Kitaev quantum double models and their generalizations Liang Chang University of California, Santa Barbara Texas A&M University December 5, 2012 1 / 25 Background Local errors, thermic noise are considered the obstacles in the realization of a quantum computer. Topological properties of physical systems probably the best answer to overcome those problems. Qubits encoded in topological states can be insensitive to local perturbations 2 / 25 Overview In 1997, Kitaev constructed a lattice model with the following features: There is a finite energy gap between the ground state and excited states. The ground state degeneracy is protected by topology from local perturbation. The excited states are anyons. 3 / 25 Lattice models Choose a lattice on a closed oriented surface. Place spins/qubits at the edges of a lattice. N Hilbert space H = V. edges Physical observable: Hermitian operators on H. 4 / 25 Lattice models V V V V V V V V V V V V Choose a lattice on a closed oriented surface. Place spins/qubits at the edges of a lattice. N Hilbert space H = V. edges Physical observable: Hermitian operators on H. 4 / 25 Kitaev models C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] G is a finite group. Group algebra C[G] is a vector space spanned by group elements. N Hilbert space H = C[G]. edges 5 / 25 Kitaev models C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] G is a finite group. Group algebra C[G] is a vector space spanned by group elements. N Hilbert space H = C[G]. edges 5 / 25 Kitaev models C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] C[G] G is a finite group. Group algebra C[G] is a vector space spanned by group elements. N Hilbert space H = C[G]. edges 5 / 25 Vertex and plaquette operators Plaquette operators Bp : g3 Bp g2 g3 + p g4 = δg1 g2 g3 g4 ,e g2 g1 + p g4 g1 Vertex operators Av : g2 + g g1 3 = Av v g4 hg2 + hg P hg1 3 1 |G| v h∈G hg4 6 / 25 Hamiltonian Hamiltonian H = − P v Av − P Bp . p All [Av , Av0 ] = 0, [Bp , Bp0 ] = 0, [Av , Bp ] = 0. They can be diagonalized simultaneously. All Av ’s and Bp ’s are projectors, i.e., A2v = Av and Bp2 = Bp . As a consequence, the eigenvalues are -1 and 1. In particular, the ground state |Ψ > is the common eigenspace of all Av ’s and Bp ’s for eigenvalue 1, i.e., Av |Ψ >= |Ψ > and Bp |Ψ >= |Ψ > for all v and p. 7 / 25 Hamiltonian Hamiltonian H = − P v Av − P Bp . p All [Av , Av0 ] = 0, [Bp , Bp0 ] = 0, [Av , Bp ] = 0. They can be diagonalized simultaneously. All Av ’s and Bp ’s are projectors, i.e., A2v = Av and Bp2 = Bp . As a consequence, the eigenvalues are -1 and 1. In particular, the ground state |Ψ > is the common eigenspace of all Av ’s and Bp ’s for eigenvalue 1, i.e., Av |Ψ >= |Ψ > and Bp |Ψ >= |Ψ > for all v and p. 7 / 25 Hamiltonian Hamiltonian H = − P v Av − P Bp . p All [Av , Av0 ] = 0, [Bp , Bp0 ] = 0, [Av , Bp ] = 0. They can be diagonalized simultaneously. All Av ’s and Bp ’s are projectors, i.e., A2v = Av and Bp2 = Bp . As a consequence, the eigenvalues are -1 and 1. In particular, the ground state |Ψ > is the common eigenspace of all Av ’s and Bp ’s