20.07.2013 Views

Noncommutative Geometry of Instanton Moduli Spaces

Noncommutative Geometry of Instanton Moduli Spaces

Noncommutative Geometry of Instanton Moduli Spaces

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

<strong>Noncommutative</strong> <strong>Geometry</strong> <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong><br />

Simon Brain<br />

(with Giovanni Landi and Walter D. van Suijlekom)<br />

IMPAN, 10th January 2011<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 1 / 28


Motivation<br />

What is the difference between a classical four-manifold and a<br />

noncommutative four-manifold?<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 2 / 28


Motivation<br />

What is the difference between a classical four-manifold and a<br />

noncommutative four-manifold?<br />

To find out, we’d like to ‘probe’ their geometric structures somehow.<br />

Donaldson did this for a classical four-manifold M by looking at the moduli<br />

space <strong>of</strong> self-dual gauge fields (instantons) on M. Can we do the same for a<br />

noncommutative manifold?<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 2 / 28


Motivation<br />

What is the difference between a classical four-manifold and a<br />

noncommutative four-manifold?<br />

To find out, we’d like to ‘probe’ their geometric structures somehow.<br />

Donaldson did this for a classical four-manifold M by looking at the moduli<br />

space <strong>of</strong> self-dual gauge fields (instantons) on M. Can we do the same for a<br />

noncommutative manifold?<br />

In this talk, we ask:<br />

◮ how do we construct instantons on a noncommutative manifold?<br />

◮ can we use them to ‘detect’ something about the noncommutative differential<br />

structure?<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 2 / 28


<strong>Instanton</strong>s in Classical <strong>Geometry</strong><br />

Let (M, g) be a compact Riemannian four-manifold.<br />

The Hodge operator ∗ : Ω 2 (M) → Ω 2 (M) obeys ∗ 2 = id and there is an<br />

eigenspace decomposition<br />

Ω 2 (M) = Ω 2 +(M) ⊕ Ω 2 −(M)<br />

into self-dual and anti-self-dual two-forms.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 3 / 28


<strong>Instanton</strong>s in Classical <strong>Geometry</strong><br />

Let (M, g) be a compact Riemannian four-manifold.<br />

The Hodge operator ∗ : Ω 2 (M) → Ω 2 (M) obeys ∗ 2 = id and there is an<br />

eigenspace decomposition<br />

Ω 2 (M) = Ω 2 +(M) ⊕ Ω 2 −(M)<br />

into self-dual and anti-self-dual two-forms.<br />

Now fix a smooth (S)U(2) vector bundle E over M.<br />

We say that a connection ∇ : Γ(E) → Ω 1 (E) is an instanton if its curvature<br />

F = ∇ 2 is a self-dual two-form, i.e. it obeys ∗F = F .<br />

NB: On M = S 4 , U(2) and SU(2) instantons are ”equivalent” objects.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 3 / 28


The <strong>Moduli</strong> Space <strong>of</strong> <strong>Instanton</strong>s<br />

The gauge group <strong>of</strong> E is the group<br />

G ⊂ Γ(End(E))<br />

<strong>of</strong> (S)U(2) endomorphisms <strong>of</strong> E which cover the identity on M.<br />

The gauge group G acts on the set C <strong>of</strong> connections on E by<br />

(U, ∇) ↦→ U∇U ∗ , U ∈ G, ∇ ∈ C.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 4 / 28


The <strong>Moduli</strong> Space <strong>of</strong> <strong>Instanton</strong>s<br />

The gauge group <strong>of</strong> E is the group<br />

G ⊂ Γ(End(E))<br />

<strong>of</strong> (S)U(2) endomorphisms <strong>of</strong> E which cover the identity on M.<br />

The gauge group G acts on the set C <strong>of</strong> connections on E by<br />

(U, ∇) ↦→ U∇U ∗ , U ∈ G, ∇ ∈ C.<br />

The set C/G <strong>of</strong> equivalence classes has the structure <strong>of</strong> an<br />

(infinite-dimensional) Banach manifold.<br />

If non-empty, the submanifold<br />

M := {[∇] ∈ C/G | ∇ is an instanton}<br />

is finite-dimensional (AHS). This is the moduli space <strong>of</strong> instantons on E<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 4 / 28


<strong>Instanton</strong>s on the Euclidean Four-Sphere<br />

For simplicity, we focus on studying instantons on the Euclidean four-sphere S 4 .<br />

Note that (S)U(2) bundles over S 4 (with ch1(E) = 0) are indexed by their<br />

‘topological charge’<br />

k = ch2(E) ∈ H 4 (S 4 , Z).<br />

Let us write Mk for the moduli space <strong>of</strong> instantons on an (S)U(2) vector<br />

bundle E over S 4 with fixed topological charge k ∈ Z.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 5 / 28


<strong>Instanton</strong>s on the Euclidean Four-Sphere<br />

For simplicity, we focus on studying instantons on the Euclidean four-sphere S 4 .<br />

Note that (S)U(2) bundles over S 4 (with ch1(E) = 0) are indexed by their<br />

‘topological charge’<br />

k = ch2(E) ∈ H 4 (S 4 , Z).<br />

Let us write Mk for the moduli space <strong>of</strong> instantons on an (S)U(2) vector<br />

bundle E over S 4 with fixed topological charge k ∈ Z.<br />

We’re interested in moduli spaces <strong>of</strong> instantons on noncommutative<br />

four-spheres....<br />

....so let’s begin by studying the construction <strong>of</strong> instantons on classical S 4<br />

from the point <strong>of</strong> view <strong>of</strong> noncommutative geometry (i.e. in terms <strong>of</strong><br />

function algebras, projective modules,...).<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 5 / 28


The SU(2) Hopf Fibration<br />

Define A(C 4 ) := A[zi, z ∗ j<br />

Write<br />

| i, j = 1, . . . , 4], then take unit sphere<br />

A(S 7 ) := A[zi, z ∗ j | <br />

Ψ =<br />

z1 z2 z3 z4<br />

−z ∗ 2 z ∗ 1 −z ∗ 4 z ∗ 3<br />

µ z∗ µzµ = 1].<br />

tr<br />

;<br />

then Ψ ∗ Ψ = 2. Define a right action <strong>of</strong> w ∈ SU(2) by Ψ ↦→ Ψw.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 6 / 28


