DBI equations and holographic dc conductivity
Alfonso Ballon-Bayona*ICTP South American Institute for Fundamental Research, Instituto de Fı´sica Teo´rica, Universidade Estadual Paulista, 01140-070 Sa˜o Paulo, Sa˜o Paulo, Brazil
Cristine N. Ferreira†
Nu´cleo de Estudos em Fı´sica, Instituto Federal de Educac¸a˜o, Cieˆncia e Tecnologia Fluminense, Rua Dr. Siqueira, 273, Campos dos Goytacazes, Rio de Janeiro CEP 28030-130, Brazil
and The Abdus Salam International Centre for Theoretical Physics Strada Costiera, 11 I-34151 Trieste Italy Victor J. Vasquez Otoya‡
Instituto Federal de Educac¸a˜o, Cieˆncia e Tecnologia do Sudeste de Minas Gerais, Rua Bernardo Mascarenhas 1283, 36080-001, Juiz de Fora, Minas Gerais, Brazil
(Received 11 March 2013; published 21 May 2013)
We provide a simple method for writing the Dirac-Born-Infeld equations of a Dp-brane in an arbitrary static background whose metric depends only on the holographic radial coordinatez. Using this method we revisit the Karch-O’Bannon procedure to calculate the dc conductivity in the presence of constant electric and magnetic fields for backgrounds where the boundary is four- or three-dimensional and satisfies homogeneity and isotropy. We find a frame-independent expression for the dc conductivity tensor. For particular backgrounds we recover previous results on holographic metals and strange metals.
DOI:10.1103/PhysRevD.87.106007 PACS numbers: 11.25.Tq, 11.25.Uv
I. INTRODUCTION
Holography provides a new approach for investigating strongly coupled field theories. The most important case is given by the AdS/CFT correspondence [1] that maps gauge theories with conformal symmetry to string/M theories compactified into anti-de Sitter spacetimes. The AdS/ CFT correspondence was conjectured by Maldacena after investigating N coincident branes in string/M theory in the large-N limit. At low energies some of these configurations describe conformal gauge theories at weak coupling, while at strong coupling they lead to supergravity solutions in-volving anti-de Sitter spacetimes.
For example, the holographic dual ofN ¼4UðNÞsuper Yang-Mills theory in four dimensions is Type IIB string theory inAdS5S5. This duality arises from a
configura-tion of N coincident D3-branes in the large-N limit. This example is very interesting for many theoretical reasons (for instance integrability) and provides new insights into the old problem of nonperturbative QCD. Another interest-ing example is the N ¼6 k-level UðNÞ UðNÞ super Chern-Simons gauge theory in three dimensions, whose holographic dual is M theory in AdS4S7=Z
k. In this case the duality arises after considering a configuration of M2-branes in M theory. This example has inspired recent holographic models for condensed matter physics.
An interesting problem in condensed matter systems is to describe charge transport in strongly coupled theories.
In holography, this is achieved by the inclusion of fermionic charge carriers in the AdS/CFT correspondence. A very simple method of including fermions in the AdS/CFT correspondence consists of the inclusion ofNf Dp-branes in the probe limitNfN. This was originally proposed in the four-dimensional case [2], where Nf D7-branes are included on top ofND3-branes. The probe limitNfN guarantees that the backreaction of the D7-branes on the bulk geometry can be ignored. The fermions are interpreted as quarks in the four-dimensional field theory, whose dual are strings stretching from the D3-branes to the D7-branes. In this paper we consider the problem of probe Dp-branes in an arbitrary static background, whose metric depends only on the holographic radial coordinate z. We provide a simple and elegant method to write the Dirac-Born-Infeld (DBI) equations that describe the dynamics of the Dp-brane. Using these results we calculate the electro-magnetic currents through holographic methods for a gen-eral gauge field configuration. We find a nice expression for the case when the gauge field strengths on the brane depend only on the radial coordinatez. As an application, we consider the case of constant electric and magnetic fields and calculate the dc conductivity tensor when the boundary is four and three dimensional and satisfies ho-mogeneity and isotropy. This is done following the Karch-O’Bannon procedure [3,4] (see also [5]), which is based on the reality of the brane action. We check our results for particular backgrounds, namely the D3-D7 model [2] and the Lifshitz background [6–10]. Our results for the D3-D7 model are consistent with those of [3–5] and are interpreted as holographic metals, while in the Lifshitz case the
*aballonb@ift.unesp.br
†cferreir@ictp.it
‡victor.vasquez@ifsudestemg.edu.br
conductivity is consistent with [11] and can be interpreted in terms of strange metals. Our results are also consistent with a recent method proposed in [12,13], which is based on the properties of an effective metric, called open string metric, that arises from the DBI action.
Our paper ends with some conclusions and an Appendix describing the Drude model and the effect of a five-dimensional Chern-Simons term on the DBI equations for constant electric and magnetic fields.
II. DP-BRANE DYNAMICS AND HOLOGRAPHIC CURRENTS
Consider the dynamics of a probe Dp-brane in an arbitrary static background of the form
ds2¼aðzÞgmnðzÞdxmdxnþbðzÞd2; (2.1)
wheregmndescribes aDþ1spacetime that includes the radial coordinate z. We will be interested in the cases D¼4 and D¼3, although the results in this section are general.
We assume for simplicity thatis a compact space. We consider the case where the Dp-brane fills the spacetime of the internal coordinatesxm and wraps a submanifoldof dimensionp Dinside. In a general class of Dp-brane models, the submanifold is specified by fixing some coordinates in to constant values while one of the coordinates, say, depends on the radial coordinatez.
