The group-theoretic approach to perfect fluid

equations with conformal symmetry

Anton Galajinsky
Laboratory of Applied Mathematics and Theoretical Physics,
arXiv:2210.14544v2 [hep-th] 9 Nov 2022

TUSUR, Lenin ave. 40, 634050 Tomsk, Russia

e-mail: [email protected]

Abstract
The method of nonlinear realizations is a convenient tool for building dynamical realiza-
tions of a Lie group, which relies solely upon structure relations of the corresponding Lie
algebra. The goal of this work is to discuss advantages and limitations of the method,
which is here applied to construct perfect fluid equations with conformal symmetry.
Four cases are studied in detail, which include the Schrödinger group, the `-conformal
Galilei group, the Lifshitz group, and the relativistic conformal group.

PACS: 11.30.-j, 02.20.Sv, 47.10.A, 47.10.ab

Keywords: perfect fluid equations, conformal symmetry
1. Introduction

Recently, there has been a considerable effort to understand the hydrodynamic limit
of the AdS/CFT-correspondence [1, 2, 3].1 The main point of the study was to establish
that the low-frequency behavior of an interacting field theory at finite temperature could be
described by fluid mechanics.
The conventional formulation of fluid dynamics relies upon an expansion scheme in which
the effects of viscosity and heat transfer are regarded as corrections to the perfect fluid equa-
tions (see e.g. [2]). As is known, for a properly chosen equation of state the (non)relativistic
perfect fluid equations exhibit conformal invariance (see e.g. [5, 6, 7]). It is then natural to
wonder whether such equations can be formulated on purely group-theoretic grounds.
A convenient tool for building dynamical realizations of a Lie group, which relies solely
upon structure relations of the corresponding Lie algebra, is provided by the method of
nonlinear realizations [8]. The scheme includes several steps. First of all, one introduces
coordinates and fields, whose number in general is equal to the number of generators of a
Lie algebra at hand. Then one builds a formal group-theoretic element g, which is a product
of exponentials of the type eiαT , where α is a coordinate (or a field) and T is a Lie algebra
generator. After that, one studies the left action of the group upon the group-theoretic
element g and establishes transformation laws of the coordinates and fields. Finally, one
computes the Maurer-Cartan one-forms g −1 dg. By construction, they hold invariant under
the left action of the group upon the group theoretic element and, hence, provide convenient
building blocks to formulate invariant equations of motion. If desirable, they can also be
used to eliminate some of the fields from the consideration by imposing constraints. Within
the method of nonlinear realizations, imposing constraints is attributed to the inverse Higgs
phenomenon [9].
An alternative formulation of fluid dynamics in terms of group-valued variables, which
relies upon Kirillov’s method of orbits [10], was proposed in [11]. The formalism is particu-
larly suitable for taking into account constituent particles, which carry nonabelian charges
or spin degrees of freedom, as well as for incorporating anomalies [12]–[16].
The goal of this work is to discuss advantages and limitations of the method of nonlinear
realizations, which is here applied to construct perfect fluid equations with conformal sym-
metry. Four cases are studied in detail, which include the Schrödinger group, the `-conformal
Galilei group, the Lifshitz group, and the relativistic conformal group.
The organization of the work is as follows. In the next section, we briefly outline key
features of the method of nonlinear realizations, which are used in later section to explore
fluid equations with conformal symmetry. For simplicity of the presentation, the construction
is illustrated by the example of so(2, 1) algebra, which was first studied in [17]. In Sect. 3,
perfect fluid equations with the Schrödinger symmetry are built in terms of the invariant
Maurer-Cartan one-forms associated with the Schrödinger group and an invariant derivative.
It is demonstrated that a proper equation of states, which links pressure to fluid density,
1
Literature on the subject is overwhelmingly large. For a review and further references see [4].

1
comes about quite naturally without the need to invoke more sophisticated arguments [5, 6].
In Sect. 4 and Sect. 5, a similar analysis is performed for the `-conformal Galilei group
and the Lifshitz group, respectively. To the best of our knowledge, the Lifshitz-invariant
equations in Sect. 5 are new. Sect. 6 is focused on the case of the relativistic conformal
group. In contrast to the nonrelativistic examples, the construction of relativistic fluid
equations invariant under the conformal group in terms of the Maurer-Cartan invariants
alone turns to be problematic and extra arguments need to be invoked. In the concluding
Sect. 7, we summarize our results and discuss possible further developments.
Throughout the paper, summation over repeated indices is understood unless otherwise
is stated.

2. The method of nonlinear realizations

In this section, we outline key features of the method of nonlinear realizations [8], which
will be used below to explore fluid equations with conformal symmetry. For simplicity of the
presentation, the construction will be illustrated by the example of so(2, 1) algebra

[H, D] = iH, [H, K] = 2iD, [D, K] = iK, (1)

which was first studied in [17]. Above H is interpreted as the temporal translation generator,
D links to dilatation, and K is associated with the special conformal transformation.
In its essence, the method of nonlinear realizations is a tool to build dynamical realizations
of a Lie group, which is based solely upon structure relations of the corresponding Lie algebra.
As the first step of the construction, one introduces coordinates and fields whose number in
general is equal to the number of generators of the Lie algebra. The choice is not unique
and it essentially depends on a dynamical realization one seeks for. For example, if one is
concerned with one-dimensional mechanics originating from (1), a temporal variable t and
a function u(t), which describes a particle dynamics, are needed. It seems natural to link
t to H and u(t) to D. In order to treat all the generators on equal footing, one introduces
one more field w(t), which is regarded as a partner of K. At this stage, it is not yet clear
whether u(t) or w(t) will be more suitable for describing a reasonable conformal mechanics
model and one anticipates an SO(2, 1)-invariant constraint, which will link the fields to each
other.
Then one introduces the group-theoretic element

g = eitH eiu(t)D eiw(t)K . (2)

In general, the expressions in (1) are regarded as formal Lie brackets (rather than commu-
tators of operators) and the exponential entering (2) is treated as the exponential map of a
Lie algebra to a neighborhood of the unit group element. Yet, nothing prevents one from
assuming that a specific representation of the Lie algebra is chosen such that (1) represents
commutators of the generators and (2) is an operator acting upon a state. In thePlatter case,
the exponential entering (2) can be regarded as a formal Taylor series eA = ∞ An
n=0 n! , in

