The mass of sound

Angelo Esposito, Rafael Krichevsky, and Alberto Nicolis

Department of Physics, Center for Theoretical Physics,
Columbia University, 538W 120th Street, New York, NY, 10027, USA

We show that the commonly accepted statement that sound waves do not transport mass is only
true at linear order. Using effective field theory techniques, we confirm the result found in [1] for
zero-temperature superfluids, and extend it to the case of solids and ordinary fluids. We show that,
in fact, sound waves do carry mass—in particular, gravitational mass. This implies that a sound
wave not only is affected by gravity but also generates a tiny gravitational field. Our findings are
valid for non-relativistic media as well, and could have intriguing experimental implications.

PACS numbers: 46.40.Cd, 47.10.-g, 47.35.Rs, 47.37.+q

arXiv:1807.08771v1 [hep-th] 23 Jul 2018

Keywords: Phonons, Sound Waves, Superfluids, Fluids, Solids

I. INTRODUCTION effect, and that is why from a linearized analysis we usu-

It is usually said that sound waves do not transport
mass. They carry momentum and energy, and lead to
temporary oscillations of the local mass density of any
region they happen to pass through, but it is an accepted
fact that the net mass transported by a sound wave van-
ishes. Here, we want to question this “fact”.
A first indication that sound waves can in fact carry a
nonzero net mass is contained in the results of [1]: there,
using an effective point-particle theory, it was shown that
phonons in zero-temperature superfluids have an effec-
tive coupling to gravity, which depends solely on their
energy (or momentum) and on the superfluid’s equation
of state. For ordinary equations of state, such a coupling
corresponds to a negative effective gravitational mass: in
the presence of an external gravitational field, such as
that of Earth, a phonon’s trajectory bends upwards.
Now, this effect is completely equivalent to standard
refraction in the form of Snell’s law: in the presence
of gravity, the pressure of the superfluid depends on
depth, and so does the speed of sound. As a result, in
the geometric acoustics limit sound waves do not prop-
agate along straight lines. Because of this, one might
be tempted to dismiss any interpretation of this phe-
nomenon in terms of “gravitational mass”. However,
since in the formalism of [1] the effect is due to a coupling
with gravity in the effective Lagrangian or Hamiltonian of
the phonon, the same coupling must affect the field equa-
tion for gravity. That is, the (tiny) effective gravitational
mass of the phonon generates a (tiny) gravitational field.
And the source of this gravitational field travels with the
phonon. Thus, in a very physical sense, the phonon car-
ries (negative) mass. Moreover, this is not due to the
usual equivalence of mass and energy in relativity: the
effect survives in the non-relativistic limit. And, finally,
it is not a quantum effect, because the formalism of [1]
applies unaltered to classical wave packets.
In this paper we want to confirm this result by comput-
ing explicitly the mass carried by a classical sound wave
packet. As we will see, from the wave mechanics stand-
point the fact that such a mass is nonzero is a non-linear
ally infer that sound waves do not carry mass. We also
want to generalize the result to sound waves in ordinary
fluids and to longitudinal and transverse sound waves in
solids. The universal result that we will find is that, in
the non-relativistic limit, the mass M carried by a sound
wave packet traveling in these media is its energy E times
a factor that only depends on the medium’s equation of
state:
ρm dcs
M =− E, (1)
cs dP
where ρm is the medium’s mass density, P its pressure,
and cs the speed of the sound wave under consideration.
(For solids, the correct formula is a minor generalization
of this.)
A final qualification, perhaps relevant for the reader
with a high-energy/particle-physics mindset: in special
relativity, for particles or excitations propagating in a
Poincaré-invariant vacuum, the invariant mass is a funda-
mental quantity that labels inequivalent irreducible rep-
resentations of the Poincaré group. As a result, it is
a completely unambiguous concept. Its value also deter-
mines directly other physical quantities—for instance the
gap, i.e. the energy of the particle or excitation at zero
momentum. Not so in a medium that breaks boost invari-
ance: there, the dispersion relation E(p) of an excitation
is not constrained by the symmetries anymore, and, as a
consequence, there is no “invariant mass”. What do we
mean by ‘M ’ in (1) then? A non-relativistic medium—
i.e. one with mass density (times c2 ) much bigger than
its energy density—has a well-defined and conserved to-
tal mass. In such a medium, the mass M carried by an
excitation is simply the fraction of this total mass that
travels with the excitation. In particular, it is perfectly
consistent with the symmetries of the system to have gap-
less excitations (E(p) → 0 for p → 0) that nevertheless
carry mass1 . For technical convenience, we use gravity
1
Because of this, we find the use of “massive” instead of “gapped”
in [2] somewhat inaccurate.
 2

