Entanglement Loss Along RG Flows Entanglement and Quantum Phase Transitions José Ignacio Latorre Dept. ECM, Universitat de Barcelona Newton Institute, Cambridge, August 2004 •Entanglement in Quantum Critical Phenomena G. Vidal, J. I. Latorre, E. Rico, A. Kitaev. Phys. Rev. Lett. 90 (2003) 227902 •Ground State Entanglement in Quantum Spin Chains J. I. Latorre, E. Rico, G. Vidal. Quant. Inf. & Comp. 4 (2004) 48 •Adiabatic Quantum Computation and Quantum Phase Transitions J. I. Latorre, R. Orús, PRA, quant-ph/0308042 •Universality of Entanglement and Quantum Computation Complexity R. Orús, J. I. Latorre, Phys. Rev. A69 (2004) 052308, quant-ph/0311017 •Fine-Grained Entanglement Loss along Renormalization Group Flows J. I. Latorre, C.A. Lütken, E. Rico, G. Vidal. quant-ph/0404120 Entanglement loss along RG flows Introduction Scaling of entropy Entanglement loss along RG flows Preview of new results HEP Condensed Matter • Black hole entropy • Conformal field theory • Spin networks • Extensions of DMRG Scaling of entropy Quantum Information • Entanglement theory • Efficient simulation Entanglement measures for many-qubit systems Few-qubit systems Formation, Distillation, Schmidt coefficients,… N=3, tangle (for GHZ-ness) out of 5 invariants Bell inequalities, correlators based measures Entropy, negativity, concurrence,… Many-qubit systems vac | Oi O j | vac connected e |i j|/ | i j | Scaling of correlators Concurrence does not scan the system We need a measure that obeys scaling and does not depend on the particular operator content of a theory Reznik’s talk Reduced density matrix entropy Schmidt decomposition H H A HB A | AB B dim H A dim H B A |u ij i 1 j 1 i A | vj B Aij U ik kV kl | AB pi | i A | i B i 1 =min(dim HA, dim HB) is the Schmidt number The Schmidt number relates to entanglement | AB pi | i A | i B i 1 Let’s compute the von Neumann entropy of the reduced density matrix A TrB | AB | pi | i i | i 1 S A Tr A log 2 A pi log 2 pi S B i 1 • =1 corresponds to a product state • Large implies large superpositions • e-bit 1 A B Tr | | I 2 1 1 1 1 S A S B log 2 log 2 1 2 2 2 2 Maximum Entropy for N-qubits 1 N N I 2N 2 1 1 S ( N ) N log 2 N N 2 i 1 2 2N Strong subadditivity theorem S ( A, B, C ) S ( B) S ( A, B) S ( B, C ) Smax=N implies concavity on a chain of spins SL+M S L M S LM SL 2 SL SL-M n→ -party entanglement Ground state reduced density matrix entropy SL measures the quantum correlations with the rest of the system Goal: Analyze SL as a function of L for relevant theories Note that ground state reduced density matrix entropy SL Measures the entanglement corresponding to the block spins correlations with the rest of the chain Depends only the ground state, not on the operator content of the theory (Relates to the energy-momentum tensor!!) Scans different scales in the system: Is sensitive to scaling!! Has been discussed in other branches of theoretical physics Black hole entropy Field Theory entanglement, conformal field theory No condensed matter computations Scaling of entropy for spin chains XY model 1 x x 1 y y z l l 1 l l 1 l 2 l 1 2 N H XY Quantum Ising model in a transverse magnetic field N H QI lx lx1 lz l 1 Heisenbeg model 1 x x l l 1 ly ly1 lz lz1 lz l 1 2 N H XXZ XY plane massive fermion massive scalar Quantum phase transitions occur at T=0. Espectrum of the XY model 1 x x 1 y y z l l 1 l l 1 l 2 2 l1 N H XY l , m 2i lm