The SU(2) Hopf Fibration<br />

Define A(C 4 ) := A[zi, z ∗ j<br />

Write<br />

| i, j = 1, . . . , 4], then take unit sphere<br />

A(S 7 ) := A[zi, z ∗ j | <br />

Ψ =<br />

z1 z2 z3 z4<br />

−z ∗ 2 z ∗ 1 −z ∗ 4 z ∗ 3<br />

µ z∗ µzµ = 1].<br />

tr<br />

;<br />

then Ψ ∗ Ψ = 2. Define a right action <strong>of</strong> w ∈ SU(2) by Ψ ↦→ Ψw.<br />

The invariant subalgebra is A(S 4 ) := A[α, α ∗ , β, β ∗ , x | αα ∗ + ββ ∗ + x 2 = 1]<br />

generated by the entries <strong>of</strong> the projection<br />

ΨΨ ∗ = 1<br />

⎛<br />

1 + x<br />

⎜<br />

2 ⎝<br />

0 α −β∗ 0 1 + x β α∗ α∗ β∗ 1 − x<br />

⎞<br />

⎟<br />

0 ⎠<br />

−β α 0 1 − x<br />

Then A(S 4 ) ↩→ A(S 7 ) is a principal SU(2)-bundle.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 6 / 28


Quaternionic Structure<br />

Note that the A(C 4 )-valued matrix<br />

Ψ =<br />

tr z1 z2 z3 z4<br />

−z ∗ 2 z ∗ 1 −z ∗ 4 z ∗ 3<br />

is made from a pair <strong>of</strong> quaternion-valued functions.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 7 / 28


Quaternionic Structure<br />

Note that the A(C 4 )-valued matrix<br />

Ψ =<br />

tr z1 z2 z3 z4<br />

−z ∗ 2 z ∗ 1 −z ∗ 4 z ∗ 3<br />

is made from a pair <strong>of</strong> quaternion-valued functions.<br />

This quaternion structure is encoded on A(C4 ) by the ∗-algebra map<br />

J : A(C 4 ) → A(C 4 ), J z1 z2 z3<br />

<br />

∗<br />

z4 = −z2 z∗ 1 −z∗ 4 z∗ <br />

3 .<br />

This gives an identification <strong>of</strong> C 4 with H 2 .<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 7 / 28


Quaternionic Structure<br />

Note that the A(C 4 )-valued matrix<br />

Ψ =<br />

tr z1 z2 z3 z4<br />

−z ∗ 2 z ∗ 1 −z ∗ 4 z ∗ 3<br />

is made from a pair <strong>of</strong> quaternion-valued functions.<br />

This quaternion structure is encoded on A(C4 ) by the ∗-algebra map<br />

J : A(C 4 ) → A(C 4 ), J z1 z2 z3<br />

<br />

∗<br />

z4 = −z2 z∗ 1 −z∗ 4 z∗ <br />

3 .<br />

This gives an identification <strong>of</strong> C 4 with H 2 .<br />

The J-invariant subalgebra <strong>of</strong> A(S 7 ) is precisely A(S 4 ), which is a<br />

coordinate-algebraic interpretation <strong>of</strong> the statement that S 4 HP 1 .<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 7 / 28


Construction <strong>of</strong> <strong>Instanton</strong>s<br />

<strong>Instanton</strong>s on S4 are constructed from monads on C4 , i.e. homomorphisms <strong>of</strong> free<br />

right A(C4 )-modules<br />

H ⊗ A(C 4 ) σz<br />

−→ K ⊗ A(C 4 )<br />

for complex vector spaces H, K <strong>of</strong> dimensions k, 2k + 2, such that:<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 8 / 28


Construction <strong>of</strong> <strong>Instanton</strong>s<br />

<strong>Instanton</strong>s on S4 are constructed from monads on C4 , i.e. homomorphisms <strong>of</strong> free<br />

right A(C4 )-modules<br />

H ⊗ A(C 4 ) σz<br />

−→ K ⊗ A(C 4 )<br />

for complex vector spaces H, K <strong>of</strong> dimensions k, 2k + 2, such that:<br />

σz is linear in the generators z1, . . . , z4 <strong>of</strong> A(C 4 );<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 8 / 28


Construction <strong>of</strong> <strong>Instanton</strong>s<br />

<strong>Instanton</strong>s on S4 are constructed from monads on C4 , i.e. homomorphisms <strong>of</strong> free<br />

right A(C4 )-modules<br />

H ⊗ A(C 4 ) σz<br />

−→ K ⊗ A(C 4 )<br />

for complex vector spaces H, K <strong>of</strong> dimensions k, 2k + 2, such that:<br />

σz is linear in the generators z1, . . . , z4 <strong>of</strong> A(C 4 );<br />

σz is injective and σ ∗ J(z)<br />

is surjective;<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 8 / 28


Construction <strong>of</strong> <strong>Instanton</strong>s<br />

<strong>Instanton</strong>s on S4 are constructed from monads on C4 , i.e. homomorphisms <strong>of</strong> free<br />

right A(C4 )-modules<br />

H ⊗ A(C 4 ) σz<br />

−→ K ⊗ A(C 4 )<br />

for complex vector spaces H, K <strong>of</strong> dimensions k, 2k + 2, such that:<br />

σz is linear in the generators z1, . . . , z4 <strong>of</strong> A(C 4 );<br />

σz is injective and σ∗ J(z)<br />

the composition σ∗ J(z) σz = 0.<br />

is surjective;<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 8 / 28


Construction <strong>of</strong> <strong>Instanton</strong>s<br />

<strong>Instanton</strong>s on S4 are constructed from monads on C4 , i.e. homomorphisms <strong>of</strong> free<br />

right A(C4 )-modules<br />

H ⊗ A(C 4 ) σz<br />

−→ K ⊗ A(C 4 )<br />

for complex vector spaces H, K <strong>of</strong> dimensions k, 2k + 2, such that:<br />

σz is linear in the generators z1, . . . , z4 <strong>of</strong> A(C 4 );<br />

σz is injective and σ∗ J(z)<br />

the composition σ∗ J(z) σz = 0.<br />

is surjective;<br />

Given such a monad, set V := <br />

σz σJ(z) ∈ Mat2k+2,2k(A(C4 )) and then<br />

P := 2k+2 − V (V ∗ V ) −1 V ∗ ∈ Mat2k+2(A(S 4 )).<br />

This P is clearly a projection, P 2 = P ∗ = P .<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 8 / 28


Construction <strong>of</strong> <strong>Instanton</strong>s<br />

<strong>Instanton</strong>s on S4 are constructed from monads on C4 , i.e. homomorphisms <strong>of</strong> free<br />

right A(C4 )-modules<br />

H ⊗ A(C 4 ) σz<br />

−→ K ⊗ A(C 4 )<br />

for complex vector spaces H, K <strong>of</strong> dimensions k, 2k + 2, such that:<br />