to probe how much mass our excitations carry, since in III. ZERO-TEMPERATURE SUPERFLUIDS
the non-relativistic limit gravity only couples to mass.
A. Effective field theory
Conventions: Throughout the paper we set ~ = 1
but keep powers of c (speed of light) explicit. This The low-energy dynamics of a zero-temperature super-
will make the non-relativistic limit straightforward. We fluid can be characterized in terms of a single scalar field
adopt a metric signature ηµν = diag(−, +, +, +). More- φ(x), with low-energy effective action [8, 9]
over, Greek indices run over all spacetime coordinates, Z
while Latin indices run over spatial coordinates only. S = dtd3 x P (X) , X ≡ −c2 ∂µ φ∂ µ φ , (2)

expanded about the ground state

II. GENERAL FRAMEWORK AND STRATEGY hφ(x)i = µt , (3)

where µ is the (relativistic) chemical potential. The func-
As a general framework, we will use the recently de- tion P above is the pressure of the superfluid as a function
veloped effective field theories for the gapless excitations of the square of the chemical potential.
of generic media. These can be characterized in terms of The superfluid phonon field π(x) corresponds to fluctu-
spontaneous symmetry breaking. In particular, all media ations of the scalar φ(x) around its background, φ(x) =
break at least part of the fundamental spacetime symme- µ(t + π(x)).
tries of Nature, which are spacetime translations, spatial Expanding the action above up to cubic order in π
rotations, and Lorentz boosts (the Poincaré group) [3]. yields
The Goldstone modes associated with the spontaneous µn̄c2 c2 ~
2

1
Z
breaking of these symmetries are gapless, and correspond S→ 2 dtd3 x π̇ 2 − s ∇π
cs 2 2
to the collective excitations of the system; the phonons 2
 2
  (4)
c c ~ 2 + g π̇ 3 .
in a solid are a prototypical example of this (see e.g. [4– 1 − 2s π̇ ∇π


−
7]). Such an approach has now been applied to a vari- 2 c 3
ety of contexts, including superfluids [8–11], fluids [12– The equilibrium number density n̄, the sound speed cs ,
17], solids [5, 12, 13, 18], and supersolids [19, 20]—see and the cubic coupling g are given by
also [21–24].
dP c2 P ′
Our general strategy will be the following. We want n̄ = √ , c2s = ′ ,
d X P + 2XP ′′
to compute the mass carried by a sound wave packet. (5)
c2 c2
 
For each of the media that we consider, we will thus use µ dcs
g = 2 1 − s2 − ,
the corresponding effective action to derive the equations 2cs c cs dµ
of motion and the energy-momentum tensor2 associated where the primes denote differentiation with respect to
with gapless excitations. In perturbation theory in the X and all the quantities above are evaluated on the back-
amplitude of our waves, we will keep the linear terms and ground. Notice that the standard relation for compres-
the leading non-linear ones, since, as we will see, the effect sional sound waves c2s = c2 dP/dρ holds, since the rela-
we are interested in vanishes at linear order. The mass of tivistic energy density is ρ = 2XP ′ − P (see below).
a wave packet will thus be—in the non-relativistic limit— By varying the action (4) with respect to π, one finds
the spatial integral of c12 T00 evaluated on a solution of the the following equation of motion to quadratic order:
equations of motion. In the case of superfluids, the effect
c2
 
will come directly from quadratic terms in T00 evaluated π̈ − c2s ∇2 π = c2 1 − 2s π̇∇2 π
on solutions of the linearized equations of motion. In the c
(6)
case of ordinary fluids and solids instead, the effect will c2
   