The induced metric on the Dp-brane takes the form
ds2
Dp¼aðzÞgmnðzÞdxmdxnþbðzÞd2 ¼:GMNdxMdxN;
(2.2)
where gzz now includes a contribution arising from d2 ¼02ðzÞdz2.
A. The DBI action
The dynamics of a single Dp-brane is described by the Abelian DBI action
SDBI¼ pZ dDxdzdp De
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffidetðGMNþ20F MNÞ
q
; (2.3)
whereGMNis the induced metric on the probe brane,FMN is the world volume gauge field strength and
p¼ ð2Þ p0 pþ1
2 (2.4)
is the brane tension.
We assume for simplicity, that the gauge fields have nonzero components in the xm directions only and these components do not depend on the coordinates of. Thus, after integrating over the compact submanifold , we obtain the Dþ1 dimensional effective action that can be written as
SDBI¼ Z dDxdz
ðzÞpffiffiffiffiffiffiffiffiE; (2.5)
whereEis the determinant of the tensor defined by Emn:¼gmnþðzÞFmn; (2.6)
and
ðzÞ ¼2a0
ðzÞ ; ðzÞ ¼pVe
ðaðzÞÞDþ1 2 ðbðzÞÞ
p D 2 : (2.7)
Here we have used the notationVto refer to the volume of the compact submanifold . Note that Fmn denotes the components of the field strength in thexm directions.
B. DBI equations and holographic currents
The relevant fields in the effective action (2.5) are the Dþ1 dimensional gauge field Am and the scalar field ðzÞ. Therefore, the variation of the action (2.5) takes the form
SDBI¼SBulk DBI þS
Bdy
DBI; (2.8)
where
SBulk DBI¼
Z
dDxdz
@m
2
ffiffiffiffiffiffiffiffi
E
p ~
Eð‘;mÞA ‘
þ
@z
2
ffiffiffiffiffiffiffiffi
E
p ~
Ezz@gzz @0
ffiffiffiffiffiffiffiffi
E
p @
@
; (2.9)
SBdyDBI¼ Z dDxdz
@m
2
ffiffiffiffiffiffiffiffi
E
p ~
Eð‘;mÞA ‘
þ@z
2
ffiffiffiffiffiffiffiffi
E
p ~
Ezz@gzz @0
; (2.10)
andE~ð‘;mÞ¼E~‘m E~m‘, withE~‘mbeing the inverse ofE ‘m. Since the field variationsA‘ andare arbitrary, the bulk term (2.9) leads to the DBI equations
@m
2
ffiffiffiffiffiffiffiffi
E
p ~
Eð‘;mÞ¼0; (2.11)
@z
2
ffiffiffiffiffiffiffiffi
E
p ~
Ezz@gzz @0
ffiffiffiffiffiffiffiffi
E
p @
@¼0: (2.12)
A natural prescription for the boundary electromagnetic currents in holography is the following:
j ¼ S A
z¼zBdy
: (2.13)
In our case the only contribution comes from the first term in the boundary term (2.10), so we find
j ¼ lim z!zBdy
2
ffiffiffiffiffiffiffiffi
E
p ~
Eð;zÞ: (2.14)
@0
2
ffiffiffiffiffiffiffiffi
E
p ~
Eðz;0Þþ@ i
2
ffiffiffiffiffiffiffiffi
E
p ~
Eðz;iÞ¼0; (2.15)
@z
2
ffiffiffiffiffiffiffiffi
E
p ~
Eðz;0Þþ@ i
2
ffiffiffiffiffiffiffiffi
E
p ~
Eð0;iÞ¼0; (2.16)
@z
2
ffiffiffiffiffiffiffiffi
E
p ~
Eðz;iÞ @
0
2
ffiffiffiffiffiffiffiffi
E
p ~
Eð0;iÞ
þ@j
2
ffiffiffiffiffiffiffiffi
E
p ~
Eði;jÞ¼0: (2.17)
In the case where the gauge field strength tensor Fmn depends only on the radial coordinate z, the equation (2.15) is automatically satisfied, while the others can be integrated out leading to the equation
2
ffiffiffiffiffiffiffiffi
E
p ~
Eðz;Þ¼j; (2.18)
where j is the holographic current associated with the DBI action, as obtained in (2.14). This is a nontrivial equation involving the field strengths on the lhs and the holographic currents on the rhs.
1. The contribution from a-term
In the D¼4 case, the boundary field theory is four dimensional and may include a-term of the form1
S¼ Z dtd3xF
F; (2.19)
where ¼ðt; ~xÞ is a four-dimensional axion field and F¼@A @A is the four- dimensional field strength.
Integrating by parts (2.19), we obtain S¼2
Z
dtd3xf@½A
F þAð@ÞFg:
(2.20)
This term provides a contribution to the holographic current,
j ¼S
A ¼
2 ð@
ÞF: (2.21)
C. The method of-expansion
The method proposed in this paper consists of a series expansion in the parameterðzÞfor the relevant quantities appearing on the DBI equations, namely the determinantE and the inverse tensorE~m;n.
