local recovery maps as duct tape for many body … · metropolis algorithm: (- start with random...
TRANSCRIPT
![Page 1: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/1.jpg)
LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY
SYSTEMS
Michael J. Kastoryano
November 14 2016, QuSoft Amsterdam
Tuesday, November 15, 16
![Page 2: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/2.jpg)
CONTENTSLocal recovery maps
Exact recovery and approximate recovery
Local recovery for many body systemsHammersley-Clifford and Gibbs sampling
State preparationEvaluating local expectation values
Efficient state preparation
Further ApplicationsTuesday, November 15, 16
![Page 3: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/3.jpg)
LOCAL RECOVERY MAPS
I⇢(A : C|B) = S(AB) + S(BC)� S(B)� S(ABC) � 0
I⇢(A : C|B) = 0 , RAB(⇢BC) = ⇢
RAB(�) = ⇢1/2AB⇢�1/2B �⇢�1/2
B ⇢1/2AB
Markov State
there exists a disentangling unitary on B.
Petz map
Strong subadditivity (SSA):
Equality
⇢ = �j⇢ABLj⌦ ⇢BR
j C
P. Hayden, et. al., CMP 246 (2004)
M. Ohya and D. Petz, (2004)
Tuesday, November 15, 16
![Page 4: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/4.jpg)
LOCAL RECOVERY MAPSApproximatelyStrengthening SSA:
I⇢(A : C|B) � �2 log2 F (⇢, RAB(⇢AB))
RAB(�) =
Zdt�(t)⇢
12+itAB ⇢
� 12�it
B �⇢� 1
2+itB ⇢
12�itAB
Rotated Petz map
ABC are arbitrary
Is the map universal?Is the conditional mutual information necessary?Other properties of the map?
O. Fawzi and R. Renner, CMP 340 (2015)
M. Junge, et. al. arXiv:1509.07127
Tuesday, November 15, 16
![Page 5: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/5.jpg)
APPLICATIONSShannon Theory and Entanglement theory
Tuesday, November 15, 16
![Page 6: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/6.jpg)
APPLICATIONSShannon Theory and Entanglement theory
Quantum Simulations (sampling)
Classical Simulations
Topological order
Quantum error correction
Renormalization Group, critical models, AdS/CFT
Tensor networks, stoquastic models
Tuesday, November 15, 16
![Page 7: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/7.jpg)
APPLICATIONSShannon Theory and Entanglement theory
Quantum Simulations (sampling)
Classical Simulations
Topological order
Quantum error correction
Renormalization Group, critical models, AdS/CFT
Tensor networks, stoquastic models
Tuesday, November 15, 16
![Page 8: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/8.jpg)
MANY-BODY SETTING
For any A, and B shielding A:
A`
B
C
Exact recovery
I⇢(A : C|B) = 0
H = H⌦N2
Tuesday, November 15, 16
![Page 9: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/9.jpg)
HAMMERSLEY-CLIFFORD
is the Gibbs state of a local commuting H
⇢ > 0
is the ground state of a local commuting H
⇢ = | ih |
For any A, and B shielding A:
A`
B
C
Exact recovery
I⇢(A : C|B) = 0
H = H⌦N2
W. Brown, D. Poulin, arXiv:1206.0755
Tuesday, November 15, 16
![Page 10: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/10.jpg)
HAMMERSLEY-CLIFFORD
is the Gibbs state of a local commuting H
⇢ > 0
is the ground state of a local commuting H
⇢ = | ih |
For any A, and B shielding A:
A`
B
C
Exact recovery
I⇢(A : C|B) = 0
Approximate recovery
For any A, and B shielding A: I⇢(A : C|B) Ke�c`
H = H⌦N2
W. Brown, D. Poulin, arXiv:1206.0755
Tuesday, November 15, 16
![Page 11: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/11.jpg)
HAMMERSLEY-CLIFFORD
is the Gibbs state of a local commuting H
⇢ > 0
is the ground state of a local commuting H
⇢ = | ih |
For any A, and B shielding A:
A`
B
C
Exact recovery
I⇢(A : C|B) = 0
Approximate recovery
For any A, and B shielding A: I⇢(A : C|B) Ke�c`
is the Gibbs state of a local non-commuting H⇢ > 0
is the ground state of a gaped local non-commuting H⇢ = | ih |
H = H⌦N2
W. Brown, D. Poulin, arXiv:1206.0755
K. Kato, F Brandao, arXiv:1609.06636
Tuesday, November 15, 16
![Page 12: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/12.jpg)
AREA LAW
A`
Further consequences
B1
B2B3
I(A : B1 · · ·Bn+1)� I(A : B1 · · ·Bn) = I(A : Bn+1|B1 · · ·Bn)
Decaying CMI provides a quantitative MI area law
Mutual info area law: I(A : Ac) c|@A|
Tuesday, November 15, 16
![Page 13: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/13.jpg)
AREA LAW
A`
Further consequences
B1
B2B3
I(A : B1 · · ·Bn+1)� I(A : B1 · · ·Bn) = I(A : Bn+1|B1 · · ·Bn)
Decaying CMI provides a quantitative MI area law
Mutual info area law: I(A : Ac) c|@A|
Can also show: Small CMI implies efficient MPS/MPO representation!
