Global anomalies on Lorentzian space-times Jochen Zahn Universit - - PowerPoint PPT Presentation

global anomalies on lorentzian space times
SMART_READER_LITE
LIVE PREVIEW

Global anomalies on Lorentzian space-times Jochen Zahn Universit - - PowerPoint PPT Presentation

Global anomalies on Lorentzian space-times Jochen Zahn Universit at Leipzig based on 1609.06562 (Ann. H. Poincar e) [joint work with A. Schenkel] LQP 41, G ottingen, February 2018 Global anomalies in the path integral Chiral SU p 2


slide-1
SLIDE 1

Global anomalies on Lorentzian space-times

Jochen Zahn

Universit¨ at Leipzig

based on 1609.06562 (Ann. H. Poincar´ e) [joint work with A. Schenkel] LQP 41, G¨

  • ttingen, February 2018
slide-2
SLIDE 2

Global anomalies in the path integral

§ Chiral SUp2q doublet: Not anomalous w.r.t. infinitesimal (local) gauge

trafos.

§ But: Anomalous w.r.t. large (global) gauge trafos [Witten 82]. § As π4pSUp2qq “ Z2, there are compactly supported gauge trafos g that

can not be deformed to the identity.

§ However, one may deform A to Ag via a path Aλ of connections that are

not gauge equivalent to A. Along such a path, the fermion path integral „ż dψd ¯ ψ expp ¯ ψi { DAλψq  1

2

“ “ det i { DAλ ‰ 1

2

changes sign as A is varied to Ag (mod 2 index theorem).

§ This implies that the full partition function

Z “ ż dA “ det i { DA ‰ 1

2 exp

ˆ ´ 1 2g 2

YM

ż tr F ^ ‹F ˙ vanishes, as the contributions from A and Ag always cancel.

§ The theory is thus inconsistent. § Non-perturbative effect, not visible in perturbation theory around single

background.

slide-3
SLIDE 3

Riemannian vs. Lorentzian

§ The computations of global anomalies involve fermions in background

fields in Riemannian signature.

§ No clear relation to Lorentzian signature. § What is an appropriate condition for global anomalies in Lorentzian

signature (based on free fermions in non-trivial backgrounds)?

§ How does a global anomaly render a theory inconsistent?

slide-4
SLIDE 4

The framework (I)

§ As in the path integral framework, we formulate a criterion for global

anomalies based on free chiral fermions in generic gauge backgrounds.

§ Gauge backgrounds described by principal bundle connection ¯

A.

§ Two backgrounds ¯

A, ¯ A1 differ by Lie-algebra valued one-form A “ ¯ A ´ ¯ A1.

§ Locally covariant field theory [Hollands, Wald 01; Brunetti, Fredenhagen, Verch 03]

adapted to the gauge theory setting [Z. 14]: Local covariance also w.r.t. principal bundle morphisms.

§ Fields provide a consistent assignment of observables to different

  • backgrounds. Example: The current

j ¯

ApAq “

δ

δ ¯ AS, A

  • “ ´

ż ¯ ψ { Aψ vol defined by point-splitting w.r.t. the Hadamard parametrix.

§ No local anomalies, i.e., the current is conserved

¯ δj ¯

ApΛq .

“ j ¯

Ap¯

dΛq “ 0. (CC) It is then unique up to charge renormalization [Z. 14] j ¯

A Ñ j ¯ A ` λ¯

δ ¯ F.

slide-5
SLIDE 5

The framework (II)

§ When two backgrounds ¯

A, ¯ A1 differ only in a compact region, there is a natural isomorphism of the corresponding algebras, the retarded variation τ r

¯ A, ¯ A1 : Ap ¯

A1q Ñ Ap ¯ Aq.

§ It acts trivially on observables localized in the past of suppp ¯

A1 ´ ¯ Aq.

§ Perturbative agreement (PA) [Hollands, Wald 05] is the requirement that it

should not matter whether one puts quadratic terms in the free or interaction part of the action: τ r

¯ A, ¯ A1pT ¯ A1peFqq “ R ¯ ApeF; eijp ¯ A1´ ¯ Aqq .

