WANDs of the Black Ring 5 0 0 V. Pravda1 and A. Pravdov´a2 2 n a Mathematical Institute, Academy of Sciences, Zˇitna´ 25, 115 67 Prague 1, Czech Republic J 1 1 Abstract v 3 Necessary conditions for various algebraic types of the Weyl tensor are determined. These conditions 0 are then used to find Weyl aligned null directions for the black ring solution. It is shown that the black 0 ring solution is algebraically special, of typeIi, while locally on the horizon the typeis II. Oneexceptional 1 subclass – the Myers-Perry solution – is of typeD. 0 5 0 1 Introduction / c q Recently a classification of tensors on Lorentzian manifolds of arbitrary dimensions was introduced [1]. When - r applied to the Weyl tensor [1, 2], in four dimensions it reduces to the well known Petrov classification, in g higher dimensions it leads to a similar, dimensionally independent classification scheme (see Table 1). This : v classification is based on the existence of preferred null directions – Weyl aligned null directions (WANDs) i and corresponding principal null congruences in a given spacetime (see Section 2 for details). In 4 dimensions X WANDs correspond to principal null directions of the Weyl tensor. r a The Petrov classification in four dimensions was very useful for generating algebraically special exact solu- tionsaswellasforphysicalinterpretationofspacetimes. Ultimately,onewouldliketousethehigherdimensional classificationfor similar purposes as in classicalrelativity, however,while in higher dimensions some properties related to the classification remain the same as in four dimensions, other differ. For example, it was proven that for all vacuum spacetimes of types III and N in arbitrary dimension the principalnullcongruenceisgeodesicasinfourdimensions[3]. Thiscongruenceinfourdimensionsisinaddition shear-freebutinhigherdimensionsshearmaynotvanish(see[3]foradditionalcommentsontheGoldberg-Sachs theorem in higher dimensions). Also all spacetimes with vanishing curvature invariants are necessarily of types III and N in arbitrary dimension, including four [4]. Black holes are another interesting example. First let us point out that the Kerr solution as well as the Myers-Perry solution in five dimensions are of type D with geodesic principal null congruences [3, 8]. More general black holes in four dimensions can be algebraically general, however, if the horizon is isolated, the spacetime is of type II locally on the horizon [5]. Recently, it was proven that locally, on the isolated horizon, the algebraic type is II also in higher dimensions [6]. 1E-mail: [email protected] 2E-mail: [email protected] 1 A vacuum, asymptotically flat, stationary black hole solution with a horizonof topologyS1 S2 - so called × black ring - was discovered recently in [10]. There exist black holes of spherical topology and black rings with thesamevaluesoftheconservedquantitiesM andJ andthusblackholesinhigherdimensionsarenotuniquely characterized by their mass and angular momenta and uniqueness theorems cannot be (straightforwardly) generalized for higher dimensions. In this paper it is shown that while the Myers-Perry solution representing a spherical black hole is of type D, black rings, while still algebraically special, belong to the more general class I . In Section 2 the i classificationoftheWeyltensorinhigherdimensionsisoverviewedandnecessaryconditionsforvariousclasses, which significantly simplify search for WANDs, are determined. In Section 3 we briefly overview the (neutral) black ring solution and analyze some of its properties. In Section 4 we classify static and stationary black rings. Infourdimensionsitispossibletodeterminealgorithmicallyanalgebraictypeofasolutionusingvarious invariants and covariants of the metric (see e.g. Chapter 9 in [9]). Since such a method is not developed in higher dimensions, it is necessary to find WANDs explicitly. Components of the Weyl tensor for the black