+ ∂t c2 1 − 2s ~ 2 − g π̇ 2 .


come from linear terms in T00 evaluated on solutions of ∇π
c
the nonlinear equations of motion. In general, for other
media, one in principle could have a combination of these We can solve this equation to non-linear order by writing
two contributions. π = π(1) +π(2) +. . . , where π(1) is a localized wave packet
that solves the linear equation of motion,
Finally, for technical convenience, we will keep a rela-
tivistic notation and take the non-relativistic limit only π̈(1) − c2s ∇2 π(1) = 0 , (7)
when needed. and π(2) a quadratic correction, which is sourced by π(1)
itself,
c2
  
π̈(2) − c2s ∇2 π(2) = ∂t c2 1 − 2s ~ (1) 2 − g π̇ 2


∇π (1)
c
2 It is in this sense that we use gravity to probe our sound waves’
c2 c2
  
mass: gravity interacts with any system through the system’s + 2 1 − 2s π̇(1) 2
, (8)
energy-momentum tensor. 2cs c
 3

where we used the equation for π(1) . The detailed so- The above result is fully relativistic. In the non-
lution is unimportant, but we will keep this approach in relativistic limit, we have that the mass carried by our
mind for the next subsection, where we study the dynam- wave packet is
ics of phonons to determine their gravitational mass.
1 R ρm dcs
M= 2
h T00 i ≃ − E, (14)
c cs dP
B. Gravitational mass
which is indeed Eq. (1), as obtained in [1] from the point-
particle effective theory. In Appendix A we also show
The energy-momentum tensor of our superfluid is that the same result can be recovered in quantum field
found by varying the effective action w.r.t. to the metric: theory for single-phonon states.
Tµν = P (X)ηµν + 2c2 P ′ (X)∂µ φ∂ν φ . (9)
IV. SOLIDS AND FLUIDS
Expanding T00 to quadratic order in small fluctuations,
we find
A. Effective field theory
µn̄c2 c2
  
2µcs dcs
T00 = 2 π̇ + 2 1 − 2 π̇ 2
cs 2cs c dµ As mentioned in the Introduction, the bending of a
(10)
c2

2c2s


2
 sound waves’s trajectory in the presence of gravity is
− 1− 2 ~
∇π . equivalent to Snell’s law. In light of this, given the gen-
2 c
erality of Snell’s law, we expect sound waves to interact
To find the mass transported by the sound wave, we with gravity in fluids and solids much in the same way as
integrate T00 over a volume containing the wave packet they do in superfluids. We will take a parallel approach
and average it over a time interval τ much larger than to analyze phonons in a solid from an effective field the-
the typical oscillation time, τ ≫ 1/ω. As far as the linear ory point of view. Moreover, since in the EFT language
term in T00 is concerned, it can receive contributions both a fluid is simply a solid with an enhanced symmetry (see
from the linearized wave packet π(1) and from the non- e.g. [3, 18]), the results of this section extend straight-
linear correction π(2) . However, the contribution from forwardly to fluids as well. The effective field theory of
π(1) vanishes, since it is a total time derivative and we are solids has been developed in [12, 13, 18], and we will in
integrating on a time interval over which the wave packet large part follow their notation.
averages to zero. The contribution from π(2) vanishes as The EFT for a three-dimensional solid can be built out
well: the non-linear source on the r.h.s. of (8) will, in of three scalar fields, φI (I = 1, 2, 3), representing the co-
general, contain both oscillatory terms as well as constant moving coordinates of the solid’s volume elements. In the
ones, but the time derivative gets rid of the latter. This ground state, they take the values
means that π(2) will be purely oscillatory as well, and 1/3
hence its time derivative will vanish in our time average. hφI (x)i = b0 xI , (15)
As far as the quadratic terms in T00 are concerned, only
where b0 is a dimensionless free parameter representing
the linearized wave packet π(1) can give a contribution to
the degree of compression or dilation of the solid. (The
this order in perturbation theory. Then, we can relate
cubic root convention has been chosen for later conve-
their volume integral and time average to the energy of
nience.)
our wave packet, that is the value of the free Hamiltonian
The low energy effective action is a generic function of
that we get from (4),
the combination
µn̄c2 1h
Z
B IJ ≡ ∂µ φI ∂ µ φJ ,