Consider the matrix representation of the metricg^ and the field strength F^. Then the tensor Emn has a matrix representation given by
^
E¼g^þF^ ¼g^ð1þX^Þ; (2.22)
where we have definedX^ :¼g^ 1F^. Note that this matrix is matrix linear in . Recalling some properties of the determinant, we can writeEin terms ofX^,
E¼detðE^Þ ¼detðg^Þdetð1þX^Þ
¼detðg^ÞexpfTr½lnð1þX^Þg: (2.23)
On the other hand, the tensor Em;n corresponds to the inverse matrix E^ 1 and can also be represented in terms ofX^,
^
E 1 ¼ ð1þX^Þ 1g 1: (2.24)
Now we consider a series expansion inX^ for the determi-nantEand the inverse tensorE~m;n. For the determinant we first expand the matrixlnð1þX^Þand take the trace
Tr½lnð1þX^Þ ¼ 1 2TrðX^
2Þ 1
4TrðX^
4Þ
1 6TrðX^
6
Þ þ ; (2.25)
where we used the identityTrðX^nÞ ¼0for odd values ofn. Substituting (2.25) into (2.23) we get
E¼detðg^Þ
1 1
2TrðX^
2Þ 1
4TrðX^
4Þ
þ18½TrðX^2Þ2 1
6TrðX^
6Þ þ1
8½TrðX^
2Þ½TrðX^4Þ
1 48½TrðX^
2
Þ3þ : (2.26)
The traces appearing in the expansion can be written in a tensorial form,
TrðX^2Þ ¼ 2FmnF mn;
TrðX^4Þ ¼4FmnF
npFpqFqm; TrðX^6Þ ¼6FmnF
npFpqFqrFrsFsm; :
(2.27)
Then we conclude that the determinant has the -expansion,
E¼detðg^Þ
1þ
2
2 F mnF
mnþ 4
4!F mnpqF
mnpq
þ
6
6!F
mnpqrsF
mnpqrsþ
; (2.28)
where
Fmnpq:¼FmnFpq FmpFnqþFmqFnp;
Fmnpqrs:¼FmnFpqrs FmpFnqrsþFmqFnprs
FmrFnpqsþFmsFnpqr; : (2.29)
It is easy to check that the covariant tensors Fmnpq and Fmnpqrsare antisymmetric in the index.2
1A-term may arise from instanton configurations in
QCD-like theories and break parity symmetry; see for instance [14].
2The antisymmetry of these tensors implies that they admit a
The expansion (2.28) involves only even powers of. It can also be truncated according to the dimensionDþ1of the spacetime relevant to the effective action. This is due to the antisymmetry of the tensors appearing in the expan-sion. Consider for instance the cases D¼4 or D¼3, where the effective bulk theory is five or four dimensional, respectively. In those cases the antisymmetry of the tensor Fmnpqrsdefined in (2.29) implies that this tensor vanishes. This is also true for the tensors appearing at higher order in. Therefore, for D¼4andD¼3, the -expansion (2.28) reduces to a polynomial or order4.
Now consider the matrix representation (2.24) of the inverse tensor Em;n. Expanding ð1þX^Þ 1 in a geometric
series and using (2.27), we get the following expansion: ~
Emn¼gmn Fmnþ2Fm
pFpn 3FmpFpqFqn þ4Fm
pFpqFqrFrn 5FmpFpqFqrFrsFsn
þ : (2.30)
We are also interested in the antisymmetric tensorE~ðm;nÞ¼ ~
Emn E~nm. The terms in the expansion (2.30) that contain even powers ofare symmetric inm$n. Therefore, the tensorE~ðm;nÞtakes the form
~
Eðm;nÞ¼ 2Fmn 23FmpF pqFqn 25FmpF
pqFqrFrsFsnþ : (2.31) If we multiply the expansion (2.31) with the determinant given in (2.28), we obtain the expansion
EE~ðm;nÞ¼detðg^Þ 2Fmn 3F
pqFmnpq
5
4 FpqFrsF
mnpqrsþ : (2.32)
1. The DBI effective action and DBI equations
The-expansion of the determinantEallows us to find a useful expansion for the effective DBI action given in (2.5). In (2.5) the Lagrangian density is proportional to pffiffiffiffiffiffiffiffiE. Using the-expansion in (2.28), we find that
ffiffiffiffiffiffiffiffi
E p
¼pffiffiffiffiffiffiffigQðz; xÞ; (2.33)
withQðz; xÞgiven by
Qðz; xÞ ¼
1þ
2
2 F mnF
mnþ 4
4!F mnpqF
mnpq
þ
6
6!F
mnpqrsF
mnpqrsþ
1=2
: (2.34)
As a consequence of (2.33) and (2.32), we find that
ffiffiffiffiffiffiffiffi
E
p ~
Eðm;nÞ¼ 2
ffiffiffiffiffiffiffig
p
Qðz; xÞ
Fmnþ2 2 FpqF
mnpq
þ
4
8 FpqFrsF
mnpqrsþ
: (2.35)
This is a very useful expansion. In the general case, where the field strengths depend not only on zbut also on time and spatial coordinates, we can use (2.35) in (2.11) to rewrite the DBI equations as
@m
2 ffiffiffiffiffiffiffig
p
Qðz; xÞ
F‘mþ2 2 FpqF
‘mpq
þ
4
8 FpqFrsF
‘mpqrsþ ¼0: (2.36)
The DBI equations written in the form (2.36) show explicitly the expansion in , which is equivalent to an expansion in the string length squared 0. This is useful if we want to truncate the DBI equations in the limit of small . In many brane models the product2 is
inde-pendent of z, so that in the limit of small the DBI equations reduce to Maxwell equations in a curved space-time with metricgmnðzÞ.
The traditional way of solving DBI equations is to choose first a gauge field ansatz and then work with an effective determinant. The advantage of Eq. (2.36) is that they are valid for a general class of gauge field configura-tions, including time- and space-dependent field strengths. As a simple application of the method described above, we will consider the case the field strengths Fmn depend only on the radial coordinate z. This case is relevant to describe the physics of constant electric and magnetic fields in the boundary. In this case we can integrate out (2.36) and obtain
j¼ 2 ffiffiffiffiffiffiffig
p
QðzÞ
Fzþ
2
2 FF z
þ
4
8 FFF
zþ : (2.37)
This equation is equivalent to (2.18) and will be used in the next section to calculate the conductivity in the presence of constant electric and magnetic fields.
