Exact exponential solutions in Einstein-Gauss-Bonnet flat anisotropic cosmology Dmitry Chirkov Sternberg Astronomical Institute, Moscow State University, Moscow 119991 Russia and Faculty of Physics, Moscow State University, Moscow 119991 Russia Sergey Pavluchenko Instituto de Ciencias Físicas y Matemáticas, Universidad Austral de Chile, Valdivia, Chile Alexey Toporensky 4 Sternberg Astronomical Institute, Moscow State University, Moscow 119991 Russia 1 0 2 Inthis paperwearelookingforthe exponentialsolutions(i.e. the solutionswith the scale n factorschangeexponentiallyovertime)intheEinstein-Gauss-Bonnetgravity. Wearguethat a J wefoundallpossiblenon-constant-volumesolutions(uptopermutations)inlowerdimensions 3 ((4+1)and (5+1)) anddeveloped a scheme which allowsone to find necessaryconditions in 1 arbitrary dimension. ] c PACS numbers: 04.20.Jb, 04.50.-h, 98.80.-k q - r g [ I. INTRODUCTION 1 v 2 6 Duetothedevelopmentsinthestringandrelatedtheoriestheinteresttothemultidimensionalgravitywas 9 2 increasing rapidly in the last decades. Indeed, the gravitational counterpart of these theories allows one to . 1 perform some kind of independent viability test on them. Thinking of possible implications of gravitational 0 4 theories one can notice that the observational data on the compact objects (including what is believed to 1 v: be “black holes”) is still too unprecise to make differences between different gravitational theories – not to i X mention, the existence of the black holes is still have not been proven. On the other hand, the gravitational r a waves are also beyond our reach – let us hope, just for now; these two facts make cosmology the best (if not the only) candidate to test the viability of the modified gravity. Mainly because of this, since 1980s the studies of the modified multidimensional cosmology were quite intense (see e.g. [1–20]). When doing multidimensional cosmology, it is naturally to do it anisotropic – indeed, if we want to link originally multidimensional universe to our observed (3+1)-dimensional Universe, one have to think of a way to reduce the number of dimensions – a way to compactify – and this is clear indication of anisotropy. Andtherearesuccessfulattemptstodescribe“natural”compactification –duetodynamical[11] or topological [20] effects. Yet, studies of the generic anisotropic solutions is also important since with wider analysis we get better understanding of the dynamics. Studies of anisotropic cosmological solutions in multidimensional Gauss-Bonnet (GB) gravity reveals several important differences between GB and General Relativity (GR). More than 20 years ago it was shown that Kasner solution of GR is replaced by similar power-law solution with different relations between power indices [5, 9]. What is more interesting, it appears that the form of solution similar to the Kasner solution in GR is not unique for a flat anisotropic background in GB. Another possible solution with two arbitrary power indices (with others equal to zero) have been obtained in the above cited paper. No counterpart of this solution existsinGR.In[15]suchsolutionhavebeengeneralized forhigherLovelock terms. Itisinteresting, however, thatthoughsuchsolutionscontaining unconstrainedvaluesofappropriatenumberofpowerindicesareexact solutions in Lovelock gravity when only the highest Lovelock term is taken into account, they disappear when subdominant terms