i
E=H= 2 ~ 2 .
d3 x π̇ 2 + c2s ∇π (11) (16)
cs 2
that is invariant under the rotational symmetries of the
Representing the combination
R of volume integral and solid, acting on the I, J indices. For simplicity, we will
time average with h . . . i, and using the virial theorem, focus on solids that are isotropic at large distances 3 . In
we have such a case, there are only three independent invariants
c2s that we can form out of B IJ . We will find it convenient
~ 2i =
h π̇ 2 i = c2s h ∇π
R R

E, (12) to work with
µn̄c2
√ trB 2 trB 3
and so, putting everything together, b≡ det B , Y ≡ 2 , Z≡ 3 , (17)
(trB) (trB)
R  µ dcs   ρ + P dcs 
h T00 i = 1 − E = 1− E, (13)
cs dµ cs dP
3
where in the last step we used the zero-temperature ther- Generalizing our analysis to a solid with discrete rotational sym-
modynamic identities ρ + P = µn, dP = ndµ. metry is straightforward, although quite tedious.
 4

where the normalizations of Y and Z have been chosen where the prime, subscript Y , and subscript Z denote,
so that b is the only quantity that retains information respectively, differentiation with respect to b, Y , and Z,
about b0 —i.e., Y and Z are invariant under an overall and all derivatives are evaluated on the background. No-
rescaling of our solid. The most general action for our tice that, with the exception of f ′ , the derivatives of f
solid will then be given by with respect to its arguments are of order c2L or c2T , and
Z not of order c2 [13]. This is what makes T00 so simple in
S = −w0 dtd3 x f (b, Y, Z) , (18) the non-relativistic limit—see below.
The action (20) implies the following equations of mo-
where w0 is a constant with the units of a mass density, tion to quadratic order:
which we pulled out for convenience. It is convenient
to set w0 to be 1/c2 times the value of the relativistic π̈i − c2L ∂i ∇ ~ · ~π = ∂j Aij + ∂t Bi ,
~ · ~π − c2T ∇2 πi − ∂i ∇


enthalpy density ρ + P on the ground state. With such (22)
a normalization, on the background one has ∂f /∂b =
c2 [13], which we will implicity use below. where
A perfect fluid can be viewed as an infinitely symmet-
~ · ~π 2
Aij = g1 δij ~π˙ 2 + g2 π̇i π̇j + 3g3 δij ∇


ric solid [3, 18], whose action is invariant under volume
preserving diffeomorphisms acting on the φI fields. Then, + g4 δij (∂k πℓ )2 + 2g4 ∂j πi ∇ ~ · ~π
the only invariant in this case is det B, and we can re-
2
cover the fluid EFT from the solid one by simply setting + g5 δij ∇ ~ × ~π + 2g5 ∂j πi ∇ ~ · ~π (23)
to zero all derivatives of f with respect to Y or Z. ~ · ~π + g6 ∂i πk ∂j πk + g6 ∂k πi ∂k πj
− 2g5 ∂i πj ∇
Sound waves will once again correspond to the fluctu-
ations of the fields about their backgrounds: + g6 ∂k πi ∂j πk + 3g7 ∂i πk ∂k πj .

φI (x) = b0 1/3 xI + π I (x) .

