Euclidean Path Integral Optimization
in CFT’s: holographic complexity and
an attempt to a derivation of AdS/CFT
Nilay Kundu
Based on the following works ..
1) Phys. Rev. Lett. 119 (2017) no.7, 071602
2) arXiv : 1706.07056 [hep-th] — (Accepted in JHEP)
With … Pawel Caputa, Masamichi Miyaji, Tadashi Takayanagi and Kento Watanabe
Euclidean Path Integral Optimization
in CFT’s: holographic complexity and
an attempt to a derivation of AdS/CFT
Euclidean Path Integral Optimization
in CFT’s: holographic complexity and
an attempt to a derivation of AdS/CFT
Euclidean Path Integral Optimization
in CFT’s: holographic complexity and
an attempt to a derivation of AdS/CFT
Towards a BETTER understanding of the basic mechanism behind
Euclidean Path Integral Optimization
in CFT’s: holographic complexity and
an attempt to a derivation of AdS/CFT
Euclidean path-integral for wave-fn in CFTs
Towards a BETTER understanding of the basic mechanism behind
Euclidean Path Integral Optimization
in CFT’s: holographic complexity and
an attempt to a derivation of AdS/CFT
OPTIMIZATION
Euclidean path-integral for wave-fn in CFTs
Towards a BETTER understanding of the basic mechanism behind
Euclidean Path Integral Optimization
in CFT’s: holographic complexity and
an attempt to a derivation of AdS/CFT
OPTIMIZATION
Towards a BETTER understanding of the basic mechanism behind
AdS/CFT Euclidean path-integral for wave-fn in CFTs
Euclidean Path Integral Optimization
in CFT’s: holographic complexity and
an attempt to a derivation of AdS/CFT
OPTIMIZATION
Holographic Complexity Towards a BETTER
understanding of the basic mechanism behind
AdS/CFT Euclidean path-integral for wave-fn in CFTs
•
After 20 years of the first proposal by Maldacena … We are still unaware of the basic mechanism behind AdS/CFT conjecture•
After 20 years of the first proposal by Maldacena … We are still unaware of the basic mechanism behind AdS/CFT conjecture•
That means can we aim to demystify this duality beyond known examples …•
To understand the basic mechanism it may be useful to know what does a suitable measure of information in CFT corresponds to that in the AdS ..•
After 20 years of the first proposal by Maldacena … We are still unaware of the basic mechanism behind AdS/CFT conjecture•
That means can we aim to demystify this duality beyond known examples …•
One such quantitative suitable measure happens to be : the entanglemententropy or von-Neumann entropy …
Holographic Entanglement Entropy
•
To understand the basic mechanism it may be useful to know what does a suitable measure of information in CFT corresponds to that in the AdS ..•
After 20 years of the first proposal by Maldacena … We are still unaware of the basic mechanism behind AdS/CFT conjecture•
That means can we aim to demystify this duality beyond known examples …•
One such quantitative suitable measure happens to be : the entanglemententropy or von-Neumann entropy …
Holographic Entanglement Entropy
•
Generalization of Hawking -Bekenstein BH entropy ..•
Idea of Holography ==> unit entanglement per Planck area•
To understand the basic mechanism it may be useful to know what does a suitable measure of information in CFT corresponds to that in the AdS ..•
After 20 years of the first proposal by Maldacena … We are still unaware of the basic mechanism behind AdS/CFT conjecture•
That means can we aim to demystify this duality beyond known examples …•
One such quantitative suitable measure happens to be : the entanglemententropy or von-Neumann entropy …
Holographic Entanglement Entropy
Interesting interpretation of HEE : “Spacetime in Gravity is built out of collections of bits of quantum entanglement”
•
Generalization of Hawking -Bekenstein BH entropy ..•
Idea of Holography ==> unit entanglement per Planck area•
To understand the basic mechanism it may be useful to know what does a suitable measure of information in CFT corresponds to that in the AdS ..Interesting interpretation of HEE : “Spacetime in Gravity is built out of collections of bits of quantum entanglement”
Interesting interpretation of HEE : “Spacetime in Gravity is built out of collections of bits of quantum entanglement”
Tensor Networks are an useful and explicit framework to realize this interpretation
Interesting interpretation of HEE : “Spacetime in Gravity is built out of collections of bits of quantum entanglement”
Tensor Networks are an useful and explicit framework to realize this interpretation
AdS/CFT or more generally holography explicitly realizes
the geometrization of dynamics in CFT’s
Tensor networks uses quantum entanglement to represents
the geometrization of quantum states in many body system
The HEE suggests the following novel interpretation: ``A spacetime in gravity = Collections of bits of quantum entanglement’’
B
A
A
Planck length ⇒ Manifestly realized in the proposed connection between AdS/CFT and tensor networks ! [Swingle 09] This entanglement/geometry duality may work for any surfaces [Bianchi‐Myers 12]AdS/CFT ( or more generally holography)
⇒ ``Geometrization’’ of Dynamics in QFTs
One important fact for this to work is
Quantum entanglement
⇒ ``Geometry’’ of Quantum States in many-body system
① Introduction
Emergent spacetime from Entanglement
Algebraically complicated !
=
Tensor Network
Quantum States
AdS
The HEE suggests that
A spacetime in gravity
= Collections of qubits of quantum entanglement
B
A
γ
A
Planck length
A useful explicit framework for this is the tensor network.
(4-1) Tensor Network [See e.g. Cirac-Verstraete 09(review)]
Tensor network states
= Efficient variational ansatz for the ground state
wave functions in quantum many-body systems.
