27.01.2015 Views

Hierarchy of Self-Consistent Approximate Methods and the Parquet ...

Hierarchy of Self-Consistent Approximate Methods and the Parquet ...

Hierarchy of Self-Consistent Approximate Methods and the Parquet ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong><br />

<strong>Parquet</strong> Formalism<br />

Shuxiang Yang<br />

April 29, 2011<br />

1


Motivation<br />

<strong>Parquet</strong> formalism<br />

• Vertex<br />

• Scattering channels<br />

• Reducibility<br />

• Equations<br />

Outline<br />

<strong>Hierarchy</strong> <strong>of</strong> self-consistent approximate methods<br />

• Hartree-Fock approximation F = 0<br />

• Second-Order perturbation <strong>the</strong>ory (SOPT)<br />

F = U<br />

• R<strong>and</strong>om Phase approximation (RPA), T-matrix approximation <strong>and</strong> Fluctuation Exchange<br />

(FLEX) approximation Γ = U<br />

• <strong>Parquet</strong> approximation (PA)<br />

Λ = U<br />

• Multi-Scale Many-Body (MSMB) approach<br />

Λ II = Λ I,QMC<br />

Numerical results for a toy model <strong>and</strong> Hubbard model<br />

Conclusion<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 2


Motivation<br />

For <strong>the</strong> strongly-correlated electron systems, <strong>the</strong> Hubbard model is a minimum<br />

model<br />

H = −t ∑ 〈i,j〉,σ (c† i,σ c j,σ + h.c.) + U ∑ i n i,↑n i,↓<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 3


Motivation<br />

For <strong>the</strong> strongly-correlated electron systems, <strong>the</strong> Hubbard model is a minimum<br />

model<br />

H = −t ∑ 〈i,j〉,σ (c† i,σ c j,σ + h.c.) + U ∑ i n i,↑n i,↓<br />

Diagrammatic method is a popular approach to analyze this model<br />

Central question:<br />

How to sum up as many as possible physically relevant Feynman diagrams<br />

=⇒ <strong>Self</strong>-consistent diagrammatic approximate methods<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 4


Green function <strong>and</strong> self-energy<br />

(x 2 , t 2 )<br />

(x 1 , t 1 )<br />

• 1-particle Green function G<br />

ψ(x 2 , t 2 ) = ∫ G(x 2 , t 2 ; x 1 , t 1 )ψ(x 1 , t 1 ) dx 1 dt 1<br />

⊲ amplitude (phase) accumulated when a particle moving around some path<br />

⊲ long-ranged<br />

• self-energy Σ<br />

⊲ renormalization due to <strong>the</strong> interaction with o<strong>the</strong>r particles<br />

⊲ short-ranged<br />

• Similarly, we can define 2-partilce Green function <strong>and</strong> 2-particle “self-energy” (vertex).<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 5


<strong>Self</strong>-<strong>Consistent</strong> Diagrammatic Calculation<br />

• Conventional perturbation calculation (Green function G based)<br />

⊲ G = G[G 0 , U], e.g., 1st-order (Hartree)<br />

• Conventional perturbation calculation (self-energy Σ based)<br />

⊲ Σ = Σ[G 0 , U]<br />

⊲ G = G 0<br />

1−ΣG 0<br />

• <strong>Self</strong>-<strong>Consistent</strong> Diagrammatic Calculation<br />

⊲ Σ = Σ[G, U, F ]<br />

⊲ G = G 0<br />

+ + +<br />

G<br />

Σ =<br />

1−ΣG 0<br />

Σ<br />

1st iter<br />

+ +<br />

+<br />

2nd iter<br />

+ + +<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 6


Schwinger-Dyson equation<br />

Schwinger-Dyson equation: Σ = Σ[G, U,F ]<br />

• relating Σ to F , a bridge between one-particle <strong>and</strong> two-particle quantities<br />

• exact, derived from <strong>the</strong> Equation <strong>of</strong> Motion<br />

Σ<br />

=<br />

+<br />

F<br />

• Two questions about F :<br />

⊲ What is F – Vertex describing <strong>the</strong> 2-particle scattering process.<br />

⊲ How to determine F — Using parquet formalism.<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 7


Vertex<br />

v<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 8


Vertex<br />

v<br />

+<br />

initial<br />

3 1<br />

+ + 4<br />

H<br />

2<br />

P<br />

F ph<br />

h<br />

final<br />

|3, 4 >→ |1, 2 ><br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 9


Vertex<br />

3<br />

P<br />

1<br />

4<br />

initial<br />

F ph<br />

h<br />

H<br />

final<br />

2<br />

|3, 4 >→ |1, 2 ><br />

• vertex: matrix used to describe scattering process<br />

• F ph<br />

h<br />

(12; 34): <strong>the</strong> amplitude <strong>of</strong> a particle-hole pair scattered from its initial state |3, 4 > to<br />