Take-home message: CMI replaces Area Law, HC program replaces the area law conjecture
Tuesday, November 15, 16
![Page 14: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/14.jpg)
AREA LAW
A`
Further consequences
B1
B2B3
I(A : B1 · · ·Bn+1)� I(A : B1 · · ·Bn) = I(A : Bn+1|B1 · · ·Bn)
Decaying CMI provides a quantitative MI area law
Mutual info area law: I(A : Ac) c|@A|
Can also show: Small CMI implies efficient MPS/MPO representation!
Take-home message: CMI replaces Area Law, HC program replaces the area law conjecture
What about dynamics and
state preparation?
Tuesday, November 15, 16
![Page 15: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/15.jpg)
MONTE-CARLO SIMULATIONSWant to evaluate: hQi =
X
x
⇡(x)Q(x) ⇡ / e��H
classical Gibbs state
Idea: - obtain a sample configuration from the distribution ⇡
- Set up a Markov chain with as an approximate fixed point
⇡
Tuesday, November 15, 16
![Page 16: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/16.jpg)
MONTE-CARLO SIMULATIONSWant to evaluate: hQi =
X
x
⇡(x)Q(x) ⇡ / e��H
classical Gibbs state
Idea: - obtain a sample configuration from the distribution ⇡
- Set up a Markov chain with as an approximate fixed point
⇡
Metropolis algorithm: (- start with random configuration)- Flip a spin at random, calculate energy- If energy decreased, accept the flip- If energy increased, accept the flip with probability pflip = e���E
- Repeat until equilibrium is reached
Tuesday, November 15, 16
![Page 17: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/17.jpg)
MONTE-CARLO SIMULATIONSWant to evaluate: hQi =
X
x
⇡(x)Q(x) ⇡ / e��H
classical Gibbs state
Idea: - obtain a sample configuration from the distribution ⇡
- Set up a Markov chain with as an approximate fixed point
⇡
Metropolis algorithm: (- start with random configuration)- Flip a spin at random, calculate energy- If energy decreased, accept the flip- If energy increased, accept the flip with probability pflip = e���E
- Repeat until equilibrium is reached Equilibrium?Tuesday, November 15, 16
![Page 18: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/18.jpg)
ANALYTIC RESULTSNote: - Glauber dynamics (Metropolis) is modeled by a
semigroup Pt = etL
Tuesday, November 15, 16
![Page 19: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/19.jpg)
ANALYTIC RESULTSNote: - Glauber dynamics (Metropolis) is modeled by a
semigroup Pt = etL
Fundamental result for Glauber dynamics:
has exponentially decaying correlations
mixes in time
independent of boundary conditions in 2D
no intermediate mixing
Pt⇡ O(log(N))
independent of specifics of the model
is gapped LF. Martinelli, Lect. Prof. Theor. Stats , Springer
A. Guionnet, B. Zegarlinski, Sem. Prob., Springer
Tuesday, November 15, 16
![Page 20: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/20.jpg)
QUANTUM GIBBS SAMPLERS
Davies maps are another generalization of Glauber dynamics
Tt = etL
L =X
j2⇤
(Rj@ � id)
Rj@ is the Petz recovery map!
Commuting Hamiltonian
MJK and K. Temme, arXiv:1505.07811
Tuesday, November 15, 16
![Page 21: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/21.jpg)
QUANTUM GIBBS SAMPLERS
Davies maps are another generalization of Glauber dynamics
Tt = etL
L =X
j2⇤
(Rj@ � id)
Rj@ is the Petz recovery map!
Commuting Hamiltonian
The exists a partial extension of the statics = dynamics theorem
MJK and F. Brandao, CMP 344 (2016)
MJK and K. Temme, arXiv:1505.07811
Tuesday, November 15, 16
![Page 22: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/22.jpg)
QUANTUM GIBBS SAMPLERS
Davies maps are another generalization of Glauber dynamics
Tt = etL
L =X
j2⇤
(Rj@ � id)
Rj@ is the Petz recovery map!