(19) The vector Bi in Eq. (22) is a quadratic combination of
fields, and, as we will see, its particular form is irrelevant
Expanding the invariants for small fluctuations and drop-
for what follows.
ping total derivatives, one obtains the action to cubic
order:
c2 ~
2 c2

1
Z
~ × ~π 2 B. Gravitational mass
S → w0 b0 dtd x ~π˙ 2 − L ∇
3
· ~π − T ∇


2 2 2

3 Let us now find the gravitational mass carried by sound
˙ 2~ ~
− g1~π ∇ · ~π − g2 π̇i π̇j ∂i πj − g3 ∇ · ~π
(20) waves in a solid (or fluid). It is convenient to take the
~ · ~π − g5 ∇
− g4 (∂i πj )2 ∇ ~ × ~π 2 ∇~ · ~π


non-relativistic limit right away, because the only O(c2 )
term in T00 turns out to be

− g6 ∂i πj ∂k πi ∂k πj − g7 ∂i πj ∂k πi ∂j πk .
~ · ~π .
T00 ≃ c2 w0 b0 ∇ (24)
We defined the longitudinal and transverse sound speeds
and cubic coupling constants as follows: In particular, with our parameterization of fluctuations,
16 fY + fZ quadratic and higher-order terms contribute to the en-
c2L = b0 F ′′ + , (21a) ergy density of the solid, but not to its mass density.
27 b0 The mass of a sound wave is therefore
4 fY + fZ
c2T = , (21b) R
~ · ~π i .
9 b0 M = w0 b0 h ∇ (25)
1 c2 c2
g1 = − − L2 + T2 , (21c) The volume integral reduces to a boundary term and,
2 2c c
for large enough volumes, one might be tempted to dis-
c2
g2 = 1 − T2 , (21d) card it. This is certainly allowed for a linearized solu-
c tion, since we can take it to be as localized as we wish.
c2 − c2T 1 10 16 fZ However, in general we must be careful about non-linear
g3 = L + b0 cL c′L − b0 cT c′T + , (21e)
3 3 9 243 b0 corrections. Splitting as before the fluctuation field as
c2 8 fZ ~π = ~π(1) + ~π(2) + . . . , from (22) one sees that for a local-
g4 = − L + 2b0 cT c′T − , (21f) ized wave packet π(1) , the nonlinear correction π(2) can
2 27 b0
c2 − c2T 4 fZ scale as 1/r2 at large distances, thus potentially giving a
g5 = L − b0 cT c′T + , (21g) finite contribution to (25).
2 27 b0 To compute such a contribution, we first take the di-
2 fZ vergence of (22), obtaining a wave equation directly for
g6 = c2T + , (21h)
9 b0 Ψ≡∇ ~ · ~π(2) , with ~π(1) -dependent sources:
2 fZ
g7 = , (21i)
27 b0 ~ · B)
(∂t2 − c2L ∇2 )Ψ = ∂i ∂j Aij + ∂t (∇ ~ . (26)
 5