2
which case the well known Baker-Campbell-Hausdorff formula
∞ n
iA −iA
X i
e T e =T+ [A, [A, . . . [A, T ] . . . ]] (3)
n! | n=1
{z }
n times

is applicable, where A and T are two arbitrary operators. The conventional property of the
exponential
eαA eβA = e(α+β)A , (4)
holds true as well, where α, β are constants and A is an operator action upon a state in
the representation chosen. Eq. (3) is the cornerstone of the whole consideration to follow.
Note that the order in which the factors contribute to the group-theoretic element (2) can
be chosen at will and different options are related by coordinate and field redefinitions.
At the second step of the construction, one analyzes the left action of the group upon
the group-theoretic element
0 0 0 0 0
g 0 = eiβH eiλD eiσK · g = eit H eiu (t )D eiw (t )K , (5)
where β, λ, σ are real transformation parameters, and makes use of (3) so as to establish
transformation laws of the coordinates and fields. For most physical applications it suffices
to consider their infinitesimal form.
When performing calculations, depending on a specific operator eL at hand, it might
prove helpful to insert the unit operator e−B eB , with B to be specified below, either to the
left or to the right of eL . The following chain of relations
eiλD eitH = eiλD eitH |e−iλD iλD iλD
{ze } = e (1 + itH + . . . ) e
−iλD iλD
e =
1

(1 + it (H + [iλD, H] + . . . ) + . . . ) eiλD = ei(1+λ)tH eiλD (6)

gives the example of how eiλD ”passes through” eitH . The Baker-Campbell-Hausdorff formula
was repeatedly applied in the second line of (6). At the same time, when computing eiσK eitH
with infinitesimal σ, it proves convenient to place eitH e−itH to the left of the operator
e−itH} eiσK eitH = eitH e−itH (1 + iσK) eitH =
eiσK eitH = e|itH{z
1
  
itH 1 2
e 1 + iσ K − [itH, K] + [itH, [itH, K]] = ei(t+σt )H eiσK e2iσtD , (7)
2!
where we used the fact that eiσA eiσB = eiσB eiσA for infinitesimal σ.
Taking into account the technicalities, from Eq. (5) one obtains the infinitesimal SO(2, 1)-
transformations acting upon the temporal variable and the fields u(t), w(t) (each transfor-
mation is separated by semicolon)
t0 = t + β, u0 (t0 ) = u(t), w0 (t0 ) = w(t);
t0 = (1 + λ)t, u0 (t0 ) = u(t) + λ, w0 (t0 ) = w(t);
t0 = t + σt2 , u0 (t0 ) = u(t) + 2σt, w0 (t0 ) = w(t) + eu(t) σ. (8)

3
 The third step of the method consists in computing the Maurer-Cartan one-forms2

g −1 dg = iωH H + iωD D + iωK K, (9)

where
ωH = e−u dt, ωD = du − 2we−u dt, ωK = dw − wdu + w2 e−u dt. (10)
By construction, they hold invariant under the group transformation (5), (8) and, hence,
provide convenient building blocks to formulate invariant equations of motion. If desirable,
they can also be used to eliminate some of the fields entering the group-theoretic element
from the consideration by imposing constraints.
In the last step of the construction, one specifies a dynamical realization of the group at
hand by choosing judiciously a combination of the Maurer-Cartan invariants which results
in a reasonable set of second order differential equations. The latter are identified with the
equations of motion of a dynamical system. This item is not straightforward and requires
guesswork. For example, if one wishes to use (10) in order to formulate one-dimensional
SO(2, 1)-invariant mechanics, one imposes the constraint ωD = 0, which links w to u

1 deu
w= , (11)
2 dt
and then postulates the equation of motion [17]

ωK − γ 2 ωH = 0, (12)

in which γ is a (coupling) constant. Implementing the field redefinition u = ln ρ2 , which is

2
meant to remove the du dt
-term from (12), one finally arrives at

d2 ρ γ2
= , (13)
dt2 ρ3

which is the conventional 1d conformal mechanics equation of motion [18].

To give another example, it is known that the Schwarzian derivative
...
ρ (t) 3 ρ̈(t) 2
 
S(ρ(t)) = − , (14)
ρ̇(t) 2 ρ̇(t)

where ρ(t) is a real function and the dot designates the derivative with respect to t, holds
invariant under the SL(2, R)-transformation acting upon the argument

aρ(t) + b
ρ0 (t) = , (15)
cρ(t) + d
2
Given g in (2), the inverse element is g −1 = e−iwK e−iuD e−itH , while the differential reads dg =
itH
e (idtH) eiuD eiwK + eitH eiuD (iduD) eiwK + eitH eiuD eiwK (idwK).

4
with ad − cb = 1. At first glance, the invariance does not seem obvious at all. Yet, because
sl(2, R) ∼ so(2, 1), one can naturally arrive at (14) by applying the group-theoretic argu-
ments similar to those above [19]. It suffices to keep the temporal variable t as an external
parameter3 and introduce the group-theoretic element
g = eiρ(t)H eis(t)K eiu(t)D , (16)
where ρ(t), s(t), u(t) are as yet unspecified functions, which gives rise to the Maurer-Cartan
invariants
ωH = ρ̇e−u dt, ωK = eu ṡ + s2 ρ̇ dt,


ωD = (u̇ − 2sρ̇) dt. (17)
Imposing the constraints ωH − µdt = 0 and ωD + 2νdt = 0, where µ and ν are arbitrary
constants, one can express u and s in terms of ρ, while substituting the result into the only
remaining one-form ωK , one arrives at the Schwarzian derivative [19].
Our primary concern in later sections will be to understand whether the construction
outlined above is powerful enough to result in perfect fluid equations with conformal sym-
metry. Specifically, the cases of the Schrödinger group, the `-conformal Galilei group, the
Lifshitz group, and the relativistic conformal group will discussed in turn.

3. Perfect fluid equations with the Schrödinger symmetry

The Lie algebra associated with the Schrödinger group includes generators of temporal
translation, dilatation, and special conformal transformation, which form so(2, 1) subal-
gebra, as well as spatial translations, the Galilei boosts, and spatial rotations. Given a
nonrelativistic spacetime parameterized by a temporal variable t and Cartesian coordinates
xi , i = 1, . . . , N , where N is the spatial dimension, they can be represented in the form4
   
∂ ∂ 1 ∂ 2 ∂ ∂
H=i , D = i t + xi , K=i t + txi ,
∂t ∂t 2 ∂xi ∂t ∂xi
(0) ∂ (1) ∂
Ci = i , Ci = it , (18)
∂xi ∂xi
which obey the commutation relations
(1) (0)
[H, D] = iH, [H, K] = 2iD, [D, K] = iK, [H, Ci ] = iCi ,
(0) i (0) (1) i (1) (0) (1)
[D, Ci ] = − Ci , [D, Ci ] = Ci , [K, Ci ] = −iCi . (19)
2 2
The finite form of the corresponding transformations acting in the nonrelativistic spacetime
reads
 0  12
0 αt + β 0 ∂t (0) (1)
t = , xi = xi ; t0 = t, x0i = xi + ai + ai t, (20)
γt + δ ∂t
3
To be more precise, one considers R × SL(2, R) and links t to the generator of the group of real numbers
under addition R.
4
Here and in what follows the conventional rotation generators are disregarded.