are added [19]. Recently another type of vacuum solution in GB gravity have been found. In contrast to any analogue of the Kasner solution, the scale factors in the solution in question change exponentially [21]. Again, no analogueofthissolutionexistsinGR.In[17]thissolutionhavebeengeneralizedtonon-vacuumcase,though detailed analysis of its existence have not been provided. In the present paper we study the conditions for the exponential solution to exist in non-vacuum case. In addition, we provide full analysis of exponential solutions in (4+1) and (5+1) dimensions in models with a cosmological constant, generalizing particular results of [17]. Thestructureofthemanuscriptisasfollows: inthesecondsection wepresenttheequations ofmotionfor thesystem underconsideration, inthird–consider thecasein low ((4+1) and(5+1)) numberof dimensions, in the following section we generalize some of the solutions obtained on the higher number of dimensions and finally in the Conclusions sum up the obtained results. II. THE SET-UP. We start from the Einstein-Gauss-Bonnet action in (N +1)-dimensional spacetime1: 1 S = dN+1x det(g) +α + , = R, = R Rαβγδ 4R Rαβ +R2, (1) E GB m E GB αβγδ αβ 16π | | L L L L L − Z q (cid:0) (cid:1) where R,R ,R are the (N +1)-dimensional scalar curvature, Ricci tensor and Riemann tensor respec- αβ αβγδ tively; is the Lagrangian of a matter; det(g) is the determinant of a metric tensor g. We introduce the m L metric components as g00 = 1, gkk = e2Hkt, gij = 0, i , j. (2) − 1 Throughout thepaperweusethesystem ofunitsin whichG=c=1, Gisthe(N+1)-dimensional gravitational constant,c is the speed of light. Greek indices run from 0 to N,Latin indices from 1 to N. 2 In the following we interest in the case of H const. In this work we deal with a perfect fluid with the i ≡ equation of state p = ωρ as a matter source. Let κ = 8πρ; the equations of motions now can be written as follows: 2 H2+2 H H +8α H2 H H +24α H H H H = ωκ, j = 1,N; (3) i i k i k l i k l m − i,j i>k ,j i,j k>l , i,j i>k>l>m ,j X { X} X { X} { } { X } 2 H H +24α H H H H = κ. (4) i j i j k l i>j i>j>k>l X X Since we have H const, it follows from Eqs. (3)–(4) that ρ const, so that the continuity equation i ≡ ≡ ρ˙+(ρ+p) H = 0 (5) i i X reduces to (ρ+p) H = 0, (6) i i X which allows several different cases: a) ρ 0 (vacuum case), b) ρ+ p = 0 (Λ-term case), c) H = 0 i i ≡ (constant volume case) and their combinations: d) H = 0 vacuum and e) H = 0 with ΛP-term. In i i i i present paper we do not consider constant volume soPlutions (CVS), leaving thePir description for a separate paper. So we left with only options – either vacuum (ρ = 0) or Λ-term (ρ+p = 0) and further we consider only these two cases. III. LOW-DIMENSIONAL CASES. Subtracting i-th dynamical equation from j-th one we obtain: 1 (H H ) + H H H = 0 (7) j i k l k − 4α  k>l , i,j k { X} { } X   It follows from (7) that 1 H = H H H = H = 0 (8) i j k l k ∨ −4α ∨ k>l , i,j k { X} { } X These are necessary conditions for a given set H1,...,HN to be a solution of Eqs. (3)–(4). The case H = 0 is CVS and will be considered in a separate paper; in this section we deal with the following k k pPossibilities only: 1 H = H H H = (9) i j k l ∨ −4α k>l , i,j { X} { } We call the left equality as type I condition, the right equality as type II condition. 