Spectral Triples
Giovanni Landi
Trieste
Noncommutative Topics
Noncommutative geometry
has roots in Heisenberg’ (and Born and Jordan and ...) matrix formulation of quantum mechanics
which resulted in the replacement of the phase space by a
‘noncommutative space’
or, more precisely
the replacement of classical observable ( usual functions )
Commutativity of algebra of functions on space X
is
localization of points of X
Quantum mechanics phase space noncommutativity:
[p, x] = 1 ⇒ ∆p∆x ≥ 1
To incorporate limitations on spatial resolution one makes posi-tions noncommuting as well
yx − xy = iθ
as a consequence
∆y ∆x ≥ θ
The idea that spacetime geometry may be noncommutative
(as a way to regularize infinities)
goes back to Schr¨odinger and Heisenberg
Heisenberg mentioned this possibility in a letter to Rudolph Peierls in the 30s.
Peierls mentions Heisenberg’s ideas to Wolfgang Pauli
Pauli explains it to Hartland Snyder
A basic duality
(commutative) Gel’fand-Naimark thm:
equivalence of categories C0(X) ↔ X,
commutative C∗-algebras ↔ locally compact Haus. spaces;
points x ∈ X are characters of C0(X)
x(f) := f(x)
or irreducible representations;
A (complex, unital) Banach algebra is a complex algebra equipped with a complete normed vector space structure ; the norm and the multiplicative structures are related by the identity
kabk ≤ kakkbk
An involutive Banach algebra satisfying the C∗-identity
ka∗ak = kak2
becomes a C∗-algebra
Pathological spaces: not a good point-set theoretical description
Equivalence relation R on X;
the quotient Y = X/R can be bad even for good X
Classically: function on the quotient
A(Y ) := {f ∈ A(X) ; f is R − invariant}
NCG approach:
the noncommutative algebra
A(Y ) := A(ΓR)
of functions on the graph ΓR ⊂ X ×X of the equivalence relation (of compact support, of rapid decay,...)
convolution product:
(f1 ∗ f2)(x, y) = X
x∼u∼y
f1(x, u)f2(u, y)
involution:
the quotient Y = X/R is a noncommutative space
with a noncommutative algebra of functions
A(Y ) := A(ΓR)
as good as X to do geometry:
exterior forms, metric, integration,
vector bundles, connections, curvature, ...
The celebrated irrational rotation algebra a.k.a. the noncommutative torus
Aθ = C∞(T2θ) ' C∞(S1/θZ) θ ∈ R − Q
a = X
(m,n)∈Z2
amn U1m U2n {amn} ∈ S(Z2)
U2 U1 = e2πiθ U1 U2
Yang-Mills theory on T2
A dictionary :
Classical Noncommutative
locally compact space C∗-algebra
compact space unital C∗-algebra
vector bundle finite projective module
smooth manifold C∗-algebra with ‘smooth’ subalgebra
partial derivative unbounded derivation
integral tracial state
spin structure spectral triple
Requirements on noncommutative geometry dictated by physics
The noncommutative geometry of the standard model
Present knowledge of spacetime is described by two theories :
• General Relativity
• The Standard Model of particle physics
Gravity minimally coupled to matter :
the gravitational potential gµν
ds2 = gµνdxµdxν SEH[gµν] = 1
G
Z
M
R√g d4x
the standard model SSM based on the gauge theory of the group
G = U(1) × SU(2) × SU(3)
The symmetries of S = SEH + SSM
G = Map(M, G) o Diff (M)
Is there a space so that G = Diff (X) ?
it can be a noncommutative space !!
Almost commutative geometry: X = M × F
M Riemannian spin 4-manifold
F finite (i.e. metric 0-dimensional) geometry of KO-dimension ( inner signature ) = 6
A(X) = C∞(M) ⊗ AF, AF = C ⊕ H ⊕ M3(C)
Full standard model minimally coupled to gravity obtained from
S(geometry) = tr(χ(DA/Λ)) + hJ ξ, DAξi
DA = D + A + ε0J AJ−1
J extra structure dictating the KO-dimension (the signature) ;
Λ a mass scale, χ is a cut-off function.
The simplest example
A = C∞(M; Mn(C)) = C∞(M) ⊗ Mn(C)
Algebra of n × n matrices of smooth functions on manifold M
The group Inn(A) of inner automorphisms locally isomorphic to the group G of smooth maps from M to the gauge group SU(n)
1 → Inn(A) → Aut(A) → Out(A) → 1
becomes identical to
(Connes-Chamseddine)
The pure gravity on this space yields Einstein gravity on M mini-mally coupled with Yang-Mills theory for the gauge group SU(n)
The gauge potential appears as the inner part of the metric,
What is a metric in spectral geometry
d(A; B) = Inf
Z
γ
q
gµν dxµdxν
Dirac’s square root of the Laplacian
(A,H, D) ds = D−1
Kinematic relations
• The algebra A is commutative.
