Persistent Superconductor Currents in Holographic Lattices

Norihiro Iizuka,
1,
Akihiro Ishibashi,
2,
and Kengo Maeda
3,
1
Interdisciplinary Fundamental Physics Team, Interdisciplinary
Theoretical Science Research Group, RIKEN, Wako 351-0198, JAPAN
2
Department of Physics, Kinki University, Higashi-Osaka 577-8502, JAPAN
3
Faculty of Engineering, Shibaura Institute of Technology, Saitama 330-8570, JAPAN
(Dated: December 24, 2013)
We consider non-linear gravitational backreactions of persistent superconductor current ow in a
holographic superconductor model with lattices. We numerically construct a novel solution of hairy
charged black brane with momentum/rotation along a direction of no translational symmetry due
to lattices. The lattice structure prevents the horizon from rotating, and the total momentum is
only carried by matter elds outside the black brane horizon. This is consistent with the black hole
rigidity theorem, and suggests that in dual eld theory with lattices, superconductor currents are
made up by composite elds, rather than fractionalized degrees of freedom.
- Introduction -
Recent developments of applying the holography, or
the gauge/gravity duality [13], to strongly coupled
quantum eld theory reveal that there is a fascinating
relations between a rich class of various new types of
black brane solutions in gravity side, and a rich class of
various infrared (IR, or vacuum) structures for the gauge
theory sides.
A famous example of these is a holographic supercon-
ductor [48]. Various symmetry breaking pattern in the
IR in gauge theory side is reected as various new hairy
black brane solutions in gravity side. In superconduc-
tors, superconductor current can ow without any exter-
nal electric eld, namely conductivity becomes innite,
and persistent current ows. Suppose the direction of the
superconductor current is along x
1
direction. In gravity
side, this is reected as normalizable mode of the gauge
boson A
=x
1 and this can backreact to the geometry.
This induces the ow of the geometry along x
1
, and g
tx
1
and T
tx
1 become nonzero, where T
tx
1 is the bulk stress-
tensor. Here nonzero g
tx
1 plays the role of rotation,
i.e., the bulk space-time is a rotating AdS black brane
space-time.
If there is a translational invariance along x
1
in the sys-
tem, then construction of such solutions is not so much
surprising since anyhow we are guaranteed to have a con-
served current along x
1
and such rotating solutions must
exist. However in real world superconductors, materials
do not allow a translational invariance due to the lattice
structures of materials. Recently there are several new
developments of understanding the holographic super-
conductors with various lattice eects [911]. With these
developments of lattice eects in holographic models, in
this paper we try to construct a gravitational dual of the
superconductor with persistent current in the presence

norihiro.iizuka@riken.jp

akihiro@phys.kindai.ac.jp

maeda302@sic.shibaura-it.ac.jp
of lattices. This corresponds, in gravity side, to have a
rotating black brane conguration with charged scalar
hair, where we have lattice structure on the direction of
rotation.
In general relativistic viewpoint, nding such a solu-
tion would be a highly non-trivial task. If lattice struc-
ture would induce an inhomogeneity, then the rotation
along the direction of such an inhomogeneity will pro-
duce radiation, making the geometry dynamical. In this
paper, we present a novel black brane solution with mo-
mentum along the direction of no translational invari-
ance. A key point of our solution is that it admits mo-
mentum along lattice without inducing inhomogeneities.
To our best knowledge, this is the rst example of such
a solution.
- Helical lattices from Bianchi type VII
0
-
In order to introduce the lattice eects in holography,
we will make use of the Bianchi type VII
0
geometry, char-
acterized by the following three Killing vectors,

1
=
x
2 ,
2
=
x
3 ,
3
=
x
1 x
3

x
2 +x
2

x
3 , (1)
which form the Lie algebra, [
i
,
j
] = C
k
ij

k
with C
1
23
=
C
1
32
= 1, C
2
13
= C
2
31
= 1 and the rest C
i
jk
= 0.
Associated with them are the following three one-forms,

