Survey
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
Introduction to Tensor Network States Sukhwinder Singh Macquarie University (Sydney) Contents • The quantum many body problem. • Diagrammatic Notation • What is a tensor network? • Example 1 : MPS • Example 2 : MERA Quantum many body system in 1-D 1 N 2 }D dim(V ) D Total Hilbert Space : V N V N Dimension = D Huge ! i1i2 iN i1i2 iN i1 i2 iN N How many qubits can we represent with 1 GB of memory? Here, D = 2. 2 8 2 N 30 N 27 To add one more qubit double the memory. But usually, we are not interested in arbitrary states in the Hilbert space. Typical problem : To find the ground state of a local Hamiltonian H, H h12 h23 h34 ... hN 1, N Ground states of local Hamiltonians are special Limited Correlations and Entanglement. C(l ) OxOxl S (l ) i log i i Properties of ground states in 1-D 1) Gapped Hamiltonian C (l ) e S (l ) const l / 2) Critical Hamiltonian C (l ) l a a0 S (l ) log(l ) l We can exploit these properties to represent ground states more efficiently using tensor networks. V N Ground states of local Hamiltonians Contents • The quantum many body problem. • Diagrammatic Notation • What is a tensor network? • Example 1 : MPS • Example 2 : MERA Tensors Multidimensional array of complex numbers b a a Ket : 1 2 3 * 2 a Matrix * 3 M 11 M 21 M 31 ik b a Bra : * 1 Ti1i2 M 12 M 22 M 32 c Rank-3 Tensor M 11 c 1 M 21 M 31 N11 c 2 N 21 N 31 M 12 M 22 M 32 N12 N 22 N 32 Contraction = a a M a b M b ab b Contraction a Q P R c = Rac a b c Pab Qbc b contraction cost a b c Contraction b b Q S = a c S abc g f P e c a P afe efg R Q fbg Regc Trace a z M aa M = b P b = a a R a c Pab Rabcc c Tensor product a e a b f cd c d e a b b (Reshaping) b a Decomposition M M T 1 Q D Q = = = U U S S V V Decomposing tensors can be useful M d d = P Q d d d Rank(M) = Number of components in M = d Number of components in P and Q = 2 2 d Contents • The quantum many body problem. • Diagrammatic Notation • What is a tensor network? • Example 1 : MPS • Example 2 : MERA Many-body state as a tensor i1i2 i1i2 iN i1 i2 iN i1 i2 iN iN Expectation values O O i1i2 i1i2 iN Ok iN contraction cost = O D N * i1i2 iN Correlators O1 O2 O1O2 contraction cost = O D N Reduced density operators Trsblock contraction cost = O D N Tensor network decomposition of a state Essential features of a tensor network 1) Can efficiently store the TN in memory Total number of components = O(poly(N)) 2) Can efficiently extract expectation values of local observables from TN Computational cost = O(poly(N)) 1 Number of tensors in TN = O(poly(N)) is independent of N Contents • The quantum many body problem. • Diagrammatic Notation • What is a tensor network? • Example 1 : MPS • Example 2 : MERA Matrix Product States MPS 1 Total number of components = N D 2 Recall! O O i1i2 i1i2 iN Ok iN contraction cost = O D N * i1i2 iN Expectation values Expectation values Expectation values Expectation values Expectation values contraction cost = O N D 4 But is the MPS good for representing ground states? But is the MPS good for representing ground states? Claim: Yes! Naturally suited for gapped systems. Recall! 1) Gapped Hamiltonian C (l ) e S (l ) const l / 2) Critical Hamiltonian C (l ) l a a0 S (l ) log(l ) l In any MPS Correlations decay exponentially Entropy saturates to a constant MPS Recall! O1 O2 O1O2 contraction cost = O D N Correlations in a MPS l l 0 1 Correlations in a MPS l Correlations in a MPS l Correlations in a MPS l Correlations in a MPS M MM l Correlations in a MPS M l L M R L QD Q l l 1 l 0 1 R L D R l l Entanglement entropy in a MPS l S const rank ( ) const Entanglement entropy in a MPS Entanglement entropy in a MPS Entanglement entropy in a MPS Entanglement entropy in a MPS Entanglement entropy in a MPS dl 2 S i log i i dl rank ( ) S 2 log( ) 2 MPS as an ansatz for ground states 1. Variational optimization by minimizing energy min MPS MPS H MPS 0 2. Imaginary time evolution ground state lim e Ht random gs t MPS Contents • The quantum many body problem. • Diagrammatic Notation • What is a tensor network? • Example 1 : MPS • Example 2 : MERA Summary • The quantum many body problem. • Diagrammatic Notation • What is a tensor network? • Example 1 : MPS • Example 2 : MERA Thanks !