Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
Topological spin liquids with PEPS
Didier Poilblanc
Laboratoire de Physique Théorique, Toulouse
• Some motivations: «trivial» vs «topological» spin liquids (on frustrated lattices)
• Simple physical ansatze using Projected Entangled Pair States
• «Holographic» framework : bulk-edge correspondence, entanglement Hamiltonian & detecting topological order from the PEPS
jeudi 13 novembre 14
COLLABORATORS
Norbert Schuch, DP, J. Ignacio Cirac, and David Pérez-García, Phys. Rev. B 86, 115108 (2012)DP, Norbert Schuch, David Pérez-García, and J. Ignacio Cirac, Phys. Rev. B 86, 014404 (2012)
DP, Norbert Schuch, J. Ignacio Cirac, Phys. Rev. B 88, 144414 (2013)DP and Norbert Schuch, Phys. Rev. B 87, 140407 (2013)
Norbert Schuch, DP, J. Ignacio Cirac, and David Perez-GarciaPhys. Rev. Lett. 111, 090501 (2013)
DP, Philippe Corboz, Norbert Schuch, J. Ignacio Cirac, PRB 2014
jeudi 13 novembre 14
Exotic «spin liquids» beyond the
«order parameter» paradigm
Topological order can also be detected by entanglement measures !
* no spontaneous broken symmetry* no local order * Topological order
X. G. Wen
- Do they exist in materials ? in simple models ?
- How to detect them ?
GS degeneracy (depends on topology of space)
jeudi 13 novembre 14
The many-body spectrum of a topological liquid
{
EExcitations are fractional (created by pairs)
Degeneracy from «topological order»
jeudi 13 novembre 14
Best candidate : spin-1/2 Heisenberg QAF on the Kagome lattice !
S. Yan, D.A. Huse & S. White, Science 2011
S. Depenbrock, I.P. McCulloch & U. Schollwock, PRL 2012
topological features ?
~S1 · ~S2
1
2
S = 1/2
Numerical «evidence» (DMRG) for a (gapped) spin liquid:
jeudi 13 novembre 14
Materials realizing a spin-1/2 Kagome QAFM
e.g. experiments by C. Broholm (MIT), Z. Hiroi (ISSP), P. Mendels (Orsay)...
jeudi 13 novembre 14
Low-energy spectrum N=36 sites
Sindzingre & Lhuillier, EPL 2009
Exponential # of singlets
Lecheminant et al., PRB 97
Jiang et al., PRL 2008Finite spin gap ?
jeudi 13 novembre 14
RVB spin liquid:
S = 0
spin-1/2 RVB
Equal-weight superposition of NN singlet coverings
P. Fazekas and P.W. Anderson Philosophical Magazine 30, 423-440 (1974)
Topological liquidHasting-Oshikawa-LSM theorem
jeudi 13 novembre 14
Two «types» of cylinder boundaries
«even» «odd»Gv = +1 Gv = �1
s s̄
cylinder geometry
jeudi 13 novembre 14
Eigenstates of a «Wilson loop» operator
no vison flux:
Z2 vison flux:
+�
wv = +1
wv = �1
cylinder geometry
jeudi 13 novembre 14
s s̄
v v
Gv = +1
Gv = �1
Wv = +1
Wv = �1 same class asKitaev’s Toric Code (fixed point )
spin liquid : topological GS inserting «spinons» and «visons»
Z2
⇠ = 0
jeudi 13 novembre 14
HAKLT =X
hi,ji
PSi+Sj=4ij
S=2... vs «Trivial» insulator
The GS is unique !
The Affleck-Kennedy-Lieb-Tasaki spin liquid
jeudi 13 novembre 14
Projected Entangled Pair States (PEPS) construction
Ex.: the S=2 AKLT
|S↵= |01
↵� |10
↵
D=2 spin-1/2 virtual states
Project onto physical subspace :
local tensor
d = 2S + 1
jeudi 13 novembre 14
The spin-1/2 RVB can also be written as a PEPS !
(D=3)
Project onto physical subspace S=1/2 (d=2)
1/2� 0virtual states:
PEPS tensor
jeudi 13 novembre 14
The PEPS as a variational ansatz
I. CiracF. Verstraete
G. Vidal↵,� = {1, . . . , D}
dimension of auxilary (or virtual) space
i = {1, · · · , dphys}
“contract” product of tensors
Coefficients of the wavefunction
C{i1,1,...,iNv,Nh}
| ↵=
XCi1,i2,...,iN |i1, i2, ..., iN
↵
N = NvNh
jeudi 13 novembre 14
⌦ |
↵ «Transfer Operator»
Build «double layer» tensor network by contracting physical variables
if D small enough exact contractions possible...
Iterate product of TM’s to build infinite cylinder
D2Nv ⇥D2Nv matrix
jeudi 13 novembre 14
use simple PEPS ansatz to investigate physical (e.g. topological, etc...) properties
optimize PEPS to construct competitive ansatz for microscopic models
Two possible «routes» :
G. Vidal, T. Xiang, R. Orus, P. Corboz,... & many more
jeudi 13 novembre 14
Simple rewriting for Kagome lattice :
map on a square lattice
«defect triangle»dphys = 8
jeudi 13 novembre 14
Improving the RVB / PEPS ...