[A tensor network diagram = A wave function]
⇒ An ansatz should respect the correct
quantum entanglement of ground state. ~Geometry of Tensor Network
④ Tensor Network and Emergent Spacetime
The HEE suggests the following novel interpretation: ``A spacetime in gravity = Collections of bits of quantum entanglement’’
B
A
A
Planck length ⇒ Manifestly realized in the proposed connection between AdS/CFT and tensor networks ! [Swingle 09] This entanglement/geometry duality may work for any surfaces [Bianchi‐Myers 12]AdS/CFT ( or more generally holography)
⇒ ``Geometrization’’ of Dynamics in QFTs One important fact for this to work is
Quantum entanglement
⇒ ``Geometry’’ of Quantum States in many-body system
① Introduction
Emergent spacetime from Entanglement
Algebraically complicated !
=
Tensor Network
Quantum States
AdS
The HEE suggests that
A spacetime in gravity
= Collections of qubits of quantum entanglement
B
A
γ
A
Planck length
A useful explicit framework for this is the tensor network.
(4-1) Tensor Network [See e.g. Cirac-Verstraete 09(review)] Tensor network states
= Efficient variational ansatz for the ground state
wave functions in quantum many-body systems. [A tensor network diagram = A wave function]
⇒ An ansatz should respect the correct
quantum entanglement of ground state. ~Geometry of Tensor Network
④ Tensor Network and Emergent Spacetime
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of ) Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of ) Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA and time slice of AdS and Entanglement entropy
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of ) Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of ) Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of ) Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA and time slice of AdS and Entanglement entropy
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Tensor Network : An efficient variational ansatz for the
ground state wave functions in quantum many body systems
The HEE suggests the following novel interpretation:
``
A spacetime in gravity
= Collections of bits of quantum entanglement’’
B
A
A
Planck length
⇒ Manifestly realized in the proposed connection
between AdS/CFT and tensor networks !
[Swingle 09]
This entanglement/geometry
duality may work
for any surfaces
[Bianchi‐Myers 12]
AdS/CFT ( or more generally holography)
⇒
``Geometrization’’ of Dynamics in QFTs
One important fact for this to work is
Quantum entanglement
⇒
``Geometry’’ of Quantum States in many-body system
① Introduction
Emergent spacetime from Entanglement
Algebraically complicated !
=
Tensor Network
Quantum States
AdS
The HEE suggests that
A spacetime in gravity
= Collections of qubits of quantum entanglement
B
A
γ
A
Planck length
A useful explicit framework for this is the
tensor network
.
(4-1) Tensor Network
[See e.g. Cirac-Verstraete 09(review)]
Tensor network states
=
Efficient variational ansatz for the ground state
wave functions in quantum many-body systems.
[A tensor network diagram = A wave function]
⇒ An ansatz should respect the correct
quantum entanglement
of ground state.
~
Geometry of Tensor Network
④ Tensor Network and Emergent Spacetime
A Tensor Network diagram = A many-body wave-fn
Requirement : A good ansatz should reproduce the correct quantum entanglement of the ground state wave-fn
A network of tensors that geometrizes quantum entanglement
Tensor networks
Multi-scale Entanglement Renormalization Anzatz(MERA)
Matrix Product State(MPS), Density, Matrix Renormalization Group(DMRG)
Efficient method to produce ground state of given Hamiltonian.
• Real space, variational method: Using Hamiltonian.
• No sign problem: No Monte-Carlo method unlike lattice
gauge theory.
• Efficiency: Computable on usual computers
• Gapped systems
• Critical or gapped systems
Examples
Efficiency of Tensor Network Descriptions
Generic States : Exponential … Polynomial !! Efficient !! # of Parameters MPS : Tensors MERA : Tensors Polynomial !! Efficient !! Sites
Tensor Network (TN) [See e.g. the review Cirac-Verstraete 09]
Recent remarkable progresses in numerical algorithms for quantum lattice models:
⇒ Tensor network states
= Efficient variational ansatz for the ground state
wave functions in quantum many-body systems.
⇒ An ansatz is good if it respects the quantum entanglement of the true ground state.
Quantum wave function ⇔ Geometry of network of entanglement
Tensor Networks …
(4-1) Tensor Network [See e.g. Cirac-Verstraete 09(review)]
Tensor network states
= Efficient variational ansatz for the ground state
wave functions in quantum many-body systems.
[A tensor network diagram = A wave function]
⇒ An ansatz should respect the correct
quantum entanglement of ground state. ~Geometry of Tensor Network
④ Tensor Network and Emergent Spacetime
Tensor Network (TN) [See e.g. the review Cirac-Verstraete 09]
Recent remarkable progresses in numerical algorithms for quantum lattice models:
⇒ Tensor network states
= Efficient variational ansatz for the ground state
wave functions in quantum many-body systems.
⇒ An ansatz is good if it respects the quantum entanglement of the true ground state.
Quantum wave function ⇔ Geometry of network of entanglement
Tensor Networks …
(4-1) Tensor Network [See e.g. Cirac-Verstraete 09(review)]
Tensor network states
= Efficient variational ansatz for the ground state
wave functions in quantum many-body systems.
[A tensor network diagram = A wave function]
⇒ An ansatz should respect the correct
quantum entanglement of ground state. ~Geometry of Tensor Network
④ Tensor Network and Emergent Spacetime
(4-2) MERA
MERA (Multiscale Entanglement Renormalization Ansatz):
⇒ An efficient variational ansatz for CFT ground states.
[Vidal 05]
To increase entanglement in a CFT, we add (dis)entanglers.
1
σ σ2 σ3 σ4 σ5 σ6 σ7 σ8 σ1 σ2 σ3 σ4 σ5 σ6 σ7 σ8
Unitary transf. between 2 spins
Ex. Matrix Product State (MPS) [DMRG: White 92,…, Rommer-Ostlund 95,..] α β σ Mαβ (σ) 1
σ
σ
2σ
n 1 α α2 α3 αn Spins n n n n M M M 2 1 , , , 2 1) ( ) ( )] , , , ( Tr[ 2 1 σ σ σ σ σ σ σ σ σ∑
= Ψ . or , 1,2,..., ↓ =↑ = i i σ χ α Spin chainTensor Network (TN) [See e.g. the review Cirac-Verstraete 09]
Recent remarkable progresses in numerical algorithms for quantum lattice models:
⇒ Tensor network states
= Efficient variational ansatz for the ground state
wave functions in quantum many-body systems.