III. DC CONDUCTIVITY IN3þ1DIMENSIONS
Now we consider the problem of charge transport in a 3þ1 dimensional strongly coupled field theory in the presence of constant electric and magnetic fields. The way of describing this system in holography is by consid-ering a five-dimensional effective DBI action of the form (2.3) with a gauge field of the form
A0ðzÞ ¼f0ðzÞ; A~ ðz;xÞ ¼f~ðzÞ Et~ þ1
2B~~x: (3.1) For simplicity, we also assume that the metric is homoge-neous and isotropy in the spatial directions ~xin the sense thatg11¼g22 ¼g33:¼gxxðzÞ.
QðzÞ 2j0 ¼
ffiffiffiffiffiffiffig
p gzzg00f½1þ2ðgxxÞ2jB~j2@
zf0
þ2ðgxxÞ2ðE~B~Þ ð@
zf~Þg; (3.2)
2
QðzÞ~j¼
ffiffiffiffiffiffiffig
p gzzgxxf½1þ2g00gxxjE~j2@
zf~
2g00gxx½ðE~@
zf~ÞE~ ð@zf0ÞE~B~
þ2ðgxxÞ2ðB~@zf~ÞB~g; (3.3)
where
QðzÞ fQ0ðzÞ þ2gzzg00½1þ2ðgxxÞ2jB~j2ð@
zf0Þ2
þ2gzzgxx½1þ2g00gxxjE~j2j@
zf~j2
4gzzg00ðgxxÞ2½ 2ð@
zf0ÞðE~B~Þ ð@zf~Þ
þ ðE~@zf~Þ2 þ4gzzðgxxÞ3ðB~@
zf~Þ2g1=2; (3.4)
and
Q0ðzÞ ¼1þ2½g00gxxjE~j2þ ðgxxÞ2jB~j2
þ4g00ðgxxÞ3ðE~B~Þ2: (3.5)
Note thatQ0ðzÞdoes not depend on the functionsf0ðzÞand
~ fðzÞ.
Taking the scalar product of (3.3) withE~,B~ andE~B~, respectively, we get the following equations:
QðzÞ 2E~ ~j¼
ffiffiffiffiffiffiffig
p gzzgxxfðE~@ zf~Þ
þ2ðgxxÞ2ðE~B~ÞðB~@
zf~Þg; (3.6)
QðzÞ 2B~ ~j¼
ffiffiffiffiffiffiffig
p gzzgxxf½1þ2g00gxxjE~j2
þ2ðgxxÞ2jB~j2ðB~@
zf~Þ
2g00gxxðE~B~ÞðE~@
zf~Þg; (3.7)
QðzÞ
2ðE~B~Þ ~j¼ ffiffiffiffiffiffiffig
p gzzgxxf ½1þ2g00gxxjE~j2
ðE~B~Þ ð@zf~Þ 2g00gxxjE~B~j2@
zf0g: (3.8)
A. Solutions of the DBI equations
We can solve the system (3.6) and (3.7) as a function of QðzÞ. The result is
ffiffiffiffiffiffiffig
p gzzgxxB~@ zf~
¼ QðzÞ 2Q
0ðzÞf
~
Bþ2g00gxxðE~B~ÞE~g ~j
¼: QðzÞ 4Q
0ðzÞ
jB; (3.9)
ffiffiffiffiffiffiffig
p gzzgxxE~@zf~
¼ Q2QðzÞ 0ðzÞ½
Q0ðzÞ~jE~ ðgxxÞ2ðE~B~ÞjB
¼: QðzÞ 4Q
0ðzÞ
jE: (3.10)
Similarly, we can solve (3.2) and (3.8) as a function of QðzÞ. The solutions are given by
ffiffiffiffiffiffiffig
p gzzgxxðE~B~Þ @ zf~
¼ Q2QðzÞ 0ðzÞf
2ðgxxÞ2jE~B~j2j0
þ ½1þ2ðgxxÞ2jB~j2ðE~B~Þ ~jg
¼: QðzÞ 5Q
0ðzÞ
jEB; (3.11)
ffiffiffiffiffiffiffig
p gzzg00@
zf0
¼Q2QðzÞ 0ðzÞf ½
1þ2g00gxxjE~j2j0
þ2g00gxxðE~B~Þ ~jg
¼: QðzÞ 3Q
0ðzÞ
~
j0: (3.12)
Using these results in (3.3), we get
ffiffiffiffiffiffiffig
p gzzgxx½1þ2g00gxxjE~j2@
zf~
¼Q2QðzÞ 0ðzÞf
Q0ðzÞ~jþg00gxxjEE~
ðgxxÞ2~j0E~B~ ðgxxÞ2j
BB~g: (3.13) Finally, using the solutions in the expansion for QðzÞ [Eq. (3.4)], we find a solution forQðzÞ,
QðzÞ ¼
2Q 0ðzÞ
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Q0ðzÞðzÞ þ ðgxxÞ3
½1þ2g00gxxjE~j2j~20 g
00ðgxxÞ4 ½1þ2g00gxxjE~j2j2B
r ;
(3.14)
where
ðzÞ:¼24þ 2g00ðgxxÞ2
½1þ2g00gxxjE~j2
½j~jj2þ2g00gxxð~jE~Þ2: (3.15)
density j0, the electromagnetic current ~j and the electric
and magnetic fieldsE~ andB~.
B. The Karch-O’Bannon procedure and the conductivity tensor in a general frame
The on-shell DBI action can be written as
So:s DBI¼
Z
d4xdzðzÞpffiffiffiffiffiffiffigQ
ðzÞ; (3.16)
withQðzÞgiven by (3.14).