l Jordan-Wigner transformation to spinless fermions Lieb, Schultz, Mattis (1961) bi zj i j i b , b i j ij i, j n 1 n H bi 1bi bi bi 1 bi 1bi bi bi 1 bi bi 2 i 1 i 1 Fourier plus Bogoliubov transformation 1 n1 ijk 2 / n ck e bj n j 0 n/2 H k n / 2 1 2k 2 2 2k k k cos sin n n For γ=0, Ek=λ-cos(2πk/n) k | 0 0 | 0 0 k k uk ck iv k ck 2 Coordinate space correlators can be reconstructed Some intuition The XY chain reduces to a gaussian hamiltonian • We have the exact form of the vacuum • We can compute exact correlators The partial trace of N-L does not imply interaction • Each k mode becomes a mixed state | 0 (1) ( 2) ( L ) L L ~ (1) ~ ( 2) ~ ( L ) (k ) ~ ( k ) 1 0 0 0 , 0 0 0 1 1 i 2 0 0 1 i 2 1 k 1 k L ck ck ck ck 2 2 k 1 L 1 k 1 k 1 k 1 k S L log 2 log 2 2 2 2 2 k 1 L Universality of scaling of entanglement entropy At the quantum phase transition point c S L L log 2 L 3 Quantum Ising XY XX Heisenberg c=1/2 c=1/2 c=1 c=1 free fermion free fermion free boson free boson Universality Logarithmic scaling of entropy controled by the central charge Conformal Field Theory A theory is defined through the Operator Product Expansion Oi ( x)O j ( y ) ij | x y| hi h j Cij | x y| k hi h j hk Ok ( y ) Scaling dimensions=anomalous dimensions Structure constants In d=1+1, the conformal group is infinite dimensional: the structure of “descendants” is fixed the theory is defined by Cijk and hi c 1 T ( z )T ( w) T ( w) 4 2 | z w| | z w| Stress tensor Central charge Away from criticality Saturation of entanglement Quantum Ising S L N / 2 c log 2 | 1 | 6 Connection with previous results Srednicki ’93 (entanglement entropy) Fiola, Preskill, Strominger, Trivedi ’94 (fine-grained entropy) Callan, Holzey, Larsen, Wilczeck ’94 (geometric entropy) Poor performance of DMRG at criticality Area law for entanglement entropy B A Schmidt decomposition SA= SB → Area Law Entropy comes from the entanglement of modes at each side of the boundary Entanglement depends on the connectivity! i A Area law Entanglement bonds Area law in d>1+1 does not depend on the mass Valence bond representation of ground state Plenio’s talk Verstraete’s talk Entanglement in higher dimensions, “Area Law”, for free theories L S c1 d 1 n d 1 d c1 is an anomaly!!!! sm2 e eff ds d / 2 s s0 c0 c R c Fs c Gs 1 2F 2G s Von Neumann entropy captures a most elementary counting of degrees of freedom Trace anomalies Kabat – Strassler Is entropy scheme dependent is d>1+1? Yes L S c1 d 1 No c1=1/6 bosons c1=1/12 fermionic component Entanglement along quantum computation Spin chains are slightly entangled → Vidal’s theorem Schmidt decomposition If A B poly(n) << en max(AB) Then The register can be classically represented in an efficient way! All one- and two-qubit gates actions are also efficiently simulated!! Quantum speed-up needs large entanglement !!! The idea for an efficient representation of states is to store and manipulate information on entanglement, not on the coefficients!! d d i1 1 in 1 | ... ci1 ... in | i1...in ci1 ... in .... [1]i1 [1] 1 1 ... 