σz is linear in the generators z1, . . . , z4 <strong>of</strong> A(C 4 );<br />

σz is injective and σ∗ J(z)<br />

the composition σ∗ J(z) σz = 0.<br />

is surjective;<br />

Given such a monad, set V := <br />

σz σJ(z) ∈ Mat2k+2,2k(A(C4 )) and then<br />

P := 2k+2 − V (V ∗ V ) −1 V ∗ ∈ Mat2k+2(A(S 4 )).<br />

This P is clearly a projection, P 2 = P ∗ = P .<br />

Moreover, E := P A(S 4 ) 2k+2 is a rank two vector bundle over S 4 and<br />

∇ := P ◦ d is an instanton with topological charge k.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 8 / 28


Gauge Freedom<br />

We say that a pair <strong>of</strong> monads are equivalent if there is a commuting diagram<br />

H ⊗ A(C4 )<br />

⏐<br />

U⊗id<br />

H ⊗ A(C 4 )<br />

σz<br />

−−−−→ K ⊗ A(C4 )<br />

⏐<br />

W ⊗id<br />

˜σz<br />

−−−−→ K ⊗ A(C 4 )<br />

for invertible linear maps U : H → H and W : K → K. The effect on the resulting<br />

projection is to map P ↦→ W P W ∗ .<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 9 / 28


Gauge Freedom<br />

We say that a pair <strong>of</strong> monads are equivalent if there is a commuting diagram<br />

H ⊗ A(C4 )<br />

⏐<br />

U⊗id<br />

H ⊗ A(C 4 )<br />

σz<br />

−−−−→ K ⊗ A(C4 )<br />

⏐<br />

W ⊗id<br />

˜σz<br />

−−−−→ K ⊗ A(C 4 )<br />

for invertible linear maps U : H → H and W : K → K. The effect on the resulting<br />

projection is to map P ↦→ W P W ∗ .<br />

Theorem (ADHM) There is a bijective correspondence<br />

Mk ∼ = {Monads with index k}/ ∼ .<br />

So these equivalences <strong>of</strong> monads account for all <strong>of</strong> the gauge freedom in the<br />

construction <strong>of</strong> instantons.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 9 / 28


The Space <strong>of</strong> Monads<br />

Note that, since σz : H ⊗ A(C 4 ) → K ⊗ A(C 4 ) is linear in the generators<br />

z1, . . . , z4, it can be written<br />

σz =<br />

4<br />

j=1<br />

M j<br />

ab ⊗ zj, σ ∗ z =<br />

4<br />

j=1<br />

for constant matrices M j ∈ Mat2k+2,k(C), j = 1, . . . , 4.<br />

M j ∗ ∗<br />

ab ⊗ zj ,<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 10 / 28


The Space <strong>of</strong> Monads<br />

Note that, since σz : H ⊗ A(C 4 ) → K ⊗ A(C 4 ) is linear in the generators<br />

z1, . . . , z4, it can be written<br />

σz =<br />

4<br />

j=1<br />

M j<br />

ab ⊗ zj, σ ∗ z =<br />

4<br />

j=1<br />

for constant matrices M j ∈ Mat2k+2,k(C), j = 1, . . . , 4.<br />

M j ∗ ∗<br />

ab ⊗ zj ,<br />

As we allow σz to vary, we can think <strong>of</strong> the matrix elements M j j<br />

ab , M<br />

coordinate functions on the space <strong>of</strong> all monads with index k.<br />

ab ∗ as<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 10 / 28


The Space <strong>of</strong> Monads<br />

Note that, since σz : H ⊗ A(C 4 ) → K ⊗ A(C 4 ) is linear in the generators<br />

z1, . . . , z4, it can be written<br />

σz =<br />

4<br />

j=1<br />

M j<br />

ab ⊗ zj, σ ∗ z =<br />

4<br />

j=1<br />

for constant matrices M j ∈ Mat2k+2,k(C), j = 1, . . . , 4.<br />

M j ∗ ∗<br />

ab ⊗ zj ,<br />

As we allow σz to vary, we can think <strong>of</strong> the matrix elements M j j<br />

, M<br />

coordinate functions on the space <strong>of</strong> all monads with index k.<br />

The monad condition σ ∗ J(z) σz = 0 translates into the relations<br />

<br />

for all j, l = 1, . . . , 4 and c, d = 1, . . . , k, where<br />

b<br />

ab<br />

ab ∗ as<br />

<br />

N j<br />

abM l bd + N l abM j<br />

<br />

bd = 0, (1)<br />

σ ∗ J(z) = N j ⊗ zj = M j ∗ ⊗ J(zj) ∗ .<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 10 / 28


The Space <strong>of</strong> Monads<br />

Note that, since σz : H ⊗ A(C 4 ) → K ⊗ A(C 4 ) is linear in the generators<br />

z1, . . . , z4, it can be written<br />

σz =<br />

4<br />

j=1<br />

M j<br />

ab ⊗ zj, σ ∗ z =<br />

4<br />

j=1<br />

for constant matrices M j ∈ Mat2k+2,k(C), j = 1, . . . , 4.<br />

M j ∗ ∗<br />

ab ⊗ zj ,<br />

As we allow σz to vary, we can think <strong>of</strong> the matrix elements M j j<br />

, M<br />

coordinate functions on the space <strong>of</strong> all monads with index k.<br />

The monad condition σ ∗ J(z) σz = 0 translates into the relations<br />

<br />

for all j, l = 1, . . . , 4 and c, d = 1, . . . , k, where<br />

b<br />

ab<br />

ab ∗ as<br />

<br />

N j<br />

abM l bd + N l abM j<br />

<br />

bd = 0, (1)<br />

σ ∗ J(z) = N j ⊗ zj = M j ∗ ⊗ J(zj) ∗ .<br />

We write A(Mk) for the algebra generated by the functions M j j<br />

ab , M<br />

modulo the relations (1).<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 10 / 28<br />

ab ∗ ,


Functorial Cocycle Deformation<br />

Now we pass to noncommutative geometry. We view the deformation as a functor<br />

LF : H C → HF C<br />

from a certain category H C (wherein lives our classical geometry) to a new<br />

category HF C, both to be determined.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 11 / 28


Functorial Cocycle Deformation<br />

Now we pass to noncommutative geometry. We view the deformation as a functor<br />

LF : H C → HF C<br />

from a certain category H C (wherein lives our classical geometry) to a new<br />

category HF C, both to be determined.<br />

Let H be a commutative unital Hopf algebra. Let F : H ⊗ H → C be a<br />

convolution-invertible unital two-cocycle on H.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 11 / 28