Commuting Hamiltonian
The exists a partial extension of the statics = dynamics theorem
Non-commuting Hamiltonian
L =X
j2⇤
(Rj@ � id)
Rj@ is the rotated Petz map!no longer frustration-freeTheorem does not holdDavies maps are non-local
MJK and F. Brandao, CMP 344 (2016)
MJK and K. Temme, arXiv:1505.07811
Tuesday, November 15, 16
![Page 23: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/23.jpg)
STATE PREPARATIONBased on : MJK, F. Brandao, arXiv:1609.07877
Tuesday, November 15, 16
![Page 24: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/24.jpg)
SETTING
Hamiltonian:
Lattice: ⇤
A
Gibbs states:
A ⇢ ⇤
hj
hZ = 0 for |Z| � K
Note:
is the Gibbs state restricted to A
HA =X
Z⇢A
hZ
⇢A = e��HA/Tr[e��HA ]
Superscript for domain of definition of Gibbs state, while subscript for partial trace.
Tuesday, November 15, 16
![Page 25: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/25.jpg)
THE MARKOV CONDITIONUniform Markov:
A`
B
C
A BB
C `
Any subset with shielding from in , we have
X = ABC ⇢ ⇤ BA C X
I⇢X (A : C|B) �(`)
Recall: ⇢X = e��HX/Tr[e��HX ]
Also must hold for non-contractible regions
⇤
Tuesday, November 15, 16
![Page 26: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/26.jpg)
CORRELATIONS
⇤
AB
Cov⇢(f, g) = |tr[⇢fg]� tr[⇢f ]tr[⇢g]|
Cov⇢X (f, g) ✏(`)
`C
Uniform Clustering:
Any subset with and
X = ABC ⇢ ⇤supp(f) ⇢ A supp(g) ⇢ B
ABC `
Note: Uniform Clustering follows from uniform Gap
Tuesday, November 15, 16
![Page 27: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/27.jpg)
if
General
⇤
e��(HA+HB) = e��HA
e��HB
[HA, HB ] = 0
e��(H+V ) = OV e��HO†
V
||OV || e�||V ||
Only works if is local!
V`
Commuting Hamiltonian
Non-commuting Hamiltonian
V
||OV �O`V || c1e
�c2` ⌘ �(`)
LOCAL PERTURBATIONS
MB. Hastings, PRB 201102 (2007)
Tuesday, November 15, 16
![Page 28: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/28.jpg)
⇤
V`
Uniform MarkovAPPROXIMATIONS
I⇢X (A : C|B) �(`)A`
B
C
AB `
CUniform clustering
Cov⇢X (f, g) ✏(`)
Local perturbations
||e��(H+V ) �O`V e
��HO`V || c1e
�c2` ⌘ �(`)
Tuesday, November 15, 16
![Page 29: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/29.jpg)
LOCAL INDISTINGUISHABILITY
⇤
Cov⇢X (f, g) ✏(`)
Result 1:
Any subset with and
X = ABC ⇢ ⇤supp(f) ⇢ A supp(g) ⇢ B
Any subset with shielding from in , if is uniformly clustering,
X = ABC ⇢ ⇤ BA C X
A`
B
C⇢
Consequence: Efficient evaluation of local expectation values
hOAi = tr[⇢⇤OA] ⇡ tr[⇢ABOA]
||trBC [⇢ABC ]� trB [⇢
AB ]||1 c|AB|(✏(`) + �(`))
Tuesday, November 15, 16
![Page 30: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/30.jpg)
LOCAL INDISTINGUISHABILITY
Cov⇢X (f, g) ✏(`)
Result 1:
Any subset with and
X = ABC ⇢ ⇤supp(f) ⇢ A supp(g) ⇢ B
Any subset with shielding from in , if is uniformly clustering,
X = ABC ⇢ ⇤ BA C X
C⇢
Proof idea:
Remove pieces of the boundary of one by oneB
A`
B
telescopic sum
Bound each term ||trBC [⇢Xj+1 � ⇢Xj ]||1 ⇡ sup
gA|tr[gA(O`
j⇢XjO`,†
j � ⇢Xj ]|
||trBC [⇢X � ⇢AB ⌦ ⇢C ]||1
X
j
||trBC [⇢Xj+1 � ⇢Xj ]||1
= Cov⇢Xj (gA, O`,†j O`
j)
||trBC [⇢ABC ]� trB [⇢
AB ]||1 c|AB|(✏(`) + �(`))
Tuesday, November 15, 16
![Page 31: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/31.jpg)
STATE PREPARATION
Cov⇢X (f, g) ✏(`)
Main Result:
Any subset with and
X = ABC ⇢ ⇤supp(f) ⇢ A supp(g) ⇢ B
If is uniformly clustering and uniformly Markov, then there exists a depth circuit of quantum channels of local range , such that
⇢
D + 1 F = FD+1 · · ·F1
O(log(L))
||F( )� ⇢||1 cLD(✏(`) + �(`) + �(`))
MJK, F. Brandao, arXiv:1609.07877
Tuesday, November 15, 16