for eigenvalue 1, i.e., Av |Ψ >= |Ψ > and Bp |Ψ >= |Ψ > for all v and p. 7 / 25 Hamiltonian Hamiltonian H = − P v Av − P Bp . p All [Av , Av0 ] = 0, [Bp , Bp0 ] = 0, [Av , Bp ] = 0. They can be diagonalized simultaneously. All Av ’s and Bp ’s are projectors, i.e., A2v = Av and Bp2 = Bp . As a consequence, the eigenvalues are -1 and 1. In particular, the ground state |Ψ > is the common eigenspace of all Av ’s and Bp ’s for eigenvalue 1, i.e., Av |Ψ >= |Ψ > and Bp |Ψ >= |Ψ > for all v and p. 7 / 25 Toric code model The surface is a torus T . G = Z2 = {1, −1}. Plaquette operators Bp : g3 Bp g2 g3 + p g4 g1 = δg1 g2 g3 g4 ,1 g2 + p g4 g1 Vertex operators Av : g2 + g g1 3 Av = v g4 1 2 g2 + −g2 + −g g g −g1 1 3 3 + v v g4 −g4 8 / 25 Eigenstates of all Bp c A common eigenstate of all Bp ’s can be represented by closed curves in the dual lattice. 9 / 25 Eigenstates of all Bp c A common eigenstate of all Bp ’s can be represented by closed curves in the dual lattice. 9 / 25 Eigenstates of all Av v c c ∼ c 0 : c is homologous to c 0 . P A common eigenstate of all Av ’s is of the form |ci > where ci ∼ cj . 10 / 25 Eigenstates of all Av v c0 c ∼ c 0 : c is homologous to c 0 . P A common eigenstate of all Av ’s is of the form |ci > where ci ∼ cj . 10 / 25 Eigenstates of all Av v c0 c ∼ c 0 : c is homologous to c 0 . P A common eigenstate of all Av ’s is of the form |ci > where ci ∼ cj . 10 / 25 Ground states of toric code The ground state space has 4 basis vetors: [c0 ], [c1 ], [c2 ], [c1 ∪ c2 ]. The ground state space has a topological interpretation as the homology group H1 (T ) with two generators c1 and c2 . c0 c2 c1 11 / 25 Generalized Kitaev models To generalize the group algebra models, we expect to construct a Hamiltonian X X H =− Av − Bp v p such that all Av ’s and Bp ’s commute with each other. It was anticipated by Kitaev to be possible for finite dimensional C ∗ Hopf algebras and was achieved by Buerschaper, et al. It can be generalized in a general quantum groupoid setting. 12 / 25 Generalized Kitaev models To generalize the group algebra models, we expect to construct a Hamiltonian X X H =− Av − Bp v p such that all Av ’s and Bp ’s commute with each other. It was anticipated by Kitaev to be possible for finite dimensional C ∗ Hopf algebras and was achieved by Buerschaper, et al. It can be generalized in a general quantum groupoid setting. 12 / 25 Generalized Kitaev models To generalize the group algebra models, we expect to construct a Hamiltonian X X H =− Av − Bp v p such that all Av ’s and Bp ’s commute with each other. It was anticipated by Kitaev to be possible for finite dimensional C ∗ Hopf algebras and was achieved by Buerschaper, et al. It can be generalized in a general quantum groupoid setting. 12 / 25 Quantum groupoids Quantum groupoid H(µ, ∆, S, 1, ε) over C is a complex vector space equipped with the following compatible maps: Multiplication µ : H ⊗ H → H; Unit i : C → H. i(1) = 1H ; Comultiplication ∆ : H → H ⊗ H. P ∆(x) = (x) x(1) ⊗ x(2) (Sweedler notation); Counit ε : H → C; Antipode S : H → H. 13 / 25 Example of quantum groupoids For a finite group G, its group algebra C(G) is Hopf algebra. For g, h ∈ G, µ(g ⊗ h) = gh, i(1) = e, ∆(g) = g ⊗ g, ε(g) = 1, S(g) = g −1 14 / 25 Exactly solvable Hamiltonian It turns out vertex and plaquette operators can be built from two cocommutative elements Λ ∈ H and λ ∈ H ∗ . 