AdS 5 /CFT 4 correspondence
4.1. Holographic s-Wave Superconductor
4.1.5. Fluctuations about background fields
According to the holographic fluctuation-dissipation theorem explained in Section 3.5.1, the linear response functions are related to the linearized equations of motions in the fluctuations about the background solutions. Thus, in this section we will derive the full probe limit and backreaction linearized equations of motion of the field fluctuations.
Fluctuation in the probe limit
We now look at fluctuations of the scalar fieldΦpuqand the gauge field Apuqabout there fixed background values with the scalar potential VpΦq “ m2Φ2. The metric will be fixed to the Schwarzschild solution since we do not consider any backreaction. The complex scalar field and the gauge field will be replaced by
ΦÝÑΦ`δφ, AaÝÑAa`aa, (4.70)
withΦ PRas stated in (4.23), butδφ PC. We allow a dependence on all coordinates for the fluctuations,i.e.δφ“δφpt,x, uqandaa “aapt,x, uq. Expanding the action (4.1) up to quadratic order in the fluctuations we find
SF“SFp0q`SFp1q`SFp2q, (4.71) where
SFp0q“SM, (4.72)
SFp1q“ 1 e2
ż
dd`1x?
´g
„
´1 4
`δFabFab`FabδFab˘
´∇aΦ∇apδφ`δφ˚q
(4.73)
´iAa∇apδφ´δφ˚qΦ`i∇aΦAapδφ´δφ˚q
´AaAaΦpδφ`δφ˚q ´2AaaaΦ2´m2Φpδφ`δφ˚q
, (4.74)
SFp2q“ 1 e2
ż
dd`1x?
´g
„
´1
4δFabδFab´∇aδφ∇aδφ˚`iAa∇apδφ˚qδφ
´iAa∇apδφqδφ˚´iaa∇apδφ´δφ˚qΦ`i∇aΦaapδφ´δφ˚q
´AaAaδφδφ˚´2ABaaaΦpδφ`δφ˚q ´aaaaΦ2´m2δφδφ˚
. (4.75)
Variations with respect to the fluctuations ofSFp0qare vanishing, while the variation ofSFp1qgives rise to the equations of motions of the background fields as expected, since this is the usual variation procedure. The variations of the quadratic action in the fluctuations yield the linearized equations of motions for the fluctuations
p∇a´iAaq p∇a´iAaqδφ´m2δφ´i∇apaaΦq ´iaa∇aΦ´2AaaaΦ“0, (4.76)
∇aδFab`i∇bΦpδφ´δφ˚q ´i∇bpδφ´δφ˚qΦ´2AbΦpδφ`δφ˚q ´2abΦ2“0, (4.77) whereδFabis given by
δFab“ BapAb`abq ´ BbpAa`aaq ´ BaAb` BbAa“ Baab´ Bbaa. (4.78) In order to work out the quasi-normal modes we apply a Fourier transformation and assume plain wave behavior for the spatial dependence
δφpt,x, uq “
ż dωdd´1k
p2πqd e´iωt`ik¨xδφpuq, aapt,x, uq “
ż dωdd´1k
p2πqd e´iωt`ik¨xaapuq.
(4.79)
Thus, we end up with the following equation of motion for the scalar fluctuations
δφ2puq ` ˆf1puq
fpuq ´d´1 u
˙
δφ1puq `
„pω`Atpuqq2 fpuq2 ´ k2
fpuq´ L2m2 u2fpuq
δφpuq
`
„ˆω`2Atpuq fpuq2
˙
atpuq ` 1
fpuqk¨apuq
Φpuq
´iΦa1upuq ´i
„ˆf1puq
fpuq ´d´1 u
˙
Φpuq `2Φ1puq
aupuq “0, (4.80)
and for the gauge field fluctuations we have a2tpuq ´d´3
u a1tpuq ´ ˆ k2
fpuq`2L2Φpuq2 u2fpuq
˙
atpuq ´ ω
fpuqk¨apuq
´ L2 u2fpuq
”`ω`2Atpuq˘
δφpuq ´`
ω´2Atpuq˘ δφ˚ı
Φpuq
`iω ˆ
a1upuq ´d´3 u aupuq
˙
“0, (4.81)
a2puq ` ˆf1puq
fpuq ´d´3 u
˙ a1puq `
ˆ ω2
fpuq2 ´ k2K
fpuq´2L2Φpuq2 u2fpuq
˙ apuq
` k fpuq
`kK¨apuq˘
` ωk
fpuq2atpuq ` L2k u2fpuq
`δφpuq ´δφ˚puq˘ Φpuq
´ik
„
a1upuq ` ˆf1puq
fpuq ´d´3 u
˙ aupuq
“0, (4.82)
ω
fpuqa1tpuq `k¨a1puq ´i ˆ
k2´ ω2
fpuq`2L2Φpuq2 u2
˙ aupuq
´L2 u2
”`δφpuq ´δφ˚puq˘
Φ1puq ´`
δφ1puq ´δφ1˚puq˘ Φpuqı
“0, (4.83) where kK denotes the transverse vector orthogonal to the directions given by the particular equations for the component ofapuq. For instance, we can look at the equation for one of the d´1spatial components,apuq “axpuqexsay, sokK“kyey`kzez`¨ ¨ ¨ and hencekK¨acouples the complementary components to the one chosen,i.e.aypuq, azpuq, . . .. Furthermore, assumingk“ kxex, we see that the equations of theaxpuqcomponent couples only with theatpuqcomponent (in this casekK “ 0), whereas the equations for all other d´2 spatial components decouple.