<strong>the</strong> final state |1, 2 >.<br />

• i ≡ P i ≡ (k i , iω ni ), or i ≡ (P i , σ i , m i ) if <strong>the</strong> internal degrees <strong>of</strong> freedom are included, say<br />

<strong>the</strong> spin σ i or b<strong>and</strong> index m i .<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 10


scattering channels<br />

p-h horizontal channel<br />

3<br />

P<br />

1<br />

4<br />

initial<br />

F ph<br />

h<br />

H<br />

final<br />

2<br />

|3, 4 >→ |1, 2 ><br />

3 1<br />

P<br />

4<br />

F ph v<br />

H<br />

2<br />

|3, 1 >→ |4, 2 ><br />

p-h vertical channel<br />

initial<br />

final<br />

3<br />

P<br />

1<br />

4<br />

F pp<br />

P<br />

p-p channel<br />

2<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 11


scattering channels<br />

p-h horizontal channel<br />

3<br />

P<br />

1<br />

4<br />

initial<br />

F ph<br />

h<br />

H<br />

final<br />

2<br />

|3, 4 >→ |1, 2 ><br />

analogous to <strong>the</strong> way to<br />

arrange small pieces <strong>of</strong> parquet<br />

3 1<br />

P<br />

4<br />

F ph v<br />

H<br />

2<br />

|3, 1 >→ |4, 2 ><br />

p-h vertical channel<br />

initial<br />

final<br />

3<br />

P<br />

1<br />

4<br />

F pp<br />

P<br />

p-p channel<br />

2<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 12


1-particle reducibility<br />

reducible<br />

+<br />

⇒T<br />

irreducible<br />

+<br />

⇒<br />

Σ<br />

cutting one line !<br />

One-particle formalism<br />

1-particle<br />

re<br />

irre<br />

G T Σ<br />

Dyson equations<br />

G = G 0 + G 0 ∗ T ∗ G 0<br />

G = G 0 + G 0 ∗ Σ ∗ G<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 13


2-particle reducibility<br />

reducible<br />

+<br />

cutting two lines !<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 14


2-particle reducibility<br />

reducible<br />

+<br />

irreducible<br />

+<br />

+<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 15


2-particle reducibility<br />

reducible<br />

+<br />

irreducible<br />

+<br />

+<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 16


2-particle reducibility<br />

reducible<br />

+<br />

irreducible<br />

+<br />

+<br />

fully irreducible<br />

+<br />

+<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 17


2-particle reducibility<br />

reducible<br />

+<br />

irreducible<br />

+<br />

+<br />

⇒<br />

⇒<br />

F<br />

Γ<br />

fully irreducible<br />

+<br />

v<br />

+<br />

⇒<br />

Λ<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 18


2-particle reducibility<br />

reducible<br />

+<br />

irreducible<br />

+<br />

+<br />

⇒<br />

⇒<br />

F<br />

Γ<br />

fully irreducible<br />

+<br />

v<br />

+<br />

⇒<br />

Λ<br />

How are F , Γ <strong>and</strong> Λ related –> Be<strong>the</strong>-Salpeter equation <strong>and</strong> <strong>Parquet</strong> equation.<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 19


Be<strong>the</strong>-Salpeter equation <strong>and</strong> <strong>Parquet</strong> equation<br />

• Be<strong>the</strong>-Salpeter equation<br />

⊲ describing how a reducible vertex is decomposed in its own scattering channel<br />

⊲ exact, derived from <strong>the</strong> categorization <strong>of</strong> <strong>the</strong> Feynman diagrams<br />

F<br />

= Γ + Γ F<br />

• <strong>Parquet</strong> equation<br />

⊲ describing how a irreducible vertex is still decomposable in <strong>the</strong> o<strong>the</strong>r two scattering<br />

channels<br />

⊲ exact, derived from <strong>the</strong> categorization <strong>of</strong> <strong>the</strong> Feynman diagrams<br />

Γ<br />

Γ v<br />

= Λ + + Γ p F p<br />

F v<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 20


Be<strong>the</strong>-Salpeter equation <strong>and</strong> <strong>Parquet</strong> equation<br />

• Be<strong>the</strong>-Salpeter equation<br />

⊲ describing how a reducible vertex is decomposed in its own scattering channel<br />

⊲ exact, derived from <strong>the</strong> categorization <strong>of</strong> <strong>the</strong> Feynman diagrams<br />

F<br />

= Γ + Γ F<br />

• <strong>Parquet</strong> equation<br />

⊲ describing how a irreducible vertex is still decomposable in <strong>the</strong> o<strong>the</strong>r two scattering<br />

channels<br />

⊲ exact, derived from <strong>the</strong> categorization <strong>of</strong> <strong>the</strong> Feynman diagrams<br />

