Anomalies, Gauss laws, and Page charges in M-theory. Gregory Moore. Strings 2004, Paris. Related works: Witten , , ,

Transcription

1 Anomalies, Gauss laws, and Page charges in M-theory Gregory Moore Strings 2004, Paris Related works: Witten , , , Diaconescu, Moore, and Witten 00 Diaconescu, Freed, and Moore hep-th/ Freed & Moore, to appear Moore, to appear

2 Introduction In 1978 Cremmer, Julia, and Scherk found the action for 11- dimensional supergravity. The action is simple, but contains a very subtle Chern-Simons term. This talk is about that Chern-Simons term. Y : 11-dimensional, oriented, spin. In topologically trivial situations C Ω 3 (Y ) G := dc Ω 4 (Y ). The action for the theory is (schematically): [ exp Y vol(g)r(g) + G G + ψ /Dψ ( Φ(C) = exp 2πi Y ] ) 1 6 CG2 CI 8 (g) Φ(C)

3 Defining The Chern-Simons term But we also want [G] 0. Existence of M2 branes (Witten 96) [G] = ā 1 2 λ, ā H 4 (Y ; Z) If Y = the usual definition of a Chern-Simons term involves an extension to a bounding 12-manifold Z: ( ) 1 Φ(C) exp 2πi 6 G3 GI 8 (g) Z [I 8 (g)] = p 2 λ 2 a priori only defined up to a 96 th root of unity. Witten 96: a H 4 (Y, Z) Principal E 8 bundle P (a). 48 Identify [G] = [trf trr2 ] 1 6 G3 GI 8 = 1 2 i( /D A ) i( /D RS ) + d( ) Index theory Φ is well-defined up to a ±1. The sign cancels against Pf( /D RS ) = det /D RS

4 Boundaries The extension to the case with boundary is nontrivial. Y = X: We must distinguish Temporal boundaries Spatial boundaries In either case we need a model for the C-field.

5 Models for the C-field For the C-field what we know for sure is the isomorphism class of C. It is fixed by the Wilson loops Σ e 2πi isomorphism classes labeled by Cheeger-Simons characters, Ȟ 4 (Y, U(1)). But the proper way to express this in terms of redundant variables is not really known. Abstractly,... Σ C Gauge potentials = objects in a category gauge transformations = morphisms global gauge transformations = automorphisms Different models for the C-field correspond to equivalent categories.

6 E 8 Model for the C-field Definition: A C-field on Y with characteristic class a a pair (A, c) in is C(Y ) := Conn(P (a)) Ω 3 (Y ) Gauge invariant fieldstrength: G = trf trr2 + dc Morally speaking C = CS(A) 1 2 CS(g) + c Note that G depends on metric.

7 Chern-Simons term [ { 1 Φ(C) = exp 2πi 4 η( /D A ) + 1 } ] 8 η( /D RS ) + 2πiI local I local = Y (c( 12 G2 I 8 ) 12 cdcg + 16 c(dc)2 ) The same formula applies with or without boundary Φ is a section of a line bundle: L C(Y ) Met(Y ) When Y = X, L has a connection: A = 2π ( 1 2 G2 I 8 )δc with nonzero curvature: F = π X GδCδC

8 Anomalies The theory of the C-field by itself is quantum inconsistent. The theory of the gravitino by itself is quantum inconsistent. But the product is consistent. Pf( /D RS ), a section of No boundary L := PF( /D RS ) Met(Y ) L is a complex line bundle with real structure holonomies = ±1: The gravitino has a global anomaly. A natural isomorphism L = L global anomaly cancelation: Pf( /D RS ) Φ is a well-defined function on C(Y ) Met(Y )/G Temporal boundary : Φ is a section of a line bundle over M et(y ), NOT over Met(X). Obstruction to defining a well-defined Hilbert space of states. Nevertheless, with APS bc s on fermions ( ) Pfaff( /D RS ) Φ Γ PF( /D A ) C(X) Met(X)