Looking at the transverse directionse.g.k“kyey`kzez`¨ ¨ ¨, the equation for the fluctuations in thex-direction will decouple from all otherd´2fluctuation equations, but the equations for the remainingd´2fluctuations will couple with each other andatpuq. The background solutions are given by the AdS-Schwarzschild solutions forfpuq(4.65) and the numerical integrated solution for the coupled system (4.68) and (4.69).
Fluctuations with backreaction
Turning on the backreaction we need to consider the fluctuations of the metric field, too,
gab“Gab`hab, (4.84)
Thus, we need to expand the determinant of the metric up to quadratic order δp2qg“detpGab`habq ´detGab
“ BpdetGq
BGab hab`1 2
B2pdetGq
BGabGcdhabhcd, (4.85)
and the square root of (4.85) δp2q?
´g“B? - detG BdetG
BdetG BGab
hab`1 4
˜ B BGab
B? - detG BGcd ` B
BGcd
B? - detG BGab
¸ habhcd
“?
´detG
„1
2Gabhab` ˆ1
8GabGcd´1 4GbcGad
˙ habhcd
, (4.86)
where we symmetrized the second partial derivative in its indices and inserted the well known relation
B? - detG BGab “1
2
?´GGab. (4.87)
Up to linear order in the fluctuations the indices of the metric fluctuationshabare raised and lowered by the background metricGab, so we can rewrite (4.86) as
adetpG`hq “? detG
„ 1`1
2trh´1 4tr`
h2˘
`1 8ptrhq2
, (4.88)
which can be compared to the expansion usingdetp1`Mq “ expptr lnp1`Mqq as given in AppendixA.1.2, (A.10). This will lead to additional terms in (4.74) and (4.75) coming form the quadratic expansion of?
´gin the fluctuations Lp1q
M “?
´G
„
´1
4GceGdfpFcdδFef `δFcdFefq `GcdLp2q
φcd
´1 4
ˆ
GcsGethstGdf`GdsGf thstGce`1
2GsthstGceGdf
˙ FcdFef
` ˆ
GcsGdthst`1
2GsthstGcd
˙ Lp0q
φcd
, (4.89)
Lp2q
M “?
´G
"
´1
4GceGdfδFcdδFef `GcdLp2q
φcd
´1 4
ˆ
GcsGethstGdf `GdsGf thstGce`1
2GsthstGceGdf
˙
pFcdδFef `δFcdFefq
` ˆ
GcsGdthst`1
2GsthstGcd
˙ Lp1q
φcd
´1 4
„
GcsGethstGdmgf nhmn`1
2Gmnhmn`
GcsGethstGdf`GdsGf thstgce˘
` ˆ1
8gstGmn´1
4GtmGsn
˙
hsthmnGceGdf
FceFdf
`
„1
2GsthstGcmGdnhmn` ˆ1
8GstGmnhsthmn´1
4GtmGsn
˙
hsthmnGcd
Lp0q
φcd
*
. (4.90)
Note that here we write the total action in the form S“ 1
2κ2 ż
dd`1x`
LEH`2α2L2LM˘
“SEH`SM, (4.91)
and the scalar field LagrangianLφcdis given by Lp0q
φcd“ ´ p∇cΦ`iAcΦq p∇dΦ´iAdΦq ´m2Φ2, Lp1q
φcd“ ´∇cδφ˚∇dΦ´∇dδφ∇cΦ`i∇cΦAdδφ´i∇dΦAcδφ˚´iAc∇dδφΦ
`iAd∇cδφ˚Φ´AcAdΦpδφ`δφ˚q ´ pAcad`AdacqΦ2´m2Φpδφ`δφ˚q, Lp2q
φcd“ ´∇cδφ∇dδφ˚´iAc∇dδφδφ˚`iAd∇cδφ˚δφ´iac∇dδφΦ`iad∇cδφ˚Φ
`i∇cΦadδφ´i∇dΦacδφ˚´AcAdδφδφ˚´ pAcad`Adacq pδφ`δφ˚q
´aaabφ2´m2δφ˚δφ.
(4.92)
The Einstein-Hilbert term expanded up to quadratic order in the fluctuations can be casted in the following form
Lp1q
EH “?
´Gδp1qR`δp1q?
´gpR´2Λq, (4.93)
Lp2q
EH “?
´Gδp2qR`δp1q?
´gδp1qR`δp2q?
´gpR´2Λq, (4.94) where
δp2qR“ ´hab∇pcGq∇pGqchab´3
4GabBahcdBbhcd´gabBaln?
´dethBbln?