3 Table I. The necessary conditions. type I type II type I type II 1 1 1 H1 =H2 A H3H4 =−4α 4 H1 =H4 D H2H3 =−4α 1 1 2 H3 =H4 B H1H2 =−4α 5 H2 =H3 E H1H4 =−4α 1 1 3 H1 =H3 C H2H4 =−4α 6 H2 =H4 F H1H3 =−4α A. (4+1)-dimensional spacetime. In this case we have 6 pairs of the type I and type II conditions. The table I lists all such pairs. We assign numbers from 1 to 6 to each type I condition and letters from to to each type II condition. A F There are the following combinations of type I and type II conditions: 1 1. 0 type I conditions, 6 type II conditions = H1 = H2 = H3 = H4 H = ; ⇒ ≡ √ 4α − 1 2. 1 (any) type I conditions, 5 type II conditions = H1 = H2 = H3 = H4 H = ; ⇒ ≡ √ 4α − 3. 2 type I conditions, 4 type II conditions: (a) typeIconditionshasnoidentical parameters, forexample, (seetableI)1 2 = − −B−D−E−F ⇒ 1 H1 = H2 ≡ H, H3 = H4 ≡ h, Hh= −4α; (b) bothof typeIconditions includeonethesameparameter, for example, 1 3 = − −C−D−E−F ⇒ 1 H1 = H2 = H3 = H4 H = ; ≡ √ 4α − 4. 3 type I conditions, 3 type II conditions: (a) a chain of conditions does not include one of the parameters, for example, 1 3 5 − − −C−D−F (H4 is absent) = H1 = H2 = H3 H = 1 , H4 h R, ⇒ ≡ √ α ≡ ∈ − 1 (b) a chain of conditions includes all the parameters = H1 =H2 = H3 = H4 H = ; ⇒ ≡ √ 4α − 1 5. 4 (or 5) type I conditions, 2 (or 1) type II conditions = H1 = H2 = H3 = H4 H = ; ⇒ ≡ √ 4α − 6. 6 type I conditions, 0 type II conditions = H1 = H2 = H3 = H4 H R. ⇒ ≡ ∈ Summarizing aforesaid we see that there are three cases: i) H1 = H2 = H3 = H4 H R (clearly, ≡ ∈ H = 1 is the subcase of that case); ii) H1 = H2 = H3 H = 1 , H4 h R; iii) H1 = H2 H, √ 4α ≡ √ 4α ≡ ∈ ≡ − − 1 H3 =H4 ≡ h, Hh= −4α. In (4+1)-dimensional spacetime Eqs. (3)–(4) have the form: 4 4 4 4 2 H2+2 H H +8α H2 H H = ωκ, j = 1,4 (10) i i k i k l − i,j i>k ,j i,j k>l , i,j X { X} X { X} { } 4 4 2 HiHj +24αH1H2H3H4 = κ (11) i>j X Now we consider solutions of the Eqs. (10)–(11) for each of the cases i)–iii). i) After substitution H1 = H2 = H3 = H4 H into Eqs. (10)–(11) one can see that they reduce to ≡ 12H2 +24αH4 +ωκ = 0, 12H2+24αH4 κ = 0 (12) − one can see that the system (12) could be resolved only if κ= 0 (vacuum case) or ω = 1 (Λ-case); both of − these cases could be reduced to the following solution: 1 2κα κ > 3 , α> 0 H2 = 1 1+ , −2α (13) 4α − ± r 3 ! κ 6 −23α, α< 0  For vacuum case (κ = 0) we obtain H = 0 or H2 = 1 , α < 0. It is interesting to note that the cases 1, −2α 3 2, 3b, 4b and 5 are realized as a subcase of the case iii) for Λ= , α < 0 (see below). −16πα ii) Substitution H1 = H2 = H3 H = 1 , H4 h R into Eqs. (10)–(11) gives a system which ≡ √ 4α ≡ ∈ − could be resolved only under ω = 1, which gives us 3 = κ. − −2α 1 iii) Substitution H1 = H2 ≡ H, H3 = H4 ≡ h with the condition Hh = −4α into Eqs. (10)–(11) leads to a system that could be resolved only under ω = 1 which could be further brought to − κ 1 2 2 H +h = + (14) 2 4α It follows from (14) immediately that κ> 1 . Substituting h = 1 into (14) we obtain −2α −4αH κ 1 1 4 2 H + H + = 0 (15) − 2 4α 16α2 (cid:18) (cid:19) Solutions of Eq. (15) are 2 1 1 1 1 1 1 1 1 H2 = κ+ κ+ , κ > κ 6 (16) 4 2α ± s 2α − α2 α − 2α ∨ − α − 2α (cid:18) (cid:19) | | | |   One can easily see that in vacuum case (κ = 0) the radical in (16) is negative so such solution do not exist in that case. We summarize the results in the tables II,III. The vacuum case admits only isotropic solution: H2 = 1 , α < 0. Let us note that solution with one −2α the Hubble parameter different from the other three (H1 = H2 = H3 = H, H4 = h) exists for the single 3 value of Λ: Λ= , α < 0. Also one can note that (H, H, h, h) solution tends smoothly to the isotropic −16πα 3 solution for Λ = and negative α. −16πα 5 Table II. Λ-term solutions, α>0. 