Γ<br />

Γ v<br />

= Λ + + Γ p F p<br />

F v<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 21


<strong>Parquet</strong> formalism<br />

1-particle<br />

2-particle<br />

re irre fully irre<br />

G T Σ<br />

S.D.<br />

χ F Γ Λ<br />

B.S.<br />

more localized<br />

<strong>Parquet</strong><br />

Be<strong>the</strong>-Salpeter equation <strong>and</strong> <strong>Parquet</strong> equation<br />

• relating F , Γ <strong>and</strong> Λ<br />

2-particle Dyson equations (recall Dyson e.q. for 1-particle formalism)<br />

• χ = χ 0 + χ 0 ∗ F ∗ χ 0<br />

• χ = χ 0 + χ 0 ∗ Γ ∗ χ<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 22


<strong>Hierarchy</strong> <strong>of</strong> self-consistent approximate methods<br />

1-particle<br />

2-particle<br />

F = 0<br />

G T Σ<br />

S.D.<br />

χ F Γ Λ<br />

H-F<br />

re irre fully irre<br />

B.S.<br />

more localized<br />

<strong>Parquet</strong><br />

Better approximation<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 23


Hartree-Fock approximation: F = 0<br />

• No vertex correction to <strong>the</strong> self-energy<br />

• 〈n i↑ n i↓ 〉 = 〈n i↑ 〉〈n i↓ 〉 (χ = G ∗ G)<br />

• Lowest-order correction: O(U 2 )<br />

Σ<br />

=<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 24


<strong>Hierarchy</strong> <strong>of</strong> self-consistent approximate methods<br />

1-particle<br />

2-particle<br />

F = 0<br />

F ≈ v<br />

G T Σ<br />

S.D.<br />

χ F Γ Λ<br />

H-F<br />

SOPT<br />

re irre fully irre<br />

B.S.<br />

more localized<br />

<strong>Parquet</strong><br />

Better approximation<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 25


<strong>Self</strong>-consistent Second-Order Perturbation Theory (SOPT): F ≈ v<br />

• using Schwinger-Dyson equation only<br />

• Lowest-order correction: O(U 3 )<br />

Σ =<br />

+<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 26


<strong>Hierarchy</strong> <strong>of</strong> self-consistent approximate methods<br />

1-particle<br />

2-particle<br />

F = 0<br />

F ≈ v<br />

Γ ≈ v<br />

re irre fully irre<br />

G T Σ<br />

S.D.<br />

χ F Γ Λ<br />

B.S.<br />

more localized<br />

H-F<br />

SOPT<br />

RPA, TMA, FLEX<br />

<strong>Parquet</strong><br />

Better approximation<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 27


RPA, TMA <strong>and</strong> FLEX: Γ ≈ v<br />

• using both Schwinger-Dyson <strong>and</strong> Be<strong>the</strong>-Salpeter equations<br />

• more diagrams are included<br />

⊲ ring-type: R<strong>and</strong>om Phase Approximation (RPA)<br />

⊲ ladder-type: T-matrix approximation (TMA)<br />

⊲ ring-type <strong>and</strong> ladder-type: Fluctuation Exchange approximation (FLEX)<br />

• Lowest-order correction: O(U 3 )<br />

+<br />

+<br />

+<br />

+<br />

+<br />

+<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 28


<strong>Hierarchy</strong> <strong>of</strong> self-consistent approximate methods<br />

1-particle<br />

2-particle<br />

F = 0<br />

F ≈ v<br />

Γ ≈ v<br />

Λ ≈ v<br />

re irre fully irre<br />

G T Σ<br />

S.D.<br />

χ F Γ Λ<br />

B.S.<br />

more localized<br />

<strong>Parquet</strong><br />

H-F<br />

SOPT<br />

RPA, TMA, FLEX<br />

<strong>Parquet</strong> Approximation<br />

Better approximation<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 29


<strong>Parquet</strong> approximation: Λ ≈ v<br />

• using Schwinger-Dyson, Be<strong>the</strong>-Salpeter <strong>and</strong> <strong>Parquet</strong> equations<br />

• more diagrams: ring+ladder-type <strong>and</strong> simply-crossed, are included<br />

• Lowest-order correction: O(U 5 )<br />

+ + +<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 30


<strong>Parquet</strong> approximation: Λ ≈ v<br />

• using Schwinger-Dyson, Be<strong>the</strong>-Salpeter <strong>and</strong> <strong>Parquet</strong> equations<br />

• more diagrams: simply-crossed <strong>and</strong> ring+ladder-type, are included<br />

• Lowest-order correction: O(U 5 )<br />

+ + +<br />

• still some contributions are missing: fully irreducible (complicatedly crossed) → QMC<br />

+<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 31


<strong>Hierarchy</strong> <strong>of</strong> self-consistent approximate methods<br />