9 Anomaly Cancellation for spatial boundaries With local (chiral) b.c. s on fermions one can still define elliptic operators and study geometric invariants (D. Freed s student M. Scholl is studying the general case, filling a gap in the math literature.). Rigorous proof of anomaly cancellation in the Horava- Witten model. The cancellation is completely local, e.g. multiple boundaries with ɛ i = ± on each boundary is allowed topologically: G Xi = ɛ i (trf 2 (A i ) 1 2 trr2 (g i )) Φ : Pfaff( /D Ai ) : Pfaff( /D RS ) : F = ɛ i i F = ɛ i F = i X i ( 1 2 i( /D A ) + 1 ) 4 i( /D RS ) 1 X i 2 i( /D Ai ) 1 ɛ i X i 4 i( /D RS )

10 The Gauge group G Our next goal is to write the Gauss law for C-field gauge invariance. In the E 8 model C = (A, c). Small gauge transformations: c c + dλ, Λ Ω 2 (Y ) All gauge transformations: c c + ω, ω Ω 3 Z (Y ) Does not properly account for global gauge transformations Λ constant. Actually, G is an extension: 0 H 2 (Y, U(1)) G Ω 3 Z (Y ) 0 H 2 (Y, U(1)) = global gauge transformations, acts as a group of automorphisms: If α H 2 (Y, U(1)) then g α : (A, c) (A, c) But it still has nontrivial physical effects.

11 Gauss law Physical wavefunctions should be gauge invariant: g Ψ(C) = Ψ(g C) g G To formulate the Gauss law we define a lift: L G L C(X) G C(X) g CX g Ψ(C X ) = ϕ(c X, g) exp( C X A) Ψ Definition of the lifting phase ϕ(c X, g): Given C X & g twisted C-field C X,g on Y = X S 1 : C X,g (x, 1) = g C X,g (x, 0) ϕ(c X, g) := Φ(C X,g ) ϕ(c X, g 1 )ϕ(g 1 C X, g 2 ) = e X iπ Gω 1ω 2 ϕ(c X, g 1 g 2 )

12 Writing out the Gauss law Recall G is an extension: 0 H 2 (Y, U(1)) G Ω 3 Z (Y ) 0 So the Gauss law consists of two statements: 1. Law for g = g α, α H 2 (X, U(1)) C-field electric charge tadpole condition. 2. Law for g Ω 3 Z (X) quantization of Page charge.

13 The Tadpole Condition A global gauge transformation g α, α H 2 (X, U(1)) acts nontrivially on quantum wavefunctions: If Ψ L A,c then g α Ψ = exp [ 2πi Q, α ] Ψ Q H 8 (X, Z) is the C-field electric charge. Gauss Law: Q 0 Ψ = 0. From the definition of the group lift we get a formula for Q. It is easy to show that Q = [ 1 2 G2 I 8 ] DR but there is indeed a torsion component to the usual tadpole condition.

14 Part 2 of the Gauss law When Q = 0 we can have nonzero gauge invariant wavefunctions Ψ(C) Γ(L) There is still further information in the statement of gauge invariance. Trivialize L. Choose a basepoint C = C + c, Ψ(C) ψ(c) Remainder of the Gauss law says: ψ(c + ω) = e ω (c)ψ(c) ω Ω 3 Z (X) e ω (c) := ϕ(c, ω) e 2πi X ( 1 2 G 1 6 dc )cω

15 Quantization of Page Charge The equation ψ(c + ω) = e ω (c)ψ(c) is equivalent to: exp ( 2πi ωp ) ψ = ψ ω Ω 3 Z (X) P = 1 2π Π + ( 1 2 G c cdc) + T where dt = 1 2 G 2 I 8 & Π = momentum, P is the Page charge, or electric flux, (expressed in a Hamiltonian formalism). Roughly speaking, P = dc 6. If [G] = 0, the quantum Gauss law for large C-field gauge transformations [P ] H 7 (X; Z) This is the expected electro-magnetic dual to the quantization of magnetic flux: [G] H 4 (X; Z)

