Consider two primitive elements x, y. We try to impose a quantum deformation xy = qyx. But the coproduct is no longer multiplicative
Δ(x)Δ(y) = xy ⊗ 1 + x ⊗ y + y ⊗ x + 1 ⊗ xy = qyx ⊗ 1 + x ⊗ y + y ⊗ x + q(1 ⊗ yx) Δ(x)Δ(y) − qΔ(y)Δ(x) = x ⊗ y + y ⊗ x − qx ⊗ y − qy ⊗ x
If we take a 3D spacetime, with particles confined to move in two spatial dimensions, the braiding of the particles becomes extremely important: we get a system of anyons.
To represent this mathematically, we need some kind of algebraic representation of the braid group on our space of states.
We first look for representations that act on tensor products individually: ϕ ⋅ (x ⊗ y) = ϕ(x) ⊗ ϕ(y).
Here's the problem: if you try to impose symmetries on this system (for instance, an SU(2) gauge symmetry), you find that the only representations are by the symmetric group (ie permutations), which don't really braid, since if you swap two particles twice you get the identity transformation.
The reason this happens is because in order for the braiding other be compatible with the symmetry, you need that the action of the braid representation commutes with the action of the gauge group.
Solution: You can fix this by replacing
the symmetric group by a quantum group. The idea is
that you still have something that will commute with the action of the
gauge group, but the way the quantum group acts on the state space is
different. Specifically, you insist that there's a particular
compatibility condition on how the quantum group acts on tensor
products: ϕ(x ⊗ y) = ∑ϕ(1)(x) ⊗ ϕ(2)(y)
This gives you a Hopf module rather than an ordinary
module.
Our original attempt at finding a suitable representation of the braid group then can be understood as looking at a grouplike Hopf algebra.
It turns out that the appropriate thing to look at are non-cocommutative quantum groups.
You look for Hopf algebras that come packaged with an element that behaves like a braiding operation
Let H be a Hopf algebra. An R matrix on H is an element R ∈ H ⊗ H such that
RΔ(x)R−1 = (τ ∘ Δ)(x), for all x ∈ H (Δ ⊗ 1)(R) = R13R23 (1 ⊗ Δ)(R) = R13R12
where Rij denotes the first tensor factor of matrix R acting in component i, and the second tensor factor acting in component j. For example R13 = ∑R(1) ⊗ 1 ⊗ R(2)
The actual braiding operation will be τ ∘ R
Take the first axiom and write it like RΔ(x) = (τ ∘ Δ)(x)R. Now let x ∈ H ⊗ H and consider (τ ∘ Δ ⊗ 1)(x), where we're expanding on the first tensor factor
Since R12 = R ⊗ 1, we have
We then set x = R and apply the second axiom to write this as (τ ∘ Δ ⊗ 1)(R)R12 = R12R13R23. And then we write the left hand side as ((τ ⊗ 1) ∘ (Δ ⊗ 1)(R))R12 = (τ ⊗ 1)(R13R23)R12 Then applying the permutation, we get (τ ⊗ 1)(R13R23)R12 = R23R13R12
We acquire the algebraic Yang-Baxter equations R23R13R12 = R12R13R23.
If you're only working with a single representation, you really don't need the full three axioms defining a quasitriangular Hopf algebra; you can get away with just assuming the Yang-Baxer equations hold
The benefit of the extra structure comes from when you want to look at a whole system of representations. It basically makes tensor products of representations work (that's the role of the first axiom), so you can get a monoidal category of representations
The second and third axioms make sure that you end up getting a braided monoidal category of representations
A dual pairing of bialgebras is a non-degenerate bilinear form such that ⟨xy, z⟩ = ∑⟨x, z(1)⟩⟨y, z(2)⟩ ⟨x, yz⟩ = ∑⟨x(1), y⟩⟨x(2), z⟩
and possibly ⟨S(x), y⟩ = ⟨x, S(y)⟩ ⟨1, y⟩ = ε(y) ⟨x, 1⟩ = ε(x)
A dual pairing between a Hopf algebra and itself is called a Laplace pairing, but I'll refer to any dual pairing as a Laplace pairing.
A Laplace pairing is called a coquasitriangular structure if it also satisfies ∑⟨x(1), y(1)⟩x(2)y(2) = ∑x(1)y(1)⟨x(2), y(2)⟩
Consider a dual pair of bialgebras ⟨A, B⟩ → k.