The Karch-O’Bannon procedure proposed in [3] and further developed in [4,5] consists of imposing the reality constraint on the on-shell DBI action. In a general class of backgrounds, the metric blows up at the boundary, while the component gtt vanishes at the horizon. Under these circumstances, it is not difficult to see from (3.5) and (3.15) that the functionsQ0ðzÞandðzÞshould have a zero at a pointz¼z andz¼z, respectively. Using the identity
½1þ2g00gxxjE~j2
1þ2ðgxxÞ2ðE~B~Þ2
~ E2
¼Q0ðzÞ 2ðgxxÞ2j
~ EB~j2
jE~j2 ; (3.17)
we conclude that 1þ2g00gxxjE~j2<0 at z¼z
. Therefore, in order to avoid an imaginary action, we find from (3.14) thatj~0andjBmust vanish atz¼z. Since1þ 2g00gxxE~2
is also positive at the boundary, it should have a zero atz¼zE2, which is closer to the boundary thanz. Analyzing all the possible localizations ofz with respect tozandzE2leads to the conclusion thatz¼z(for more details see [5]).
In summary, we find from the reality constraint on the action that at some pointz¼z, the following equations must be satisfied:
Q0 ¼ 1þ 2g00gxxjE~j2 þ2ðgxxÞ2jB~j2
þ 4g00ðgxxÞ3ðE~ B~Þ2
¼ 0; (3.18)
jB¼ 2½~jB~þ2g00gxxðE~B~Þ~jE~ ¼0; (3.19)
~
j0¼f ½1þ2g00gxxjE~j2j0þ2g00gxx~j ðE~B~Þg
¼0; (3.20)
¼24þ 2g00ðgxxÞ2
½1þ2g00gxxjE~j2½j~jj
2þ2g00gxxð~jE~Þ2
¼0: (3.21)
From Eqs. (3.18), (3.19), and (3.20) we get
1þ2ðgxxÞ2jB~j2 ¼ 2g00gxx½jE~j2þ2ðgxxÞ2ðE~B~Þ2;
(3.22)
~j ðE~B~Þ ¼
2ðgxxÞ2jE~B~j2j0
½1þ2ðgxxÞ2jB~j2; (3.23)
~jB~ ¼ 2g00gxxðE~B~Þ~jE:~ (3.24)
The spatial current ~jcan be decomposed into projections onE~,B~ andE~B~ as
~j¼ 1
jE~B~j2f½ð~jE~ÞjB~j
2 ð~jB~ÞðE~B~ÞE~
þ ½ð~jB~ÞjE~j2 ð~jE~ÞðE~B~ÞB~þ~j ðE~B~ÞE~B~g:
(3.25)
Using (3.23), (3.24), and (3.25) we get ~j¼
2gxx
½1þ2ðgxxÞ2jB~j2fg
00ð~jE~Þ½E~þ2ðgxxÞ2ðE~B~ÞB~
gxxj0E~B~g: (3.26)
Using (3.21) and (3.26) we can solve ~jE~. Assuming ~j ~
E0, we get
~jE~ ¼ g00 2
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðgxxÞ3½1þ2ðgxxÞ2jB~j224þ2ðj0Þ2 q
:
(3.27)
Writing the current asji¼ijEj, we get the conductivity tensor
ij¼ g xx
½1þ2ðgxxÞ2jB~j2
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðgxxÞ3½1þ2ðgxxÞ2jB~j224þ2ðj0Þ2
q
½ijþ2ðgxxÞ2BiBjþ2gxxj0 ijkBk
: (3.28)
This result is consistent with the one obtained in [3,5] for the D3-D7 brane model. In that case the induced metric can be written as
ds2
D7
R2 ¼
1 z2
ð1 z4 z4
HÞ 2
ð1þz4 z4
HÞ
dt2þ 1
z2
1þz
4
z4
H
d ~x2
þ
1
z2þ ð@zÞ2
dz2þcos2ðzÞd2
3: (3.29)
The black brane horizon is related to the field theory temperature byzH ¼pffiffiffi2=ðTÞ. The dilaton is juste ¼ g 1
s , and the AdS radius is given byR¼ ð4Þ1=4
ffiffiffiffiffi
0 p
ðzÞ ¼20 R2 ¼
ffiffiffiffi
p
1=2;
ðzÞ ¼7 gsVS3R
8cos3ðzÞ ¼Nc
43cos3ðzÞ;
gxxðzÞ ¼ 1 z2
1þz
4
z4
H
:
(3.30)
Our result (3.28) is also consistent with that obtained recently in [12,13], using the method of open string metric.
C. The Drude limit and holographic metals
As observed in [3,5], the Drude limit is obtained by neglecting the effects of pair production (large mass limit) and higher powers in the electric field. In our framework, this corresponds to the limit !0 and g00!0. Then (3.28) reduces to
D ij ¼
j0
M½1þjB2~Mj22
ijþ BiBj 2M2þ ijk
Bk M
; (3.31)
where
M¼ 1
gxxðzÞ: (3.32)
As shown in Appendix A, this is exactly the Drude conductivity that describes a metallic behavior for particles with densityj0and massM. Note that in the case of a black
brane the induced metric contains a horizon defined by g00¼0. Then in the Drude limit the pointz¼zcoincides with the horizon zH. In particular, for the D3-D7 brane model we get from (3.30),
M¼2
ffiffiffiffi
p
ffiffiffiffi
p z2
H ¼ 2
ffiffiffiffi
p
T2; (3.33)
where¼4. As discussed in [3,5], this result can also be obtained by computing the drag force on the charge carriers in the holographic metals.
The temperature dependence in (3.33) is typical of metals. In the next section we will consider backgrounds with Lifshitz symmetry, where the conductivity has a different dependence on temperature.
