forward and always involve new phenomena. 5 In the commutative world the power tools are homology and the fundamental group. They don't have a straightforward generalization to the noncommutative world. On the otherhand topological K-theory (which classi¯es vector bundles ) is the most powerfull tool that passes immediatly to the noncommutative world. Swan's theorem states that K0(C(M )) = K 0(M ); for M a compact Hausdor® space In the frame work of noncommutative geometry, one can regard the elements of the group K0(A), where A is a noncommutative algebra of functions on noncommutative space as vector bundles over the noncommutative space. 6 Chern -Connes character: The classical Chern character is a natural transformation from K- Theory to ordinary cohomology theory with rational coe±cients. Ch : K 0(X) ! ©i¸0H 2i(X; Q); X is compact Hausdor® space When X is smooth manifold there is an alternative construction of the map Ch, called the Chern - Weil construction, that uses the di®erential geometric notions of connection and curvature on vector bundles. - E is a complex vector bundle on X - r is a connection on E - R the curvature form. The Chern character of E is then de¯ned to be the class of non homogeneous even form Ch(E) = T r(eR ) With all the advances made on the K-theory, the lack of cohomological companion to K-theory and an e®ective computational devise remained a serious obstruction, and noncommutative topology was now in need of its own homology theory which would allow making concrete calculations. The breakthrough came in 1981 with the discovery by Alain Connes of cyclic cohomology and of spectral sequence relating it to Hochschild cohomology . 7 4- QUANTIZED CALCULUS: With the translation of the problem from geometry to algebra and analysis. The need of new calculus appeared. Connes developed a calculus which would replace the usual di®erential and integral calculus. The following dictionary shows the noncommutative analogues of some of the classical theories and concepts originally conceived for spaces: CLASSICAL NOTION OF CALCULUS Complex variable Real variable In¯nitesimal P @f Di®erential of f is df = @x dxi i Integral of in¯nitesimal of order 1 QUANTIZED CALCULUS Operator in H Self adjoint operator in H Compact operator in H df = [f; F ] Dixmier trace, T r(T ) 8 Fredholm Module Let A be a unital * algebra, an odd Fredholm module is a triple (H; ¼; F ) where H is a seperable Hilbert space on which A acts by a ¤ representation ¼ and F = F ¤ is a bounded operator on H such that F 2 ¡ I and [F; ¼(a)] are compact operators for all a 2 A. A Ferdholm module is called even if in addition the Hilbert space is Z2 graded. Key Operator : Dirac Operator - M a smooth compact oriented Riemannian manifold of dimension 2n - E a spinor bundle over M . The Dirac operator D is a ¯rst order di®erential operator D : L2(E) ¡! L2(E) Locally in terms of a local orthonomal frame ei of T M , D= 2n X i=1 ei ² rei ; where - r is the covariant deriv. on E determined by the connection. - \:" denotes Cli®ord module multiplication, where the ei 's satisfy the Cli®ord multiplication relation ei:ej = ¡ej :ei ; and e2i = ¡1 There is a natural inner product for the spinor bundle, which turns the L2 sections of E on M into a Hilbert space and the algebra C 1(M ) acts on the Hilbert space. Connes calls the system (H; C 1(M ); D) a spectral triple. 9 SOME APPLICATIONS Applications Include the Following Areas: 1. Index theory of elliptic operators. 2. Algebraic and di®erential topology. 3. Generalization of Riemannian manifold 4. Deformation quantization and quantum geometry 5. Number theory 6. Theory of Foliation 7. Quantum ¯eld theory, and Theoretical physics. 1 - INDEX THEORY Classi¯cation of manifolds is an important thing. Theory of homology and cohomology was developed to help classify spaces. For example, the torus T 1 has genus one (one hole) where as the sphere S 2 has genus zero (no hole). Thus T 1 is not di®eomorphic to S 2. Let M be a manifold with genus k. The Euler charactristic Â(M ) is given by Â(M ) = X (¡)pdimH p(M; C) = 2 ¡ 2k: Thus , Â(S 2) = 2 and Â(T 1) = 0. 