Functorial Cocycle Deformation<br />

Now we pass to noncommutative geometry. We view the deformation as a functor<br />

LF : H C → HF C<br />

from a certain category H C (wherein lives our classical geometry) to a new<br />

category HF C, both to be determined.<br />

Let H be a commutative unital Hopf algebra. Let F : H ⊗ H → C be a<br />

convolution-invertible unital two-cocycle on H.<br />

Then there is a new Hopf algebra HF which is the same as H as a coalgebra,<br />

but has the new product<br />

h •F g := F (h (1), g (1))h (2)g (2)F −1 (h (3), g (3)).<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 11 / 28


Functorial Cocycle Deformation<br />

Now we pass to noncommutative geometry. We view the deformation as a functor<br />

LF : H C → HF C<br />

from a certain category H C (wherein lives our classical geometry) to a new<br />

category HF C, both to be determined.<br />

Let H be a commutative unital Hopf algebra. Let F : H ⊗ H → C be a<br />

convolution-invertible unital two-cocycle on H.<br />

Then there is a new Hopf algebra HF which is the same as H as a coalgebra,<br />

but has the new product<br />

h •F g := F (h (1), g (1))h (2)g (2)F −1 (h (3), g (3)).<br />

In fact, since H = HF as a coalgebra, H-comodules are exactly the same<br />

thing as HF -comodules.<br />

So there is an invertible functor LF : H C → HF C which simultaneously<br />

converts all H-covariant constructions into HF -covariant ones (it’s just the<br />

identity functor).<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 11 / 28


The Deformation Functor<br />

More interesting is the fact that the trivial braiding on H C,<br />

Ψ : A ⊗ B → B ⊗ A, Ψ(a ⊗ b) = b ⊗ a,<br />

is twisted into a new braiding on HF C,<br />

ΨF : AF ⊗ BF → BF ⊗ AF , ΨF (a ⊗ b) = F −2 (b (−1) , a (−1) )b (0) ⊗ a (0) .<br />

If A is a left H-comodule algebra,<br />

∆L : A → H ⊗ A, ∆(a) = a (−1) ⊗ a (0) ,<br />

then the algebra AF := LF (A) with product a ·F b := F (a (−1) , b (−1) )a (0) b (0)<br />

is a left HF -comodule algebra.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 12 / 28


The Deformation Functor<br />

More interesting is the fact that the trivial braiding on H C,<br />

Ψ : A ⊗ B → B ⊗ A, Ψ(a ⊗ b) = b ⊗ a,<br />

is twisted into a new braiding on HF C,<br />

ΨF : AF ⊗ BF → BF ⊗ AF , ΨF (a ⊗ b) = F −2 (b (−1) , a (−1) )b (0) ⊗ a (0) .<br />

If A is a left H-comodule algebra,<br />

∆L : A → H ⊗ A, ∆(a) = a (−1) ⊗ a (0) ,<br />

then the algebra AF := LF (A) with product a ·F b := F (a (−1) , b (−1) )a (0) b (0)<br />

is a left HF -comodule algebra.<br />

Theorem (Majid-Oeckl) The functor LF : H C → HF C is an isomorphism <strong>of</strong><br />

braided monoidal categories.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 12 / 28


Example: Twisting by a Torus Action<br />

Take H = A(T 2 ), generated by t1, t2 with tjt ∗ j = t∗ j tj = 1,<br />

∆(tj) = tj ⊗ tj, S(tj) = t ∗ j , ɛ(tj) = 1.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 13 / 28


Example: Twisting by a Torus Action<br />

Take H = A(T 2 ), generated by t1, t2 with tjt ∗ j = t∗ j tj = 1,<br />

Define a cocycle by<br />

∆(tj) = tj ⊗ tj, S(tj) = t ∗ j , ɛ(tj) = 1.<br />

F (ti, ti) = 1, F (t1, t2) := exp( 1<br />

2 iπθ),<br />

extended as a Hopf bicharacter. Then H = HF as a Hopf algebra, but the<br />

category <strong>of</strong> H-comodules is twisted.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 13 / 28


Example: Twisting by a Torus Action<br />

Take H = A(T 2 ), generated by t1, t2 with tjt ∗ j = t∗ j tj = 1,<br />

Define a cocycle by<br />

∆(tj) = tj ⊗ tj, S(tj) = t ∗ j , ɛ(tj) = 1.<br />

F (ti, ti) = 1, F (t1, t2) := exp( 1<br />

2 iπθ),<br />

extended as a Hopf bicharacter. Then H = HF as a Hopf algebra, but the<br />

category <strong>of</strong> H-comodules is twisted.<br />

Example: The coproduct ∆ : H → H ⊗ H makes H into an H-comodule algebra<br />

in the category H C. The comodule-twisted torus has algebra relations<br />

t1 ·F t2 = F (t1, t2)t1t2 = F 2 (t1, t2)t2 ·F t1 = µt2 ·F t1, µ = exp(iπθ)<br />

i.e. we get the noncommutative torus A(T 2 θ ) as an algebra in HF C.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 13 / 28


Example: The Connes-Landi Sphere<br />

Similarly, we have a coaction<br />

A(C 4 ) → H ⊗ A(C 4 ), zj ↦→ τj ⊗ zj, (τj) := (t1, t ∗ 1, t2, t ∗ 2).<br />

This gives twisted algebras A(C 4 θ ), A(S7 θ ), generated by zj, z ∗ l<br />

zjzl = ηljzlzj,<br />

zjz ∗ l = ηjlz ∗ l zj,<br />

⎛<br />

1<br />

⎜<br />

where (ηjl) = ⎜1<br />

⎝µ<br />

1<br />

1<br />

¯µ<br />

¯µ<br />

µ<br />

1<br />

⎞<br />

µ<br />

¯µ ⎟<br />

1⎠<br />

,<br />

subject to<br />

µ := exp (iπθ).<br />

¯µ µ 1 1<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 14 / 28


Example: The Connes-Landi Sphere<br />

Similarly, we have a coaction<br />

A(C 4 ) → H ⊗ A(C 4 ), zj ↦→ τj ⊗ zj, (τj) := (t1, t ∗ 1, t2, t ∗ 2).<br />

This gives twisted algebras A(C 4 θ ), A(S7 θ ), generated by zj, z ∗ l<br />

zjzl = ηljzlzj,<br />

zjz ∗ l = ηjlz ∗ l zj,<br />

⎛<br />

1<br />

⎜<br />

where (ηjl) = ⎜1<br />

⎝µ<br />

1<br />

1<br />

¯µ<br />

¯µ<br />

µ<br />

1<br />

⎞<br />

µ<br />

¯µ ⎟<br />

1⎠<br />

,<br />

subject to<br />

µ := exp (iπθ).<br />

¯µ µ 1 1<br />

The SU(2)-invariant subalgebra A(S4 θ ) generated by entries <strong>of</strong> the projection<br />