5
 (0) (1)
where α, β, γ, δ, ai , ai are transformation parameters, the first four of which obey the
restriction αδ − βγ = 1.
In order to describe a fluid, one introduces the density5 ρ(t, x) and the velocity vector
field υi (t, x), i = 1, . . . , N . The transformation law of ρ(t, x) under the Schrödinger group
is obtained by fixing a value of the temporal variable t and demanding the mass of an
N -dimensional volume element V to be invariant under (20)
Z Z
0 0 0 0
dx ρ (t , x ) = dxρ(t, x), (21)
V0 V
where dx = dx1 . . . dxN . This gives
 0  N2
∂t
ρ(t, x) = ρ0 (t0 , x0 ); ρ(t, x) = ρ0 (t0 , x0 ), (22)
∂t
where the first relation corresponds to the SO(2, 1)-transformations, while the second links
to the spatial translations and the Galilei boosts.
Considering an orbit of a liquid particle parameterized by xi (t), for which
dxi (t)
= υi (t, x(t)), (23)
dt
from Eq. (20) one can determine the transformation laws of the velocity vector field6
 0  12  − 12
∂t 0 0 0 ∂ ∂t0 (1)
υi (t, x) = υi (t , x ) + x0i ; υi (t, x) = υi0 (t0 , x0 ) − ai , (24)
∂t ∂t ∂t
where again the former equality corresponds to the SO(2, 1)-transformations, while the latter
links to the spatial translations and the Galilei boosts.
In what follows, we will need infinitesimal form of the Schrödinger transformations acting
upon the coordinates and fields. Substituting α = 1, δ = 1, γ = 0 into (20), (22), (24) and
regarding β as infinitesimal parameter, one obtains the infinitesimal form of the temporal
translation
t0 = t + β, x0i = xi ,
ρ0 (t0 , x0 ) = ρ(t, x), υi0 (t0 , x0 ) = υi (t, x). (25)
λ λ
Choosing α = e 2 , δ = e− 2 , β = 0, γ = 0, setting λ to be infinitesimal parameter and
Taylor expanding in λ up to the first order, one gets the dilatation transformation
 
0 0 λ
t = (1 + λ)t, xi = 1 + xi ,
2
   
0 0 0 Nλ 0 0 0 λ
ρ (t , x ) = 1 − ρ(t, x), υi (t , x ) = 1 − υi (t, x), (26)
2 2
5
Throughout the paper, we use units in which ρ is dimensionless. This can be done by choosing a reference
value ρ0 and rescaling ρ → ρρ0 .
6
As is usual in classical dynamics, when restricting transformations similar to (20) to a particle orbit,
one replaces x0i with x0i (t0 ) and xi with xi (t).

6
where, as above, N denotes the spatial dimension.
Infinitesimal form of the special conformal transformation is found by setting α = 1,
δ = 1, β = 0, γ = −σ in Eq. (20), regarding σ as infinitesimal parameter, and Taylor
expanding in σ up to the first order

t0 = t + σt2 , x0i = (1 + σt)xi ,

ρ0 (t0 , x0 ) = (1 − N σt)ρ(t, x), υi0 (t0 , x0 ) = (1 − σt)υi (t, x) + σxi . (27)

The spatial translations and the Galilei boosts maintain their form
(0) (1)
t0 = t, x0i = xi + ai + ai t,
(1)
ρ0 (t0 , x0 ) = ρ(t, x), υi0 (t0 , x0 ) = υi (t, x) + ai . (28)

Having fixed transformation laws of the fields which characterize a fluid, we are now
in a position to construct invariant equations of motion within the method of nonlinear
realizations. As the first step, one has to introduce a proper group-theoretic element. The
building blocks, which are at our disposal, are the coordinates t, xi and the fields ρ(t, x),
(0) (1)
υi (t, x) on the one hand, and the generators of the Schrödinger algebra H, D, K, Ci , Ci
on the other hand. It seems natural to link the generator of temporal translation H and the
(0)
generator of spatial translations Ci to the temporal variable t and the Cartesian coordinates
xi , respectively. The remaining generators of the algebra should be accompanied by specific
combinations of ρ(t, x) and υi (t, x) in a way compatible with the left action of the Schrödinger
(1)
group upon a group-theoretic element, which will be introduced shortly. Because Ci carries
a vector index, its partner in a group-theoretic element should be proportional to the velocity
vector field υi (t, x). For symmetry reasons (see below), a coefficient of proportionality can
be taken to be just the unity. Denoting companions of D, K by u(t, x), w(t, x), respectively,
and postponing for later a clarification of their connection to ρ(t, x), υi (t, x), one finally gets
the group-theoretic element
(0) (1)
g = eitH eixi Ci eiυi (t,x)Ci eiu(t,x)D eiw(t,x)K . (29)

As was mentioned is Sect. 2, factors contributing to a group-theoretic element in general do

not commute. The choice (29) proves to be most convenient for our subsequent analysis.
As the next step, one considers the left action of the Schrödinger group upon the group-
theoretic element
(0) (0) (1) (1) (0) (1)
eiβH eiai Ci
eiai Ci
eiλD eiσK · eitH eixi Ci eiυi (t,x)Ci eiu(t,x)D eiw(t,x)K =
0 0 (0) 0 0 0 (1) 0 0 0 0 0 0
eit H eixi Ci eiυi (t ,x )Ci eiu (t ,x )D eiw (t ,x )K , (30)
(0) (1)
where β, λ, σ, ai , ai are transformation parameters, and makes use of the Baker-Campbell-
Hausdorff formula (3) to determine transformation laws of the coordinates and fields. For

7
simplicity of the presentation, we focus on their infinitesimal form (each transformation is
separated by semicolon)

t0 = t + β, x0i = xi ,
υi0 (t0 , x0 ) = υi (t, x), u0 (t0 , x0 ) = u(t, x), w0 (t0 , x0 ) = w(t, x);
 
λ
t0 = (1 + λ)t, 0
xi = 1 + xi ,
2
 
0 0 0 λ
υi (t , x ) = 1 − υi (t, x), u0 (t0 , x0 ) = u(t, x) + λ, w0 (t0 , x0 ) = w(t, x);
2
t0 = t + σt2 , x0i = (1 + σt)xi ,
υi0 (t0 , x0 ) = (1 − σt)υi (t, x) + σxi , u0 (t0 , x0 ) = u(t, x) + 2σt, w0 (t0 , x0 ) = w(t, x) + σeu(t,x) ;
(0)
t0 = t, x0i = xi + ai ,
υi0 (t0 , x0 ) = υi (t, x), u0 (t0 , x0 ) = u(t, x), w0 (t0 , x0 ) = w(t, x);
(1)
t0 = t, x0i = xi + tai ,
(1)
υi0 (t0 , x0 ) = υi (t, x) + ai , u0 (t0 , x0 ) = u(t, x), w0 (t0 , x0 ) = w(t, x). (31)