We have 4 different actions:
⟨a, a′ ⇀ b⟩ = ⟨aa′, b⟩ = ∑⟨a, b(1)⟩⟨a′, b(2)⟩
a′ ⇀ b = ∑b(1)⟨a′, b(2)⟩
Mnemonic: a' passes through one factor
of the coproduct and then pairs with the second factor.
⟨a, b ↼ a′⟩ = ⟨a′a, b⟩ = ∑⟨a′, b(1)⟩⟨a, b(2)⟩
b ↼ a′ = ∑⟨a′, b(1)⟩b(2)
Mnemonic: a' passes through one factor
of the coproduct and then pairs with the first factor.
⟨b ⊳ a, b′⟩ = ⟨a, b′b⟩ = ∑⟨a(1), b′⟩⟨a(2), b⟩
since this holds for all b' we have b ⊳ a = ∑a(1)⟨a(2), b⟩
Mnemonic: b passes through one factor
of the coproduct and then pairs with the second factor.
⟨a ⊲ b, b′⟩ = ⟨a, bb′⟩ = ∑⟨a(1), b⟩⟨a(2), b′⟩
a ⊲ b = ∑⟨a(1), b⟩a(2)
Mnemonic: b passes through one factor
of the coproduct and then pairs with the first factor.
Let's check that the ordering is appropriate to give us left/right actions
⟨a, a′ ⇀ (a″ ⇀ b)⟩ = ⟨aa′, a″ ⇀ b⟩ = ⟨aa′a″, b⟩
⟨a, (b ↼ a′) ↼ a″⟩ = ⟨a″a, b ↼ a′⟩ = ⟨(a′a″)a, b⟩
⟨b′ ⊳ (b ⊳ a), b″⟩ = ⟨b ⊳ a, b″b′⟩ = ⟨a, b″(b′b)⟩
⟨(a ⊲ b) ⊲ b′, b″⟩ = ⟨a ⊲ b, b′b″⟩ = ⟨a, bb′b″⟩
So it's actually pretty easy to tell when you have to swap the order of things in the pairing; it happens when you have to move a and b past each other.
Let H be a finite-dimensional Hopf algebra and consider its algebraic dual H*. We have the evaluation pairing ⟨ϕ, g⟩ = ϕ(g)
The left and right coregular actions of H on H* are defined by ⟨g ⋅ ϕ, h⟩ = ⟨ϕ, hg⟩ and ⟨ϕ ⋅ g, h⟩ = ⟨ϕ, gh⟩. In other words, (g ⋅ ϕ)(h) = ϕ(hg) and (ϕ ⋅ g)(h) = ϕ(gh).
Since the pairing is non-degenerate, g ⋅ ϕ and ϕ ⋅ g are determined completely by how they pair with all of the h ∈ H.
Note: Part of the confusion with these
doubles is that you can define them with either orientation: if you have
a pairing ⟨A, B⟩, you
can get a double A ⊗ B and one B ⊗ A.
The Heisenberg doubles of a dual pair ⟨A, B⟩ are algebra structures on A ⊗ B and B ⊗ A given by
On A ⊗ B: [this is like a position representation] [Standard left Heisenberg double] (1 ⊗ b)(a ⊗ 1) = ∑(b(1) ⊳ a) ⊗ b(2) = ∑a(1) ⊗ b(2)⟨a(2), b(1)⟩ [Standard right Heisenberg double] (1 ⊗ b)(a ⊗ 1) = ∑(a ⊲ b(1)) ⊗ b(2) = ∑⟨a(1), b(1)⟩a(2) ⊗ b(2) [Standard right contraction] (1 ⊗ b)(a ⊗ 1) = ∑a(1) ⊗ (b ↼ a(2)) = ∑⟨a(2), b(1)⟩a(1) ⊗ b(2) [Standard left contraction] (1 ⊗ b)(a ⊗ 1) = ∑a(1) ⊗ (a(2) ⇀ b) = ∑⟨a(2), b(2)⟩a(1) ⊗ b(1)
Note: The 1st and 3rd options are
the same (but i don't think they are in the graded
case?)