0<Λ6 1 Λ> 1 16πα 16πα (H,H,h,h) No H2 = 1 8πΛ+ 1 8πΛ+ 1 2 1 ,h= 1 4 2α 2α α2 4αH ± − − (cid:18) q (cid:19) (H,H,H,h) No (cid:0) (cid:1) (H,H,H,H) H2 = 1 1+ 1+ 16παΛ 4α − 3 (cid:18) q (cid:19) Table III. Λ-term solutions, α<0. 3 3 3 Λ< Λ= Λ> 16πα 16πα 16πα − − − H2 = 1 8πΛ+ 1 8πΛ+ 1 2 1 , (H,H,h,h) No H2 =h2 = 1 4 2α ± 2α − α2 −4α (cid:18) h= q(cid:0)1 (cid:1) (cid:19) 4αH − H2 = 1 , (H,H,H,h) No −4α No h R ∈ H2 = 1 1 1+ 16παΛ 4α − ± 3 (H,H,H,H) (cid:18) q (cid:19) H2 = 1 No solutions with positive square root −4α exist forΛ<0only B. (5+1)-dimensional spacetime. In this case we have 10 pairs of the type I and type II conditions (9). Note that, as distinct from the (4+1)-dimensional case, type II conditions are the sums of three pairwise products of Hubble parameters but it does not affect follow-up reasoning: as in the (4+1)-dimensional case one can check that there are the following necessary conditions for a given set of the Hubble parameters H1,...,H5 to be a solution of the dynamical equations: i) H1 = H2 = H3 = H4 = H5 H; ii) H1 = H2 = H3 = H4 H, H5 h; ≡ ≡ ≡ iii) H1 = H2 = H3 = H4 H, H5 h; iv) H1 = H2 = H3 H, H4 = H5 h. Similarly to (4+1)- − − ≡ ≡ ≡ ≡ dimensional case, i) could be resolved only if κ = 0 or ω = 1; ii)–iv) have solution with ω = 1 and the − − particular case of H = 0 (once they fulfill the conditions for the solutions of this branch, it is enough for us for now); additiPonally, iv) has vacuum (κ = 0) solution. Solutions for Λ cases of i),ii),iii) are presented in the table IV. The last case iv) is more complicated; it leads to the following equations: 4H2α+1 6 3 4 2 2 192H α 112H α +(256Λπα+4)H α 1 =0, h = (17) − − − 8Hα It is easy to check that when α > 0 Eqs. (17) has at least one solution for any Λ; when α < 0 Eqs. (17) 5 has at least one solution iff Λ > . Additionally, if we require Λ > 0 (which seems more physically −48πα motivated) we can impose an additional constraint on the solution of the Eq. (17). In that case we require 6 Table IV. Λ-term solutions. α>0 α<0 H2 = 1 1 1+ 48παΛ , Λ< 5 12α 5 48πα (H,H,H,H,H) H2 = 121α(cid:18)−1+ q1+ 48π5αΛ(cid:19), Λ>0 H2 = 12(cid:18)1α− −−1q+ 1+ 48π5α(cid:19)Λ , Λ−<0 (H,H,H,H,h) No H2 = (cid:18) 1 , hq R, Λ= (cid:19) 5 −12α ∈ −48πα (H,H, H, H,h) H2 = 1 , h R, Λ= 1 No − − 4α ∈ 16πα (192ξ3 112ξ2 +4ξ 1) < 0 where ξ = αH2. This equation has only one real root ξ0 = (4√3 10+ √3 100+ − − 7)/36 0.56276, so that the constraint is αH2 < ξ0. ≈ The vacuum case admits: • isotropic solution: H2 = 1 , α< 0; −6α • solution with (3+2) spatial splitting: H1 = H2 = H3 =H, H4 = H5 =ξH and f(ξ) ξ2+6ξ +3 2 H = , f(ξ)= (18) 4α (cid:12) 3 3 − 3ξ(3ξ +2) (cid:12)ξ= √10 √100 2 0.722 (cid:12) 3 − 6 −3≈− (cid:12) (cid:12) As well as in the (4+1)-dimensional case solution with one the Hubble parameter different from the others 5 (H1 = H2 = H3 = H4 =H, H5 = h) exists for the single value of Λ: Λ =−48πα, α< 0. IV. HIGHER DIMENSIONS Equation (7) is valid for any number of dimensions and it is necessary (but not sufficient) condition for solution to exist. Previously we found all solutions with non-constant volume in (4+1) and (5+1) dimensions, but formally we can extend an analysis