Then, we time-average this equation, thus getting rid of and strain tensors in an isotropic solid, one can also trade
the total time derivative terms: ρm d/dρm for Kd/dP , where K is the bulk modulus.)
This is the generalization of Eq. (1) to the case of both
1
∇2 hΨi = − h∂i ∂j Aij i , (27) longitudinal and transverse waves.
c2L As mentioned above, this result holds equally well for
where Aij is understood to be a function of the linear the sound waves of a perfect fluid, for which fY = fZ =
solution ~π(1) . The resulting Poisson’s equation is easily c2T = 0, in which case it reduces directly to Eq. (1). In
solved using Green’s function methods, ultimately lead- fact, for a perfect fluid we also checked our result by
ing to the volume-integrated and time-averaged expres- directly taking the non-relativistic continuity and Euler
equations for ρm and a potential velocity field ~v = ∇ψ,~
sion we are after:
expanding them to quadratic order in the perturbations
~ · ~π i ≃ − 1 h Aii i . δρm and ψ about a staticR and homogeneous background,
R R
h ∇ (28)
3c2L and computing M ≡ d3 x δρm for a perturbative so-
lution of such equations of motion. We again obtain
(To get to this, we used that Eq. (1). We spare the reader the details of that tedious
1 1 4π 3 analysis.
∂i ∂j = 3 (3r̂i r̂j − δ ij ) − δ (~x)δ ij , (29)
r r 3
and the fact that the first piece integrates to zero in any V. CONCLUSIONS
spherical region centered at the origin.)
The right-hand side of (28) can be evaluated straight- We showed that, contrary to common belief, sound
forwardly from the definitions in Eqs. (21) and (23). It is waves carry gravitational mass, in a standard Newtonian
convenient to split ~π(1) into longitudinal (L) and trans- sense: they are affected by gravity, but they also source
verse (T ) components, for which the energies read gravity. Our results show that this effect goes hand-in-
Z
1 h˙2
i hand with the non-linear interactions of sound, and that
EL = w0 b0 d3 x ~ · ~πL 2 ,
~πL + c2L ∇ (30a) this is true for superfluids, ordinary fluids, and solids. In
2 particular, for all these media, in the non-relativistic limit
1 h˙2
Z

i
ET = w0 b0 d3 x ~ × ~πT 2 ,
~πT + c2T ∇ (30b) the mass transported by a sound wave is proportional to
2 its energy times a coefficient that only depends on the
medium’s equation of state.
and for which the obvious generalizations of the virial
Naturally, the transported mass is in general quite
relation (12) hold.
small, of order M ∼ E/c2s . For example, a very ener-
Putting together Eqs. (25), (28), (30a) and (30b) and
getic phonon in superfluid helium-4 with momentum k ∼
using the analogs of (12), in the non-relativistic limit we
1 keV/c carries a mass roughly of order M ∼ 1 GeV/c2 ,
find
i.e. that of a single helium atom.
w0 b2 Nevertheless, it is possible to envision experimental se-
 
M ≃ − 2 0 cL c′L h ∇ ~ · ~π 2 i + cT c′ h ∇~ × ~π 2 i
R
R

cL T tups where this effect could be detected. One possibility
 ′ is to employ ultra-cold atomic or molecular gases. In
c′

b0 c
= − 2 L EL + T ET . (31) these systems, in fact, not only might one be able to
cL cL cT achieve very small sound speeds and enhance the effect,
but also use suitable trappings to simulate strong gravi-
Longitudinal and transverse sound waves in general
have different propagation speeds—in fact, in general one tational potentials [27].
has c2L > 43 c2T [25]. So, even if initially one has a wave Moreover, atomic clocks and quantum gravimeters can
packet made up of both components, over time these currently detect tiny changes in the gravitational accel-
will split into two spatially separated wave packets, each eration of Earth, up to fractions of nm/s2 [28–30]. It is
propagating at its own speed. For each of these, one possible to imagine that, in a not too distant future, such
simply has techniques will reach the sensitivity necessary to detect
the gravitational field associated with seismic waves.
1 ρm dcs The effects we are considering may also be of rele-
M =− E, (32) vance to neutron star dynamics, since gravity would af-
c2L cs dρm
fect phonon-mediated transport properties in the super-
where cs is the propagation speed for the type (L or T ) fluid stellar interior [31, 32]. The sound speed inside the
of sound wave in question. We used that, in the non- core is expected to be relativistic [33, 34], but our results
relativistic limit, the mass density is proportional to the hold nonetheless, with minor generalizations.
level of compression, ρm ∝ b [26], and the derivative with Finally, since sound waves both source gravity and
respect to ρm has to be taken in the absence of pure are affected by it, they can interact through gravity. In
shear—that is, it corresponds to hydrostatic compression particular, for ordinary equations of state (higher sound
[25]. (Using the standard relationships between the stress speeds at higher pressures), their gravitational mass is
 6

