+ All Categories
Home > Documents > The - uni-osnabrueck.de · Christian Ba rtsch, Jo chen Gemmer Universit y of Osnab rück, Quantum E...

The - uni-osnabrueck.de · Christian Ba rtsch, Jo chen Gemmer Universit y of Osnab rück, Quantum E...

Date post: 25-Jan-2021
Category:
Upload: others
View: 0 times
Download: 0 times
Share this document with a friend
11
Transcript
  • Boltzmann approah to transport in weakly interating, nearintegrable, 1-d fermioni systemsChristian Bartsh, Johen GemmerUniversity of Osnabrük,Quantum E�ienyFreiburg, Nov. 22., 2011C. Bartsh, J.Gemmer Boltzmann for 1-d

  • ContentsThe model and some of its transport propertiesHow to set up a (linearized) Boltzmann equation?Transition rates and sattering operatorsDi�usion oe�ientsThe model without next nearest neighbor hopping: integrableSpetrum of deay rates of the urrent orrelation funtionSummary

    C. Bartsh, J.Gemmer Boltzmann for 1-d

  • The model and some of its transport propertiesspinless fermions Hubbard-type tight-binding model plus next nearest neighborhoppingH = J NXn=1 h − 12 (a†nan+1 + b · a†nan+2 + h..) + ∆ · a†nana†n+1an+1iWhat is known (more or less for sure)?for b = 0 and any ∆ the model is �integrable�: Bethe ansatz applies,in�nitely many loal onserved quantities, no Wigner-Dyson levelstatistis, et.linear response roughly states that transport is ballisti if PnTr(ĴQ̂n) 6= 0⇒ integrable systems tend do be ballisti (≈ feature in�nite ondutivity,di�usion onstants, et.)Even for b = 0 the transport properties may depend on interation ∆, �lling fand temperature T .for ∆ > 1 the system is �gapped� and non-ballisti, i.e., di�usive for all(most) �llings and temperaturesfor ∆ < 1 the system is most likely ballisti for all ases exept thefollowing:for ∆ < 1, half �lling f = 0.5 and �nite temperature T 6= 0 is urrentlyunder debate C. Bartsh, J.Gemmer Boltzmann for 1-d

  • How to set up a (linearized) Boltzmann equation ?We intend to investigate transport properties in the regime ∆

  • How to set up a (linearized) Boltzmann equation ?...something that may be interpreted as the linearized sattering operator of aBoltzmann equation:̇di (t) = Xk 6=i Rik (t)dk(t) − Xk 6=i Rki (t)di (t)where the orresponding rates (T = ∞, f = 0.5) are given byRk,i(t) = 12N2~2 Xl (Re(W (i− k)) − Re(W (k− l)))2tsin[ 1~ (εi + εl − εk − εi−k+l)t]−12 (Re(W (l− i)) − Re(W (k− l)))2tsin[ 1~ (εk + εi − εi+k−l − εl)t]In the limit of large times and large system sizes the rates should beometime-independent (for non-pathologial ases, i.e., no multiple zeroes in thesin-argument) C. Bartsh, J.Gemmer Boltzmann for 1-d

  • Rates and sattering operators 0

    0.002

    0.004

    0.006

    0.008

    0.01

    0.012

    0.014

    0.016

    0 20 40 60 80 100 120

    R47

    5,50

    0(t)

    tabove:time evolution of one rate forb = 1, ∆ = 0.01,T = ∞,N = 600right:sattering operator Rb = 1, ∆ = 0.01,T = ∞,N = 200 -2e-05 0 2e-05 4e-05 6e-05 8e-05 0.0001

    0 100 200

    0

    100

    200

    C. Bartsh, J.Gemmer Boltzmann for 1-d

  • Di�usion oe�ientsFor regular linear Boltzmann equations (at T = ∞) the di�usion oe�ientreads D = − 1N vlR−1lk vk veloities : vk = ∂ǫ∂k�instantanous di�usion oe�ients� 4

    5

    6

    7

    8

    50 100 150 200 250 300

    D(t

    ) ∆2

    /J

    t [J]

    b=1.0b=0.8b=0.6b=0.5b=0.4b=0.3b=0.2

    �long time di�usion oe�ients� 1

    10

    100

    0.1 1

    D(b

    ) ∆2

    /Jb

    4.0/b1/2

    It appears that with b → 0:the timesale on whih the sattering operator beomes onstant divergesthe di�usion oe�ient diverges as D ∝ b−1/2C. Bartsh, J.Gemmer Boltzmann for 1-d

  • Anisotropi hain without next nearest neighbor hoppingTaking a loser look at b → 0 we �nd thatthe argument in the sin funtion in the rate-expression indeed appraohesmultiple zeroesespeially for b = 0,N → ∞, t → ∞, t → ∞ the sattering operator anbe found analytiallyR(j , k) ∝ (os(k − j) + os(k + j))2|sin(k) − sin(i)|-0.0001

    -5e-05

    0

    5e-05

    0.0001

    0.00015

    0 100 200

    0

    100

    200��nite time sketh�At t → ∞the sattering operator divergesat some transitions:j + k = π/2, j + k = −π/2the sattering operator hasat least one zero eigenvalue sine

    R R(j , k)(δ(k − π/2) − δ(k + π/2))dk = 0C. Bartsh, J.Gemmer Boltzmann for 1-d

  • Spetrum of deay rates of the urrent-orrelation funtionThe b = 0 sattering operator has zero but also in�nitely large (negative)eigenvalues. Whih dominate the urrent-orrelation funtion? Within thisframework all urrents relax multi-exponentially:ĵ = Xk vka†kak 〈̂j(t )̂j(0)〉 := Xe O(e)2exp{−et}�spetrum of deay rates� 0

    0.005

    0.01

    0.015

    0.02

    0.025

    0.03

    0.035

    0.04

    -16 -14 -12 -10 -8 -6 -4 -2 0

    O(e

    )

    e

    a substantial fration of theurrent relaxes arbitrarily slowthis graph appears to beidependent of the �oarsegraining�one may alulate a timedependent di�usion oe�ientfromD(t) = Z t0 dτ 〈̂j(τ )̂j(0)〉Proeeding this way yields for large times D(t) ∝ ln(t)C. Bartsh, J.Gemmer Boltzmann for 1-d

  • SummaryWe investigated the di�usion oe�ient of a 1-d (spinless) fermioni systemwith a nearest neighbor hopping of J = 1, a next-nearest neighbor hopping ofb, and a nearest neighbor interation of ∆.We �nd in the limit of small ∆ for the inverse di�usion oe�ient(∝ resistivity): D−1 ≈ 0.25 √b ∆2+?∆4 + · · ·Within this approah transport oe�ients sale at least like ∆−4 for nonext-nearest neighbor hopping.All of this has been done for in�nte temperature, T = ∞, however, the overallbehavior is expeted to remain unaltered at �nite temperaturesC. Bartsh, J.Gemmer Boltzmann for 1-d

  • Thanks !If you are interested in Refs. just ask me.We thank you, the audiene, for your interest and patiene!

    C. Bartsh, J.Gemmer Boltzmann for 1-d


Recommended