1 P For group algebra C[G], λ (g) = δg,e and Λ = |G| g. g∈G For general quantum groupoids, λ and Λ play a similar role as trace on H and H ∗ . 15 / 25 Exactly solvable Hamiltonian Plaquette operators Bp : c(1) c Bp b + p d P = λ (d(2) c(2) b(2) a(2) ) b(1) (a,b,c,d) + p d(1) a(1) a Vertex operators Av : b + a c Av v d = P (Λ) Λ(2) b + Λ c Λ(1) a (3) v Λ(4) d 16 / 25 Fibonacci category F Simple objects: 1 and τ; Self dual: 1∗ = 1 and τ ∗ = τ; Quantum dimension: d1 = 1 and dτ = φ = Fusion rule: τ2 √ 1+ 5 2 (golden ratio); = 1 + τ. F-moves: τ τ τ τ = φ −1 1 τ τ τ +φ τ τ τ 1 τ τ τ τ τ 1 φ−2 τ − 21 τ = τ τ 1 τ τ τ τ τ −φ −1 τ τ τ 17 / 25 Fibonacci algebra In 2010, Kitaev and Kong constructed a C ∗ -quantum groupoid HC from a unitary fusion category C. Here we illustrate their construction by using the example of Fibonacci category F. As vector space, HF is spanned by a ab ei;cd c i = b d where a, b, c, d, i ∈ {1, τ}. Multiplication a b a0 c i · d b0 i0 c0 = d0 a i0 δc,a0 δd,b0 δi,i 0 √ c0 di b d0 18 / 25 Unit P√ η= di a a i a,b,i b b Comultiplication a i ∆ b c ! = P j,k p,q d a √ di dp dq iap idq √ Fk ,jk Fj,kc p c j ⊗ db dc p q q k b d Counit a i ε b c ! = δa,b δc,d δi,1 d 19 / 25 Antipode a c ! i S d √ db dc =√ b i da dd b c d a Star a i b c !∗ c = d a i d b 20 / 25 Examples of quantum groupoid structure 1 τ τ τ ∆ τ τ 1 ! 1 1 τ ⊗ τ τ − 12 τ 1 τ + 1 τ 1 ⊗ τ τ τ τ τ τ τ ⊗ τ τ τ τ 1 τ −φ τ τ 1 τ τ √1 φ τ 1 1 τ = τ τ = τ 1 τ · τ τ 1 τ τ 21 / 25 Solvable Hamiltonian for HF By choosing a Λ= 1 1+φ 2 P da db a,b a a λ= 1 1+φ 2 P p a,b,µ b 1 b ∧ a µ dµ b We get an exactly solvable Hamiltonian HK = − b P v AKv − P p BpK . 22 / 25 Ground states The ground state is k1 a2 c1 a2 6 P Q dbn kn in µ k c b n n n √ Fi ,i a Fj ,j i djn n+1 n n+1 n+1 n n+1 j,k ,µ n=1 a,b,c c2 k2 a3 c2 b1 i2 c1 i1 a1 b6 b1 a1 c6 c6 b6 b2 b2 a3 i3 i6 a6 b5 b5 b3 c3 c3 k3 k6 a4 b3 a4 b4 i4 c4 b4 i5 a5 c5 a6 k5 + c5 a5 c4 k4 23 / 25 Main result and future work Theorem Given a trivalent lattice Γ on a closed oriented surface Σ, the ground space LK (Σ, Γ) of the Kitaev model based on HF is canonically isomorphic to the ground space LLW (Σ, Γ) of Levin-Wen Models based on F. As a consequence, LK (Σ, Γ) is canonically isomorphic to the target space ZTV (Σ) of the Turaev-Viro TQFT based on F. Excitation(Anyons). Non-semisimple generalization. 24 / 25 Main result and future work Theorem Given a trivalent lattice Γ on a closed oriented surface Σ, the ground space LK (Σ, Γ) of the Kitaev model based on HF is canonically isomorphic to the ground space LLW (Σ, Γ) of Levin-Wen Models based on F. As a consequence, LK (Σ, Γ) is canonically isomorphic to the target space ZTV (Σ) of the Turaev-Viro TQFT based on F. To do: Excitation(Anyons). Non-semisimple generalization. 24 / 25 Thank You! 25 / 25