• No results found

Memory Storage and Thermalization in Multi-Mode Systems

N/A
N/A
Protected

Academic year: 2022

Share "Memory Storage and Thermalization in Multi-Mode Systems"

Copied!
27
0
0

Loading.... (view fulltext now)

Full text

(1)

Memory Storage and Thermalization in Multi-Mode Systems

Oleg Kaikov

LMU Munich, Germany

Supervisor: Prof. Dr. Georgi Dvali Co-supervisor: Dr. Sebastian Zell

44 th IMPRS Workshop

MPP Munich, July 15 th 2019

(2)

Outline

Motivation

System with high entropy Specific sub-system Solution of Sub-System

Coupling sensitivity Stability

Equilibration

Thermalization

Outlook: ETH

Summary

(3)

Motivation Solution of Sub-System Outlook: ETH Summary

System with high entropy

Application to black holes

G. Dvali [Phys. Rev. D 97, 105005 (2018)]

(1) large number of micro-states

Ω ∼ e S

H = ˆ

 1 − n ˆ 0

M



S

X

k=1

ε k n ˆ k

(4)

System with high entropy

Application to black holes

G. Dvali [Phys. Rev. D 97, 105005 (2018)]

(1) large number of micro-states

Ω ∼ e S

(2) all S modes ˆ n k gapless to fit within infinitesimal energy gap H = ˆ

 1 − n ˆ 0

M

 S

X

k=1

ε k n ˆ k

(5)

General system

Individual mass gaps

Interactions among ˆ a j modes Interaction with external ˆ b 0 mode

H = ˆ X

j

 1 − n ˆ 0

M j



ε j n ˆ j + X

j ,k j 6=k

A jk

 ˆ

a j a ˆ k + h.c.

 + λ

 ˆ

a 0 b ˆ 0 + h.c.



j , k = 1, . . . , D

(6)

Specific sub-system

Common mass gap

H = ˆ X

j

 1 − n ˆ 0

M



εˆ n j + X

j ,k j 6=k

A jk  ˆ

a j a ˆ k + h.c.  + λ 

ˆ

a 0 b ˆ 0 + h.c. 

j , k = 1, . . . , D

(7)

Solution of sub-system

D = 2 solved

G. Dvali [arXiv:1810.02336]

Our generalization: arbitrary D

Observation: h ˆ n j i (t) independent of (time-varying) common mass gap

(8)

Time-evolution

|ini = |N 1 , . . . , N D i ; N j ∈ [0, N] , P

j

N j = N

Figure: h ˆ n j i (t) (solid) and h ˆ n j i (t) (dashed) for j = 1, 2, 10 for D = 10, N = 10.

(9)

Coupling sensitivity

A jk ∈ [−1, +1]

Figure: h ˆ n j i (t) for j = 1, 2, 10 vs A 12 for D = 10, N = 10.

(10)

Stability

N

1

= 0, N

2

= 3

Figure: h ˆ n j i (t) for j = 1, 2 for various systems A i ∈ [−1, +1] D×D .

(11)

Equilibration

D = 10, N = 100 ; |ini = |N 1 , . . . , N D i

Figure: h ˆ n j i (t) vs N j with slope k as “memory of |ini”.

(12)

Thermalization

“Memory of |ini” k:

• indep. of N

• lim

D→∞ k = 0

Figure: Mean of k over A i ’s and |ini’s vs D at N = 10.

(13)

Thermalization

h ˆ Oi (t) = h ˆ Oi mc

h ˆ Oi (t) = lim

T →∞

1 T

T

Z

0

h ˆ Oi (t) dt

h ˆ Oi mc = 1 N ε

X

|E −ε|<δ ε

hε| ˆ O |εi

(14)

Outlook: Eigenstate Thermalization Hypothesis (ETH)

J. M. Deutsch [Phys. Rev. A 43, 2046 (1991)], M. Srednicki [Phys. Rev. E 50, 888 (1994)]

1 of 3 conditions: ε j ' ε k ⇒ hε j | ˆ O |ε j i ' hε k | ˆ O |ε k i

Figure: hε| ˆ n j |εi vs eigenstate energy ε for D = 6, N = 18.

(15)

Summary

Analytic generalization of high S system to arbitrary dim. D

Thermalization for large systems

Indications for thermalization mechanism distinct from that of ETH

(16)

Any questions?

(17)

Appendix

(18)

Formulae (solutions)

h ˆ n j i (t) = X

k,l ,s

N s Q jl Q jk Q sl Q sk cos [(ω l − ω k )t]

ω j = Q T AQ 

jj = X

i ,k

A ik Q ij Q kj

h ˆ n j i (t) = X

k,s

N s Q jk 2 Q sk 2

h ˆ n a

0

i (t) = N a

0

+  λ ϕ

 2

(N b

0

− N a

0

) sin 2 (ϕt)

h ˆ n b

0

i (t) = N b

0

−  λ ϕ

 2

(N b

0

− N a

0

) sin 2 (ϕt)

ϕ = 1 2

v u u

t (2λ) 2 + X

i

ε M N i

! 2

(19)

Formulae (unitarity)

H ∼ A ˆ X

i ,j

ˆ

a j a ˆ i , N i ∼ N D

⇒ M i →j = hN j | ˆ H |N i i ∼ A p N i pN j

⇒ Γ i =

D

X

j =1

M i →j

2 ∼ A 2 N 2 D

∼ const !

⇒ A ∼ √ D/N

τ =

P

k,l ,s |N s Q jl Q jk Q sl Q sk | P

k,l ,s |N s Q jl Q jk Q sl Q sk (ω l − ω k )|

(20)

k vs D at N = 10

Figure: k vs D at N = 10.

(21)

k vs N at D = 10

Figure: k vs N at D = 10.

(22)

k vs D and N

Figure: Mean of k over A i ’s and |ini’s vs D and N.

(23)

τ vs D at N = 10

Figure: τ vs D at N = 10 for A jk ∈ h

− √

D/N, + √ D/N i

.

(24)

τ vs N at D = 10

Figure: τ vs N at D = 10 for A jk ∈ h

− √

D/N, + √ D/N i

.

(25)

Equilibration time scale τ

Figure: τ vs D and N for A jk ∈ h

− √

D/N, + √ D/N i

.

(26)

ETH: scatter of hε| ˆ n j |εi

Figure: Scatter σ of hε| ˆ n j |εi within appropriate energy bins vs D and N.

(27)

Integrability

“A system is quantum integrable:

(i) If it exhibits n physically meaningful mutually commuting quantities that are in some sense independent [...] or depend linearly on some parameter of the Hamiltonian. [...]”

[C. Gogolin, J. Eisert, Rep. Prog. Phys. 79, 056001 (2016)]

References

Related documents

Since we consider travel time and service time as well, a trade-o ff between the number of served stations and the ratio of relocated bikes and served stations is given.. I.e., on

In this study, CHD4 rs74790047, TSC2 rs2121870, and AR rs66766408, were found to be common exonic mutations in both lung cancer patients and normal individuals exposed to high

T -score &lt; 2.5), American Society of Anesthesiologists (ASA) score, Charlson Comorbidity Index (CCI), metabolic equivalents (METs), surgical indication and diagnosis, whether

perceived needs and preferences. Such an approach introduces participant bias or preference, skewing findings and reducing the chances that researchers are comparing

The purpose of this chapter is to show how approximations of stochastic epidemic models can be used with disease data for fast inference to support public health authorities.. Our

Here, we describe the synthesis and SAR of novel 3,5-diamino-piperidine derivatives as aminoglycoside mimetics, and show they act as inhibitors of bacterial translation and

[r]