IV. DC CONDUCTIVITY IN2þ1DIMENSIONS
In2þ1dimensions the magnetic field is a scalar, so the gauge field ansatz takes the form
A0ðzÞ ¼f0ðzÞ; A~ ðz;xÞ ¼f~ðzÞ Et~ þ
1
2B ijxix^j; (4.1)
where x^i is the unit vector in the spatial directions i¼ ð1;2Þ, and ijis the Levi-Civita symbol. Since the field strengths depend only on the z coordinate, we can use (2.37) to get the DBI equations
j0 ¼2 ffiffiffiffiffiffiffig
p
QðzÞ g
zzg00f½1þ2ðgxxÞ2B2@
zf0
þ2ðgxxÞ2ðE~~@
zf~ÞBg (4.2)
~j¼
2pffiffiffiffiffiffiffig
QðzÞ g
zzgxxf½1þ2g00gxxjE~j2@
zf~
2g00gxx½ðE~@
zf~ÞE~ Bð@zf0ÞE~~g; (4.3)
whereE~~ ¼"ijEjx^iandQðzÞis given by
QðzÞ ¼ fQ0ðzÞ þ2gzzg00½1þ2ðgxxÞ2B2ð@
zf0Þ2
þ2gzzgxx½1þ2g00gxxjE~j2j@
zf~j2
4gzzg00ðgxxÞ2½ 2@
zf0ðE~~@zf~ÞB
þ ðE~@zf~Þ2g1=2; (4.4)
with
Q0ðzÞ ¼1þ2g00gxxjE~j2þ2ðgxxÞ2B2: (4.5)
Following a procedure similar to that described in the previous section, we solve (4.2) and (4.3) in terms ofQðzÞ. We find in this case
ffiffiffiffiffiffiffig
p gzzg00@
zf0 ¼
QðzÞ 2Q
0ðzÞf ½
1þ2g00gxxjE~j2j0
þ2g00gxxB ~jE~~g
¼: QðzÞ 3Q
0ðzÞ
~
j0; (4.6)
ffiffiffiffiffiffiffig
p gzzgxxE~@ zf~¼
QðzÞ
2 ~jE;~ (4.7)
ffiffiffiffiffiffiffig
p gzzgxxE~~@ zf~¼
QðzÞ 2Q
0ðzÞf½
1þ2gzzgxxB2~jE~~
2ðgxxÞ2BjE~j2j0g; (4.8)
ffiffiffiffiffiffiffig
p gzzgxx½1þ2g00gxxjE~j2@
zf~
¼ Q2QðzÞ 0ðzÞ½
Q0ðzÞ~jþ2g00gxxQ
0ðzÞð~jE~ÞE~
þðgxxÞ2j~
0B ~E~: (4.9)
Finally, using the results (4.6), (4.7), (4.8), (4.9), and (4.4) we find a solution for QðzÞ in terms of the current and electromagnetic fields,
QðzÞ ¼
2Q 0ðzÞ
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Q0ðzÞðzÞ þ ðg xxÞ2 ½1þ2g00gxxjE~j2~j
2 0
r ; (4.10)
ðzÞ:¼24þ 2g00gxx
½1þ2g00gxxjE~j2
½j~jj2þ2g00gxxð~jE~Þ2: (4.11)
A. The conductivity tensor
Following the Karch-O’Bannon procedure, described in Sec.III, we find that the reality constraint of the on-shell DBI action leads to the following conditions atz¼z:
Q0 ¼1þ2g00gxxjE~j2þ2ðgxxÞ2B2 ¼0; (4.12)
~
j0¼f ½1þ2g00gxxjE~j2j0þ2g00gxxB ~jE~~g ¼0;
(4.13)
24þ 2g00gxx
½1þ2g00gxxjE~j2½j~jj
2þ2g00gxxð~jE~Þ2 ¼0:
(4.14)
From Eqs. (4.12) and (4.13), we get
1þ2ðgxxÞ2B2 ¼ 2g00gxxjE~j2; (4.15)
~jE~~¼ 2ðgxxÞ2jE~j2Bj0
½1þ2ðgxxÞ2B2: (4.16)
The current can be decomposed as
~j¼½ð~j ~ ~
EÞE~~þ ð~jE~ÞE~
jE~j2 : (4.17)
Using (4.15), (4.16), and (4.17), we obtain ~j¼
2gxx
½1þ2ðgxxÞ2B2fg00ð~jE~ÞE~ gxxBj0
~ ~
Eg: (4.18)
Using (4.14) and (4.18) we solve ~jE~. Assuming ~jE~0, we get
~jE~ ¼ g00 2
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðgxxÞ2½1þ2ðgxxÞ2B224þ2ðj0Þ2 q
:
(4.19)
Writing the current as ji¼ijEj, we get the conduc-tivity tensor
ij ¼ ðg xxÞ
½1þ2ðgxxÞ2B2
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiqðgxxÞ2½1þ2ðgxxÞ2B224þ2ðj0Þ2
ij
þ2gxxBj0
ij
: (4.20)
In this case the conductivity has two identical diagonal components 11¼22 and two opposite nondiagonal components 12¼ 21. Remember that in (4.20) the parametersandas well as the metricgxxare evaluated at the special pointz satisfyingQ0ðzÞ ¼0.
Our result (4.20) is also consistent with that obtained recently in [12,13], using the method of open string metric.
B. Lifshitz symmetry and strange metals
Here we consider the finite temperature background considered in [11] in a holographic approach to strange metals. The induced metric of a Dp-brane in the massless case can be written as
ds2 ¼R2 fðvÞdt2
v2z þ dv2
fðvÞv2þ
dx2þdy2
v2
þd2;
(4.21)
where is a compact submanifold and v is the radial direction that extends from the boundary v¼0 to the horizonvþ.