1
= cos(x
1
)dx
2
+ sin(x
1
)dx
3
,

2
= sin(x
1
)dx
2
+ cos(x
1
)dx
3
,
3
= dx
1
, (2)
each of which is invariant under all the Killing vectors
i
.
Using these, in this paper we make the following metric
ansatz;
ds
2
= f(r)dt
2
+
dr
2
f(r)
+e
2v3(r)
(
3
(r)dt)
2
+e
2v1(r)
(
1
)
2
+e
2v2(r)
(
2
)
2
. (3)
where f(r), (r), v
i
(r), with i = 1, 2, 3, are functions of
the radial coordinate r only. Under a similarbut more
generic homogeneous but anisotropicmetric ansatz, the
a
r
X
i
v
:
1
3
1
2
.
6
1
2
4
v
1


[
h
e
p
-
t
h
]


2
0

D
e
c

2
0
1
3
2
near horizon geometries for generic Bianchi type of static,
i.e., (r) = 0, black brane solutions are classied and
studied in [12, 13]. With (r) ,= 0, there is a ow of the
geometry along x
1
, i.e., the black branes can rotate
along the x
1
direction, and there is a nonzero bulk stress-
tensor component T
tx
1.
Note that if v
1
(r) = v
2
(r), then, due to (
1
)
2
+(
2
)
2
=
(dx
2
)
2
+ (dx
3
)
2
, we have translational invariance along
x
1
. However, as long as v
1
,= v
2
, there is no translational
invariance along x
1
direction, i.e.,
x
1 is not a Killing
vector of the geometry. Since x
1
x
1
+ 2n (with n
integer) is a discrete symmetry, this implies that there
is a helical lattice along x
1
direction [14]. The metric
ansatz (3) is a homogeneous metric which admits helical
lattice, and this homogeneity enables us to reduce the
Einstein equations to a tractable set of ordinary dieren-
tial equations.
In holographic superconductor model [511] for ex-
ample, optical conductivity is calculated from the ratio
between non-normalizable (NN) mode and normalizable
(N) mode of the vector eld A

. If there is a large nor-

malizable mode, say along x
1
as A
=x
1, then large A
=x
1
implies large current and the back reaction induces g
tx
1
component of the metric beyond the linear response the-
ory. This is the non-linear gravitational back reaction by
the holographic superconductor current. In the presence
of lattices, DC conductivity for superconductor is shown
to be innite in the framework of linear response the-
ory [911]. However, the innite DC conductivity does
not by itself mean the presence of persistent current, and
this is seen in the normal states at zero temperature [15].
Therefore it is interesting to ask if there is a stationary
ow of superconductor current in eld theory, and we try
to answer this from the holographic dual gravity side.
The purpose of this paper is to consider such a holo-
graphic gravity model by providing a novel stationary
black brane solution with momentum along the direction
of no translational symmetry due to lattices. Note that
even though our metric ansatz (3) is homogeneous, it is
shown in [14] that there is a Drude peak and nonzero
resistance in a normal state.
Before discussing our explicit model and showing its
solutions, we can, at this stage, argue generic nature of
our geometry. Under the metric ansatz (3), we are con-
cerned with black branes whose event horizon H is given
by f(r) = 0. Then, the tangent vector l along the null
geodesics of such an H is
l =
t
+
h

x
1 ,
h
[
r=r
h
, (4)
where here and hereafter r = r
h
denotes the root of
f(r) = 0, a horizon. Then, we have, on H [23]
R

= 2
2
h
(sinh(v
1
v
2
))
2
. (5)
In order to satisfy the null energy condition, i.e.,
R

0, we need either
h
= 0 or v
1
= v
2
at the
horizon. This implies that it is impossible to have a ro-
tating horizon with the lattice eects along x
1
under the
metric ansatz (3).
This is nothing but what is implied by the black hole
rigidity theorem, which claims that under the analyticity
assumption, a stationary rotating black hole, must be ax-
isymmetric [1618]. This is routed by the fact that for a
stationary black hole, there must exist a null Killing vec-
tor K on H. However if the black hole is rotating, then

t
becomes by denition spacelike on H, and therefore
there must exist another Killing vector that, combined
together with
t
, provides K.
- A Holographic Model -
The model we consider is ve-dimensional action S =

d
5
x

g / where the Lagrangian has U(1)U(1) gauge

symmetry;
/ = R +
12
L
2

1
4
F
2

1
4
W
2
[D[
2
m
2
[[
2
, (6)
where R is Einstein-Hilbert term, L represents AdS scale
and A

, B

are one-form gauge potentials, and their eld

strengths are F = dA, W = dB. is a complex scalar
eld, which is charged under only the gauge potential A

,
but is neutral to B

. Covariant derivative acting on is

D

iqA

.
The one-form A

is to introduce a chemical potential.