![Page 32: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/32.jpg)
STATE PREPARATION
Cov⇢X (f, g) ✏(`)
Main Result:
Any subset with and
X = ABC ⇢ ⇤supp(f) ⇢ A supp(g) ⇢ B
If is uniformly clustering and uniformly Markov, then there exists a depth circuit of quantum channels of local range , such that
⇢
D + 1 F = FD+1 · · ·F1
Cov⇢X (f, g) ✏(`)
Corollary:
Any subset with and
X = ABC ⇢ ⇤supp(f) ⇢ A supp(g) ⇢ B
If is uniformly clustering and uniformly Markov, then there exists a depth circuit of strictly local quantum channels , such that
⇢
O(log(L))
F = FM · · ·F1
M = O(log(L))
||F( )� ⇢||1 cLD(✏(`) + �(`) + �(`))
||F( )� ⇢||1 cLD(✏(`) + �(`) + �(`))
MJK, F. Brandao, arXiv:1609.07877
Tuesday, November 15, 16
![Page 33: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/33.jpg)
PROOF OUTLINEStep 1: Cover the lattice in concentric
squares⇤
A
A+
A� ⇢ A ⇢ A+
By the Markov condition`A�
By Local indistinguishability||trA[⇢
Ac�
Ac ]� ⇢Ac ]||1 NA✏(`)
Local cpt map FA ⌘ R⇢A+
trA
||R⇢A+
(⇢Ac)� ⇢||1 NA(�(`) + �(`))
||FA(⇢Ac
�)� ⇢||1 NA(✏(`) + �(`) + �(`))
If we can build the lattice with holes, then we can reconstruct the original lattice.
Ac�
Tuesday, November 15, 16
![Page 34: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/34.jpg)
PROOF OUTLINEStep 2: Break up the connecting regions⇤
By the Markov condition
By Local indistinguishability
Local cpt map
If we can build the lattice , then we can reconstruct the original lattice.
B�
B+
B `
B� ⇢ B ⇢ B+
||R⇢Ac
�
B+(⇢
Ac�
Bc )� ⇢Ac� ||1 NB(�(`) + �(`))A�
||trB [⇢(A�B�)c ]� ⇢Ac
�Bc
�]||1 NB✏(`)
FB ⌘ R⇢Ac
�
B+trB
||FBFA(⇢(A�B�)c)� ⇢||1 (NA +NB)(✏(`) + �(`) + �(`))
(A�B�)c
Tuesday, November 15, 16
![Page 35: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/35.jpg)
PROOF OUTLINEStep 3:
Project onto
By locality
C
⇢C
FC( ) = ⇢ctrC [ ]
Finally
The entire lattice can be built from a local circuit of cpt maps.
||FCFBFA( )� ⇢||1 (NC +NA +NB)(✏(`) + �(`) + �(`))
Tuesday, November 15, 16
![Page 36: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/36.jpg)
GROUND STATES
Proof ingredients (uniform) Local indistinguishability(uniform) Markov conditionLocal definition of states
For injective PEPS, proof can be reproduced exactly.
We can show that the conditions of the theorem hold it the topological entanglement entropy is zero.
Tuesday, November 15, 16
![Page 37: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/37.jpg)
SPECTRAL GAPWe showed:
Define FA = etLA LA =X
j
(FAi � id)
If had the same fixed point, then is gaped, by the reverse detectability lemma. FA,FB ,FC L = LA + LB + LC
The same strategy works for proving gaps of parent Hamiltonians of injective PEPS
New strategy for proving the gap of the 2D AKLT model!!!
All about boundary conditions
||FCFBFA( )� ⇢||1 LDe�`/⇠
A. Anshu, et. al., Phys. Rev. B 93, 205142 (2016)
Tuesday, November 15, 16
![Page 38: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/38.jpg)
OUTLOOK
Approximate Quantum error correction
Renormalization Group, critical models, AdS/CFT
Tradeoff bounds
Spectral gap analysis, entanglement spectrumNew classification for many-body systems
New codes?S. Flammia, J. Haah, MJK, I. Kim, arXiv:1610.06169
Tuesday, November 15, 16
![Page 39: LOCAL RECOVERY MAPS AS DUCT TAPE FOR MANY BODY … · Metropolis algorithm: (- start with random configuration) - Flip a spin at random, calculate energy - If energy decreased, accept](https://reader031.vdocuments.mx/reader031/viewer/2022011919/601c99fa30e703299e54849a/html5/thumbnails/39.jpg)
THANK YOU!
Tuesday, November 15, 16