The function fðvÞ is model dependent but satisfies the universal condition fðvþÞ ¼0. As a consequence of this condition, we get
T¼jf0ðvþÞj 4vzþ 1
1
vzþ : (4.22)
The metric (4.21) realizes Lifshitz symmetry through the following scaling transformations:
t!zt; ~x! ~x; v!v: (4.23)
The parameter z is then interpreted as the dynamical critical exponent of the quantum critical sector in the dual field theory.
The Dp-brane action corresponding to (4.21) can be written as in (2.5) with
ðvÞ ¼20; ðvÞ ¼
pVe ¼:eff;
gxxðvÞ ¼R
2
v2: (4.24)
Therefore, from (4.20) we get for the Lifshitz background the conductivity tensor
ij¼
1
h
1þ 20
R2
2
v4
B2
i 8 <
:
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1þ
20
R2 2
v4
B2
2
effð20Þ4þ 20
R2 2
v4
ðj0Þ2
s
ijþ
20
R2 2
v4
Bj0 ij
9 =
This result is consistent with the one obtained in [11]. In the limit of large densities, the conductivity tensor reduces to
ij ¼
20
R2
v2
j0
h
1þ 20
R2
2
v4
B2
i
ijþ
20
R2
v2
B ij
¼:h 1 1þ B2
2M2
i
ijþ B M ij
: (4.26)
The parameteris interpreted as the dc resistivity, which has the following temperature dependence:
¼ R
2
20v2
j0 T2=z
j0 : (4.27)
For the casez¼2, the dc resistivity is linear in the tem-perature, as expected for strange metals. Note that the second line in (4.26) is the same obtained in a 2þ1-dimensional Drude model withMgiven by
M¼ R
2
20v2
T2=z: (4.28)
V. CONCLUSIONS
We have proposed a simple expansion method to inves-tigate DBI equations and define holographic currents in a general class of backgrounds (whose metric depends on the radial coordinate only) and Dp-brane configurations. Our method is alternative to the traditional approach where an effective determinant is calculated for each field configu-ration. In our method, we write the DBI equations for arbitrary gauge field configurations before choosing any ansatz. We expect that our results would be useful in several holographic models that are derived from string theory. A future direction in our approach would be to extend the expansion method for the case of backgrounds with conical deficits (see for instance [15]).
As an application of our method we have discussed thoroughly the situation where constant electric and mag-netic fields are turned on. Following the Karch-O’Bannon method, we found for a general background, satisfying homogeneity and isotropy, a frame-independent expression for the conductivity tensor in3þ1and2þ1dimensions. For particular backgrounds we have recovered previous results on holographic metals and strange metals. An in-teresting point in that analysis was that the presence of a horizon guarantees the existence of a solution that satisfies the real constraint.
It would be interesting also to investigate physical systems, where the electric or magnetic fields have a time or spatial dependence. We leave this problem for future work.
ACKNOWLEDGMENTS
The authors would like to thank the ICTP in Trieste (Italy) and the ICTP-SAIFR in Sao Paulo (Brazil) for their
hospitality. The work of A. B. B. is supported by a Capes fellowship. The work of C. N. F is partially supported by a CNPq fellowship.
APPENDIX A: THE DRUDE MODEL
The Drude equation for an electromagnetic current can be written as
j0E~ M ~jþ~jB~ ¼0; (A1)
where is the drag coefficient andM is the mass of the charge carrier. We can write this equation in a matrix form,
Mijjj¼ j0E
i; (A2)
where
Mij¼ Mijþ ijkBk: (A3)
The inverse of this object is
M 1
ij ¼
1
Mh1þjB2~Mj22
i
ijþ BiBj 2M2þ ijk
Bk M
:
(A4) Using this result we get the Drude conductivity
D ij¼
j0
Mh1þjB2~Mj22
i
ijþ BiBj 2M2þ ijk
Bk M
: (A5)
APPENDIX B: THE EFFECT OF A FIVE-DIMENSIONAL CHERN-SIMONS TERM
In some brane models that reduce to five dimensions, we obtain an effective Chern-Simons (CS) action,
SCS ¼
4
‘mnpqZ d4xdzA
‘FmnFpq: (B1)
In the special case of (3.1), where we have constant electric and magnetic fields, the DBI-CS equations reduce to
0¼ ~EB;~ (B2)
j0 ¼2
QðzÞ
ffiffiffiffiffiffiffig
p gzzg00f½1þ2ðgxxÞ2B~2
@zf0
þ2ðgxxÞ2ðE~B~Þ ð@zf~Þg 6 ~Bf;~ (B3)
~j¼
2
QðzÞ
ffiffiffiffiffiffiffig
p gzzgxxf½1þ2g00gxxjE~j2@
zf~
2g00gxx½ðE~@
zf~ÞE~þ ð@zf0ÞE~B~
þ2ðgxxÞ2ðB~@
zf~ÞB~g þ6½f0B~þf~E~: (B4)
Here the quantities j0 and ~j are integration constants, not
necessarily the electromagnetic current.3
3The definition of holographic currents in the presence of a
According to Eq. (B2) the electric and magnetic field must be orthogonal.4Note that we can rewrite the last term in (B4) as
6½f0B~þf~E~
¼6
½f0jB~j2þ ðE~B~Þ f~ B~ jB~j2
ðB~f~Þ jB~j2 E~B~
:
(B5)
Here we used the vectorial decomposition
~
f¼ 1
jE~B~j2f½ðf~E~ÞjB~j 2
ðf~B~ÞðE~B~ÞE~þ ½ðf~B~ÞjE~j2
ðf~E~ÞðE~B~ÞB~þf~ ðE~B~ÞE~B~g (B6)
and the constraintE~B~ ¼0.
