Indeed, linear Langevin theory (1.1)-(1.2) could be generalized in a number of ways [2–6].
First, linear Langevin equation (1.1) could be made nonlinear by adding general polynomial
terms such as f1(q, ˙q,χ), which contains self-interactions for dynamical variable qand nonlinear
interactions between dynamical variable qand noise ξ. One special case is multiplicative noise,
which amounts to making a replacement χ→f2(q, ˙q)χin (1.1). Second, the noise would
obey a non-Gaussian distribution, and might be coloured as well, which requires to go beyond
(1.2). Third, isotropy would be broken by an external field, such as in magnetized thermal
medium [7,8]. Then, beyond linear level, dynamics of transverse and longitudinal modes (with
respect to external field) would get mixed. These corrections may become relevant and/or
important for more realistic systems. A natural question arises: what is a more systematic
way of organizing these extensions? This will be pursued here through two complementary
approaches.
In this work we search for an effective description for a Brownian particle in a magnetized
plasma, with potential applications in heavy-ion collisions in mind. The main purpose is to
reveal nonlinear corrections to linear Langevin theory (1.1)-(1.2) in a systematic way. This will
be achieved by non-equilibrium effective field theory (EFT) formalism for a quantum many-
body system at finite temperature [9–12] (see [13–15] for an alternative approach). Within
such a formalism, dynamics of Brownian particle is entirely dictated by an effective action,
which will be constructed on a set of symmetries. The effective action may be thought of
as generalization of the MSRDJ action for a linear theory (1.1)-(1.2). Moreover, the effective
action contains “free parameters” representing UV physics and information of the state as well.
Generically, it is challenging to compute those free parameters from an underlying UV theory
(here, it is a closed system consisting of the Brownian particle and the magnetized plasma).
Given that quark-gluon plasma produced in heavy-ion collisions is strongly coupled, we turn
to a microscopic holographic model and derive the effective action (including values of free
parameters).
Holography [16–18] is insightful in understanding symmetry principles underlying non-
equilibrium effective action. Of particular importance is the dynamical Kubo-Martin-Schwinger
(KMS) symmetry [9–11] acting on dynamical variable of the effective action, which guarantees
the generalized fluctuation-dissipation theorem [19,20] at full level. In [9–11], dynamical KMS
symmetry is implemented in the classical statistical limit where ~→0, which corresponds to
neglecting quantum fluctuations in the effective theory. However, for a holographic theory, the
mean free path is ∼~/T , which implies that gradient expansion would generally inevitably
bear quantum fluctuations [21]. Via the example of Brownian motion, we will elaborate on
this point from both non-equilibrium EFT approach and holographic calculation. Intriguingly,
imposition of a constant translational invariance (i.e., q→q+cwith ca constant) renders
resultant effective theory to be of classical statistical nature.
While effective action formalism is more systematic in covering nonlinear corrections alluded
above, it turns out to be inconvenient to convert non-Gaussian effective action into Langevin
type equation [9]. The main obstacle stems from non-Gaussianity in a-variable (to be defined
below), which prohibits from carrying out Hubbard-Stratonovich transformation [2]. Interest-
ingly, we are able to convert the non-Gaussian effective action constructed in present work into
3