ΨΨ ∗ = 1<br />

2<br />

⎛<br />

1 + x 0 α −¯µ β<br />

⎜<br />

⎝<br />

∗<br />

⎞<br />

⎟<br />

⎠ .<br />

0 1 + x β µ α ∗<br />

α ∗ β ∗ 1 − x 0<br />

−µ β ¯µ α 0 1 − x<br />

The inclusion A(S4 θ ) ↩→ A(S7 θ ) is a noncommutative principal SU(2)-bundle.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 14 / 28


Differential and Hodge Structure on S 4 θ<br />

The coaction A(S 4 ) → H ⊗ A(S 4 ) is by isometries, i.e. it is an intertwiner<br />

for the exterior derivative d : A(S 4 ) → Ω 1 (S 4 ).<br />

This means that the differential calculus Ω • (S 4 ) is also an object in H C. So<br />

we can deform it to get a differential calculus Ω • (S 4 θ ) on S4 θ .<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 15 / 28


Differential and Hodge Structure on S 4 θ<br />

The coaction A(S 4 ) → H ⊗ A(S 4 ) is by isometries, i.e. it is an intertwiner<br />

for the exterior derivative d : A(S 4 ) → Ω 1 (S 4 ).<br />

This means that the differential calculus Ω • (S 4 ) is also an object in H C. So<br />

we can deform it to get a differential calculus Ω • (S 4 θ ) on S4 θ .<br />

Similarly, the Hodge operator ∗ : Ω 2 (S 4 ) → Ω 2 (S 4 ) is H-equivariant and so<br />

it is a morphism in H C. Similarly for the map J : A(C 4 ) → A(C 4 ). Their<br />

images under LF are a Hodge operator and a quaternion structure<br />

∗θ : Ω 2 (S 4 θ ) → Ω 2 (S 4 θ ), J : A(C 4 θ) → A(C 4 θ).<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 15 / 28


Differential and Hodge Structure on S 4 θ<br />

The coaction A(S 4 ) → H ⊗ A(S 4 ) is by isometries, i.e. it is an intertwiner<br />

for the exterior derivative d : A(S 4 ) → Ω 1 (S 4 ).<br />

This means that the differential calculus Ω • (S 4 ) is also an object in H C. So<br />

we can deform it to get a differential calculus Ω • (S 4 θ ) on S4 θ .<br />

Similarly, the Hodge operator ∗ : Ω 2 (S 4 ) → Ω 2 (S 4 ) is H-equivariant and so<br />

it is a morphism in H C. Similarly for the map J : A(C 4 ) → A(C 4 ). Their<br />

images under LF are a Hodge operator and a quaternion structure<br />

∗θ : Ω 2 (S 4 θ ) → Ω 2 (S 4 θ ), J : A(C 4 θ) → A(C 4 θ).<br />

This is all we need to discuss instantons on S 4 θ .<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 15 / 28


Differential and Hodge Structure on S 4 θ<br />

The coaction A(S 4 ) → H ⊗ A(S 4 ) is by isometries, i.e. it is an intertwiner<br />

for the exterior derivative d : A(S 4 ) → Ω 1 (S 4 ).<br />

This means that the differential calculus Ω • (S 4 ) is also an object in H C. So<br />

we can deform it to get a differential calculus Ω • (S 4 θ ) on S4 θ .<br />

Similarly, the Hodge operator ∗ : Ω 2 (S 4 ) → Ω 2 (S 4 ) is H-equivariant and so<br />

it is a morphism in H C. Similarly for the map J : A(C 4 ) → A(C 4 ). Their<br />

images under LF are a Hodge operator and a quaternion structure<br />

∗θ : Ω 2 (S 4 θ ) → Ω 2 (S 4 θ ), J : A(C 4 θ) → A(C 4 θ).<br />

This is all we need to discuss instantons on S 4 θ .<br />

Remark: In fact this works for the action <strong>of</strong> any locally compact Abelian group -<br />

so all <strong>of</strong> the following works in particular for the Moyal plane R4 as well.<br />

Can we use this to find a construction <strong>of</strong> instantons on S 4 θ ?<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 15 / 28


A <strong>Noncommutative</strong> Family <strong>of</strong> Monads over C 4 θ<br />

The map<br />

σz : H ⊗ A(C 4 ) → A(Mk) ⊗ K ⊗ A(C 4 ), σz = <br />

is a morphism in H M provided A(Mk) carries the H-coaction<br />

A(Mk) → H ⊗ A(Mk), M j<br />

ab ↦→ τ ∗ j ⊗ M j<br />

ab .<br />

j<br />

j<br />

Mab ⊗ zj<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 16 / 28


A <strong>Noncommutative</strong> Family <strong>of</strong> Monads over C 4 θ<br />

The map<br />

σz : H ⊗ A(C 4 ) → A(Mk) ⊗ K ⊗ A(C 4 ), σz = <br />

is a morphism in H M provided A(Mk) carries the H-coaction<br />

A(Mk) → H ⊗ A(Mk), M j<br />

ab ↦→ τ ∗ j ⊗ M j<br />

ab .<br />

j<br />

j<br />

Mab ⊗ zj<br />

Under the deformation functor, we get a new algebra A(Mk,θ) generated by<br />

j<br />

, M<br />

the matrix elements M j<br />

ab<br />

ab ∗ , but now subject to the twisted relations<br />

M j<br />

ab M l cd = ηljM l cdM j<br />

ab .<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 16 / 28


A <strong>Noncommutative</strong> Family <strong>of</strong> Monads over C 4 θ<br />

The map<br />

σz : H ⊗ A(C 4 ) → A(Mk) ⊗ K ⊗ A(C 4 ), σz = <br />

is a morphism in H M provided A(Mk) carries the H-coaction<br />

A(Mk) → H ⊗ A(Mk), M j<br />

ab ↦→ τ ∗ j ⊗ M j<br />

ab .<br />

j<br />

j<br />

Mab ⊗ zj<br />

Under the deformation functor, we get a new algebra A(Mk,θ) generated by<br />

j<br />

, M<br />

the matrix elements M j<br />

ab<br />

ab ∗ , but now subject to the twisted relations<br />

M j<br />

ab M l cd = ηljM l cdM j<br />

ab .<br />

We interpret the underlying ‘space’ Mk,θ as parameterising a<br />

noncommutative family <strong>of</strong> monads over C4 θ with index k.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 16 / 28