ring are quite complicated and thus corresponding necessary conditions are also complicated even with the use of Maple (however, much simpler than alignment equations). We thus limit ourselves to presenting only their solutions and indicating the procedure how to obtainthem. We also explicitly show that locally on the horizon the Weyl tensor is of type II. 2 Algebraic classification of the Weyl tensor in higher dimensions In this section, algebraic classification of the Weyl tensor in higher dimensions developed in [1, 2] is briefly summarized and at the end necessary conditions for various algebraic types are introduced. We will work in the frame m(0) =n, m(1) =ℓ, m(i), i,j,k =2...D 1, − with two null vectors n, ℓ ℓaℓ =nan =0, ℓan =1, a=0...D 1, a a a − and D 2 spacelike vectors m(i) − m(i)am(j) =δ , m(i)aℓ =0=m(i)an , i,j,k=2...D 1, a ij a a − where the metric has the form g =2ℓ n +δ m(i)m(j). ab (a b) ij a b The group of ortochronous Lorentz transformations is generated by null rotations 1 ℓˆ=ℓ+z m(i) ziz n, nˆ =n, mˆ(i) =m(i) z n (1) i i i − 2 − spins and boosts, respectively, ℓˆ=ℓ, nˆ =n, mˆ(i) =Xi m(j); ℓˆ=λℓ, nˆ =λ−1n, mˆ(i) =m(i). j Aquantityqhasaboostweightbifittransformsunderaboostaccordingtoqˆ=λbq. Boost orderofatensorTis definedas the maximumboostweightofits framecomponents anditcanbe shownthatitdepends only onthe choiceofanulldirectionℓ(see PropositionIII.2 in[1]). ForagiventensorT,b denotesthe maximumvalue max of b(k) taken over all null vectors k . Then a null vector k is aligned with the tensor T whenever b(k) < b max and the integer b b(k) 1 is called order of alignment. The classification of tensors [1] is based on the max − − existenceofsuchalignednullvectorsofvariousorders. Namely,fortheWeyltensorthe primary alignment type is G if there are no null vectors aligned with the Weyl tensor and the primary alignment type is 1, 2, 3, 4 if the maximally aligned null vector has order of alignment 0, 1, 2, 3, respectively. Once ℓ is fixed as an aligned nullvectorofmaximalorderofalignment,onecansearchfornwith maximalorderofalignmentsubjecttothe constraint n ℓ=1 and similarly define secondary alignment types. · 2 Let us introduce the operation which allows us to construct a basis in the space of Weyl-like tensors by { } 1 w x y z (w x y z +w x y z ). (2) {a b c d} ≡ 2 [a b] [c d] [c d] [a b] Now we can decompose the Weyl tensor in its frame components and sortthem by their boost weight(see [1]): 2 1 C = 4C n m(i)n m(j) +8C n ℓ n m(i) +4C n m(i)m(j)m(k) abcd 0i0j {a b c d} 010i {a b c d} 0ijk {a b c d} z }| { z }| { 0 + 4C n ℓ n ℓ + C n ℓ m(i)m(j) +8C n m(i)ℓ m(j) +C m(i)m(j)m(k)m(l) 0101 {a b c d} 01ij {a b c d} 0i1j {a b c d} ijkl {a b c d} z }| { −1 −2 + 8C ℓ n ℓ m(i) +4C ℓ m(i)m(j)m(k) +4C ℓ m(i)ℓ m(j) . 101i {a b c d} 1ijk {a b c d} 1i1j {a b c d} z }| { z }| { Additional constraints follow from symmetries of the Weyl tensor C =0, 0[i|0|j] C =C =0, 0i(jk) 0{ijk} C =C , C =0, C =2C , (3) ijkl {ijkl} i{jkl} 01ij 0[i|1|j] C =C =0, 1i(jk) 1{ijk} C =0 1[i|1|j] and from its tracelessness C =C =0, (4) 0i0i 1i1i C =C , C =C , 010i 0jij 101i 1jij 1 2C =C C , C = C . 0i1j 01ij ikjk 0101 ijij − −2 NowtheWeyltensorisoftypeIifthereexistsuchz thatwecansetallcomponentsofboostweight2,C , i 0i0j to zero using the transformation (1). It is of type II if we can set boost order 2 and boost order 1 components to zero andso on. Resulting polynomialequations for z depend on the originalchoice of the frame ℓ, n, m(i) i and can be considerably simplified if the frame is chosen appropriately. However, there is no general method how to make such a choice. For this reason we investigate another method, which involves only the vector ℓ instead of the frame ℓ, n, m(i). In four dimensions, the following equivalences for principal null directions hold (see e.g. [9]) ℓbℓcℓ C ℓ =0 4D ℓ is PND, at most Petrov type I; [e a]bc[d f] ⇐⇒ ℓbℓcC ℓ =0 4D ℓ is PND, at most Petrov type