⇒ An ansatz is good if it respects the quantum entanglement of the true ground state.
Quantum wave function ⇔ Geometry of network of entanglement
Tensor Networks …
(4-1) Tensor Network [See e.g. Cirac-Verstraete 09(review)]
Tensor network states
= Efficient variational ansatz for the ground state
wave functions in quantum many-body systems. [A tensor network diagram = A wave function]
⇒ An ansatz should respect the correct
quantum entanglement of ground state. ~Geometry of Tensor Network
④ Tensor Network and Emergent Spacetime
2 type of tensors ==>
Polynomial parameters ==> efficient
Tensor Network (TN) [See e.g. the review Cirac-Verstraete 09]
Recent remarkable progresses in numerical algorithms for quantum lattice models:
⇒ Tensor network states
= Efficient variational ansatz for the ground state
wave functions in quantum many-body systems.
⇒ An ansatz is good if it respects the quantum entanglement of the true ground state.
Quantum wave function ⇔ Geometry of network of entanglement
Tensor Networks …
(4-1) Tensor Network [See e.g. Cirac-Verstraete 09(review)] Tensor network states
= Efficient variational ansatz for the ground state
wave functions in quantum many-body systems.
[A tensor network diagram = A wave function]
⇒ An ansatz should respect the correct
quantum entanglement of ground state. ~Geometry of Tensor Network
④ Tensor Network and Emergent Spacetime
Polynomial parameters ==> efficient
Tensor Network (TN) [See e.g. the review Cirac-Verstraete 09]
Recent remarkable progresses in numerical algorithms for quantum lattice models:
⇒ Tensor network states
= Efficient variational ansatz for the ground state wave functions in quantum many-body systems.
⇒ An ansatz is good if it respects the quantum entanglement of the true ground state.
Quantum wave function ⇔ Geometry of network of entanglement
Tensor Networks …
(4-1) Tensor Network [See e.g. Cirac-Verstraete 09(review)]
Tensor network states
= Efficient variational ansatz for the ground state
wave functions in quantum many-body systems.
[A tensor network diagram = A wave function]
⇒ An ansatz should respect the correct
quantum entanglement of ground state. ~Geometry of Tensor Network
④ Tensor Network and Emergent Spacetime
Exponential parameters ==> inefficient
Efficiency of Tensor Network Descriptions
Generic States : Exponential … Polynomial !! Efficient !! # of Parameters MPS : Tensors MERA : Tensors Polynomial !! Efficient !! Sites
Tensor Network Descriptions of Quantum States
Optimization of the Energy
with
Ex: MPS (Matrix Product States)
good for gapped systems
Efficient variational ansatz for ground state wave functions :
Tensor Networks
A “Geometrization” of Quamtum States
( respect the correct Quantum Entanglement )
[DMRG: White 92,…, Rommer-Ostlund 95,… ]
Tensor Networks …
(4-2) MERA
MERA (Multiscale Entanglement Renormalization Ansatz):
⇒ An efficient variational ansatz for CFT ground states.
[Vidal 05]
To increase entanglement in a CFT, we add (dis)entanglers.
1
σ σ2 σ3 σ4 σ5 σ6 σ7 σ8 σ1 σ2 σ3 σ4 σ5 σ6 σ7 σ8
Unitary transf. between 2 spins
Ex. Matrix Product State (MPS) [DMRG: White 92,…, Rommer-Ostlund 95,..] α β σ Mαβ(σ) 1 σ σ2 σn 1 α α2 α3 αn Spins n n n n M M M 2 1 , , , 2 1) ( ) ( )] , , , ( Tr[ 2 1 σ σ σ σ σ σ σ σ σ ∑ = Ψ . or , 1,2,..., ↓ =↑ = i i σ χ α Spin chain
Effective variational ansatz for ground state wave-fn : Tensor Networks that Optimizes the energy
Efficiency of Tensor Network Descriptions
Generic States : Exponential … Polynomial !! Efficient !! # of Parameters MPS : Tensors MERA : Tensors Polynomial !! Efficient !! Sites
Tensor Network Descriptions of Quantum States
Optimization of the Energy
with
Ex: MPS (Matrix Product States)
good for gapped systems
Efficient variational ansatz for ground state wave functions : Tensor Networks
A “Geometrization” of Quamtum States
( respect the correct Quantum Entanglement )
[DMRG: White 92,…, Rommer-Ostlund 95,… ]
Tensor Networks …
(4-2) MERA
MERA
(Multiscale Entanglement Renormalization Ansatz):⇒ An efficient variational ansatz for CFT ground states.
[Vidal 05]
To increase entanglement in a CFT, we add
(dis)entanglers
.
1
σ σ 2 σ3 σ 4 σ5 σ6 σ7 σ8 σ1 σ 2 σ3 σ 4 σ5 σ6 σ7 σ8
Unitary transf. between 2 spins
Ex. Matrix Product State (MPS) [DMRG: White 92,…, Rommer-Ostlund 95,..] α β σ Mαβ (σ) 1 σ σ2 σn 1 α α2 α3 αn Spins n n n n M M M 2 1 , , , 2 1) ( ) ( )] , , , ( Tr[ 2 1 σ σ σ σ σ σ σ σ σ ∑ = Ψ . or , 1,2,..., ↓ =↑ = i i σ χ α Spin chain
MERA Network
- Isometry MERA (Multi-scale Entanglement Renormalization Ansatz) network
- 2 kinds of Tensors : For CFT ground states, a good TN is ….