At this stage, comparing (31) with (25), (26), (27), (28) and taking into account the
identities
 0  0  0  0
∂ ∂t ∂ ∂xi ∂ ∂ ∂t ∂ ∂xj ∂
= + 0
, = + , (32)
∂t ∂t ∂t 0 ∂t ∂xi ∂xi ∂xi ∂t 0 ∂xi ∂x0j

one can unambiguously link u, w to ρ and υi

N 1 − 2 ∂υi
ρ = e− 2 u , w= ρ N . (33)
N ∂xi
A comparison of (31) with (25), (26), (27), (28) also confirms our earlier identification of a
(1)
companion of the Galilei boost generator Ci entering the group-theoretic element with the
fluid velocity vector field υi .
As the final step, one computes the invariant Maurer-Cartan one-forms
(0) (0) (1) (1)
g −1 dg = iωH H + iωD D + iωK K + iωi Ci + iωi Ci , (34)

where

ωH = e−u dt, ωD = du − 2we−u dt, ωK = dw − wdu + w2 e−u dt,

(0) u (1) u u
ωi = e− 2 (dxi − υi dt), ωi = e 2 dυi − we− 2 (dxi − υi dt), (35)

8
and takes notice of the fact that the derivative
1 ∂
∇i = ρ− N , (36)
∂xi
is invariant under the action of the Schrödinger group.7
The invariants (35) and (36) are all one needs to formulate perfect fluid equations with
the Schrödinger symmetry within the group-theoretic approach. First of all, a comparison
(0)
of ωi with one of the key ingredients of fluid mechanics (23) suggests a further specification
in the group-theoretic element (29)
xi → xi (t), (37)
where xi (t) is now interpreted as parameterizing an orbit of a liquid particle. The substitution
(37) also links the differential d to the material derivative D, which is commonly used within
fluid mechanics
∂ ∂
d = dtD, D= + υi (t, x) . (38)
∂t ∂xi
Imposing the Schrödinger-invariant constraint
(0)
ωi = 0, (39)

one thus reproduces Eq. (23) and confirms once again that the identification of υi in (29)
with the fluid velocity vector field was correct.
Substituting (33) into ωD , one gets
 
2 Dρ ∂υi
ωD = − + dt. (40)
N ρ ∂xi
Hence, demanding ωD to vanish, one naturally arrives at the continuity equation
∂ρ ∂(ρυi )
+ = 0, (41)
∂t ∂xi
which is the equation of motion for ρ in fluid mechanics.
It remains to determine the equation for υi . Note that after imposing the constraint (39),
(1)
the one-form ωi simplifies to  1 
(1)
ωi = ρ− N Dυi dt. (42)
In particular, it involves the material derivative of the velocity vector field υi , which is the
fluid mechanics analog of the acceleration vector in Newtonian mechanics. Because the
remaining Maurer-Cartan forms in (35) do not carry vector indices, one is led to use the
invariant derivative (36) so as to specify a potential term. Imposing the simplest Schrödinger-
invariant restriction
(1)
ωi + α∇i ωH = 0, (43)
7
The invariance of the derivative is most easily verified by making recourse to the finite transformations
(20), (22) and taking into account the identity (32).

9
where α is a real constant, one gets the equation of motion
2
∂ρ N
Dυi = −α . (44)
∂xi
The latter can be put into the conventional Euler form
∂p
ρDυi = − , (45)
∂xi
by introducing the pressure p(t, x) which obeys the equation of state
2
p = νρ1+ N , (46)

with ν = N2α+2
. Eqs. (41), (45), (46) reproduce the perfect fluid equations invariant under
the action of the Schrödinger group, which were originally introduced in [5, 6] (for conserved
charges and the energy-momentum tensor see [6]). Within the group-theoretic approach,
(0) (1)
they result from imposing the invariant constraints ωi = 0, ωD = 0, ωi + α∇i ωH = 0
upon the Maurer-Cartan one-forms. The advantage of the method is that the equation
of state (46) comes about quite naturally without the need to invoke more sophisticated
arguments (cf. [5, 6]).
Concluding this section, we note that some statements in the literature regarding con-
formal symmetries of nonrelativistic fluid mechanics contradict each other. For a detailed
account and further references see [20].

4. Perfect fluid equations with the `-conformal Galilei symmetry

As is well known, the Schrödinger algebra is a particular instance of the so called `–

conformal Galilei algebra [21, 22]. The latter includes generators of (temporal) translation,
dilatation, special conformal transformation, spatial rotations, spatial translations, Galilei
boosts and constant accelerations. They obey the structure relations8

[H, D] = iH, [H, K] = 2iD, [D, K] = iK,

(n) (n−1) (n) (n) (n) (n+1)
[H, Ci ] = inCi , [D, Ci ] = i(n − l)Ci , [K, Ci ] = i(n − 2l)Ci , (47)

where i = 1, . . . , N and n = 0, . . . , 2`. Here N designates the spatial dimension and ` is an

arbitrary (half)integer real parameter, which gives the name to the algebra.
A conventional realization of the algebra in terms of differential operators acting in a
nonrelativistic spacetime reads
   
∂ ∂ ∂ 2 ∂ ∂ (n) ∂
H = i , D = i t + `xi , K=i t + 2`txi , Ci = itn . (48)
∂t ∂t ∂xi ∂t ∂xi ∂xi
8
As above, in this section we disregard spatial rotations.

10
Note that the value of the parameter ` specifies the number of acceleration generators at
(n)
hand (Ci with n > 1) and for ` = 12 one reveals the Schrödinger algebra discussed in the
preceding section. A finite form of the transformations is given by (no sum over repeated
index n)
 0 `
0 αt + β 0 ∂t (n)
t = , xi = xi ; t0 = t, x0i = xi + ai tn , (49)
γt + δ ∂t
(0) (1) (2`)
where n = 0, . . . , 2`, and α, β, γ, δ, ai , ai , . . . , ai are transformation parameters, the
first four of which obey the restriction αδ − βγ = 1.
In a very recent work [23], the perfect fluid equations with the Schrödinger symmetry
were generalized so as to accommodate the `–conformal Galilei symmetry

∂ρ ∂(ρυi ) ∂p 1
+ = 0, ρD2` υi = − , p = νρ1+ `N , (50)
∂t ∂xi ∂xi
where ν is a constant. Conserved charges and the energy-momentum tensor were built as
well [23]. Our goal in this section, is to demonstrate that the group-theoretic approach allows
one to arrive at (50) in a rather natural and efficient way.
As the first step, one has to determine transformation laws of the fluid density and
the velocity vector field under the action of the `-conformal Galilei group. Repeating the
arguments in the preceding section, one finds (no sum over repeated index n)
`N 1−`  −`
∂t0 ∂t0 ∂ ∂t0
 