We make an ansatz for the other one-form B to be pro-
portional to type VII
0
Bianchi form, so that it induces
holographic helical lattice eects of the Bianchi type
VII
0
. If we set B = 0, then it makes our model the same
type as [5, 6].
The Einstein equation, equations of motion for gauge
bosons A

, B

and a charged scalar are

R

=
4
L
2
g

+
1
2
(F

+W

1
12
g

(F
2
+W
2
) +
m
2
3
g

[[
2
+
1
2
[D

(D

+D

(D

] , (7)
1

gF

) = iq[

(D

] , (8)
1

gW

) = 0 , D

= m
2
. (9)
We solve these with the metric ansatz (3) and
= (r) , A

dx

= A
x
1(r)
3
+A
t
(r)dt , (10)
B

dx

= b(r)
1
, (11)
where
i
s are invariant one-forms given by (2), and we
have set the phase of complex scalar to be zero by the
gauge transformation, which one can easily check from
the equation of motion.
The equation of motion for the metric component
v
1
(r) v
2
(r) , (12)
3
is given by
f

+f

+f(v

1
+v

2
+v

3
)

2e
2v3

1 e
2v3
f
1

sinh 2
=
1
2
e
2(v2+v3)

1 e
2v3
f
1

b
2

1
2
e
2v1
fb
2
, (13)
where

is the derivative with respect to r. The gauge
potential b plays the role of source term for evolu-
tion [14]. It is then clear that nonzero b only introduces
the helical lattice eects, i.e., nonzero ,= 0. There
is no matter eld other than b which plays the role of
source [24] to induce the disparity between
1
and
2
for
the metric ansatz (3) under the matter eld ansatz for
A

and , (10). Therefore, we seek for the solution with

b(r) ,= 0, and therefore (r) ,= 0; One form B with ansatz
(11) is our source for holographic helical lattice eects.
The equation of motion for b is
fb

+f

+f(v

3
+v

2
v

1
)b

e
2(v1v2)

e
2v3
f
1

b = 0 . (14)
This clearly allows b = 0 as a trivial solution, and one
can therefore perform a linear perturbation analysis by
rst setting b = 0, obtaining a superconductor solution
with = 0, and then by solving the equations of motion
for nonzero b and [9]. However we do not proceed in
that way, as we are here interested in non-linear gravita-
tional backreaction of our superconductor current. Our
treatment of the lattice eects is fully non-perturbative.
- Near horizon analysis -
Let us analyze the near horizon of (14) in the nite
temperature case. Near the horizon, the metric should
have a single zero f(r) (r r
h
) ( > 0). Furthermore
let us assume that v
i
is at least C
2
; thus in particular,
v
i
[
r=r
h
= nite, v

i
[
r=r
h
= nite [25], and set
lim
rr
h
(r)
h
, lim
rr
h
(r) (r
h
) . (15)
Now suppose that
h
is nonzero for a moment. Then, as
the right-hand side of (5) must be non-negative, we need
to impose (r
h
) = 0, and thus reduce (14) to

2
(r r
h
)
2
b

+
2
(r r
h
) b

+
2
h
b 0 , (16)
in the r r
h
limit. This admits solutions
b(r) (r r
h
)

i
h

. (17)
Since both solutions are singular at the horizon, we have
to choose their coecients to vanish and thus b(r) =
b

(r) = 0 at r r
h
, in order to have a smooth solution.
However, this means that b(r) must vanish identically
in all the radius as b(r) obeys the 2nd order dierential
equation (14), and as a consequence (r) = 0. This itself
does not cause any problem with the rigidity theorem,
since it implies that either
h
or must vanish at the
horizon. However since we are interested in constructing
a holographic model with lattices, we are interested in a
solution with b(r) ,= 0, and this forces us to choose
lim
rr
h
(r) = 0 , (18)
Therefore we conclude that in our set-up, the lattice ef-
fects and smoothness condition force us to have ows (or
rotation) only outside of the horizon. Black branes can-
not be rotating. Then there is no obstacle from the rigid-
ity theorem to seeking for a solution with
lim
rr
h
b(r) b(r
h
) ,= 0 , (19)
which induces lattice eects [26],
(r
h
) ,= 0 . (20)
We choose for A
t
at the horizon
lim
rr
h
A
t
(r) = 0 , (21)
as in the static case [27].
Explicit equations of motion in terms of the com-
ponent f, , v
1
, v
3
, , A
t
, A
x
1, b, are summarized in the
Appendix A (A1) - (A9).
From the equation of motion for (r) given in (A2),
one can show that