- (Dis-) Entangler
Coarse-graining IR (Less Entangled)
UV (More Entangled)
Add (Remove) Entanglement
- Layer by layer RG transformation
Lattice scale changes exponentially
Entanglement Renormalization
[Vidal 05, 06]
MERA Network
- Isometry MERA (Multi-scale Entanglement Renormalization Ansatz) network
- 2 kinds of Tensors : For CFT ground states, a good TN is ….
- (Dis-) Entangler
Coarse-graining IR (Less Entangled)
UV (More Entangled)
Add (Remove) Entanglement
- Layer by layer RG transformation
Lattice scale changes exponentially
Entanglement Renormalization [Vidal 05, 06]
MERA Network
- Isometry MERA (Multi-scale Entanglement Renormalization Ansatz) network
- 2 kinds of Tensors : For CFT ground states, a good TN is ….
- (Dis-) Entangler
Coarse-graining IR (Less Entangled)
UV (More Entangled)
Add (Remove) Entanglement
- Layer by layer RG transformation
Lattice scale changes exponentially
Entanglement Renormalization
[Vidal 05, 06]
MERA Network
- Isometry MERA (Multi-scale Entanglement Renormalization Ansatz) network
- 2 kinds of Tensors : For CFT ground states, a good TN is ….
- (Dis-) Entangler
Coarse-graining IR (Less Entangled)
UV (More Entangled)
Add (Remove) Entanglement
- Layer by layer RG transformation
Lattice scale changes exponentially
Entanglement Renormalization [Vidal 05, 06]
Multiscale Entanglement Renormalization Ansatz (MERA)
MERA Network
- Isometry MERA (Multi-scale Entanglement Renormalization Ansatz) network
- 2 kinds of Tensors : For CFT ground states, a good TN is ….
- (Dis-) Entangler
Coarse-graining IR (Less Entangled)
UV (More Entangled)
Add (Remove) Entanglement
- Layer by layer RG transformation
Lattice scale changes exponentially
Entanglement Renormalization [Vidal 05, 06]
MERA and time slice of AdS and Entanglement entropy
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA and time slice of AdS and Entanglement entropy
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA and time slice of AdS and Entanglement entropy
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA and time slice of AdS and Entanglement entropy
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
MERA, Hyperbolic Space and Holographic EE
: size of
steps (a time slice of )
Hyperbolic space
Holographic EE saturates this bound
Entanglement Renormalization
AdS/CFT and tensor network
⇥ log
number of bonds of surface γ
Significant similarity between geometric expression of entanglement entropy in AdS/CFT and MERA.
SA =
Area( ) 4GN
Ryu-Takayanagi formula: Entanglement entropy is given by area of minimal surface γ devided by Newton constant.
MERA:
Entanglement entropy is bounded by number of bond of minimal surface γ. This bound is often approximately saturated.
A A
SA
What we learn from tensor networks :
What we learn from tensor networks :
MERA geometry is equal to time slice of AdS
What we learn from tensor networks :
MERA geometry is equal to time slice of AdS
It sort of qualitatively works … but several doubts remains … Evidences :
1) A real-space RG structure of MERA :: The extra radial direction and evolution in bulk AdS
2) If the entanglement bound in MERA saturates :: it coincides with the HEE
What we learn from tensor networks :
MERA geometry is equal to time slice of AdS
It sort of qualitatively works … but several doubts remains … Evidences :
1) A real-space RG structure of MERA :: The extra radial direction and evolution in bulk AdS
2) If the entanglement bound in MERA saturates :: it coincides with the HEE
Doubts :
1) Is it possible to explore locality beyond the AdS scale ..
2) Why do we need to have saturation of entanglement entropy ..
3) Is the full conformal symmetry structure realizable ?
4) Ultimately we want to go beyond the lattice formulation and see the AdS in true continuum sense ..
What we learn from tensor networks :
MERA geometry is equal to time slice of AdS
It sort of qualitatively works … but several doubts remains … Evidences :
1) A real-space RG structure of MERA :: The extra radial direction and evolution in bulk AdS
2) If the entanglement bound in MERA saturates :: it coincides with the HEE
Doubts :
1) Is it possible to explore locality beyond the AdS scale ..
2) Why do we need to have saturation of entanglement entropy ..
3) Is the full conformal symmetry structure realizable ?
4) Ultimately we want to go beyond the lattice formulation and see the AdS in true continuum sense ..
We propose an alternative method using OPTIMIZATION of
An “OPTIMIZATION’’ of the Euclidean Path Integral ( ˜'(x)) = Z Y x Y ✏<z<1 D'(x, z) ! e SCF T(') Y x ('(✏, x) '(x))˜
An “OPTIMIZATION’’ of the Euclidean Path Integral ( ˜'(x)) = Z Y x Y ✏<z<1 D'(x, z) ! e SCF T(') Y x ('(✏, x) '(x))˜
The ground state
wave-fn as a functional of the UV boundary
value of the fields in the CFT
The measure of the path-integration over the fields in the CFT
The CFT action
The delta-fn ensuring the boundary value for the field at UV
An “OPTIMIZATION’’ of the Euclidean Path Integral x z= 0 0 0 0 0 0 0 0 0 0 Hyperbolic Space z Flat Space
Euclidean path Integral can be rewritten as MERA
network but we would like to get rid of the possible lattice artifact
Another motivation: Tensor network renormalization
[Evenbly-Vidal 14, 15]
We can rewrite the Euclidean path-integral into a MERA
tensor network.
Picture taken from
Evenbly-Vidal
arXiv:1502.05385
In this talk, we would like to have a general argument
which does not assume any particular tensor networks to
avoid any lattice artifacts.
( ˜'(x)) = Z Y x Y ✏<z<1 D'(x, z) ! e SCF T(') Y x ('(✏, x) '(x))˜
Optimization of Path-Integral
Tensor Network
Renormalization(TNR) = Optimization of TN
Optimization of
Path-integral
Euclidean
Time (-z)
Space (x)
Hyperbolic Space = Time slice of AdS3
ε
Lattice Constant
[Evenbly-Vidal 14, 15] [Miyaji-Watanabe-TT 16]
⑤ Optimization of Path-Integral and AdS/CFT
[Miyaji-Watanabe-TT 2016 + work in progress]
How to find the ground state of a given quantum many-body system ?