Taking the scalar product of (B4) withE~,B~ andE~B~, respectively, we get the equations:
ffiffiffiffiffiffiffig
p gzzgxxE~@zf~¼ QðzÞ
2 ~jE;~ (B7)
ffiffiffiffiffiffiffig
p gzzgxxB~@zf~
¼ Q2QðzÞ 0ðzÞf
~jB~þ6½f0jB~j2þ ðE~B~Þ f~g;
(B8)
ffiffiffiffiffiffiffig
p gzzgxxf½1þ2g00gxxjE~j2ðE~B~Þ @
zf~
2g00gxxjE~B~j2@
zf0g
¼QðzÞ
2½~j ðE~B~Þ 6ðB~f~ÞjE~j2: (B9)
Here we usedE~B~¼0. From (B3) and (B9) we get
ffiffiffiffiffiffiffig
p gzzgxxðE~B~Þ @ zf~
¼ Q2QðzÞ 0ðzÞf
2ðgxxÞ2jE~B~j2j0
þ ½1þ2ðgxxÞ2jB~j2~j ðE~B~Þ 6ðf~B~ÞjE~j2g
¼: QðzÞ 5Q
0ðzÞf
^
jEB 63ðf~B~ÞjE~j2g and (B10)
ffiffiffiffiffiffiffig
p gzzg00@
zf0 ¼
QðzÞ 2Q
0ðzÞf ½
1þ2g00gxxjE~j2j0
þ2g00gxx~j ðE~B~Þ þ6 ~fB~g
¼: QðzÞ 3Q
0ðzÞf
^
j0þ6 ~fB~g: (B11)
Using (B5), (B7), (B8), and (B11) in (B4), we get
ffiffiffiffiffiffiffig
p gzzgxx½1þ2g00gxxjE~j2@
zf~ ðB~@zf~Þ ~ B jB~2j
¼Q4QðzÞ 0ðzÞ
Q0ðzÞ
2~jþ2ð~jB~Þ B~
jB~j2
4Q
0ðzÞg00gxxð~jE~ÞE~
2ðgxxÞ2j^0 6
jB~j2½1þ
2g00gxxjE~j2ðB~f~ÞE~B~: (B12)
We can take the derivative of (B8) and use (B10) and (B11) to get a second-order differential equation forB~f~,
ffiffiffiffiffiffiffig
p gzzgxx5Q0ðzÞ
QðzÞ
@z
2Q 0ðzÞ
QðzÞ
ffiffiffiffiffiffiffig
p gzzgxx@zðB~f~Þ
¼ 6fg00gxx2jB~j2j^
0 j^EBþ6½g00gxx2jB~j2þ2jE~j2 ~Bf~g: (B13)
Finally, using (B7) and (B10)–(B12) in the definition ofQðzÞ, we get
4If we want to consider parallel electric and magnetic fields, we would need to introduce a time dependence on theA
0component
QðzÞ 2Q
0ðzÞ¼
ffiffiffiffiffiffiffiffiffiffiffiffi
Q0ðzÞ
q
1þ2gzzgxxðB~@zf~Þ2 jB~j2
1=2
f24Q2
0ðzÞ þ ðgxxÞ3½1þ2ðgxxÞ2jB~j2½j^0þ6 ~Bf~2
2g00ðgxxÞ4½j^
0þ6 ~Bf~½j^EB 63ðB~f~ÞjE~j2 ðg00Þ2ðgxxÞ34Q20ðzÞð~jE~Þ2
þ g
00ðgxxÞ2
2½1þ2gzzg00jE~j2 Q0ðzÞ
2~jþ2ð~jB~Þ B~ jB~j2
4Q
0ðzÞg00gxxð~jE~ÞE~
2½ðgxxÞ2j^0 6
jB~j2½1þ
2g00gxxjE~j2ðB~f~ÞE~B~j2g 1=2: (B14)
[1] J. M. Maldacena, Adv. Theor. Math. Phys.2, 231 (1998). [2] A. Karch and E. Katz,J. High Energy Phys. 06 (2002) 043. [3] A. Karch and A. O’Bannon, J. High Energy Phys. 09
(2007) 024.
[4] A. O’Bannon,Phys. Rev. D76, 086007 (2007).
[5] M. Ammon, T. H. Ngo, and A. O’Bannon,J. High Energy Phys. 10 (2009) 027.
[6] S. Kachru, X. Liu, and M. Mulligan, Phys. Rev. D 78, 106005 (2008).
[7] U. H. Danielsson and L. Thorlacius,J. High Energy Phys. 03 (2009) 070.
[8] R. B. Mann,J. High Energy Phys. 06 (2009) 075. [9] G. Bertoldi, B. A. Burrington, and A. Peet,Phys. Rev. D
80, 126003 (2009).
[10] K. Balasubramanian and J. McGreevy, Phys. Rev. D80, 104039 (2009).
[11] S. A. Hartnoll, J. Polchinski, E. Silverstein, and D. Tong,
J. High Energy Phys. 04 (2010) 120.
[12] K.-Y. Kim, J. P. Shock, and J. Tarrio,J. High Energy Phys. 06 (2011) 017.
[13] K.-Y. Kim and D.-W. Pang, J. High Energy Phys. 09 (2011) 051.
[14] D. E. Kharzeev, Ann. Phys. (Amsterdam) 325, 205 (2010).
[15] C. A. B. Bayona, C. N. Ferreira, and V. J. V. Otoya,
Classical Quantum Gravity28, 015011 (2011).
[16] A. Rebhan, A. Schmitt, and S. A. Stricker,J. High Energy Phys. 01 (2010) 026.
[17] A. Gynther, K. Landsteiner, F. Pena-Benitez, and A. Rebhan,J. High Energy Phys. 02 (2011) 110.