´deth, (4.95) with ∇pGq denoting the background curved space covariant derivative constructed only out of the background metricGab. The full backreacted equations of motions are listed in AppendixC, where we applied the Fourier transformation for all fluctuationsi.e.
δφpt,x, uq “
ż dωdd´1 k
p2πqd e´iωt`ik¨xδφpuq, aapt,x, uq “
ż dωdd´1 k
p2πqd e´iωt`ik¨xaapuq, habpt,x, uq “
ż dωdd´1 k
p2πqd e´iωt`ik¨xhabpuq.
(4.96)
For most of the computations, conducted in Section 4.3 and 4.4, involved in applications to linear response and Homes’ law, the normal phase equations of motions are sufficient. The normal phaseΦ“0andχ“0equations of motion for the scalar field fluctuations are given by
δφ2puq ` ˆf1puq
fpuq ´d´1 u
˙
δφ1puq `
„pω`Atq2 fpuq2 ´ k2
fpuq´ L2m2 u2fpuq
δφpuq “0. (4.97)
The gauge field equations are extended by metric fluctuations as well, wherehtpuq,hupuqdenotes the spatial entries of the corresponding row/column
a2tpuq ´d´3 u a1tpuq ´
ˆ k2 fpuq
˙
atpuq ´ ω
fpuqk¨apuq `iω ˆ
a1upuq ´d´3 u aupuq
˙
´3u2A1tpuq
2L2fpuqh1ttpuq `u2A1tpuq 2L2
dÿ´1 i“1
h1iipuq `3u2fpuqA1tpuq 2L2 h1uupuq
` u L2A1tpuq
dÿ´1 i“1
hiipuq `iuA1tpuq L2
dÿ´1 i“1
huipuqki`3 2
u2fpuq L2
ˆf1puq fpuq ` 2
u
˙
A1tpuqhuupuq “0, (4.98)
a2puq ` ˆf1puq
fpuq ´d´3 u
˙ a1puq `
ˆ ω2
fpuq2´ k2K fpuq
˙
apuq ` k fpuq
`kK¨apuq˘
` ωk
fpuq2atpuq ´ik
„
a1upuq ` ˆf1puq
fpuq ´d´3 u
˙ aupuq
´u2A1tpuq
L2fpuqh1tpuq ´2uA1tpuq
L2fpuqhtpuq ´iu2ωA1tpuq
L2fpuq hupuq “0, (4.99)
ω
fpuqa1tpuq `k¨a1puq ´i ˆ
k2´ ω2 fpuq
˙ aupuq
´3 2
u2ωA1tpuq
L2fpuq2 httpuq ´u2A1tpuq L2fpuq
dÿ´1 i“1
htiki`u2ωA1tpuq 2L2fpuq
dÿ´1 i“1
hiipuq `3 2
u2ωA1tpuq
L2 huupuq “0.
(4.100) Note that in the normal phase the background solution for the metric is given by the AdS-Reissner-Nordström black brane, sofpuqis given by (4.49), as well asAtpuqby (4.48).
Numerical solution
Numerical solutions are integrated by adapting the Mathematica CodeD.2for fluctuations. In particular, we need to modify the asymptotic expansion at the horizon to use an index associated with infalling boundary conditions, as explained in detail in Section4.3.6, in order to extract the retarded response function according to the holographic fluctuation-dissipation theorem outlined in
i
toiv
on page111. The numerical response function is defined as the ratio of the subleading coefficient to the leading coefficient in the boundary expansion. The numerically obtained key features of holographic superconductors are shown in Figure4.110
0 2 4 6 8
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Superconducting Order Parameter
T{Tc xOyrTcs
axO2y xO1y
0 50 100 150 200 250
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Optical Conductivity
ω{T ReσrTcs
Figure 4.1. In the left panel, the expectation value of the dual operator is shown which serves as a order parameter for the superconducting phase transition for α “ 0, d “ 3 andm2L2 “ ´2. In this case we may define two dual operators,
O
∆ andO
d´∆, since ´9{4 ă m2L2 ă ´5{4 as explained below (3.77). In order to compare the curves for the dimension one and two operator we need to take the square root ofO
∆“2. In the following we will only work with the dimension two operatorO
∆with RG scaling fieldd´∆“1. The continuous phase transition is of mean-field type with the critical exponentβ “1{2,c.f. Table2.6. The right panel shows the response function of the spatial gauge field fluctuation corresponding to the optical conductivity σpωq. The real part measures the dissipation of the system, c.f. Section4.3.2, which is exponentially suppressed in the superconducting phase.The normal phase solutions for d “ 3 is given by a frequency independent opti-cal conductivity. This is a general properties of conformal theories [181]. For IR deformation induced by the non-trivial profile of the scalar field solution in the limitω !T, the conformal solution is lost and a superconducting condensate al-lows for nearly dissipationless transport. The curves are color coded as folal-lows:
T{Tc “ 1,0.5274,0.2464,0.165,0.1212,0.0974. Generically, upon approaching the conformal limit in the UV,ω "T, the optical conductivity is proportional toωd´3 fordą2as derived in (4.190). Deeper implications of the gap in the real part of the optical conductivity and its connection to the superfluid strength is discussed extensively in Section4.4.1.