⇒(1)Variational Method: First constrain the possible quantum states and next minimize its energy.
Ex. Look at behaviors of quantum entanglement
(e.g. Area law) → The idea of tensor networks
⇒(2)Path-integral: Path-integrate for a long time in
Euclidean theory:
lim
β→∞e
−βHψ
=
ψ
0⑤ Optimization of Path-Integral and AdS/CFT
[Miyaji-Watanabe-TT 2016 + work in progress]
How to find the ground state of a given quantum many-body system ?
⇒(1)Variational Method: First constrain the possible quantum states and next minimize its energy.
Ex. Look at behaviors of quantum entanglement
(e.g. Area law) → The idea of tensor networks
⇒(2)Path-integral: Path-integrate for a long time in
Euclidean theory:
lim
β→∞e
−βHψ
=
ψ
0⑤ Optimization of Path-Integral and AdS/CFT
[Miyaji-Watanabe-TT 2016 + work in progress]
How to find the ground state of a given quantum many-body system ?
⇒(1)Variational Method: First constrain the possible quantum states and next minimize its energy.
Ex. Look at behaviors of quantum entanglement
(e.g. Area law) → The idea of tensor networks
⇒(2)Path-integral: Path-integrate for a long time in
Euclidean theory:
lim
β→∞e
−βHψ
=
ψ
0Tensor Network Renormalization (TNR) : OPTIMIZATION of Tensor Networks
( ˜'(x)) =Z Y x Y ✏<z<1 D'(x, z) ! e SCF T(')Y x ('(✏, x) '(x))˜
Discretized Path-integral and Its Optimization
Time
τ
Space
x
[
(
)
]
(
,
)
( )(
(
)
(
,
0
)
)
0=
−
⋅
=
Ψ
− ∞ < < ∞ −∞< < −∫ ∏
φ
τ
δ
ϕ
φ
τ
ϕ
φ τx
x
e
x
D
x
SCFT x[
ϕ
(x
)
]
Ψ
τ=0
τ=-∞
Optimize
High momentum modes
k>1/τ is not important !
= AdS !
(hyperbolic space)[
(
)
]
Minϕ
x
Ψ
(4-3) AdS/MERA duality ? [Swingle 09]
0
=
u
u
=
−
1
A
A
A
Min[Area]
Aγ
. , ) ( 2 2 2 2 2 2 2 2 2 u where z e u z x d dt dz x d dt e ds + − + = − + = ε ⋅ − ε 2 + dAdS
1 + dCFT
Aγ
Links]
Min[#
? Equivalent ) ( uIR u = −∞ ≡ The idea: Tensor Network of MERA (∃scale inv.) = a time slice of AdS space Qualitative evidences (i) Real space RG = radial evolution in AdS → something we usually expect in AdS/CFT. (ii) The bound of EE in MERA: If this is saturated, it seems to agree with the HEE . log ] Min[ Int A N S . ) Area( Min 4 1 A A N A G S(4-3) AdS/MERA duality ? [Swingle 09]
0 = u u = −1 A A A
Min[Area]
Aγ
. , ) ( 2 2 2 2 2 2 2 2 2 u where z e u z x d dt dz x d dt e ds + − + = − + = ε ⋅ − ε 2 + dAdS
1 + dCFT
Aγ
Links]
Min[#
? Equivalent ) ( uIR u = −∞ ≡(4-3) AdS/MERA duality ? [Swingle 09]
0
=
u
u
=
−
1
A
A
A
Min[Area]
Aγ
. , ) ( 2 2 2 2 2 2 2 2 2 u where z e u z x d dt dz x d dt e ds + − + = − + = ε ⋅ − ε 2 + dAdS
1 + dCFT
Aγ
Links]
Min[#
? Equivalent ) ( uIR u = −∞ ≡ The idea: Tensor Network of MERA (∃scale inv.) = a time slice of AdS space Qualitative evidences (i) Real space RG = radial evolution in AdS → something we usually expect in AdS/CFT. (ii) The bound of EE in MERA: If this is saturated, it seems to agree with the HEE . log ] Min[ Int A N S . ) Area( Min 4 1 A A N A G S(4-3) AdS/MERA duality ? [Swingle 09]
0 = u u = −1 A A A
Min[Area]
Aγ
. , ) ( 2 2 2 2 2 2 2 2 2 u where z e u z x d dt dz x d dt e ds + − + = − + = ε ⋅ − ε 2 + dAdS
1 + dCFT
Aγ
Links]
Min[#
? Equivalent ) ( uIR u = −∞ ≡(4-3) AdS/MERA duality ? [Swingle 09]
0
=
u
u
=
−
1
A
A
A
Min[Area]
Aγ
. , ) ( 2 2 2 2 2 2 2 2 2 u where z e u z x d dt dz x d dt e ds + − + = − + = ε ⋅ − ε 2 + dAdS
1 + dCFT
Aγ
Links]
Min[#
? Equivalent ) ( uIR u = −∞ ≡ The idea: Tensor Network of MERA (∃scale inv.) = a time slice of AdS space Qualitative evidences (i) Real space RG = radial evolution in AdS → something we usually expect in AdS/CFT. (ii) The bound of EE in MERA: If this is saturated, it seems to agree with the HEE . log ] Min[ Int A N S . ) Area( Min 4 1 A A N A G S(4-3) AdS/MERA duality ? [Swingle 09]
0 = u u = −1 A A A
Min[Area]
Aγ
. , ) ( 2 2 2 2 2 2 2 2 2 u where z e u z x d dt dz x d dt e ds + − + = − + = ε ⋅ − ε 2 + dAdS
1 + dCFT
Aγ
Links]
Min[#
? Equivalent ) ( uIR u = −∞ ≡Path-Integral Wave-fn : it’s Discretization and Optimization
(Miyaji, Watanabe, Takayanagi-2016)
Another way of looking at it : Free scalar Field
Euclidean Path-Integral for Ground State
CFT2 on
EOM & regular at Free Scalar
At fixed ,only modes with
contribute in the -integral : UV cutoff x z= 0 0 0 0 0 0 0 0 0 0 Hyperbolic Space z Flat Space (Miyaji, Watanabe, Takayanagi-2016)
AdS/CFT and tensor network
⇥ log
number of bonds of surface γ
Significant similarity between geometric expression of entanglement entropy in AdS/CFT and MERA.