“ T ¯

Apeijp ¯ A1´ ¯ Aqq´1T ¯ ApeF beijp ¯ A1´ ¯ Aqq § The infinitesimal retarded variation around ¯

A in the direction of A is denoted by δr

¯ ApAq. § (PA) can be fulfilled provided that

E ¯

ApA1, A2q .

“ δr

¯ ApA1qjpA2q ´ δr ¯ ApA2qjpA1q ´ irjpA2q, jpA1qs “ 0.

In dimension d ď 4, (CC) implies E ¯

ApA1, A2q “ 0 [Z. 15].

slide-6
SLIDE 6

The phase of the S matrix

§ Our criterion for the occurrence of a global anomaly will be a non-trivial

phase of the S matrix for ¯ A Ñ ¯

  • Ag. Need to fix the phase of the S matrix.

§ Formally, the S matrix for ¯

A Ñ ¯ A1 “ ¯ A ` A is given by and fulfills S ¯

ApAq “ T ¯ ApeijpAqq

“ T ¯

ApeijpA1qqT ¯ ApeijpA1qq´1T ¯ ApeijpA´A1q b eijpA1qq

“ T ¯

ApeijpA1qqR ¯ ApeijpA´A1q; eijpA1qq

“ S ¯

ApA1qτ r ¯ A, ¯ A`A1pS ¯ A`A1pA ´ A1qq § With the further constraints

S ¯

Ap0q “ ✶,

BλS ¯

ApAλq|λ“0 “ ij ¯ Ap 9

A0q, we may integrate S matrix for any path r0, 1s Q λ ÞÑ Aλ from 0 to A: S ¯

ApAq “ ¯

P exp ˆ i ż 1 τ r

¯ A, ¯ A`Aλpj ¯ A`Aλp 9

Aλqqdλ ˙ (PO)

§ Path independence is equivalent to E “ 0. § Unique up to

S ¯

ApAq Ñ exp

ˆ iλ ż “ LYMp ¯ A ` Aq ´ LYMp ¯ Aq ‰˙ S ¯

ApAq.

slide-7
SLIDE 7

Hilbert space representation

§ A representation ¯

π : Ap ¯ Aq Ñ Endp ¯ Hq naturally induces representations πA . “ ¯ π ˝ τ r

¯ A, ¯ A`A : Ap ¯

A ` Aq Ñ Endp ¯ Hq.

§ In the representation, (PO) reads

UpA, A1q . “ πApS ¯

A`ApA1 ´ Aqq “ ¯

P exp ˆ i ż 1 πAλpj ¯

A`Aλp 9

Aλqqdλ ˙ , with Aλ a path from A to A1.

§ Q: Is πpjq self-adjoint? Is U well-defined and unitary? § Assuming it is,

UpA, A1qUpA1, A2q “ UpA, A2q, UpA, A1q´1 “ UpA1, Aq.

§ Furthermore, V pgq .

“ Upp ¯ A ` Aqg ´ ¯ A, Aq “ eiφg id is independent of A, and thus provides a representation of the gauge group Γ8

c pM, P ˆAd Gq. § If g is deformable to the identity, then, by (PO) and (CC), V pgq “ id. § If V pgq ‰ id for some g, then no gauge invariant vector, a global anomaly. § Same topological obstructions as in the path integral formalism and similar

computation via gauge non-equivalent connections.

slide-8
SLIDE 8

Global anomalies in a Hamiltonian framework

§ Following [Witten 82], assume that the Hilbert space is given by sections over

the space of 3d gauge fields in temporal gauge. The gauge group is then G “ C 8

c pR3, Gq with homotopy group

π1pGq “ π4pGq.

§ Physical states are annihilated by the generators QpΛq of G. § The non-trivial element of π1pGq must be represented by the identity,

  • therwise there are no physical states.