0 0 0
ρ(t, x) = ρ (t , x ); υi (t, x) = υi0 (t0 , x0 ) + x0i ;
∂t ∂t ∂t ∂t
(n)
ρ(t, x) = ρ0 (t0 , x0 ), υi (t, x) = υi0 (t0 , x0 ) − nai tn−1 , (51)

where the first line corresponds to the SO(2, 1)-transformations, while the second line de-
scribes the accelerations. Note that, similarly to the Schrödinger case, the derivative
1 ∂
∇i = ρ− N , (52)
∂xi
holds invariant under the transformations (49) and (51). Eqs. (51) will be used below in
order to link fields contributing to a group-theoretic element to ρ and υi , while (52) will help
to formulate invariant equations of motion.
Then one considers a natural generalization of the group theoretic element (29)
(0) (1) (1) (2`) (2`)
g = eitH eixi Ci eiυi (t,x)Ci
. . . eiυi (t,x)Ci
eiu(t,x)D eiw(t,x)K , (53)
(1)
where t and xi are the temporal and spatial coordinates, υi is identified with the fluid
(2) (2`)
velocity vector field υi , new vector fields υi , . . . , υi will be later linked to the material
derivatives of υi , whereas u, w are scalars to be expressed in terms of ρ and υi in accord
with the way in which they transform under the `-conformal Galilei group.

11
 Analyzing the left action of the group upon the element (53)
(0) (0) (1) (1) (2`) (2`)
g 0 = eiβH eiai Ci
eiai (t,x)Ci
. . . eiai (t,x)Ci
eiλD eiσK · g, (54)
(0) (2`)
where β, λ, σ, ai , . . . , ai are infinitesimal transformation parameters, one gets (each
transformation is separated by semicolon; no sum over repeated indices k and s)

t0 = t + β, x0i = xi ,
0(k) (k)
υi (t0 , x0 ) = υi (t, x), u0 (t0 , x0 ) = u(t, x), w0 (t0 , x0 ) = w(t, x);
t0 = (1 + λ)t, x0i = (1 + λ`) xi ,
0(k) (k)
υi (t0 , x0 ) = (1 − λ(k − `)) υi (t, x), u0 (t0 , x0 ) = u(t, x) + λ, w0 (t0 , x0 ) = w(t, x);
t0 = t + σt2 , x0i = (1 + 2σ`t)xi ,
0(k) (k)
υi (t0 , x0 ) = (1 − 2σ(k − `)t)υi (t, x) u0 (t0 , x0 ) = u(t, x) w0 (t0 , x0 ) = w(t, x)
(k−1)
− σ(k − 1 − 2`)υi (t, x), + 2σt + σeu(t,x) ;
(k)
t0 = t, x0i = xi + tk ai ,
0(k−s) (k−s) (k)
υi (t0 , x0 ) = υi (t, x) + ts Cks ai , u0 (t0 , x0 ) = u(t, x), w0 (t0 , x0 ) = w(t, x), (55)
(0) k!
where k = 0, . . . , 2`, s ≤ k, υi = xi , and Cks are the binomial coefficients Cks = s!(k−s)! .
Comparing (55) with the infinitesimal form of (51) (see [23]), one can express u, w in
terms of ρ and υi
(1)
1 − 1 ∂υi
ρ = e−N `u , w= ρ N` . (56)
2N ` ∂xi
(1)
Recall that υi was earlier identified with the fluid velocity vector field υi .
Afterwards, one computes the Maurer-Cartan invariants
(0) (0) (1) (1) (2`) (2`)
g −1 dg = iωH H + iωD D + iωK K + iωi Ci + iωi Ci + · · · + iωi Ci , (57)

then attends to an orbit of a liquid particle xi → xi (t), and finally imposes the constraints
(0) (1) (2`−1) (2`)
ωD = 0, ωi = 0, ωi = 0, ... ωi = 0, ωi + α∇i ωH = 0, (58)
(0)
where α is a real constant and ∇i is the invariant derivative (52). The equations ωi = 0,
(2`−1) (k)
. . . , ωi = 0 allow one to link υi , with k = 1, . . . , 2` − 1, to the material derivatives of
xi
(k) 1
υi = D k x i , (59)
k!

12
with D specified in (38). Substituting (56) into ωD = 0, one obtains the continuity equation,
(2`)
while the restriction ωi + α∇i ωH = 0 reproduces the generalized Euler equation entering
Eqs. (50), in which
(2`)!α
ν= . (60)
1 + N`
Like in the preceding section, the equation of state exposed in (50) arises automatically.
Note that for an arbitrary value of ` the explicit form of the Maurer-Cartan invariants is
rather complicated. As an illustration, we expose the ` = 23 case

ωH = e−u dt, ωD = du − 2we−u dt, ωK = dw − wdu + w2 e−u dt,

(0) 3u (1) (1) u (1) (2) 3u (1)
ωi = e− 2 (dxi − υi dt), ωi = e− 2 (dυi − 2υi dt) − 3we− 2 (dxi − υi dt),
(2) u (2) (3) u (1) (2) 3u (1)
ωi = e 2 (dυi − 3υi dt) − 2we− 2 (dυi − 2υi dt) + 3w2 e− 2 (dxi − υi dt),
(3) 3u (3) u (2) (3) u (1) (2)
ωi = e 2 dυi − we 2 (dυi − 3υi dt) + w2 e− 2 (dυi − 2υi dt)
3u (1)
−w3 e− 2 (dxi − υi dt). (61)

To summarize, within the method of nonlinear realizations, both the perfect fluid equa-
tions, which hold invariant under the action of the `-conformal Galilei group, and the equa-
tion of state come about naturally. It suffices to consider the group-theoretic element (53)
and then impose the constraints (58).

5. Perfect fluid equations with the Lifshitz symmetry

If one omits the special conformal transformation generator K in the Schrödinger algebra
(1)
(19), the commutators [H, D] and [D, Ci ] can be modified so as to include an arbitrary
constant z known as the dynamical critical exponent (see e.g. [24])

(1) (0) (0) i (0)

[H, D] = izH, [H, Ci ] = iCi , [D, Ci ] = − Ci ,
2
 
(1) 1 (1)
[D, Ci ] = i z − Ci , (62)
2

where, as before, i = 1, . . . , N . The resulting algebra is known as the Lifshitz algebra.9 It is

conventionally represented by the differential operators acting in a nonrelativistic spacetime
parameterized by (t, xi )

∂ ∂ i ∂ (0) ∂ (1) ∂
H=i , D = izt + xi , Ci =i , Ci = it . (63)
∂t ∂t 2 ∂xi ∂xi ∂xi
9
To be more precise, in modern literature by the Lifshitz algebra one usually means (62), in
 which the gen-

(1) ∂ ∂
erator of Galilei boosts Ci is discarded, while the generators of spatial rotations Mij = i xi ∂x j
− xj ∂x i
are reinstated.