S
A=
Area( )
4G
NRyu-Takayanagi formula: Entanglement entropy is
given by area of minimal
surface γ devided by
Newton constant.
MERA:
Entanglement entropy is bounded by number of bond of minimal
surface γ. This bound is often approximately saturated.
A
A
S
A
[Evenbly, Vidal(2014)]
AdS/CFT and tensor network
⇥ log
number of bonds of surface γ
Significant similarity between geometric expression of entanglement entropy in AdS/CFT and MERA.
S
A=
Area( )
4G
NRyu-Takayanagi formula: Entanglement entropy is
given by area of minimal
surface γ devided by
Newton constant.
MERA:
Entanglement entropy is bounded by number of bond of minimal
surface γ. This bound is often approximately saturated.
A
A
S
A
[Evenbly, Vidal(2014)]
AdS/CFT and tensor network
⇥ log
number of bonds of surface γ
Significant similarity between geometric expression
of entanglement entropy in AdS/CFT and MERA.
S
A=
Area( )
4G
NRyu-Takayanagi formula:
Entanglement entropy is
given by
area of minimal
surface γ
devided by
Newton constant.
MERA:
Entanglement entropy is bounded
by
number of bond of minimal
surface γ
. This bound is often
approximately saturated.
A
A
S
A
[Evenbly, Vidal(2014)]
x z= 0 0 0 0 0 0 0 0 0 0 Hyperbolic Space z Flat Space
What we learn so far:
In the language of Tensor networks ==> Eliminating extra tensors in TN is creating most efficient TN
==> The algorithm at work for this, is the “OPTIMIZATION” of TN for a given state. For free fields ==> Introducing momentum dependent cut-off.
x z= 0 0 0 0 0 0 0 0 0 0 Hyperbolic Space z Flat Space
What we learn so far:
In the language of Tensor networks ==> Eliminating extra tensors in TN is creating most efficient TN
==> The algorithm at work for this, is the “OPTIMIZATION” of TN for a given state. For free fields ==> Introducing momentum dependent cut-off.
Our first Key Insight :
Rearrangement of the lattice-structure for the tensors in TN, can be engineered by changing the background metric in the Euclidean Path Integral, keeping the boundary condition for the fields at UV unchanged.
x z= 0 0 0 0 0 0 0 0 0 0 Hyperbolic Space z Flat Space
What we learn so far:
In the language of Tensor networks ==> Eliminating extra tensors in TN is creating most efficient TN
==> The algorithm at work for this, is the “OPTIMIZATION” of TN for a given state. For free fields ==> Introducing momentum dependent cut-off.
Our first Key Insight :
Rearrangement of the lattice-structure for the tensors in TN, can be engineered by changing the background metric in the Euclidean Path Integral, keeping the boundary condition for the fields at UV unchanged. What we are still lacking :
What is the counterpart of the OPTIMIZATION algorithm in TN, that will be the guiding principle to determine the changed (optimized)
This takes us to the “COMPLEXITY” part of our story .. Recently an interesting question, that people are after : How to define complexity of a state in CFT’s ..
This takes us to the “COMPLEXITY” part of our story .. Recently an interesting question, that people are after : How to define complexity of a state in CFT’s ..
Computational complexity of a quantum state = Min. no. of quantum gates required to
prepare it from a given “simple” reference state
This takes us to the “COMPLEXITY” part of our story .. Recently an interesting question, that people are after : How to define complexity of a state in CFT’s ..
Computational complexity of a quantum state = Min. no. of quantum gates required to
prepare it from a given “simple” reference state
= Min. no. of tensors in the TN description .
In the literature holographic formulas for computing complexity is already proposed :
1) Complexity = Max. Volume in AdS (Standford-Susskind .. )
2) Complexity = Gravity action in WDW patch of AdS (Brown, Roberts, Susskind, Myers …)
“PATH-INTEGRAL COMPLEXITY”
= We define the CFT analogue of the Complexity
This motivates us to consider its QFT counterpart
We introduce
``Path-integral Complexity’’
.
Our guiding principle 2
・Lattice structure (= arrangement of tensors) in TN
Background metric g
abin Euclid path-integral
・Optimization of TN for a state Ψ
Minimizing Path-integral Complexity
w.r.t the metric
]
[
g
abI
Ψ abg
This motivates us to consider its QFT counterpart We introduce ``Path-integral Complexity’’ .
Our guiding principle 2
・Lattice structure (= arrangement of tensors) in TN Background metric gab in Euclid path-integral
・Optimization of TN for a state Ψ
Minimizing Path-integral Complexity w.r.t the metric
]
[
g
abI
Ψ abg
“PATH-INTEGRAL COMPLEXITY”
= We define the CFT analogue of the Complexity
This motivates us to consider its QFT counterpart
We introduce
``Path-integral Complexity’’
.
Our guiding principle 2
・Lattice structure (= arrangement of tensors) in TN
Background metric g
abin Euclid path-integral
・Optimization of TN for a state Ψ
Minimizing Path-integral Complexity
w.r.t the metric
]
[
g
abI
Ψ abg
This motivates us to consider its QFT counterpart We introduce ``Path-integral Complexity’’ .