negative. Still, for like charges gravity is attractive, and eigenstates of the full Hamiltonian (the phonon decay
therefore two sound wave packets running parallel to rate is suppressed at low energies as Γ ∼ k 5 ).
each other should start converging. Very, very slowly, With the normalizations of our paper, the quantized
of course. free phonon field reads

cs /c d3 k
Z h i
i~
k·~
x−iω~k t
π(x) = √ √ a~k e + h.c. , (A2)
ACKNOWLEDGMENTS µn̄ (2π)3 2ωk

where
We are grateful to N. Bigagli, G. Chiriacò, S. Garcia-
Saenz, L. Hui, A. Lam, R. Mcnally, R. Penco, [aq~ , a~† ] = (2π)3 δ 3 (~q − ~k) . (A3)
k
P. Richards, C. Warner and S. Will for illuminating dis-
cussions. This work has been partially supported by the For the quadratic terms of (10), we simply have
US Department of Energy grant de-sc0011941.
c2 ω~
Z
d3 x h~q |π̇ 2 (x)|~k i = s k2 h~q |~k i , (A4a)
µn̄c
ω~k
Z
Appendix A: The mass of a single phonon
2
3 ~
d x h~q | ∇π(x) |~k i = h~q |~k i . (A4b)
µn̄c2
For a superfluid, Eq. (1) can also be recovered for a
single phonon. In this case, we want to compute Using these expressions in Eq. (A1) and taking the non-
relativistic limit leads once again to Eq. (1).
Z
h~k|T00 (x)|~ki Note that one could in principle take a similar ap-
M≡ d3 x , (A1) proach for the case of a solid as well. However, in the non-
h~k|~ki relativistic limit, the first nonzero contribution comes
from the first perturbative correction in the expectation
where |~ki is a single-phonon state of momentum ~k. value of Eq. (24) on a single-phonon state. To com-
In this case, the leading contribution comes from the pute such an expectation value amounts to evaluating
quadratic terms in T00 , just as in Eq. (12). The reason is an index-heavy six-point function on the ground state,
that the linear term in T00 is the time derivative of a local which is a rather tedious calculation. The quantization
operator; its expectation value on |~ki must thus vanish of the EFT for a fluid is instead a much trickier task—see
in the approximation in which single-phonon states are e.g. [13, 17, 35–37].