h
(r r
h
) due to the condition
(18).
These constraint equations at r = r
h
, combined with
(B1) - (B8), and (21) shows that there are the 9 free
parameters;
A
x
1(r
h
), A

t
(r
h
), (r
h
), (r
h
),
v
1
(r
h
), v
3
(r
h
), ,

h
, b(r
h
) . (22)
We will tune these nonzero parameters in such a way that
we will have asymptotically AdS geometry with no non-
normalizable modes. For that purpose, rst, we impose
at r ,
lim
r
(r) = 0 , (23)
Then v
1
= v
2
and the lattice eect disappears and we
have translational invariance restored at UV.
By (A2), we nd the asymptotic behavior of as

0
+

N
r
4
, (24)
at r . Therefore, we impose
lim
r
(r) = 0 , (i.e.,
0
= 0 .) (25)
because we are interested in a solution in which non-
normalizable mode for g

is zero.
Near the boundary r , the scalar eld behaves
C
+
r
+
+C

= 2

4 +m
2
L
2
. (26)
4
For numerics purpose, we will choose the value of mass
as m
2
L
2
=
15
4
. In this case C
+
r

5
2
+C

3
2
, and
both are normalizable modes. As usual, we can impose
the boundary condition that either C
+
or C

becomes
zero [6]. In this paper, we shall impose
C

= 0 . (27)
These conditions (23), (25), (27) are achieved by tuning
the parameters (22). Under these conditions, following
requirement of vanishing non-normalizable mode is sat-
ised,
lim
r

f(r)
r
2
L
2

= O

1
r
2

. (28)
- Numerical Solution Interpolating IR and UV-
The numerical solution is shown in gures 1 - 3. These
are all for the parameter choice,
q = = 1 , L
2
= 2 , m
2
= 15/8 , (r
h
) = 0.1, (29)
v
1
(r
h
) = v
3
(r
h
) = (r
h
) = 1 ,

h
= 0.1 , r
h
= 1.635
A
x
1(r
h
) = 0.7385 , A

t
(r
h
) = 2.608 , b(r
h
) = 7.552 .
2 3 4 5 6 7 8
r 0
5
10
15

v
i
r
FIG. 1: e
v
1
(r)
(solid, blue), e
v
2
(r)
(dotted, red), and e
v
3
(r)
(dashed, green) for the parameter choice (29). r = 1.635 is a
horizon.
In our solution, asymptotically r ,
e
v1
= e
v2
e
v3
r , = v
1
v
2
0 , 0 (30)
are satised, so that lattice disappears at the boundary
and geometry approaches AdS metric [28].
The asymptotic behavior of A
x
1 and are
A
x
1 a
x0
+
a
xN
r
2
. (31)
According to the AdS/CFT dictionary, a
xN
corresponds
to the current in the dual eld theory, while
N
in (24)
corresponds to T
tx
1 component of the expectation
value of the energy momentum tensor T