to higher dimensions as well. “Formally” because there are doubts about using Gauss-Bonnet in higher dimensions – from one hand, in (6+1) and higher in the spiteofLovelock theory cubiccorrection comes into play andequations of motion becomemorecomplicated. However, on the other hand it is Gauss-Bonnet but not Lovelock gravity which is motivated by the string theory, so some motivation for considering higher dimensional EGB gravity still remains. There are two cases for the Hubble parameters distribution – all of them could be distinct or there could be multiple values. In the first case neither of H H terms could be zero and only the case with a special i j − sum of pairwise products remains. In addition to these sums, if we sum all possible pairwise products, we get one more condition: D(D 1) HiHj = − . (19) −4α(D 2)(D 3) i>j − − P 7 The case with possible repetitions in Hubble parameters is a bit more tricky: difference equations for multiple values are fulfilled since (H H ) = 0, so what remain are the difference equations for distinct i i − Hubbleparameters,whosenumberislowerthanthenumberofthedimensions. Thusthedifferenceequations forthecasewithmultiplevaluesforHubbleparametersareequivalenttothecasewithsubsetoftheseHubble parameters made of all distinct Hubble parameters. For instance, the case of D dimensions with D distinct Hubble parameters (H1,..., HD) and the case of (D +1) with two Hubble parameters are equal to each other (say, H1, H1,..., HD – so that the number of distinct Hubble parameters is D) are equivalent to each other in the terms of the difference equation. And now let us stress an attention again on the fact that these are necessary conditions – because the sufficient conditions are different for the example considered above. Generally, with increase in the number of dimensions the analysis becomes more and more complicated for each particular case, making finding solutions a difficult task. Yet, some of the solutions found could be generalized. One of them is the isotropic solution; substituting H H into (3)–(4) we obtain two i ≡ independent equations D(D 1)H2 +αD(D 1)(D 2)(D 3)H4 = κ, D(D 1)H2+αD(D 1)(D 2)(D 3)H4 = ωκ; − − − − − − − − − (20) similar to the lower-dimensional cases one can see that (20) could be resolved only in either vacuum (κ = 0) or in Λ (ω = 1) cases. The former of them admits either trivial or − 1 H2 = (21) −α(D 2)(D 3) − − solution; one can easily verify that (21) is also valid in lower-dimensional cases. TheΛ-solution of (20) could be split into two subcases – with positive or negative α. For α > 0 the solution for H and constraint on Λ are (D 2)(D 3) 1+32απΛ − − 1 H2 = s D(D−1) − ; Λ> 0. (22) 2α(D 2)(D 3) − − For α < 0 we have up to two branches (D 2)(D 3) 1+32απΛ − − 1 H2 = ±s D(D−1) − ; Λ< D(D−1) ; (23) 2α(D 2)(D 3) −32πα(D 2)(D 3) − − − − and “+” branch exists only if Λ < 0. One can easily check that expressions in (22) and (23) coincide with those in lower dimensions (tables II, III, and IV). 