• The commutator [[D, a], b] = 0 for a, b ∈ A
• It holds that P
α aα0 h
[D, aα1][D, aα2], . . . ,[D, aαn]i = 1
for some aαj ∈ A [T1, T2, . . . Tn] = P
σ (σ)Tσ(1)Tσ(2)· · ·Tσ(n)
Even case: 1 → γ5 (grading)
this means that the determinant of the metric gµν does not vanish
and its square root multiplied by the volume form
P
Spectral conditions
• The k-th characteristic value of the resolvent of D goes like
λk ∼ k−1/n
this condition gives the metric dimension n
• Regularity for the geodesic flow expit|D|
The reconstruction (Connes)
Let (A,H, D) be a spectral geometry fullfing the kinematic rela-tions and the spectral condirela-tions
Assume that the multiplicity of the action of A in H is 2n/2
Then there exists a smooth oriented compact (spinc) manifold
M such that A = C∞(M).
To get spin manifolds one needs additional data, a real structure i.e. an antilinear unitary operator J acting in H
It plays the same role and has the same algebraic properties as the charge conjugation operator in physics .
There are additional relations D, J and γ
J2 = ε(n)1, J D = ε0(n)DJ, J γ = ε00(n)γJ ε(n) = (1, 1, −1, −1, −1, −1, 1, 1)
ε0(n) = (1, −1, 1, 1, 1, −1, 1, 1)
ε00(n) = (1, , −1, , 1, , −1, )
n = 0,1, . . . ,7.
In the classical case of spin manifolds
there is thus a relation between the metric (or spectral) dimen-sion given by the rate of growth of the spectrum of D
and
the integer modulo 8 which appears in the above table
The dimension modulo 8 is called the KO-dimension because of its origin in K-theory ; it is a signature
With an ordinary spin geometry M of dimension n and taking the product space M × F, has metric dimension which is still n
but KO-dimension the sum of n with the KO-dimension of F
For the Standard Model with neutrino mixing favors
Noncommutative spin geometries
A spectral triple (A,H, D) is:
• a unital ∗-algebra A with a faithful ∗-rep. π : A → B(H),
H a (separable) Hilbert space,
• a self-adjoint operator D on H (“the Dirac operator”)
(i) (D + i)−1 is compact
(ii) [D, π(a)] is bounded for all a ∈ A
The spectral triple is graded (or even) if there exists a Z2-grading operator γ on H, γ = γ∗, γ2 = 1, s.t.
With 0 < µ < ∞, the spectral triple is µ+-summable (of metric dimension µ) if (D2 + 1)−1/2 is in the Dixmier ideal Lµ+(H);
roughly µk(D) ∼ k1/µ as k → ∞
A tracial state (an integral)
Z
− T = limω 1 logN
N−1 X
k=0
• M a compact oriented manifold (no boundary)
the spectral triple on M: (C∞(M),H, D)
H := L2(M,V∗
(M)) ; square integrable differential forms;
D = d + d∗ the signature operator M;
• M a compact Riemannian spin manifold (no boundary)
the spectral triple on M: (C∞(M),H, D/ , J)
H := L2(M,S); square integrable spinors;
D/ the Dirac operator of the metric of M;
(D2 + 1)−1/2 ∈ Lm+(H) is Weyl’s formula for the eigen.s of |D|:
µk(|D|) ∼ 2π m
Ωmvol(M)
!1/m
k1/m as k → ∞
The geodesic distance between any two points on M:
d(p, q) = sup
f∈C∞(M)
{|f(p) − f(q)| ; k[D/ , f]k ≤ 1}
The Riemannian measure on M: R
M dµg f ∼
R
(f|D/|−m)
The right hand sides in both furmulæ above make sense for any spectral triple ; in particular p, q are states of the algebra A
The (Eucledian) Einstein-Hilbert action obtained as:
Z
− |D/|2−m = −(m − 2)2
[m/2]Ω
m
12m(2π)m
Z
M R dµg
An evolution is the Spectral action :
S(geometry) = tr(χ(|D/|/Λ))
with Λ a mass scale, χ is a cut-off function.
The action counts the eigenvalues of D smaller than Λ.