1 [ 2 ]i2 [ 2 ] 1 2 2 [ 3]i3 2 3 [ n ]in n1 n1 Low entanglement iff αi=1,…, and << en • Representation is efficient • Single qubit gates involve only local update • Two-qubit gates involve only local update Impressive performance when simulating d=1+1 quantum systems! Holy Grail=Extension to higher dimensions Cirac,Verstraete - Vidal Entanglement in Shor’s algorithm (Orús) r r small = easy = small entanglement r large = hard = large entanglement no need for QM QM exponential speed-up Entanglement and 3-SAT 3-SAT 0 1 1 0 0 1 1 0 instance For every clause, one out of eight options is rejected 3-SAT is NP-complete K-SAT is hard for k > 2.41 3-SAT with m clauses: easy-hard-easy around m=4.2n Exact Cover A clause is accepted if 001 or 010 or 100 Exact Cover is NP-complete Adiabatic quantum evolution (Farhi-Goldstone-Gutmann) H(s(t)) = (1-s(t)) H0 + s(t) Hp s(0)=0 Inicial hamiltonian t Problem hamiltonian Adiabatic theorem: if E E1 gmin t E0 s(T)=1 Adiabatic quantum evolution for exact cover |0> |1> |0> |1> |1> |0> |0> |1> (|0>+|1>)(|0>+|1>) (|0>+|1>)…. (|0>+|1>) Typical gap for an instance of Exact Cover Scaling is consistent with gap ~ 1/n If correct, all NP problems could be solved efficiently! Be cautious Scaling of entropy for Exact Cover A quantum computer passes nearby a quantum phase transition! n=6-20 qubits 300 instances n/2 partition S ~ .2 n Entropy seems to scale maximally! Scaling of entropy of entanglement summary Non-critical spin chains S ~ ct Critical spin chains S ~ log2 n Spin networks in d-dimensions S ~ nd-1/d “Area Law” NP-complete problems S~n What has Quantum Information achieved? “Cleaned” our understanding of entropy Rephrased limitations of DMRG Focused on entanglement Represent and manipulate states through their entanglement Opened road to efficient simulations in d>1+1 Next? Entanglement loss along RG flow =ofloss of information RG flowRG = loss Quantum information 1. Global loss of entanglement along RG 2. Monotonic loss of entanglement along RG 3. Fine-grained loss of entanglement along RG Global loss of entanglement along RG + c-theorem Monotonic loss of entanglement along RG -1 SLUV SLIR Majorization theory Entropy provides a modest sense of ordering among probability distributions Muirhead (1903), Hardy, Littlewood, Pólya,…, Dalton Consider d x, y R d such that k i i 1 i 1 p are probabilities, P permutations k x y i 1 x y i 1 x p j Pj y xy d i i 1 i d cumulants are ordered x Dy D is a doubly stochastic matrix x y H x H (y) Fine-grained loss of entanglement L L t Lt RG t’ Lt’ 1 ’1 1 + 2 ’1 + ’2 1 + 2 + 3 ’1 + ’2 + ’3 …….. Strict majorization !!! Recent sets of results I Lütken, R. Orús, E. Rico, G. Vidal, J.I.L. Analytical majorization along Rg Exact results for XX and QI chains based on Calabrese-Cardy hep-th/0405152, Peschel cond-mat/0403048 e H L Tre H L 1 Z L (1 e k 0 L 1 H k dk dk k 0 k ) k 2 2 LogL 2k 1 • Exact eigenvalues, equal spacing • Exact majorization along RG • Detailed partition function Efficient computations in theories with c1/2,1 Recent sets of results II R. Orús, E. Rico, J. Vidal, J.I.L. Lipkin model 1 H N i j i x j x yi yj h zi i Full connectivity (simplex) → symmetric states → SL<Log L =1 3/ 2 SL LogL 3 1 1 S L LogL 3 • Entropy scaling characterizes a phase diagram as in XY + c=1/2 !!! • Underlying field theory? SUSY? • Effective connectivity of d=1 Conclusion: A fresh new view on RG RG on Hamiltonians Wilsonian Exact Renormalization Group RG on correlators Flow on parameters from the OPE RG on states Majorization controls RG flows? Lütken, Rico, Vidal, JIL Cirac, Verstraete, Orús, Rico, JIL The vacuum by itself may reflect irreversibility through a loss of entanglement RG irreversibility would relate to a loss of quantum information