Statistical Physics of active matter Tutorial session 1
Tutorial session 1
Discussion on January 6, 9:00-11:00, Room SP F2.04
1) Velocity correlations of a passive Brownian particle
In this exercise, you will be walked through to calculate the velocity autocorrelaiton function of a Brownian
particle starting from the Langevin equation:
dv(t) 1
= −v(t)/τm + ξ(t), (1)
dt m
in which τm = m/γ and ξ is a random stochastic force with zero mean and two-point correlation ⟨ξα (t)ξβ (t′ )⟩ =
Γδαβ δ(t − t′ ).
a) Integrate this equation to obtain its solution for v(t) given by
1 t − (t−s)
Z
− τt t
v(t) = v0 e m + e τm ξ(s) ds. (2)
m 0
Hint: use the change of variable u(t) = v(t)et/τm
b) Use this solution and statistical properties of random force ξ to show that the time autocorrelation
function of velocity in dimension d is given by
Γ Γ −s/τm
⟨v(t) · v(t + s)⟩ = (⟨v02 ⟩ − )e−(2t+s)/τm + e , (3)
2γm 2γm
c) Obtain the auto-time correlation function of velocity in the steady-state limit, i.e. when t ≫ 1.
d) As shown in the class, using equi-partion theorem, we can identify the noise strength as Γ = 2γ kB T .
Use these informations to simplify the Eq. (3).
2) Mean-squared displacement of Brownian particle from the Langevin equation
2
Show that the mean-squared displacement ⟨ r(t) − r(0) ⟩ ≡ ⟨∆r2 (t)⟩ can be obtained as ⟨∆r2 (t)⟩ =
Rt
2⟨ 0 r · v(s) ds⟩, assuming that r(0) = 0.
Hint: Start by multiplying the Langevin equation Eq. (1) for velocity by r, then taking ensemble average
and using equi-partition theorem. Note that ⟨r · ξ⟩ = 0.
3) Mean-squared displacement and Green-Kubo relation for diffusion In this guided exercise, you
get to know the Green–Kubo relation for diffusion derived from the Langevin theory. Consider a Brownian
particle with an initial velocity v(0). Given that ṙ = v, can obtain the position as a function of time:
Z t Z t
r(t) = r(0) + v(s)ds → ∆r(t) = v(s)ds (4)
0 0
a) Use this equation to obtain the mean-squared displacement in terms of double integral of velocity
correlation function.
b) Assume that system is in equilibrium, so that the stochastic process is stationary. As a consequence,
the velocity correlation function depends only on the time difference and has the form
⟨v(s) · v(s′ )⟩ = Cv (|s − s′ |), (5)
Statistical Physics of active matter Tutorial session 1
where Cv (τ ) is the velocity autocorrelation function. Using this form of velocity auto-correlation function,
simplify the double integral to obtain:
Z t
2
⟨|∆r(t)| ⟩ = 2 dτ (t − τ ) Cv (τ ) . (6)
0
c) Now use the outcome of exercise 1d, to calculate explicitly the form of mean-square displacement,
verifying that you get the same result as that of Ex. 2.
1
d) More generally, we can obtain the diffusion coefficient as 2d limt→∞ ∂t ⟨|∆r(t)|2 ⟩. Apply this definition
to obtain the Green–Kubo relation for diffusion in d dimensions:
1 ∞ 1 ∞
Z Z
D= Cv (τ ) dτ = ⟨v(τ ) · v(0)⟩ dτ,
d 0 d 0
which for the Langevin model yields D = kB T /γ. This result is equivalent to the Einstein relation D =
µkB T with mobility µ = 1/γ, which is a special case of Fluctuation-dissipation relations that will be
discussed during the last week of lectures.