and swapping (1) and (2): (1 ⊗ b)(a ⊗ 1) = ∑(b(2) ⊳ a) ⊗ b(1) (1 ⊗ b)(a ⊗ 1) = ∑(a ⊲ b(2)) ⊗ b(1) (1 ⊗ b)(a ⊗ 1) = ∑a(2) ⊗ (b ↼ a(1)) (1 ⊗ b)(a ⊗ 1) = ∑a(2) ⊗ (a(1) ⇀ b)
On B ⊗ A: [this is like a momentum representation] (1 ⊗ a)(b ⊗ 1) = ∑b(1) ⊗ (b(2) ⊳ a) (1 ⊗ a)(b ⊗ 1) = ∑b(1) ⊗ (a ⊲ b(2)) (1 ⊗ a)(b ⊗ 1) = ∑(a(1) ⇀ b) ⊗ a(2) (1 ⊗ a)(b ⊗ 1) = ∑(b ↼ a(1)) ⊗ a(2)
and the twists (1 ⊗ a)(b ⊗ 1) = ∑b(2) ⊗ (b(1) ⊳ a) (1 ⊗ a)(b ⊗ 1) = ∑b(2) ⊗ (a ⊲ b(1)) (1 ⊗ a)(b ⊗ 1) = ∑(a(2) ⇀ b) ⊗ a(1) (1 ⊗ a)(b ⊗ 1) = ∑(b ↼ a(2)) ⊗ a(1)
I'm going to stick with the right Heisenberg dual: it keeps A and B in the right order, and the numbers match up nicely
(1 ⊗ b)(a ⊗ 1) = ∑(a ⊲ b(1)) ⊗ b(2) = ∑⟨a(1), b(1)⟩a(2) ⊗ b(2)
H* ⊗ H is the quantum analogue of the cotangent bundle or the Weyl algebra. It doesn't naturally have a Hopf algebra structure, but we can give it one.
We want (ϕ ⊗ 1)(1 ⊗ g) = ϕ ⊗ g and (1 ⊗ g)(ϕ ⊗ 1) = ∑⟨ϕ(2)g(1)⟩ϕ(1) ⊗ g(2)
(1 ⊗ a)(b ⊗ 1) = ∑⟨b(), S(a())⟩⟨b
Unlike the Heisenberg double, the Drinfel'd double is designed to be a quasitriangular Hopf algebra
(1 ⊗ g)(ϕ ⊗ 1) = ∑⟨ϕ(1)S(g(1))⟩⟨ϕ(3)g(3)⟩ϕ(2) ⊗ g(2)
Let H be a Hopf algebra generated as an algebra by its primitive elements. Then the set of grouplike elements is trivial.
This is Lemma 2.6.13 in Hopf Algebras, Quantum Groups, and Topological Field Theory, by Schweigert
We take a 4D vector space generated by {1, g, x, gx}, where g2 = 1, x2 = 0, gx = −xg.
Δx = x ⊗ 1 + g ⊗ x and Δg = g ⊗ g.
Consider the action of a 2 × 2 matrix on the quantum plane. We have e1e2 = qe2e1.
If the coefficients of our matrix are Mij, with
Mx = ∑jMijxjei
We must map back to the quantum plane, so
(Mx)1(Mx)2 = q(Mx)2(Mx)1
Let M11 = α, M12 = β, M21 = γ, M22 = δ.
Then our condition is
So αγ = qγα , βδ = qδβ, and βγ + qαδ = q2γβ + qδα. We can rewrite this latter equation as αδ − δα = qγβ − q−1βγ.
This is the bosonic version of the quantum deformation. In order to get the supersymmetric version (which is what people usually use) we also need to preserve fermionic coordinates: x2 = y2 = 0, qxy = −yx
Note: you could instead require xy = −qyx,
but you'd get relations like βα = −qαβ,
so I'm just picking the defining equations in such a way as to get nice
relations in the end.
Enforcing these gives us (αe1 + βe2)2 = 0 + (−q−1αβ + βα)e2e1 = 0 (γe1 + δe2)2 = 0 + (−q−1γδ + δγ)e2e1 = 0 (αe1 + βe2)(γe1 + δe2) = (αδ − qβγ)e1e2 (γe1 + δe2)(αe1 + βe2) = (γβ − qδα)e1e2
So αβ = qβα, γδ = qδγ and −qαδ + q2βγ = γβ − qδα
The last equation implies αδ − δα = qβγ − q−1γβ
Combining this with αδ − δα = qγβ − q−1βγ, we have
qγβ − q−1βγ = qβγ − q−1γβ, so (q + q−1)γβ = (q + q−1)βγ and thus βγ = γβ
Determinant: The final equation we have
to impose is on the quantum determinant, which is defined to be αδ − qβγ.
If we set this to one, we get SL2q;
if we set it to just be non-zero, we get GL2q.