13
In particular, D in (63) gives rise to the anisotropic scaling of the temporal and spatial
coordinates
λ
t0 = eλz t, x0i = e 2 xi , (64)
λ being the transformation parameter. The goal of this section is to obtain perfect fluid
equations invariant under the action of the Lifshitz group and the corresponding equation
of state within the method of nonlinear realizations.
Guided by the analysis in the preceding sections, one directly attends to the group-
theoretic element (0) (1)
g = eitH eixi Ci eiυi (t,x)Ci eiu(t,x)D , (65)
and then establishes (finite) transformation laws of the coordinates and fields under the
Lifshitz group (each transformation is separated by semicolon)
t0 = t + β, x0i = xi ,
υi0 (t0 , x0 ) = υi (t, x), u0 (t0 , x0 ) = u(t, x);
λ
t0 = eλz t, x0i = e 2 xi ,
1
υi0 (t0 , x0 ) = e−λ(z− 2 ) υi (t, x), u0 (t0 , x0 ) = u(t, x) + λ;
(0)
t0 = t, x0i = xi + ai ,
υi0 (t0 , x0 ) = υi (t, x), u0 (t0 , x0 ) = u(t, x);
(1)
t0 = t, x0i = xi + tai ,
(1)
υi0 (t0 , x0 ) = υi (t, x) + ai , u0 (t0 , x0 ) = u(t, x), (66)
where β and λ are (finite) parameters associated with the temporal translation and anisotropic
(0) (1)
scaling transformation, respectively, while ai , ai are related to the spatial translations and
the Galilei boosts. Then one computes the Maurer-Cartan invariants
(0) (0) (1) (1)
g −1 dg = iωH H + iωD D + iωi Ci + iωi Ci , (67)
where
1
= e(z− 2 )u dυi .
(0) u (1)
ωH = e−zu dt, ωD = du, ωi = e− 2 (dxi − υi dt), ωi (68)
In order to make contact with fluid mechanics, one introduces the fluid density ρ(t, x) and
the velocity vector field υi (t, x). By making recourse to (21), one finds the transformation
law of the density under the dilatation
λN
ρ(t, x) = e 2 ρ0 (t0 , x0 ), (69)
while ρ(t, x) = ρ0 (t0 , x0 ) for other transformations from the Lifshitz group. Likewise, con-
sidering an orbit of a liquid particle as in Eq. (23) above and attending to (66), one can
identify υi (t, x) in (65), (66) with the fluid velocity vector field.

14
 Comparing the transformation laws (66) and (69), one can link u to ρ
N
ρ = e− 2 u . (70)

The invariant zero-form

2 − 2z ∂υi
Ω=− ρ N , (71)
N ∂xi
and the invariant derivative
1 ∂
∇i = ρ− N (72)
∂xi
can be found as well.
Finally, imposing the constraints built from the invariants
(1)
ωD + ΩωH = 0, ωi + α∇i ωH = 0, (73)

where α is a constant, and taking into account the field redefinition (70), one reduces the first
equation in (73) to the continuity equation, while the second condition gives the conventional
Euler equation and the equation of state
∂ρ ∂(ρυi ) ∂p 2(2z−1)
+ = 0, ρDυi = − , p = νρ1+ N . (74)
∂t ∂xi ∂xi
2αz
Here ν is a constant, which links to α in (73) via ν = N +2(2z−1) . At z = 1 one reproduces
the result in Sect. 2.
Note that, because each factor contributing to the left hand side of the Euler equation
scales in a concrete way under the dilatation transformation, it comes as no surprise that
choosing the pressure to be a power function of the density one can fix the exponent so
as to ensure the invariance of the entire equation. Similar reasoning worked out in two
preceding sections. To the best of our knowledge, Eqs. (74) have not yet been discussed in
the literature.
A few comments are in order. Taking into account the transformation laws (66), (69)
and the fact that the material derivative D introduced in (38) is invariant under each trans-
formation except for the dilatation, for which D = eλz D0 , one can verify that the first two
equations in (74) do hold invariant under the Lifshitz group. One reveals a subtlety in
constructing conserved charges, however. Using the relation
Z  Z
∂
dxρA = dxρDA, (75)
∂t V V

where A(t, x) is an arbitrary function, D is the material derivative, and dx = dx1 . . . dxN ,
which holds true due to the continuity equation and the assumption that ρ vanishes at the
boundary of the volume element V , one can check that
Z Z Z  
(0) (1) (0) 1
Ci = dxρυi , Ci = tCi − dxρxi , H = dx ρυi υi + V (ρ) , (76)
V V V 2

15
are conserved over time. Here V (ρ) is the potential, which via the Legendre transform gives
the pressure [6]

N
p(ρ) = ρV 0 (ρ) − V (ρ) ⇒ V (ρ) = p(ρ). (77)
2(2z − 1)

However, when trying to build a conserved charge associated with the dilatation, which is
typically of the form [6] Z
1
D = tH − dxρxi υi , (78)
2 V
one reveals a problem. In view of (77), D in (78) is conserved for z = 1 only.
When equations of motion exhibit invariance under a given transformation group, the
corresponding action functional may fail to do so. A well known example is the mechanical
similarity [25], which is illustrated here by the Lifshitz analog [26] of the 1d conformal
mechanics (13)
(2z − 1)γ 2
ρ̈ = . (79)
ρ4z−1
Whereas the equation of motion is invariant under the anisotropic conformal transformation
λ
t0 = eλz t, ρ0 (t0 ) = e 2 ρ(t), (80)

the corresponding action functional

γ2
Z  
1 2
S= dt ρ̇ − 4z−2 (81)
2 ρ

scales as S 0 = eλ(1−z) S. Only for z = 1 the action is invariant and the conserved charge
1
D = tH − ρρ̇, (82)
2
 2

where H = 21 ρ̇2 + γρ2 is the energy, can be constructed via Noether’s theorem.
It is customary to link conservation of energy and momentum to conservation of the
energy-momentum tensor T µν , with µ = (0, i) and i = 1, . . . , N , which for the case at hand
reads
 
00 1 i0 1 0
T = ρυi υi + V (ρ), T = ρυi υi υi + V (ρ) , ∂µ T µ0 = 0,
2 2
T 0i = ρυi , T ji = pδji + ρυj υi , ∂µ T µi = 0. (83)

Within the fluid mechanics, the scale invariance is sometimes imposed by representing the
dilatation generator in the form D = iξ µ ∂µ and demanding the current T µν ξν , where ξν =
ηνµ ξ µ and ηνµ = diag(+, −, . . . , −), to be conserved ∂µ (T µν ξν ) = 0. As was explained above,

16
for the case at hand such a current exists for z = 1 only, whereas the equations of motion
(74) hold invariant for an arbitrary value of z.