The <strong>Noncommutative</strong> ADHM Construction<br />

Although the space Mk,θ has fewer classical points than Mk, we can nevertheless<br />

work with the whole noncommutative family <strong>of</strong> monads at once.<br />

The algebra-valued matrix σz = <br />

j M j ⊗ zj is now thought <strong>of</strong> as a map<br />

σz : H ⊗ A(C 4 θ) → A(Mk,θ) ⊗ K ⊗ A(C 4 θ).<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 17 / 28


The <strong>Noncommutative</strong> ADHM Construction<br />

Although the space Mk,θ has fewer classical points than Mk, we can nevertheless<br />

work with the whole noncommutative family <strong>of</strong> monads at once.<br />

The algebra-valued matrix σz = <br />

j M j ⊗ zj is now thought <strong>of</strong> as a map<br />

σz : H ⊗ A(C 4 θ) → A(Mk,θ) ⊗ K ⊗ A(C 4 θ).<br />

For each character ɛ : A(Mk,θ) → C (i.e. for each classical point <strong>of</strong> Mk,θ)<br />

there is a corresponding monad<br />

(ɛ ⊗ id) ◦ σz : H ⊗ A(C 4 θ) → K ⊗ A(C 4 θ).<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 17 / 28


The <strong>Noncommutative</strong> ADHM Construction<br />

Although the space Mk,θ has fewer classical points than Mk, we can nevertheless<br />

work with the whole noncommutative family <strong>of</strong> monads at once.<br />

The algebra-valued matrix σz = <br />

j M j ⊗ zj is now thought <strong>of</strong> as a map<br />

σz : H ⊗ A(C 4 θ) → A(Mk,θ) ⊗ K ⊗ A(C 4 θ).<br />

For each character ɛ : A(Mk,θ) → C (i.e. for each classical point <strong>of</strong> Mk,θ)<br />

there is a corresponding monad<br />

(ɛ ⊗ id) ◦ σz : H ⊗ A(C 4 θ) → K ⊗ A(C 4 θ).<br />

The ADHM construction goes through just as before. This time take<br />

<br />

∈ Mat2k+2,2k(A(Mk,θ) ⊗ A(C 4 θ)).<br />

V := σz σ ∗ z<br />

Then set P := 2k+2 − V (V ∗ V ) −1 V ∗ ∈ Mat2k+2(A(Mk,θ) ⊗ A(S 4 θ )).<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 17 / 28


A <strong>Noncommutative</strong> Family <strong>of</strong> <strong>Instanton</strong>s<br />

From this family <strong>of</strong> projections P parameterised by the space Mk,θ we obtain a<br />

noncommutative family <strong>of</strong> instantons.<br />

Theorem (SB-GL) (Generalises k = 1 case <strong>of</strong> Landi-Pagani-Reina-van Suijlekom)<br />

The finitely generated projective right A(Mk,θ) ⊗ A(S 4 θ )-module<br />

E := P A(Mk,θ) ⊗ A(S 4 θ ) 2k+2<br />

is a noncommutative family <strong>of</strong> rank two vector bundles over S 4 θ ,<br />

parameterised by the noncommutative space Mk,θ.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 18 / 28


A <strong>Noncommutative</strong> Family <strong>of</strong> <strong>Instanton</strong>s<br />

From this family <strong>of</strong> projections P parameterised by the space Mk,θ we obtain a<br />

noncommutative family <strong>of</strong> instantons.<br />

Theorem (SB-GL) (Generalises k = 1 case <strong>of</strong> Landi-Pagani-Reina-van Suijlekom)<br />

The finitely generated projective right A(Mk,θ) ⊗ A(S 4 θ )-module<br />

E := P A(Mk,θ) ⊗ A(S 4 θ ) 2k+2<br />

is a noncommutative family <strong>of</strong> rank two vector bundles over S 4 θ ,<br />

parameterised by the noncommutative space Mk,θ.<br />

The operator ∇ := P ◦ (id ⊗ d) is a noncommutative family <strong>of</strong> instantons<br />

with topological charge k, parameterised by the noncommutative space Mk,θ.<br />

The latter statement means that the curvature F = ∇ 2 <strong>of</strong> the family obeys<br />

(id ⊗ ∗θ)F = F,<br />

where ∗θ : Ω 2 (S 4 θ ) → Ω2 (S 4 θ ) is the Hodge operator on S4 θ .<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 18 / 28


Gauge Freedom in the <strong>Noncommutative</strong> Case<br />

Even though the map<br />

σz : H ⊗ A(C 4 θ) → A(Mk,θ) ⊗ K ⊗ A(C 4 θ)<br />

is a noncommutative family <strong>of</strong> module homomorphisms, we still have the<br />

freedom to change bases in the vector spaces H and K. This is the same<br />

gauge freedom that we had in the classical case.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 19 / 28


Gauge Freedom in the <strong>Noncommutative</strong> Case<br />

Even though the map<br />

σz : H ⊗ A(C 4 θ) → A(Mk,θ) ⊗ K ⊗ A(C 4 θ)<br />

is a noncommutative family <strong>of</strong> module homomorphisms, we still have the<br />

freedom to change bases in the vector spaces H and K. This is the same<br />

gauge freedom that we had in the classical case.<br />

However, we now have more gauge freedom. The deformation functor<br />

canonically equips A(Mk,θ) with a Z 2 -action defined by<br />

(m1, m2) ⊲ a = F −2 (a (−1) , t m1<br />

1 tm2<br />

2 ) a(0) , (m1, m2) ∈ Z 2 , a ∈ A(Mk,θ).<br />

This action becomes trivial in the classical limit.<br />

This action is by gauge transformations (i.e. it generates projections which<br />

are unitarily equivalent to P ).<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 19 / 28


Gauge Freedom in the <strong>Noncommutative</strong> Case<br />

Even though the map<br />

σz : H ⊗ A(C 4 θ) → A(Mk,θ) ⊗ K ⊗ A(C 4 θ)<br />

is a noncommutative family <strong>of</strong> module homomorphisms, we still have the<br />

freedom to change bases in the vector spaces H and K. This is the same<br />

gauge freedom that we had in the classical case.<br />

However, we now have more gauge freedom. The deformation functor<br />

canonically equips A(Mk,θ) with a Z 2 -action defined by<br />

(m1, m2) ⊲ a = F −2 (a (−1) , t m1<br />

1 tm2<br />

2 ) a(0) , (m1, m2) ∈ Z 2 , a ∈ A(Mk,θ).<br />

This action becomes trivial in the classical limit.<br />

This action is by gauge transformations (i.e. it generates projections which<br />

are unitarily equivalent to P ).<br />

Thus the correct (?) parameter space for the construction is the quotient <strong>of</strong><br />

the space Mk,θ by Z 2 , described by the crossed product A(Mk,θ)>⊳ Z 2 .<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 19 / 28