Our guiding principle 2
・Lattice structure (= arrangement of tensors) in TN Background metric gab in Euclid path-integral
・Optimization of TN for a state Ψ
Minimizing Path-integral Complexity w.r.t the metric
]
[
g
abI
Ψ abg
x z= 0 0 0 0 0 0 0 0 0 0 Hyperbolic Space z Flat SpaceOur first Key Insight :
Rearrangement of the lattice-structure for the tensors in TN, can be engineered by changing the background metric in the Euclidean Path Integral, keeping the
“PATH-INTEGRAL COMPLEXITY”
= We define the CFT analogue of the Complexity
This motivates us to consider its QFT counterpart
We introduce
``Path-integral Complexity’’
.
Our guiding principle 2
・Lattice structure (= arrangement of tensors) in TN
Background metric g
abin Euclid path-integral
・Optimization of TN for a state Ψ
Minimizing Path-integral Complexity
w.r.t the metric
]
[
g
abI
Ψ abg
This motivates us to consider its QFT counterpart We introduce ``Path-integral Complexity’’ .
Our guiding principle 2
・Lattice structure (= arrangement of tensors) in TN Background metric gab in Euclid path-integral
・Optimization of TN for a state Ψ
Minimizing Path-integral Complexity w.r.t the metric
]
[
g
abI
Ψ abg
x z= 0 0 0 0 0 0 0 0 0 0 Hyperbolic Space z Flat SpaceOur first Key Insight :
Rearrangement of the lattice-structure for the tensors in TN, can be engineered by changing the background metric in the Euclidean Path Integral, keeping the
boundary condition for the fields at UV unchanged. Our second Key Insight (This is our proposal) :
Optimization of TN for a given state is equivalent to Minimizing the path-integral complexity with respect to the back-ground metric.
“PATH-INTEGRAL COMPLEXITY”
= We define the CFT analogue of the Complexity
This motivates us to consider its QFT counterpart
We introduce
``Path-integral Complexity’’
.
Our guiding principle 2
・Lattice structure (= arrangement of tensors) in TN
Background metric g
abin Euclid path-integral
・Optimization of TN for a state Ψ
Minimizing Path-integral Complexity
w.r.t the metric
]
[
g
abI
Ψ abg
This motivates us to consider its QFT counterpart We introduce ``Path-integral Complexity’’ .
Our guiding principle 2
・Lattice structure (= arrangement of tensors) in TN Background metric gab in Euclid path-integral
・Optimization of TN for a state Ψ
Minimizing Path-integral Complexity w.r.t the metric
]
[
g
abI
Ψ abg
x z= 0 0 0 0 0 0 0 0 0 0 Hyperbolic Space z Flat SpaceOur first Key Insight :
Rearrangement of the lattice-structure for the tensors in TN, can be engineered by changing the background metric in the Euclidean Path Integral, keeping the
boundary condition for the fields at UV unchanged. Our second Key Insight (This is our proposal) :
Optimization of TN for a given state is equivalent to Minimizing the path-integral complexity with respect to the back-ground metric.
The final crucial element :
Do we have a good answer for how to define the path-integral complexity given one CFT ..
FOR a 2D Euclidean CFT We have a very good answer !! In a 2D CFT with coordinates z (Euclidean time), x (space) we can always write any metric in the conformally flat form
② AdS from Optimization of Path-Integrals
(2-1) Formulation
A Basic Rule: Simplify a path-integral s.t.
it produces the correct UV wave functional.
Consider 2D CFTs for simplicity. ( z=- Euclidean time, x=space)
Deformation of discretizations in path-integral
= Curved metric such that one cell (bit) = unit length.
Note: The original flat metric is given by (ε is UV cutoff):
).
(
2 2 ) , ( 2 2e
dx
dz
ds
=
φ x z+
). ( 2 2 2 2 dx dz ds =ε
− ⋅ +Remember, the unoptimized metric is flat space (with UV cutoff given by “epsilon”)…
② AdS from Optimization of Path-Integrals
(2-1) Formulation
A Basic Rule: Simplify a path-integral s.t.
it produces the correct UV wave functional.
Consider 2D CFTs for simplicity. ( z=- Euclidean time, x=space)
Deformation of discretizations in path-integral
= Curved metric such that one cell (bit) = unit length.
Note: The original flat metric is given by (ε is UV cutoff):
).
(
2 2 ) , ( 2 2e
dx
dz
ds
=
φ x z+
). ( 2 2 2 2 dx dz ds =ε
− ⋅ +Optimization of Path-Integral
Tensor Network
Renormalization(TNR) = Optimization of TN
Optimization of
Path-integral
Euclidean
Time (-z)
Space (x)
Hyperbolic Space = Time slice of AdS3
ε
Lattice Constant
[Evenbly-Vidal 14, 15] [Miyaji-Watanabe-TT 16]
A Sketch: Optimization of Path-Integral
Time
-z
Space x[
( )]
Flat x UV Φ Ψz=0
z=∞
Optimize UV modes k>1/z are not important != Hyperbolic Space H2 2 2 2 2 z dz dx ds = +
∝
ΨUVg e2[
Φ(x)]
= φ 0 0 0 0 0 0 0 0 0 0 MERAA Sketch: Optimization of Path-Integral
Time
-z
Spacex
[
( )]
Flat x UV Φ Ψz=0
z=∞
Optimize UV modes k>1/z are not important !=
Hyperbolic Space H2 2 2 2 2 z dz dx ds = +∝
ΨUVg e2[
Φ(x)]
= φ 0 0 0 0 0 0 0 0 0 0 MERAA Sketch: Optimization of Path-Integral
Time
-z
Spacex
[
( )]
Flat x UV Φ Ψz=0
z=∞
Optimize UV modes k>1/z are not important !=
Hyperbolic Space H2 2 2 2 2 z dz dx ds = +∝
ΨUVg e2[
Φ(x)]
= φ 0 0 0 0 0 0 0 0 0 0 MERAA Sketch: Optimization of Path-Integral
Time
-z
Spacex
[
( )]
Flat x UV Φ Ψz=0
z=∞
Optimize UV modes k>1/z are not important !=
Hyperbolic Space H2 2 2 2 2 z dz dx ds = +∝
ΨUVg e2[
Φ(x)]
= φ 0 0 0 0 0 0 0 0 0 0 MERAA Sketch: Optimization of Path-Integral
Time
-z
Spacex
[
( )]
Flat x UV Φ Ψz=0
z=∞
Optimize UV modes k>1/z are not important !=
Hyperbolic Space H2 2 2 2 2 z dz dx ds = +∝
ΨUVg e2[
Φ(x)]
= φ 0 0 0 0 0 0 0 0 0 0 MERAA Sketch: Optimization of Path-Integral
Time -z Space x [ ( )] Flat x UV Φ Ψ z=0 z=∞ Optimize UV modes k>1/z are not important !