6. Perfect fluid equations invariant under relativistic conformal group

We now proceed to studying the case of the relativistic conformal algebra, which is
specified by the structure relations
[D, Pi ] = −iPi , [D, Ki ] = iKi ,
[Pi , Kj ] = 2iηij D − 2iMij , [Mij , Pl ] = −iηil Pj + iηjl Pi ,
[Mij , Kl ] = −iηil Kj + iηjl Ki , [Mij , Mkl ] = −iηik Mjl − iηjl Mik + iηjk Mil + iηil Mjk , (84)
where i = 0, . . . , N − 1 and ηij = diag(+, −, . . . , −) is the Minkowski metric. Above D
is identified with the dilatation generator, Pi links to translations in relativistic spacetime,
Ki is related to special conformal transformations, while Mij determines the Lie algebra
of the Lorentz group. A conventional realization of the generators in Lorentzian spacetime
parameterized by coordinates xi reads
 
i ∂ ∂ j ∂ j ∂
D = ix , Pi = i i , Ki = i 2xi x − x xj i ,
∂xi ∂x ∂xj ∂x
 
∂ ∂
Mij = i xi j − xj i . (85)
∂x ∂x
As usual, the indices are raised and lowered with the use of the Minkowski metric and its
inverse. Our objective in this section is to inquire whether the group-theoretic approach is
capable of producing relativistic fluid equations with conformal symmetry.
Given the algebra (84), it seems natural to start with the group-theoretic element
i i ij (x)M
g = eix Pi eiυ (x)Ki eiu(x)D eif ij
, (86)
where xi are identified with the coordinates parameterizing an N -dimensional Minkowski
spacetime and υ i (x), u(x), f ij (x) = −f ji (x) are fields on it. Proceeding similarly as above,
one obtains the infinitesimal conformal transformations acting upon the coordinates and
fields (each transformation is separated by semicolon)
x0i = xi + β i , υ 0i (x0 ) = υ i (x), u0 (x0 ) = u(x), f 0ij (x0 ) = f ij (x);
x0i = (1 + λ)xi , υ 0i (x0 ) = (1 − λ)υ i (x), u0 (x0 ) = u(x) + λ, f 0ij (x0 ) = f ij (x);
x0i = xi − σ i (xx) + 2xi (σx), υ 0i (x0 ) = (1 − 2(σx))υ i (x) − 2σ i (xυ) + 2xi (συ) + σ i ,
0ij (x0 )M k σp M ij (x)M
u0 (x0 ) = u(x) + 2(σx), eif ij
= e−2ix kp
eif ij
;
x0i = xi − ξ ij xj , υ 0i (x0 ) = υ i (x) − ξ ij υj (x), u0 (x0 ) = u(x),
0ij (x0 )M i kp M ij (x)M
eif ij
= e2ξ kp
eif ij
, (87)

17
where β i , λ, σ i , ξ ij = −ξ ji are transformation parameters corresponding to translations,
dilatations, special conformal transformations, and Lorentz transformations, respectively,
and we abbreviated (ab) = ai bi . The Maurer-Cartan invariants g −1 dg = iωD D + iωPi Pi +
i ij
iωK Ki + iωM Mij read

ωD = du − 2(dxυ), ωPi = e−u dxj (exp(−2f ))j i ,

i
= eu dυ j − dxj (υυ) + 2υ j (dxυ) (exp(−2f ))j i


ωK
 
ij ij i j −if kp Mkp if ls Mls
ωM Mij = df Mij + 2dx υ e Mij e , (88)

where (exp(−2f ))j i stands for the conventional matrix exponent.

At this point one reveals a problem. For one thing, the transformation laws of xi in (87)
reproduce the generators (85), which means that the identification of xi in the group-theoretic
element (86) with the coordinates parameterizing a Lorentzian manifold was correct. For
another thing, among the Maurer-Cartan one-forms (88) one does not find an invariant,
which would link dxi to υ i . For each nonrelativistic example above, there was such a relation,
which helped us to identify υi with the fluid velocity vector field. Furthermore, from(87) it
0i
0i 0
follows that υ does not transform as a contravariant vector field, i.e. υ (x ) 6= ∂x
i
∂xj
υ j (x)
with x0i = x0i (x) taken from (87), and, hence, it cannot be identified with the fluid velocity
vector field. Rather, ωD suggests that υi can be linked to the gradient of u. Thus, the
construction of relativistic fluid equations invariant under the conformal group in terms of
the invariants (88) alone appears to be problematic.
Yet, discarding υi and focusing solely on xi , u(x) and their transformation laws displayed
in (87), one can build the desired equations. Firstly, one considers the conformal time s

ds2 = e−2u(x) ηij dxi dxj , (89)

which remains intact under the transformations (87). Then one introduces the velocity vector
i
field V i (x), which is identified with dx
ds
when restricted upon a liquid particle orbit i i
 x = x (s).
∂x0i
By definition, V i (x) transforms as a contravariant vector field V 0i (x0 ) = ∂xj
V j (x) if
x0i = x0i (x) is taken from (87). In view of (89), it is constrained to obey

V i Vi = e2u , (90)

meaning that V i is a time-like vector. In particular, V 0 can be expressed in terms of V α ,

with α = 1, . . . , N − 1, and u. Note that imposing (90) is consistent with the conformal
transformations (87).
Taking into account that within the relativistic framework the operator V i ∂i , where
∂i = ∂x∂ i , is an analog of the material derivative (38) and analyzing the way in which V j ∂j V i
is changed under the conformal transformations (87), one finds that introducing two more

18
terms of the type V i V j ∂j u and e2u η ij ∂j u yields the invariant equation10

V j ∂j V i − 2V i V j ∂j u + e2u η ij ∂j u = 0. (91)

Note that, in view of (90), a contraction of (91) with Vi gives zero meaning that only
N − 1 equations are independent. This correlates with the fact that V 0 links to the spatial
components V α and u via (90). In a similar fashion, one can build a relativistic counterpart
of the continuity equation
∂i e−N u V i = 0,


(92)
where N is the spacetime dimension, which holds invariant under the conformal transfor-
mations (87).
Finally, introducing the energy-momentum tensor
1 −N u ij
T ij = e−(N +2)u V i V j − e η , ∂i T ij = 0, (93)
N
which is suggested by Eq. (91) and focusing on the T 00 -component
 
00 −(N +2)u α α N − 1 −N u
T =e V V + e , (94)
N
where V α , with α = 1, . . . , N − 1, are the spatial components of the fluid velocity vector
field, one concludes that  
N − 1 −N u
ε= e (95)
N
can be identified with the energy density (measured in the comoving frame). Likewise,
analyzing the stress tensor T αβ in the comoving frame, one determines the pressure
1 −N u
p= e , (96)
N
and establishes the equation of state
1
p(ε) = ε. (97)
N −1
The equations obtained above reproduce the ultrarelativistic limit of the relativistic fluid
mechanics [27].