on the
2 3 4 5 6 7 8
r
0.010
0.005
0.000
0.005
0.010
r,
r
10

FIG. 2: (r) (solid, blue) and (r)/10 (dotted, red) for the
parameter choice (29).
2 3 4 5 6 7 8
r
1
0
1
2
Air, r
FIG. 3: At(r) (solid, blue) and A
x
1(r) (dotted, red) and (r)
(dashed, green) for the parameter choice (29). For (r), the
condition (27) is satised.
boundary theory. We numerically nd that a
xN
1.7
and
N
0.57 under the boundary condition
0
= 0.
Since (r) 0 at the horizon, black branes are not
rotating. However, (r) ,= 0 between the horizon and
innity, and therefore our solution is not static. This
may be viewed that the matter eld outside the horizon
is rotating. Note also that there is no ergosphere with
respect to
t
outside of the horizon, as g
tt
(r) < 0 for all
the range between horizon to innity.
- Summary and Discussion -
In this paper, we have numerically constructed station-
ary, non-static hairy black brane solutions with momen-
tum/rotation which is dual to the persistent supercon-
ductor current along the direction which has no trans-
lational invariance due to lattices. The importance of
the backreaction must be clear; the DC conductivity di-
verges and therefore current can be large without exter-
nal electric eld. This means in the bulk that A
x
1 induces
nonzero g
tx
1 by the eects of non-linear, superconductor
current, i.e., beyond the linear response.
One of the key features in constructing our solutions
is the rigidity theorem, which forces us one of the below
5
choices; 1. Holographic lattice eects survive near the
horizon, but black brane cannot be rotating. Rotation
is carried by the matter eld surrounding black brane.
2. Black brane is rotating, but holographic lattice ef-
fects disappear. Note that in our metric ansatz, we have
shown that in case of 2, lattice eects actually disappear
in all of the radius due to the regularity assumption, and
therefore, we are forced to choose 1. Actually we expect
that this is a generic nature. We conjecture that on the
gravitational action without any articial sources, holo-
graphic lattices prevent black brane from rotating and
momentum must be carried by the matter elds outside.
It is interesting to ask what it implies in dual eld
theory that holographic lattices kill the rotation of black
branes. This might suggest that in dual eld theory, in
the presence of the lattices, fractionalized [19, 20] de-
grees of freedom cannot have current along the lattice
direction in the stationary limit. Fractionalized de-
grees of freedom are thought to be responsible for the
non-Fermi liquid behavior, and actually they are degrees
of freedom violating Luttinger theorem [21]. It is inter-
esting to ask what is the implication of the rigidity theo-
rem and how much it gives restriction on the dynamics of
fractionalized degrees of freedom, in condensed matter
physics.
In our solution, the rotation, carried by the matter
surrounding, does not produce ergosphere, i.e., metric
g
tt
0 outside of the horizon. However we have not been
able to show if this is a generic feature of the solutions
or not, and it is certainly interesting to ask if one can
construct such a solution where ergosphere is produced
outside of the black brane. It is also interesting to nd
critical current which breaks the superconducting phase.
Our solution is stationary non-rotating but not static.
This property and our metric ansatz appear to be quite
similar to the solution found in [22], besides the obvious
dierence of matter elds considered. However, it should
be emphasized that the solution of [22] contains a cross-
term of dt and
1
or
2
and therefore its non-staticity
is due to the matter ow in a direction of translational
symmetry
1
=
x
2 and
2
=
x
3. In contrast, our so-
lution, having the cross-term of dt and
3
, contains the
matter elds that ow in a direction of no translational
symmetry,
x
1, and thus is a novel black brane solution.
Finally, in this paper we constructed a homogeneous
black brane solution, on which there is momentum ow
along the lattices with charged scalar hair. However from
the holographic superconductor, there must exit similar
solutions even inhomogeneous cases. It would be inter-
esting to investigate this more as a future work.
Acknowledgments
The work of NI was supported by RIKEN iTHES
Project. This work was also supported in part by JSPS
KAKENHI Grant Number 25800143 (NI), 22540299
(AI), 23740200 (KM).
Appendix A: Equations of motion
Einstein equations for metric components, = v
1
v
2
,
, v
1
, v
3
become
f

+f

+f(v

1
+v

2
+v

3
)

e
2v3
f
1

sinh 2

1
2
e
2(v2+v3)

1 e
2v3
f
1

b
2
+
1
2
e
2v1
fb
2
= 0 ,
(A1)

+ (v

1
+v

2
+ 3v

3
)

4
e
2v3
f
sinh
2
(v
1
v
2
)
e
2v3
A

x
1(A

t
+ A

x
1)
e
2(v2+v3)
f
b
2

2q
2
f
1
e
2v3
A
x
1
2
(A
t
+ A
x
1) = 0 , (A2)
fv

1
+f

+f(v

1
+v

2
+v

3
)v

1
+
1
6
A
2
t
+
1
3
A

x
1A

t
+
1
6
(
2
fe
2v3
)A
2
x
1 + sinh2(v
1
v
2
)f
1

2
e
2v3
sinh2(v
1
v
2
)
4
L
2
+
1
3
e
2v1
fb
2

1
6
e
2(v2+v3)
b
2

1 f
1
e
2v3

+
m
2
3

2
= 0 , (A3)
fv

3
+f

+f(v

1
+v

2
+v

3
)v

3
+
1
6
(A

t
+ A

x
1)
2
+
1
3
fe
2v3
A
2
x
1 + 2e
2v3
sinh
2
(v
1
v
2
) +
e
2v3
2

2

4
L
2

1
6
fe
2v1
b
2
+
b
2
6
e
2(v2+v3)