Bosonisation<br />

There is another way to arrive at this crossed product algebra.<br />

The construction <strong>of</strong> instantons and, in particular, the parameter algebra<br />

A(Mk,θ) live in the braided tensor category HF C.<br />

To get back to the category <strong>of</strong> vector spaces, we apply Majid’s ‘bosonisation’<br />

construction, which just means taking the smash product with HF = A(T 2 ),<br />

yielding the algebra A(Mk,θ)>⊳ A(T 2 ).<br />

Now applying the Fourier transform on T 2 gives an isomorphism<br />

since Z 2 is the Pontryagin dual <strong>of</strong> T 2 .<br />

A(Mk,θ)>⊳A(T 2 ) → A(Mk,θ)>⊳ Z 2 ,<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 20 / 28


Understanding the <strong>Noncommutative</strong> Parameter <strong>Spaces</strong><br />

How can we make sense <strong>of</strong> these noncommutative parameter spaces?<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 21 / 28


Understanding the <strong>Noncommutative</strong> Parameter <strong>Spaces</strong><br />

How can we make sense <strong>of</strong> these noncommutative parameter spaces?<br />

Theorem (SB-GL-WvS)<br />

The algebra A(Mk,θ)>⊳ Z 2 has a commutative subalgebra, denoted A(Mk).<br />

There is a canonical action <strong>of</strong> Z 2 on A(Mk) by gauge transformations<br />

There is an isomorphism<br />

A(Mk,θ)>⊳ Z 2 A(Mk)>⊳ ′ Z 2 .<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 21 / 28


Understanding the <strong>Noncommutative</strong> Parameter <strong>Spaces</strong><br />

How can we make sense <strong>of</strong> these noncommutative parameter spaces?<br />

Theorem (SB-GL-WvS)<br />

The algebra A(Mk,θ)>⊳ Z 2 has a commutative subalgebra, denoted A(Mk).<br />

There is a canonical action <strong>of</strong> Z 2 on A(Mk) by gauge transformations<br />

There is an isomorphism<br />

A(Mk,θ)>⊳ Z 2 A(Mk)>⊳ ′ Z 2 .<br />

The algebra A(Mk) is the (commutative) coordinate algebra <strong>of</strong> a space Mk <strong>of</strong><br />

bona fide monads on C 4 θ , i.e. homomorphisms <strong>of</strong> free A(C4 θ )-modules<br />

H ⊗ A(C 4 θ) σz<br />

−→ K ⊗ A(C 4 θ)<br />

for complex vector spaces H, K <strong>of</strong> dimensions k, 2k + 2.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 21 / 28


The <strong>Noncommutative</strong> ADHM Equations<br />

Now that we have discovered a classical space <strong>of</strong> instantons, we can describe it<br />

more explicity.<br />

Theorem (SB-WvS) The space Mk/ ∼ <strong>of</strong> equivalence classes <strong>of</strong> monads on C 4 θ is<br />

given by the set <strong>of</strong> matrices<br />

satisfying the equations<br />

B1, B2 ∈ Matk(C), I ∈ Mat2,k(C), J ∈ Matk,2(C)<br />

¯µB1B2 − µB2B1 + IJ = 0,<br />

[B1, B ∗ 1] + [B2, B ∗ 2] + II ∗ − J ∗ J = 0<br />

(where µ = exp(iπθ)), modulo the action <strong>of</strong> g ∈ U(k) given by Bj ↦→ gBjg −1 ,<br />

I ↦→ gI, J ↦→ Jg −1 .<br />

The classical limit recovers the ordinary ADHM equations on classical S 4 .<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 22 / 28


The Moyal-Groenewold Plane R 4 <br />

If we work in a local patch R4 <strong>of</strong> S4 and twist by the group <strong>of</strong> translation<br />

symmetries, we get the Moyal plane A(R4 ). The twisting Hopf algebra is<br />

H = A(R4 ).<br />

The deformation is again functorial, so we get a noncommutative family <strong>of</strong><br />

monads A(Mk,) on the deformed space C 4 .<br />

This time there is a canonical action <strong>of</strong> R 4 by gauge A(Mk,) by gauge<br />

transformations, so we obtain the cross product A(Mk,)>⊳ R 4 .<br />

We can also get this by bosonisation and Fourier transform,<br />

A(Mk,)>⊳A(R 4 ) → A(Mk,)>⊳ R 4 .<br />

Again we recover a classical space <strong>of</strong> monads; the resulting ADHM equations are<br />

those <strong>of</strong> Nekrasov-Schwarz:<br />

[B1, B2] + IJ = 0,<br />

[B1, B ∗ 1] + [B2, B ∗ 2] + II ∗ − J ∗ J = −i<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 23 / 28


A Geometric Interpretation<br />

In both cases, we’re twisting the geometry by the action <strong>of</strong> a locally compact<br />

Abelian group Γ. Then we find isomorphic crossed product spaces<br />

A(Mk,Γ)>⊳ Γ ∼ = A(Mk)>⊳ Γ.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 24 / 28


A Geometric Interpretation<br />

In both cases, we’re twisting the geometry by the action <strong>of</strong> a locally compact<br />

Abelian group Γ. Then we find isomorphic crossed product spaces<br />

A(Mk,Γ)>⊳ Γ ∼ = A(Mk)>⊳ Γ.<br />

Morally, we could view this as a pair <strong>of</strong> quotient maps<br />

Mk,Γ<br />

Mk<br />

<br />

<br />

Mk,Γ/ Γ = Mk/ Γ<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 24 / 28<br />

.


A Geometric Interpretation<br />

In both cases, we’re twisting the geometry by the action <strong>of</strong> a locally compact<br />

Abelian group Γ. Then we find isomorphic crossed product spaces<br />

A(Mk,Γ)>⊳ Γ ∼ = A(Mk)>⊳ Γ.<br />

Morally, we could view this as a pair <strong>of</strong> quotient maps<br />

Mk,Γ<br />

Mk<br />

<br />

<br />

Mk,Γ/ Γ = Mk/ Γ<br />

So the moduli problem has a ”stacky” behaviour - both Mk,Γ and Mk are valid<br />

parameter spaces <strong>of</strong> charge k instantons, but one is quantum and the other is<br />

classical.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 24 / 28<br />

.