1-particle<br />

2-particle<br />

F = 0<br />

F ≈ v<br />

Γ ≈ v<br />

Λ ≈ v<br />

re irre fully irre<br />

G T Σ<br />

S.D.<br />

χ F Γ Λ<br />

B.S.<br />

more localized<br />

<strong>Parquet</strong><br />

H-F<br />

SOPT<br />

RPA, TMA, FLEX<br />

<strong>Parquet</strong> Approximation<br />

Λ II ≈ Λ I,DCA/QMC<br />

Better approximation<br />

DCA/QMC+<strong>Parquet</strong><br />

MSMB<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 32


DCA/QMC + parquet:<br />

Λ II ≈ Λ I,DCA/QMC<br />

• Multi-Scale Many-Body (MSMB) approach: SciDAC project<br />

⊲ short length scales treated explicitly with QMC methods<br />

⊲ intermediate length scales treated diagrammatically using fully irreducible vertices obtained<br />

from QMC<br />

⊲ long length scales treated in <strong>the</strong> mean field.<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 33


summary <strong>of</strong> <strong>the</strong> different approximate methods<br />

G<br />

Σ<br />

⇐ F = 0<br />

HF<br />

G<br />

Σ<br />

⇐ F ≈ v<br />

SOP T<br />

G<br />

⇐ Γ ≈ v<br />

RP A, T MA, F LEX<br />

Σ<br />

F<br />

G<br />

Γ<br />

P A<br />

⇐ Λ ≈ v<br />

MSMB<br />

Λ II ≈ Λ I,DCA/QMC<br />

Σ<br />

F<br />

• iterate until self-consistency is satisfied<br />

• It is also a hierarchy <strong>of</strong> methods with different computational costs<br />

• <strong>the</strong>re are two levels <strong>of</strong> self-consistency for <strong>the</strong> parquet calculation<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 34


application I: toy model<br />

The Hamiltonian for <strong>the</strong> classical anharmonic oscillator is<br />

Its partition function is<br />

H = p2<br />

2m + 1 2 mω2 0x 2 + ux 4 (1)<br />

Z c<br />

= 1 ∫ ∞<br />

Z c,0 Z c,0<br />

−∞<br />

dp<br />

∫ ∞<br />

−∞<br />

dxe −βH =<br />

1<br />

√<br />

2πG<br />

0<br />

∫ ∞<br />

−∞<br />

dψe − ψ2<br />

2G 0 e − g 4! ψ4 (2)<br />

where G 0 = 1/(βmω 2 0), βu = g/4! <strong>and</strong> ψ ≡ x. From this, we can construct <strong>the</strong><br />

conventional perturbation or <strong>the</strong> self-consistent field formalism.<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 35


esult<br />

3<br />

−Σ<br />

2<br />

1<br />

Hartree<br />

2nd-order<br />

self-consistent 2nd-order<br />

FLEX<br />

parquet approximation<br />

exact<br />

0<br />

0 5 10 15 20 25 30<br />

g<br />

• from 2nd-order approximation, FLEX, to parquet approximation, <strong>the</strong> results become better.<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 36


application II: Hubbard model (N c = 1)<br />

Hubbard model: H = −t ∑ 〈i,j〉,σ (c† i,σ c j,σ + h.c.) + U ∑ i n i,↑n i,↓<br />

-0.46<br />

-0.47<br />

Exact<br />

SC 2nd-order<br />

FLEX<br />

PA<br />

G(τ)<br />

-0.48<br />

-0.49<br />

U/T=1.5, Nc=1, Nl=64<br />

-0.5<br />

0 0.2 0.4 0.6 0.8 1<br />

τ/β<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 37


application II: Hubbard model (N c = 4 × 4)<br />

U = 2t, T = 0.3t<br />

S. X. Yang et al., PRE 80, 046706 (2009).<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 38


Conclusion<br />

• <strong>Parquet</strong> formalism<br />

• <strong>Hierarchy</strong> <strong>of</strong> different approximate methods<br />

1-particle<br />

2-particle<br />

F = 0<br />

F ≈ v<br />

Γ ≈ v<br />

Λ ≈ v<br />

re irre fully irre<br />

G T Σ<br />

S.D.<br />

χ F Γ Λ<br />

B.S.<br />

more localized<br />

<strong>Parquet</strong><br />

H-F<br />

SOPT<br />

RPA, TMA, FLEX<br />

<strong>Parquet</strong> Approximation<br />

Λ II ≈ Λ I,DCA/QMC<br />

Better approximation<br />

DCA/QMC+<strong>Parquet</strong><br />

MSMB<br />

<strong>Hierarchy</strong> <strong>of</strong> <strong>Self</strong>-<strong>Consistent</strong> <strong>Approximate</strong> <strong>Methods</strong> <strong>and</strong> <strong>the</strong> <strong>Parquet</strong> Formalism 39

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!