[1] A. Nicolis and R. Penco, Phys. Rev. B97, 134516 (2018), [14] M. Geracie and D. T. Son, JHEP 11, 004 (2014),
arXiv:1705.08914 [hep-th]. arXiv:1402.1146 [hep-th].
[2] H. Watanabe, T. Brauner, and H. Mu- [15] G. Ballesteros, D. Comelli, and
rayama, Phys. Rev. Lett. 111, 021601 (2013), L. Pilo, Phys. Rev. D94, 025034 (2016),
arXiv:1303.1527 [hep-th]. arXiv:1605.05304 [hep-th].
[3] A. Nicolis, R. Penco, F. Piazza, and R. Rattazzi, [16] G. Ballesteros, JCAP 1503, 001 (2015),
JHEP 06, 155 (2015), arXiv:1501.03845 [hep-th]. arXiv:1410.2793 [hep-th].
[4] R. V. Lange, PRL 14 (1965). [17] B. Gripaios and D. Sutherland,
[5] H. Leutwyler, Helv. Phys. Acta 70, 275 (1997), Phys. Rev. Lett. 114, 071601 (2015),
arXiv:hep-ph/9609466 [hep-ph]. arXiv:1406.4422 [hep-th].
[6] R. V. Lange, Phys. Rev. 146 (1966). [18] S. Endlich, A. Nicolis, and J. Wang, JCAP 10, 49 (2013).
[7] A. Nicolis and F. Piazza, [19] D. T. Son, Phys. Rev. Lett. 94, 175301 (2005),
Phys. Rev. Lett. 110, 011602 (2013), [Addendum: Phys. arXiv:cond-mat/0501658 [cond-mat].
Rev. Lett.110,039901(2013)], arXiv:1204.1570 [hep-th]. [20] C.-D. Yoo and A. T. Dorsey,
[8] F. Greiter, F. Wilczek, and E. Witten, Mod. Phys. Lett. Phys. Rev. B 81, 134518 (2010),
B, 903 (1989). arXiv:1001.0621 [cond-mat.other].
[9] D. T. Son, (2002), arXiv:hep-ph/0204199. [21] A. Nicolis, R. Penco, and R. A.
[10] C. Hoyos, S. Moroz, and D. T. Rosen, Phys. Rev. D89, 045002 (2014),
Son, Phys. Rev. B89, 174507 (2014), arXiv:1307.0517 [hep-th].
arXiv:1305.3925 [cond-mat.quant-gas]. [22] L. V. Delacrétaz, A. Nicolis, R. Penco, and
[11] S. Moroz and C. Hoyos, Phys. Rev. B91, 064508 (2015), R. A. Rosen, Phys. Rev. Lett. 114, 091601 (2015),
arXiv:1408.5911 [cond-mat.quant-gas]. arXiv:1403.6509 [hep-th].
[12] S. Dubovsky, T. Gregoire, A. Nicolis, and R. Rattazzi, [23] M. Celoria, D. Comelli, and L. Pilo,
JHEP 03, 025 (2006), arXiv:hep-th/0512260 [hep-th]. JCAP 1709, 036 (2017), arXiv:1704.00322 [gr-qc].
[13] S. Endlich, A. Nicolis, R. Rattazzi, and J. Wang, [24] I. Arav, I. Hason, and Y. Oz, JHEP 10, 063 (2017),
JHEP 04, 102 (2011), arXiv:1011.6396 [hep-th]. arXiv:1702.00690 [hep-th].
 7

[25] L. Landau, E. Lifshitz, A. Kosevich, J. Sykes, W. Reid, J. M. Robinson, W. Zhang, B. J. Bloom, and J. Ye,
and L. Pitaevskiı̆, Theory of Elasticity, Course of theo- Science 358, 90 (2017).
retical physics (Butterworth-Heinemann, 1986). [31] A. B. Migdal, Sov. Phys. JETP 10, 176 (1960).
[26] S. Dubovsky, L. Hui, A. Nicolis, and [32] V. L. Ginzburg and D. A. Kirzhnits, Sov. Phys. JETP
D. T. Son, Phys. Rev. D85, 085029 (2012), 20, 1346 (1965).
arXiv:1107.0731 [hep-th]. [33] Y. B. Zeldovich, Sov. Phys. JETP 14, 1143 (1961).
[27] S. Will, N. Bigagli, A. Lam, and C. Warner, private [34] P. Haensel, A. Y. Potekhin, and D. G. Yakovlev, Neu-
communication. tron Stars 1: Equation of State and Structure (Springer,
[28] C. Freier, M. Hauth, V. Schkolnik, B. Leykauf, 2007).
M. Schilling, H. Wziontek, H.-G. Scherneck, J. Müller, [35] M. Crossley, P. Glorioso, and H. Liu,
and A. Peters, in Journal of Physics: Conference Series, JHEP 09, 095 (2017), arXiv:1511.03646 [hep-th].
Vol. 723 (IOP Publishing, 2016) p. 012050. [36] P. Gao, P. Glorioso, and H. Liu, (2018),
[29] C.-W. Chou, D. Hume, T. Rosenband, and D. Wineland, arXiv:1803.10778 [hep-th].
Science 329, 1630 (2010). [37] P. Glorioso, M. Crossley, and H. Liu,
[30] S. L. Campbell, R. B. Hutson, G. E. Marti, A. Goban, JHEP 09, 096 (2017), arXiv:1701.07817 [hep-th].
N. Darkwah Oppong, R. L. McNally, L. Sonderhouse,