U(2) or SU(2) <strong>Instanton</strong>s?<br />

On classical S 4 , the ADHM construction produces either U(2) instantons or<br />

SU(2) instantons - they are equivalent.<br />

This means ∇ = d + ω, where ω ∈ Ω 1 (P ) ⊗ g for either g = u(2) or su(2).<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 25 / 28


U(2) or SU(2) <strong>Instanton</strong>s?<br />

On classical S 4 , the ADHM construction produces either U(2) instantons or<br />

SU(2) instantons - they are equivalent.<br />

This means ∇ = d + ω, where ω ∈ Ω 1 (P ) ⊗ g for either g = u(2) or su(2).<br />

For U(2) instantons, the gauge group is just<br />

U(E) :=<br />

<br />

U ∈ End A(S 4 θ )(E) | U ∗ U = UU ∗ = idE<br />

It turns out that the moduli space is computable in this case.<br />

Theorem (SB-GL-WvS) The moduli space <strong>of</strong> U(2) instantons with charge k on<br />

the quantum sphere S 4 θ is a smooth manifold <strong>of</strong> dimension 8k − 3.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 25 / 28<br />

<br />

.


U(2) or SU(2) <strong>Instanton</strong>s?<br />

On classical S 4 , the ADHM construction produces either U(2) instantons or<br />

SU(2) instantons - they are equivalent.<br />

This means ∇ = d + ω, where ω ∈ Ω 1 (P ) ⊗ g for either g = u(2) or su(2).<br />

For U(2) instantons, the gauge group is just<br />

U(E) :=<br />

<br />

U ∈ End A(S 4 θ )(E) | U ∗ U = UU ∗ = idE<br />

It turns out that the moduli space is computable in this case.<br />

Theorem (SB-GL-WvS) The moduli space <strong>of</strong> U(2) instantons with charge k on<br />

the quantum sphere S 4 θ is a smooth manifold <strong>of</strong> dimension 8k − 3.<br />

What happens when you try to look for SU(2) instantons? Recall that, in the<br />

classical case, it doesn’t matter.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 25 / 28<br />

<br />

.


The Trouble with the Trace<br />

If you want connections with ω ∈ Ω 1 (P ) ⊗ su(2), things get tricky. What is<br />

the gauge group here?<br />

We could try to define it as<br />

SU(E) = U ∈ U(E) | UωU ∗ + U(dU ∗ ) ∈ Ω 1 (P ) ⊗ su(2) for all ω .<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 26 / 28


The Trouble with the Trace<br />

If you want connections with ω ∈ Ω 1 (P ) ⊗ su(2), things get tricky. What is<br />

the gauge group here?<br />

We could try to define it as<br />

SU(E) = U ∈ U(E) | UωU ∗ + U(dU ∗ ) ∈ Ω 1 (P ) ⊗ su(2) for all ω .<br />

But there aren’t so many <strong>of</strong> these (only the torus-invariant part <strong>of</strong> the<br />

classical SU(2) gauge group survives).<br />

To get the full ‘set’ <strong>of</strong> SU(2) gauge symmetries, one needs a quantum group<br />

- this partly explains the origin <strong>of</strong> the noncommutative gauge parameters.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 26 / 28


The Trouble with the Trace<br />

If you want connections with ω ∈ Ω 1 (P ) ⊗ su(2), things get tricky. What is<br />

the gauge group here?<br />

We could try to define it as<br />

SU(E) = U ∈ U(E) | UωU ∗ + U(dU ∗ ) ∈ Ω 1 (P ) ⊗ su(2) for all ω .<br />

But there aren’t so many <strong>of</strong> these (only the torus-invariant part <strong>of</strong> the<br />

classical SU(2) gauge group survives).<br />

To get the full ‘set’ <strong>of</strong> SU(2) gauge symmetries, one needs a quantum group<br />

- this partly explains the origin <strong>of</strong> the noncommutative gauge parameters.<br />

In fact the new ‘quantum’ gauge symmetries given by the Pontryagin dual Z 2<br />

seem to interfere with the U(1) part <strong>of</strong> the gauge group - further evidence<br />

that you have to build them in to the ADHM construction.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 26 / 28


Summary<br />

In obtaining an ADHM construction <strong>of</strong> instantons on S4 θ , we were naturally<br />

led to noncommutative parameter spaces.<br />

Nevertheless, by incorporating the ‘quantum’ gauge symmetries into the<br />

construction, we were able to recover a classical space <strong>of</strong> parameters.<br />

The quotient <strong>of</strong> this parameter space by the ‘classical’ gauge symmetries<br />

gives the ’coarse’ moduli space.<br />

But it’s not the full story: we should not ignore the ‘quantum’ gauge<br />

symmetries. Dividing by these as well leads to the noncommutative quotient<br />

space A(Mk)>⊳ Z 2 .<br />

There seems to be a big difference between the moduli problem on classical<br />

spaces and on noncommutative spaces...but we’re getting closer to<br />

understanding it!<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 27 / 28


Relevant Papers<br />

S.Brain, G.Landi: Families <strong>of</strong> Monads and <strong>Instanton</strong>s from a<br />

<strong>Noncommutative</strong> ADHM Construction. Clay Math. Proc 11, 55-84 (2010).<br />

arXiv:math.qa/0901.0772<br />

S. Brain, G. Landi: <strong>Moduli</strong> <strong>Spaces</strong> <strong>of</strong> <strong>Noncommutative</strong> <strong>Instanton</strong>s: Gauging<br />

Away <strong>Noncommutative</strong> Parameters. Quart. J. Math., to appear<br />

arXiv:math.qa/0909.4402<br />

S. Brain, W.D. van Suijlekom: The ADHM Construction <strong>of</strong> <strong>Instanton</strong>s on<br />

<strong>Noncommutative</strong> <strong>Spaces</strong>.<br />

arXiv:math.ph/1008.4517<br />

G. Landi, C. Pagani, C. Reina, W.D. van Suijlekom: <strong>Noncommutative</strong><br />

Families <strong>of</strong> <strong>Instanton</strong>s. IMRN 12 (2008), Art. ID rnn038<br />

arXiv:math.qa/0710.0721<br />

S. Brain, G.Landi, W.D. van Suijlekom: <strong>Moduli</strong> <strong>Spaces</strong> <strong>of</strong> <strong>Instanton</strong>s on Toric<br />

<strong>Noncommutative</strong> Manifolds. In preparation.<br />

S. Brain (UL) NCG <strong>of</strong> <strong>Instanton</strong> <strong>Moduli</strong> <strong>Spaces</strong> IMPAN, 10th January 2011 28 / 28

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!