An expansion in Λ :
tr(χ(D/ /Λ)) ∼ Λ0vol(M) + Λ2 Z
M
A real structure J of KO-dimension n ∈ Z/8Z; an antilinear isometry J on H ,
J2 = ε(n)1, J D = ε0(n)DJ, J γ = ε00(n)γJ ε(n) = (1, 1, −1, −1, −1, −1, 1, 1)
ε0(n) = (1, −1, 1, 1, 1, −1, 1, 1)
ε00(n) = (1, , −1, , 1, , −1, )
n = 0,1, . . . ,7 .
J maps to the commutant
[π(a), J π(b)J−1] = 0, a, b ∈ A,
the first order condition on D
Recal the noncommutative torus T2θ.
Let M be a compact Riemannian spin manifold (no boundary) whose isometry group has rank r ≥ 2. Then M admits natural isospectral deformations to noncommutative geometries Mθ.
θ = (θjk = −θkj) θjk ∈ R j, k = 1, . . . , r
idea: deform the standard spectral triple describing the Rie-mannian geometry of M along a torus embedded in the isometry group to get an isospectral triple
gauge theory on Mθ
E a noncommutative vector bundle over Mθ:
C∞(Mθ)N = E + E0
a connection on E: a linear map satisfying a Leibniz rule
∇ : E 7→ E ⊗C∞(M
θ) Ω
1(
Mθ), ∇(ηa) = (∇η)a + η ⊗A da
F = ∇2 is its curvature
on E a natural C∞(Mθ)-valued Hermitian metric a connection ∇ is metric compatible if
the Yang-Mills functional
YM(∇) = Z
− ∗θtr(F ∗θ F)
tr is the C∞(Mθ)-valued trace on EndC∞(M
θ)(E) from the
Hermi-tian metric on E; the Hodge operator ∗θ via the Dirac operator
the equations for critical points
[∇,∗θF] = 0
(antiself )self-dual configurations solutions from Bianchi identity
∗θF = ±F
a family of gauge-nonequivalent SU(2) instantons on a Sθ4
noncommutative parameters
The noncommutative geometry of quantum groups and quantum homogeneous spaces
The guiding principle:
equivariance with respect to a ‘quantum symmetry’ this will build all the geometry from scratch
Symmetries
implemented by the action of a Hopf ∗-algebra Uq(g), a quantum universal enveloping algebra;
main example :
Symmetries
U = (U,∆, S, ε) a Hopf ∗-algebra
A a left U-module ∗-algebra: there is a left action . of U on A,
h . xy = (h(1) . x)(h(2) . y),
h . 1 = ε(h)1, (h . x)∗ = S(h)∗ . x∗,
notation ∆(h) = h(1) ⊗ h(2).
A U-equivariant ∗-representation π of A on V (a dense subspace of H) is a ∗-representation of the left crossed product ∗-algebra
defined as the ∗-algebra generated by the two ∗-subalgebras A
and U with crossed commutation relations
hx = (h(1) . x)h(2), h ∈ U, x ∈ A
that is, there is also a ∗-representation λ of U on V s. t.
λ(h) π(x) ξ = π(h(1) . x) λ(h(2)) ξ,
for all h ∈ U, x ∈ A and ξ ∈ V.
A linear operator D on V is equivariant if it commutes with λ(h), for all h ∈ U and ξ ∈ V
The nc geometry of the quantum group A(SUq(2))
The algebra:
A = A(SUq(2)) is ∗-algebra generated by a and b, with relations:
ba = qab, b∗a = qab∗, bb∗ = b∗b,
a∗a + q2b∗b = 1, aa∗ + bb∗ = 1
these state that the defining matrix is unitary
U = a b
−qb∗ a∗
!
A = A(SUq(2)) is a Hopf ∗-algebra (a quantum group) with
• coproduct:
∆ : A → A ⊗ A; ∆ a b
−qb∗ a∗
!
:= a b
−qb∗ a∗
!
.
⊗ a b
−qb∗ a∗
!
∆(a) = a ⊗ a − qb ⊗ b∗, ....
• antipode:
S : A → A; S a b −qb∗ a∗
!
= a
∗ −qb
b∗ a
!
• counit:
ε : A → C; ε a b −qb∗ a∗
!
:= 1 0 0 1
These dualize classical operations
A0 = A(SU(2), polynomial functions on SU(2)
∆ : A0 → A0 ⊗ A0; (∆f)(x ⊗ y) = f(xy)
S : A0 → A0; (Sf)(x) = f(x−1)
An isospectral nc geometry on the ‘manifold underlying’ SUq(2)
The equivariance is with respect to an action of the quantum universal envelopping algebra U = Uq(su(2))
The spectral triple (A(SUq(2)),H, D) is regular and 3+-summable.
It has simple dimension spectrum given by Σ = {1,2,3}.
Its KO-dimension is 3.