§ The matter contribution to the generators is QmatterpΛq “ j ¯ ApBq with

Ba

µpxq “ δ0 µΛap

xqδpx0q.

§ E “ 0 ensures

rQpΛq, QpΛ1qs “ iQprΛ, Λ1sq.

§ In the case of a global anomaly, there are no physical states, as integrating

up QpΛq along a non-trivial cycle does not yield the identity.

slide-9
SLIDE 9

Perturbative agreement and the Wess-Zumino consistency condition

§ Assume there is a local anomaly, i.e., (CC) does not hold. Can we still

  • btain E ¯

ApA, A1q “ 0 by giving up the requirement that j is a field? § We fix a flat reference connection ¯

A0 and specify any other background ¯ A “ ¯ A0 ` ¯ A by a vector potential ¯

  • A. Allow j ¯

A to depend on ¯

  • A. We have

E ¯

Apd ¯ AΛ, d ¯ AΛ1q “

δ

δ ¯ A ¯

δj ¯

ApΛ1q, d ¯ AΛ

  • ´

δ

δ ¯ A ¯

δj ¯

ApΛq, d ¯ AΛ1

´ ¯ δj ¯

AprΛ, Λ1sq !

“ 0. (WZ) This is the Wess-Zumino consistency condition.

§ For d “ 4 and flat space-time [Z. 14],

¯ δj ¯

ApΛq “

i 8π2 ż tr Λ ¯ F ^ ¯ F.

§ With [Bardeen & Zumino 84]

j ¯

ApAq ÞÑ j ¯ ApAq `

i 24π2 ż tr “ A ^ p ¯ A ^ ¯ F ` ¯ F ^ ¯ A ´ 1

2 ¯

A ^ ¯ A ^ ¯ Aq ‰

  • ne obtains E ¯

ApA1, A2q “ 0 and the consistent anomaly

¯ δj ¯

ApΛq “

i 24π2 ż tr “ Λpd ¯ A ^ d ¯ A ` 1

2dp ¯

A ^ ¯ A ^ ¯ Aqq ‰ .

§ For G “ Up1q and flat space-time, one can obtain (CC) and (WZ), but

then E ¯

ApA1, A2q ‰ 0. Hence, (PA) is stronger than (WZ).

slide-10
SLIDE 10

Computation of the SUp2q anomaly

§ Following [Witten 83; Elitzur & Nair 84], compute SUp2q anomaly by embedding

G “ SUp2q Ă SUp3q “ H with π4pHq “ 0. May connect the nontrivial g P π4pGq by a path in C 8

c pR4, Hq to the identity. With (PO), the global

anomaly of G is computed by integrating the consistent anomaly of H: S ¯

Ap ¯

Ag ´ ¯ Aq “ exp ˆ 1 48π2 ż 1 dλ ż tr ´ h´1 9 h ^ A ^ A ^ A ^ A ¯˙ “ exp ˆ 1 240π2 ż

r0,1sˆR4 h˚pµ5 Hq

˙ where hp0q “ id, hp1q “ g, A “ h´1dh, and ¯ A is flat.

§ h defines an element of π5pH{Gq and rhs ÞÑ 1 240π2

ş

S5 h˚pµ5 Hq is a group

homomorphism, which for the generator h1 of π5pHq is normalized to 1 240π2 ż

S5 h˚ 1 pµ5 Hq “ 2πi. § We have the exact sequence

π5pHq “ Z Ñ π5pH{Gq “ Z Ñ π4pGq “ Z2 Ñ π4pHq “ 0. Hence

1 240π2

ş

S5 h˚pµ5 Hq is odd multiple of iπ, so that S ¯ Ap ¯

Ag ´ ¯ Aq “ ´id.

slide-11
SLIDE 11

Summary & Outlook

Summary:

§ Interpreted global anomalies in a Lorentzian setting. § Phase of the S matrix. § Pivotal role of perturbative agreement (E “ 0). § Relation of perturbative agreement and WZ consistency.

Open issues:

§ Unitarity of implementers in representation. § Effect of non-trivial topologies.