7. Conclusion

To summarize, in this work a possibility to construct perfect fluid equations with con-
formal symmetry within the method of nonlinear realizations was studied. Four cases were
10
To be more precise, under the conformal transformations (87) the expression on the left hand side of
(91) transforms as a contravariant vector field. Setting it to zero, one obtains the invariant equation.

19
discussed in detail, which included the Schrödinger group, the `-conformal Galilei group,
the Lifshitz group, and the relativistic conformal group. While the method proved rather
efficient for the nonrelativistic groups, in particular yielding new results for the Lifshitz
case, within the relativistic framework the construction faced certain difficulties and extra
arguments needed to be invoked.
Turning to possible further developments, it is interesting to inquire whether thermody-
namic description of perfect fluids with conformal symmetry can be accommodated within
the group-theoretic framework.
A possibility to combine the method of nonlinear realizations with the approach in [11]
is worth studying as well.
In classical mechanics, having a potential energy, which is a homogeneous function of an
argument, results in the mechanical similarity [25], Kepler’s third law being the celebrated
example. Physical implications of the anisotropic conformal scaling symmetry of the Lifshitz
perfect fluid need to be better understood.
A Hamiltonian formulation for the equations of motion in Sect. 4 and 5 is worth exploring
as well. Of particular interest here are nontrivial Poisson brackets among the fields and their
origin [6].

Acknowledgements

The author thanks V.P. Nair for a useful correspondence.

References
[1] G. Policastro, D.T. Son, A.O. Starinets, From AdS/CFT correspondence to hydrody-
namics, JHEP 0209 (2002) 043, hep-th/0205052.

[2] R. Baier, P. Romatschke, D.T. Son, A.O. Starinets, M.A. Stephanov, Relativistic
viscous hydrodynamics, conformal invariance, and holography, JHEP 04 (2008) 100,
arXiv:0712.2451.

[3] S. Bhattacharyya, V.E. Hubeny, S. Minwalla, M. Rangamani, Nonlinear fluid dynamics

from gravity, JHEP 0802 (2008) 045, arXiv:0712.2456.

[4] M. Rangamani, Gravity and hydrodynamics: Lectures on the fluid-gravity correspon-

dence, Class. Quant. Grav. 26 (2009) 224003, arXiv:0905.4352.

[5] L. O’Raifeartaigh, V.V. Sreedhar, The maximal kinematical invariance group of

fluid dynamics and explosion-implosion duality, Annals Phys. 293 (2001) 215, hep-
th/0007199.

[6] R. Jackiw, V.P. Nair, S.Y. Pi, A.P. Polychronakos, Perfect fluid theory and its exten-
sions, J. Phys. A 37 (2004) R327, arXiv:hep-ph/0407101.

20
 [7] I. Fouxon, Y. Oz, CFT hydrodynamics: symmetries, exact solutions and gravity, JHEP
0903 (2009) 120, arXiv:0812.1266.

[8] S.R. Coleman, J. Wess, B. Zumino, Structure of phenomenological Lagrangians. I, Phys.

Rev. 177 (1969) 2239.

[9] E.A. Ivanov, V.I. Ogievetsky, The inverse Higgs phenomenon in nonlinear realizations,
Theor. Math. Phys. 25 (1975) 1050.

[10] A.A. Kirillov, Lectures on the orbit method, Graduate Studies in Mathematics, vol. 64,
2004.

[11] B. Bistrovic, R. Jackiw, H. Li, V.P. Nair, S.-Y. Pi, Non-abelian fluid dynamics in
Lagrangian formulation, Phys. Rev. D 67 (2003) 025013, hep-th/0210143.

[12] V.P. Nair, R. Ray, S. Roy, Fluids, anomalies and the chiral magnetic effect: A group-
theoretic formulation, Phys. Rev. D 86 (2012) 025012, arXiv:1112.4022.

[13] D. Capasso, V.P. Nair, J. Tekel, The isospin asymmetry in anomalous fluid dynamics,
Phys. Rev. D 88 (2013) 085025, arXiv:1307.7610.

[14] D. Karabali, V.P. Nair, Relativistic particle and relativistic fluids: Magnetic moment
and spin-orbit interactions, Phys. Rev. D 90 (2014) 105018, arXiv:1406.1551.

[15] G.M. Monteiro, A.G. Abanov, V.P. Nair, Hydrodynamics with gauge anomaly: Vari-
ational principle and Hamiltonian formulation, Phys. Rev. D 91 (2015) 125033,
arXiv:1410.4833.

[16] V.P. Nair, Topological terms and diffeomorphism anomalies in fluid dynamics and sigma
models, Phys. Rev. D 103 (2021) 085017, arXiv:2008.11260.

[17] E. Ivanov, S. Krivonos, V. Leviant, Geometry of conformal mechanics, J. Phys. A 22

(1989) 345.

[18] V. de Alfaro, S. Fubini, G. Furlan, Conformal invariance in quantum mechanics, Nuovo

Cim. A 34 (1976) 569.

[19] A. Galajinsky, Schwarzian mechanics via nonlinear realizations, Phys. Lett. B 795
(2019) 277, arXiv:1905.01935.

[20] P.A. Horvathy, P.-M. Zhang, Non–relativistic conformal symmetries in fluid mechanics,
Eur. Phys. J. C 65 (2010) 607, arXiv:0906.3594.

[21] M. Henkel, Local scale invariance and strongly anisotropic equilibrium critical systems,
Phys. Rev. Lett. 78 (1997) 1940, cond-mat/9610174.

21
[22] J. Negro, M.A. del Olmo, A. Rodriguez-Marco, Nonrelativistic conformal groups, J.
Math. Phys. 38 (1997) 3786.

[23] A. Galajinsky, Equations of fluid dynamics with the `-conformal Galilei symmetry, Nucl.
Phys. B 984 (2022) 115965, arXiv:2205.12576.

[24] M. Taylor, Lifshitz holography, Class. Quant. Grav. 33 (2016) 033001, arXiv:1512.03554.

[25] L.D. Landau, E.M. Lifshitz, Mechanics, 3rd. ed., Pergamon Press, 1976.

[26] A. Galajinsky, Dynamical realizations of the Lifshitz group, Phys. Rev. D 105 (2022)
106023, arXiv:2201.10187.

[27] L.D. Landau, E.M. Lifshitz, The Classical Theory of Fields, Pergamon Press, 1951.

22