= Hyperbolic Space H2 2 2 2 2 z dz dx ds = + ∝ ΨUVg=e2φ [Φ(x)] 0 0 0 0 0 0 0 0 0 0 MERA
In CFTs, owing to the Weyl invariance, we have
Our Proposal (Optimization of Path-integral for CFTs):
Minimize w.r.t
with the boundary condition
)]
,
(
[
x
z
I
φ
φ
( z
x
,
)
.
|
2 2 − =ε=
ε
φ ze
[
(
)
]
exp
(
[
(
,
)]
)
Flat[
(
)
]
.
2x
z
x
I
x
UV e g UVab abΦ
=
⋅
Ψ
Φ
Ψ
= φδφ
[
( )]
( , ) ( )(
( ) ( , 0))
0 = Φ − Φ ⋅ Φ = Φ Ψ − Φ ∞ < < ∞ −< <∞∫ ∏
D x z e x x z x SCFT x z g UV δThe wave functional for CFT vacuum is given by
gab(x,z): background metric
Original wf. Optimized wf.
The OPTIMIZATION should
FOR a 2D Euclidean CFT We have a very good answer !!
In a 2D CFT path-integral for the wave-fn of vacuum ==
In CFTs, owing to the Weyl invariance, we have
Our Proposal (Optimization of Path-integral for CFTs):
Minimize w.r.t
with the boundary condition
)]
,
(
[
x
z
I
φ
φ
( z
x
,
)
.
|
2 2 − =ε=
ε
φ ze
[
(
)
]
exp
(
[
(
,
)]
)
Flat[
(
)
]
.
2x
z
x
I
x
UV e g UVab abΦ
=
⋅
Ψ
Φ
Ψ
= φδφ
[
(
)
]
(
,
)
( )(
(
)
(
,
0
)
)
0=
Φ
−
Φ
⋅
Φ
=
Φ
Ψ
− Φ ∞ < < ∞ −< <∞∫ ∏
D
x
z
e
x
x
z
x
SCFT x z g UVδ
The wave functional for CFT vacuum is given by
gab(x,z): background metric
Original wf. Optimized wf.
background metric
In CFTs, owing to the Weyl invariance, we have
Our Proposal (Optimization of Path-integral for CFTs):
Minimize w.r.t
with the boundary condition
)]
,
(
[
x
z
I
φ
φ
( z
x
,
)
.
|
2 2 − =ε=
ε
φ ze
[
(
)
]
exp
(
[
(
,
)]
)
Flat[
(
)
]
.
2x
z
x
I
x
UV e g UVab abΦ
=
⋅
Ψ
Φ
Ψ
= φδφ
[
( )]
( , ) ( )(
( ) ( , 0))
0 = Φ − Φ ⋅ Φ = Φ Ψ − Φ ∞ < < ∞ −< <∞∫ ∏
D x z e x x z x SCFT x z g UV δThe wave functional for CFT vacuum is given by
gab(x,z): background metric
Original wf. Optimized wf.
FOR a 2D Euclidean CFT We have a very good answer !!
In a 2D CFT path-integral for the wave-fn of vacuum ==
In CFTs, owing to the Weyl invariance, we have
Our Proposal (Optimization of Path-integral for CFTs):
Minimize w.r.t
with the boundary condition
)]
,
(
[
x
z
I
φ
φ
( z
x
,
)
.
|
2 2 − =ε=
ε
φ ze
[
(
)
]
exp
(
[
(
,
)]
)
Flat[
(
)
]
.
2x
z
x
I
x
UV e g UVab abΦ
=
⋅
Ψ
Φ
Ψ
= φδφ
[
(
)
]
(
,
)
( )(
(
)
(
,
0
)
)
0=
Φ
−
Φ
⋅
Φ
=
Φ
Ψ
− Φ ∞ < < ∞ −< <∞∫ ∏
D
x
z
e
x
x
z
x
SCFT x z g UVδ
The wave functional for CFT vacuum is given by
gab(x,z): background metric
Original wf. Optimized wf.
background metric
In CFTs, owing to the Weyl invariance, we have
Our Proposal (Optimization of Path-integral for CFTs):
Minimize w.r.t
with the boundary condition
)]
,
(
[
x
z
I
φ
φ
( z
x
,
)
.
|
2 2 − =ε=
ε
φ ze
[
(
)
]
exp
(
[
(
,
)]
)
Flat[
(
)
]
.
2x
z
x
I
x
UV e g UVab abΦ
=
⋅
Ψ
Φ
Ψ
= φδφ
[
( )]
( , ) ( )(
( ) ( , 0))
0 = Φ − Φ ⋅ Φ = Φ Ψ − Φ ∞ < < ∞ −< <∞∫ ∏
D x z e x x z x SCFT x z g UV δThe wave functional for CFT vacuum is given by
gab(x,z): background metric
Original wf. Optimized wf.
OPTIMIZED wave-fn UN-OPTIMIZED wave-fn