10 As we go higher in dimension the spaces gets more complicated, and we need new tools to compute invariants. Invariants of manifolds can be studied through invariants of certain type of di®erential operators, if we do so, then we have moved the problem from topology to analysis and geometry. The concept of index associated to certain type of operators, arose in functional analysis. It came as a study of the relation between analytic and topological invariants of a certain class of linear maps between certain spaces. Index of operator: - M a smooth compact oriented manifold - E, F complex vector bundles over M - d an elliptic linear di®erential operator d : ¡1(M; E) ¡! ¡1(M; F ) Then, ker d and coker d are ¯nite dimensional. The Fredholm index of d is ind(d) = dim(kerd) ¡ dim(cokerd) 2 Z The index of an elliptic operator is a stable object , it remaines constant under continuous perturbation, i.e it depends only on the homotopy class of the operator. As Dirac operator is an elliptic operator, thus it has an index, which is homotopy invariant. 11 Example: @ is the Dirac operator on L2(S 1), Let M = S 1 and D = i @µ and ind(D) = 0. In 1963, Atiyah and Singer in their famous theorem expressed the index formula for elliptic di®erential operator over smooth compact manifold M in terms of topological data related to M and its curvature. Theorem - M be a smooth compact Riemannian spin manifold of even dimension - E a spinor bundle over M . - D is the Dirac operator D : ¡1(M; E) ¡! ¡1(M; E) Then the Atiyah-Singer Index theorem takes the form ind(D) = Z M ^ ): A(M A^ genus is a charactristic class which involves the curvature, it is a di®erential invariant Theorem: Let M be a compact spin manifold of dimension 4n. If M ^ ) = 0. admits a metric of positive scalar curvature, then A(M 12 Chern - Connes character Let (H; ¼; F ) be an even Fredholm module over an algebra A. Associated with this Fredholm module is a fundamental cyclic cocycle class Ch(F ), called the Chern character. It is given by Ch(F; H)(a0; a1; :::; an) = T r(°a0[F; a1] ¢ ¢ ¢ [F; an]): This general concept of the Chern character leads to an index theorem which is an extension of the classical Atiyah-Singer index theorem. Remark: Consider the following data: - M is a compact spin manifold with spinor bundle E - G is a discrete group which acts properly on the manifold , and A = C ¤(G; M ) is a smooth crossed product algebra completed in certain norm. - H = L2(E) - D is an odd Dirac operator on H The equivariant Chern- Connes character of the module (D; H) as an entire cyclic cocycle is given by; X ((Ch(D; H); ( fg0 g0; : : : = = X Z g0 g0 ;:::;g2k ¢2k X X X g2k fg2k g2k )) 2 2 T rs (fg0 g0e¡t1 D ¢ ¢ ¢ [D; fg2k g2k ]e¡(1¡t2k )D )dt1 ¢ ¢ ¢ dt2k g2G g=g0 ¢¢¢g2k ½X Z j (Fg )j [L2k (g)]j 13 ¾ CONNES' VERSION OF NONCOMMUTATIVE GEOMETRY : 1- Geometry for Connes is metric geometry, i.e the study of manifolds with a Riemannian structure given by a metric tensor. 2- He considers compact manifolds of arbitrary dimension with Riemannian structure, which gives rise to a ¯rst order di®erential operator, the Dirac operator. He shows that the manifold, including the metric tensor can be reconstructed from the discrete eigenvalues of this operator. He encodes the properties of the spectrum in a mathematical object called by him a spectral triple, which conatins a number of data and completely describes the Riemannian manifold. Thus formulating the ordianry Riemannian geoemtry as a commutative Riemannian geometry. COMMUTATIVE SPECTRAL TRIPLE : Consider the spectral triple (H; A; D) as de¯ned above, where M is an n dimensional compact spin manifold, and the Hilbert space H is formed from the spinor bundle E over M and A = C 1(M ). Connes exhibits a number of algebraic properties of this system, characterizing the various characteristics of the manifold and the metric and puts them in a list of seven properties. He then proves a theorem which might be called the \ GelfandNaimark theorem for compact Riemannian manifolds and spectral triple ". 14 THEOREM : 1- For every compact spin manifold (M; g) there is an associated spectral triple (H; A; D) as de¯ned above. 2- For every spectral triple (H; A; D) with H a Hilbert space, A is a commutative algebra of operators in H, and D is a linear operator satisfying the seven properties, then there exist a unique compact manifold (M; g) with metric g such that (H; A; D) is the spectral triple associated with (M; g). Moreover the manifold M and the metric tensor can be constructed in an explicit way from (H; A; D) The notion of the spectral triple can be generalized to a version in which one has a noncommutative algebra A . 15