2 +f
1
e
2v3

+
1
3
(m
2
+ 3q
2
e
2v3
A
2
x
1)
2
= 0 . (A4)
In addition, we have a Hamiltonian constraint,
1
2
e
2v3

1
2
fe
2v3
A
2
x
1 +
1
2
(A

t
+ A

x
1)
2

12
L
2
+ 2 sinh
2
(v
1
v
2
)

e
2v3
f
1

+f

(v

1
+v

2
+v

3
)
+ 2f(v

1
v

2
+v

2
v

3
+v

3
v

1
)
f
2
e
2v1
b
2
+
e
2(v2+v3)
2

1 e
2v3
f
1

b
2
+e
2v3

2

q
2
A
2
x
1 +e
2v3

m
2
q
2
f
1
(A
t
+ A
x
1)
2

f
2
= 0 . (A5)
For A
t
, A
x
1, b, , the equations of motion become
fA

x
1 + (f

+f(v

1
+v

2
v

3
) +e
2v3

)A

x
1
e
2v3

t
2q
2
A
x
1
2
= 0 , (A6)
fA

t
+f(v

1
+v

2
+v

3
)A

t
+ (f

+ (2fv

3
f

))A

x
1
+e
2v3

(A

t
+ A

x
1) + 2q
2
A
t

2
= 0 , (A7)
f

+ (f

+f(v

1
+v

2
+v

3
))

q
2
e
2v3
A
2
x
1 +q
2
(A
t
+ A
x
1)
2
f
m
2
= 0 , (A8)
fb

+f

+f(v

3
+v

2
v

1
)b

e
2(v1v2)

e
2v3
f
1

b = 0 . (A9)
6
Appendix B: Horizon boundary condition
With the assumption of regularity at the horizon, by
plugging

h
(r r
h
) , (B1)
into (A8), (A6), (A9), (A1), (A3), (A4), we obtain,

(r
h
) =q
2
e
2v3(r
h
)
A
2
x
1(r
h
)(r
h
) +m
2
(r
h
) , (B2)
A

x
1(r
h
) =e
2v3(r
h
)

h
A

t
(r
h
) + 2q
2
A
x
1(r
h
)
2
(r
h
), (B3)
b

(r
h
) =e
2((r
h
)v3(r
h
))
b(r
h
) , (B4)

(r
h
) =2e
2v3(r
h
)
sinh 2(r
h
)
+
1
2
e
2(v2(r
h
)+v3(r
h
))
b
2
(r
h
) , (B5)
v

1
(r
h
) =
4
L
2
+e
2v3(r
h
)
sinh 2(r
h
)
1
6
A
2
t
(r
h
)
+
1
6
e
2(v2(r
h
)+v3(r
h
))
b
2
(r
h
)
m
2
3

2
(r
h
),
(B6)
v

3
(r
h
) =
4
L
2
2e
2v3(r
h
)
sinh
2
(r
h
)
1
6
A
2
t
(r
h
)

b
2
(r
h
)
3
e
2(v2(r
h
)+v3(r
h
))