16 Page charges don t commute However, when [G] 0, things are quite different. For φ HDR 3 (X) define P (φ) = X φ P Easy computation: [P (φ 1 ), P (φ 2 )] = i 2π φ 1 φ 2 G [P ] is not gauge invariant: U(ω)P (φ)u(ω) 1 = P (φ) ωφg U(ω) := exp[2πi ωp ]

17 The Page Charge Group When [G] 0 part of the lattice H 7 (X, Z) collapses to a finite group. The gauge invariant objects are for φ such that: 2πiP (φ) W (φ) := e φωg Z ω H 3 (X, Z) These W (φ) form the Page charge group. N.B. This group is nonabelian! W (φ 1 )W (φ 2 ) = e iπ φ 1 φ 2 G W (φ 1 + φ 2 ) = e 2πi φ 1 φ 2 G W (φ 2 )W (φ 1 )

18 Analogy to 3D Chern-Simons A closely related theory is 3d massive abelian CS: S = Σ IR 1 2e 2 F F + 2π kada Σ IR On Σ IR the dynamics of the topological (flat) modes of A is that of an electron on a torus H 1 (Σ; U(1)) in a constant magnetic field. Similarly - for harmonic modes of C: c = a c aω a, ( H eff = h ab i πb aa )( c a i πb bb ) c b c a c b h ab = ω a ω b B ab = X Gω a ω b k 1 2 [G] Landau levels Wavefunction ψ(c) Magnetic translation operator Page charge Verlinde, t Hooft, Wilson operator W (φ)

19 A derivation of the 5-brane partition function As an application, we derive Witten s prescription for the 5-brane partition function. X = D 6 S 4 at conformal infinity for an asymptotically AdS space ds 2 (k 2/3 l 2 ) [dr 2 + e 2r ds 2D ] ds2s4 G G = kω S 4 + Ḡ AdS/CFT (k 1) Z [ M theory/y ] = Z [ U(k) (2, 0) theory/d 6 ] We can say something about the singleton sector U(k) = SU(k) U(1) Z k U(1) C.O.M. for k 5-branes - couples to the harmonic modes c,h.

20 Page charge is dual to t Hooft sector To write the dependence on c,h we use the symplectic splitting H 3 (D 6, Z) = Λ 1 Λ 2 ω 1, ω 2 = ω 1 ω 2 D 6 Then Z [ U(k) (2, 0) theory ] = β Λ 1 /kλ 1 ζ β Ψ β (c,h ) ζ β is the contribution of the SU(k)/ Z k (0, 2) theory. β labels t Hooft sectors of the SU(k)/ Z k theory (Witten). A formula for Ψ β in terms of theta functions shows that: W (φ 1 )Ψ β = Ψ β+φ1 φ 1 Λ 1 /kλ 1 W (φ 2 )Ψ β = e 2πik φ 2,β Ψ β φ 2 Λ 2 /kλ 2 Page charge group = a discrete Heisenberg group on H 3 (D 6, Z/k Z). t Hooft sector = Page charge.

21 Formula for Ψ β Ψ β are derived from chiral splitting of ω H 3 Z (D 6) { ϕ(c, ω) exp πk 2 = } (c + ω) (c + ω) iπk D 6 c ω D 6 β Λ 1 /kλ 1 ζ β Ψ β (c) Explicitly: Ψ β (c) = e πk 2 D 6 c c+c (1,0) Imτc (1,0) Θ β,k/2 (c (1,0) + θ + τφ, τ) The characteristics follow from ϕ(c, ω) and depend on the the metric. ϕ(c, ω) exp[2πi ω CS(g)] D 6

22 Conclusions The E 8 formalism for the C-field can be useful, but it has shortcommings. Is there a better model for the C-field? How do we formulate M-theory on unoriented manifolds? The naive classification of electric+ magnetic fluxes: H 4 (X, Z) H 7 (X, Z) is replaced by a nonabelian group This has implications for the K-theoretic classification of type II string fluxes. Five-brane partition functions have metric-dependent characteristics. This can lead to suppression of instanton effects.