8 The second solution which could be generalized is the solution with (D 1)+1 spatial splitting – i.e. − with Hubble parameters in the form (H, H, ..., H, h). This solution is 1 D(D 1) 2 H = , Λ = − . (24) −2α(D 2)(D 3) −32απ(D 2)(D 3) − − − − Once again one can make sure that lower-dimensional values from tables II, III, and IV meet general expressions in (24) and the solution follows the same rules: it exist only for a single value of Λ, only for α < 0 and h R. ∈ Let us note that the general higher-dimensional case do not, of course, limits with above-mentioned solutions – but only these two could be easily generalized for higher dimensional case. V. CONCLUSIONS We argue that we obtained all possible non-constant-volume exponential solution in the Einstein-Gauss- Bonnet gravity in (4+1) and (5+1) dimensions. Some of these solutions were generalized on the arbitrary number of dimensions. The starting point is the Eq. (7) which determine the necessary conditions for the solutions to exist. Therewecanclearly separateconstant volumecase(last condition)andconsideritseparately –wearegoing to do so in incoming paper. The first two conditions become (9) and in lower number of dimensions ((4+1) and(5+1)) they could besolved explicitly. Itappearsthatthesector with negative αis moreabundantwith the solutions – say, vacuum isotropic or Λ-solutions with ((D 1)+1) spatial splitting could be obtained − only if α < 0 in any number of dimensions; for other solutions generally the case with α < 0 allows wide variety of branches. This could explain why in previous numerical studies [14, 16–18] so little number of “good” solutions were found – most of them correspond to α < 0 case which is believed to possess serious problems on a quantum level. In the higher number of dimensions (9) become more complicated so we decided to leave complete investigation for further analysis. Still, some of the solutions retrieved for (4+1) and (5+1) could be generalized for any number of dimensions, and that was done as well. Acknowledgments.– This work was supported by RFBR grant No. 14-02-00894. S.A.P. was supported by FONDECYT via grant No. 3130599. [1] F. Müller-Hoissen, Phys. Lett. 163B, 106 (1985). [2] J. Madore, Phys. Lett. 111A, 283 (1985). [3] J. Madore, Class. Quant. Grav. 3, 361 (1986). 9 [4] F. Müller-Hoissen, Class. Quant. Grav. 3, 665 (1986). [5] N. Deruelle, Nucl. Phys. B327, 253 (1989). [6] T. Verwimp, Class. Quant. Grav. 6, 1655 (1989). [7] G. A. Mena Marugán, Phys. Rev. D 42, 2607 (1990). [8] G. A. Mena Marugán, Phys. Rev. D 46, 4340 (1992). [9] N. Deruelle and L. Fariña-Busto, Phys. Rev. D 41, 3696 (1990). [10] J. Demaret, H. Caprasse, A. Moussiaux, P. Tombal, and D. Papadopoulos,Phys. Rev. D 41, 1163 (1990). [11] E. Elizalde, A.N. Makarenko,V.V. Obukhov, K.E. Osetrin, and A.E. Filippov, Phys. Lett. B644, 1 (2007). [12] M. Farhoudi, Gen. Rel. Grav. 41, 117 (2009). [13] A. Toporensky and P. Tretyakov,Gravitation and Cosmology 13, 207 (2007). [14] S.A. Pavluchenko and A.V. Toporensky, Mod. Phys. Lett. A24, 513 (2009). [15] S.A. Pavluchenko,Phys. Rev. D 80, 107501 (2009). [16] I.V. Kirnos, A.N. Makarenko, S.A. Pavluchenko, and A.V. Toporensky, General Relativity and Gravitation 42, 2633 (2010). [17] I.V. Kirnos, S.A. Pavluchenko, and A.V. Toporensky, Gravitation and Cosmology 16, 274 (2010) arXiv: 1002.4488. [18] S.A. Pavluchenko,Phys. Rev. D 82, 104021 (2010). [19] S.A. Pavluchenko and A.V. Toporensky, Grav. Cosmol. submitted; arXiv: 1212.1386. [20] F. Canfora, A. Giacomini, and S.A. Pavluchenko,Phys. Rev. D 88, 064044(2013). [21] V. Ivashchuk, Int. J. Geom. Meth. Mod. Phys. 7 797 (2010) arXiv: 0910.3426. 10

Exact exponential solutions in Einstein-Gauss-Bonnet flat anisotropic cosmology

Similar Exact exponential solutions in Einstein-Gauss-Bonnet flat anisotropic cosmology