e
2v3(r
h
)
2

2

1
3
(m
2
+ 3q
2
e
2v3
A
2
x
1
2
(r
h
)) , (B7)
where near horizon metric f is
f (r r
h
) . (B8)
[1] J. M. Maldacena, Adv. Theor. Math. Phys. 2, 231 (1998)
[Int. J. Theor. Phys. 38, 1113 (1999)] [hep-th/9711200].
[2] E. Witten, Adv. Theor. Math. Phys. 2, 253 (1998) [hep-
th/9802150].
[3] S. S. Gubser, I. R. Klebanov and A. M. Polyakov, Phys.
Lett. B 428, 105 (1998) [hep-th/9802109].
[4] S. S. Gubser, Phys. Rev. D 78, 065034 (2008)
[arXiv:0801.2977 [hep-th]].
[5] S. A. Hartnoll, C. P. Herzog and G. T. Horowitz, Phys.
Rev. Lett. 101, 031601 (2008) [arXiv:0803.3295 [hep-th]].
[6] S. A. Hartnoll, C. P. Herzog and G. T. Horowitz, JHEP
0812, 015 (2008) [arXiv:0810.1563 [hep-th]].
[7] C. P. Herzog, J. Phys. A A 42, 343001 (2009)
[arXiv:0904.1975 [hep-th]].
[8] G. T. Horowitz, arXiv:1002.1722 [hep-th].
[9] N. Iizuka and K. Maeda, JHEP 1211, 117 (2012)
[arXiv:1207.2943 [hep-th]].
[10] J. A. Hutasoit, G. Siopsis and J. Therrien,
arXiv:1208.2964 [hep-th].
[11] G. T. Horowitz and J. E. Santos, arXiv:1302.6586 [hep-
th].
[12] N. Iizuka, S. Kachru, N. Kundu, P. Narayan, N. Sir-
car and S. P. Trivedi, JHEP 1207, 193 (2012)
[arXiv:1201.4861 [hep-th]].
[13] N. Iizuka, S. Kachru, N. Kundu, P. Narayan, N. Sircar,
S. P. Trivedi and H. Wang, arXiv:1212.1948 [hep-th].
[14] A. Donos and S. A. Hartnoll, Nature Phys. 9, 649 (2013)
[arXiv:1212.2998].
[15] A. Ishibashi and K. Maeda, Phys. Rev. D 88, 066009
(2013) [arXiv:1308.5740 [hep-th]].
[16] S.W. Hawking and G.F.R. Ellis, The large scale structure
of space-time, Cambridge University Press, Cambridge
(1973)
[17] S. Hollands, A. Ishibashi and R.M. Wald, Commun.
Math. Phys. 271, 699 (2007) [arXiv:gr-qc/0605106].
[18] V. Moncrief and J. Isenberg, Class. Quantum Grav. 25,
195015 (2008) [arXiv:0805.1451[gr-qc]].
[19] T. Senthil, S. Sachdev and M. Vojta, Phys. Rev. Lett.
90, 216403 (2003) [arXiv:cond-mat/0209144].
[20] S. Sachdev, Phys. Rev. Lett. 105, 151602 (2010)
[arXiv:1006.3794 [hep-th]].
[21] S. A. Hartnoll, D. M. Hofman and A. Tavanfar, Eu-
rophys. Lett. 95, 31002 (2011) [arXiv:1011.2502 [hep-
th]]; L. Huijse and S. Sachdev, Phys. Rev. D 84,
026001 (2011) [arXiv:1104.5022 [hep-th]]; S. A. Hartnoll,
arXiv:1106.4324 [hep-th]; S. Sachdev, Phys. Rev. D 84,
066009 (2011) [arXiv:1107.5321 [hep-th]]; S. A. Hart-
noll and L. Huijse, Class. Quant. Grav. 29, 194001
(2012) [arXiv:1111.2606 [hep-th]]; L. Huijse, S. Sachdev
and B. Swingle, Phys. Rev. B 85, 035121 (2012)
[arXiv:1112.0573 [cond-mat.str-el]]; N. Iqbal and H. Liu,
Class. Quant. Grav. 29, 194004 (2012) [arXiv:1112.3671
[hep-th]]; K. Hashimoto and N. Iizuka, JHEP 1207, 064
(2012) [arXiv:1203.5388 [hep-th]].
[22] A. Donos and J. P. Gauntlett, Phys. Rev. D 86, 064010
(2012) arXiv:1204.1734 [hep-th].
[23] Regularity of the metric at the horizon can be manifestly
seen by working in the Eddinton-Finkelstein coordinates
dened as u := t

r
f
1
dr and x
1
+
:= x
1

r
f
1
dr,
in which l = u +
h
x
+
, yielding the same result (5).
For simplicity, here and in the following we stick to the
coordinates introduced in (3).
[24] If b = 0, one can check that cannot have regular solution
at horizon.
[25] We seek for the solution whose area-density does not di-
verge at nite temperature, otherwise, dual eld theory
entropy density diverges at nite temperature.
[26] This eect has been investigated in [14] in a linear re-
sponse theory.
[27] Otherwise, Euclidean Wilson loop becomes non-trivial
around a vanishing circle [8]. This condition can also be
derived from the non-divergence of AA

at the horizon.
[28] The dierence between v1(= v2) and v3 can be eliminated
by the coordinate re-denition since (r) 0. Therefore,
metric is exactly AdS5 form asymptotically.