«defect triangles» cost energy !
dressing by introducing NNN singlets
(Zeng & Elser 90’)
The «Simplex RVB»
jeudi 13 novembre 14
Easy formalism to construct all topological GS !
(+)e
(�)e
(+)o
(�)o
orthogonal in the limit of infinite cylinders(Nh = 1)
= Z
jeudi 13 novembre 14
diag blocks
Off-diagonal blocks
«mixed» TM
Misguich et al. PRL ‘02
! 1
< 1
RK RVB
jeudi 13 novembre 14
Finite size scaling of RVB energy
YC16 YC12 YC8 cylinders
s s̄
v v
Gv = +1
Gv = �1
Wv = +1
Wv = �1
dynamical fluctuations of vison/spinon pairs
jeudi 13 novembre 14
Gs energy splittings (semi-log scale)
Very short coherence lengths
(Nh ! 1)
⇠ < 1 unit cell
YC20 infinite cylinder
(NN RVB)
(NN RVB)
(Simplex RVB)
(Simplex RVB)
jeudi 13 novembre 14
Holon doping: could a d-wave superconductor emerge (on the square lattice)
?
A Little digression:
What happens under doping the RVB state
?
Spinon doping might be relevant to describe field induced phases...
jeudi 13 novembre 14
RVB superconductors relevant for high-Tc superconductivity ?
Idea #2: magnetic frustration is the key parameter to stabilize a (RVB) spin liquid on the 2D square lattice Ling Wang, DP, Zheng-Cheng Gu, Xiao-Gang Wen, Frank Verstraete, PRL 111 037202 (2013)
Idea #1: the RVB spin liquid is the «parent» insulator of the high-Tc superconductor P.W. Anderson, T.M. Rice, etc...
jeudi 13 novembre 14
RVB spin liquid
0
X
VB
fermion repn
ii
ii
s-wave d+is-wavebosonic repn fermionic repn
for « fPEPS » see e.g. :Corboz, Evenbly, Verstraete, Vidal, PRA 2010
RVB spin liquid
X
VB
1st step:
=
jeudi 13 novembre 14
RVB spin liquid
doping
X
VB
X
VB+holes
Mott insulator Superconductor ?
+
+
· · ·
2nd step :
i
c⇥
i
jeudi 13 novembre 14
hole kinetic energy⌦c†i,�cj,�
↵
jeudi 13 novembre 14
�ij =⌦ci,"cj,#
↵
c = 0
c = �0.65
Superconducting pairing amplituded+is symmetry
(almost) pure d-wave symmetry
jeudi 13 novembre 14
Variational energy : fPEPS vs iPEPS
iPEPS
fPEPS
Very good ansatz !
jeudi 13 novembre 14
BA
«Holographic» framework
Reduced density matrix
�A = TrB |�⇥�
�|
bulk
edge?
jeudi 13 novembre 14
Basic formula:
�2b
�A = U⇥2bU†
isometry: maps 2D onto 1D
“lives” on the boundary
lA = Nh/2
L R
A B
�A L�L
J. Ignacio Cirac, DP, Norbert Schuch, Frank VerstraetePhys. Rev. B 83, 245134 (2011)
jeudi 13 novembre 14
Entanglement entropy
“area” law
Levin & Wen, 2006
Kitaev & Preskill, 2006
subleading correction to area law:topological entropy
SVN ⇠ CNv � lnD
(Von Neumann)
SVN = �Tr{⇢A ln ⇢A} = Tr{�b ln�b}-
jeudi 13 novembre 14
Numerical results
STE ' � ln 2 Z2 spin liquid
jeudi 13 novembre 14
Entanglement Hamiltonian (acting on the edge)
�2b = exp (�Hb)
Is it local ?
The spectrum of Hb is in one-to-one with the true edge spectrum !
Li & Haldane
Hb
local operator (on the edge):D �D matrix ⇥ basis of D2 operators
edge site operators :
x�
jeudi 13 novembre 14
LT
Short-range entanglement Hamiltonian
Trivial gapped spin liquid
A(r) ⇥ exp (�r/�b)
AKLT cylinders(D=2)
⇠b traks bulk correlation length
· · ·
jeudi 13 novembre 14
Entanglement Hamiltonian highly non-local
Topological gapped spin liquid
supported by the non-zero eigenvalue sector of the RDM
projector characterizing topological sectors
Exemple: Kagome RVB
jeudi 13 novembre 14
gapless ES !
Entanglement spectrum
should reflect the physical edge
spectrumLi & Haldane conjecture
jeudi 13 novembre 14
Summary & Outlook
• Simple and physical understanding of important correlated phases of matter like :
• topological and trivial SLs,
• unconventional SC, ...
• Chiral SLs ?
• Virtual degrees of freedom play «physical role» at boundary : edge states
THANK YOU !
jeudi 13 novembre 14