II; abc[d e] ⇐⇒ ℓcC ℓ =0 4D ℓ is PND, at most Petrov type III; (5) abc[d e] ⇐⇒ ℓcC =0 4D ℓ is PND, at most Petrov type N. abcd ⇐⇒ By substituting a generalformofthe Weyltensorof variousprimarytypes in previousequations(5)we can find that in arbitrary dimension ℓbℓcℓ C ℓ =0 = ℓ is WAND, at most primary type I; [e a]bc[d f] ⇐ ℓbℓcC ℓ =0 = ℓ is WAND, at most primary type II; abc[d e] ⇐ (6) ℓcC ℓ =0 = ℓ is WAND, at most primary type III; abc[d e] ⇐ ℓcC =0 = ℓ is WAND, at most primary type N. abcd ⇐ 3 In fact for the type I equivalence holds in arbitrary dimension [1] but it is not so for more special types. For example, it can be shown that the most general Weyl tensor satisfying ℓcC =0 has the form abcd C m(i)m(j)m(k)m(l) +4C ℓ m(i)m(j)m(k) +4C ℓ m(i)ℓ m(j). (7) ijkl {a b c d} 1ijk {a b c d} 1i1j {a b c d} Its quadratic curvature invariant is C Cabcd =C Cijkl =Σ(C )2, which is non-zero as long as C has abcd ijkl ijkl ijkl a non-vanishing component. Note that due to symmetries of C it can have a non-vanishing component only ijkl for dimensions D 6. Since the Weyl tensor possesses a non-vanishing invariant, it cannot be of type N or III ≥ and thus the equivalence in general does not hold. Let us conclude with a table comparing the classification of the Weyl tensor in four and higher dimensions. D>4 dimensions 4 dimensions Petrov type alignment type Petrov type G G I (1) I (1,1) I i II (2) II (2,1) II i D (2,2) D III (3) III (3,1) III i N (4) N Table 1: Comparisonofthe algebraicclassificationofthe Weyltensorinfour andhigherdimensions. Note that in four dimensions alignment type (1) is necessarily equivalent to the type (1,1), (2) to (2,1) and (3) to (3,1) and that type G does not exist. 3 Rotating black ring - overview The rotating black ring solution was found in [10], here we will use a slightly different form of the metric in coordinates t, x, y, φ, ψ introduced in [11] { } F(x) 2 ds2 = dt+R√λν(1+y)dψ −F(y) (cid:16) (cid:17) R2 F(y) dx2 G(x) + F(x) G(y)dψ2+ dy2 +F(y)2 + dφ2 , (8) (x y)2 − G(y) G(x) F(x) − (cid:20) (cid:18) (cid:19) (cid:18) (cid:19)(cid:21) where F(ξ)=1 λξ, G(ξ)=(1 ξ2)(1 νξ). (9) − − − Note that only certain regions in the (x,y) plane have signature +3. These regions can be determined by analyzing eigenvalues λ ...λ of the metric 1 5 R2F(y)2 λ = , 2 G(x)(x y)2 − R2F(x)F(y) λ = , 3 −G(y)(x y)2 − R2F(y)2G(x) λ = , 4 F(x)(x y)2 − R2F(x)2G(y) λ λ = . 1 5 −F(y)(x y)2 − 4 ∞ A2 1/ν A3 1/λ B y 1 C2 C1 1 − A1 1 1 1/λ 1/ν −∞ − ∞ x Figure1: Variousregionsintheblackringsolution: regionswithsignature+3areshaded;curvaturesingularities are located at x = 1/λ, y = 1/λ, and x = (dashed lines); the metric is flat at x = y (see the text for ±∞ details). We do not give explicit forms of λ and λ but only their product which is much simpler. For λ λ <0, values 1 5 1 5 ofλ ...λ havetobepositiveandweobtain3regions( 1,1) ( , 1),( 1,1) (1,1/λ),( 1,1) (1/ν, ) 2 4 − × −∞ − − × − × ∞ with signature +3. For λ λ >0, λ has to be positive and one of the eigenvalues λ ...λ negative. We arrive 1 5 5 2 4 at following three regions ( 1,1) (1/λ,1/ν), (1/λ,1/ν) ( 1,1) and (1/ν, ) ( 1,1) (see Figure 1). − × × − ∞ × − The curvature invariant R Rabcd has the form abcd (x y)4P (x,y) λν − , (10) R4( 1+λx)4( 1+λy)6 − − where the polynomial P (x,y) is quadratic in y and a 6th degree polynomial in x, which does not vanish at λν x = 1/λ = y or y = 1/λ = x. Consequently, there are curvature singularities located at x = 1/λ, y = 1/λ, 6 6 and x= . The invariant (10) as well as other curvature invariants vanish at x=y. This indicates that the ±∞ spacetime is flat there. Let us now summarize basic properties of various regions in the black ring solution: Region : Here signs of λ ...λ are -++++. This region is asymptotically flat, ∂ is a timelike Killing A1 1 5 ∂t vector and thus spacetime is static here. Moreover, the norm of the Killing vector ∂ approaches 1 at the ∂t − “flat point” (x,y)=(-1,-1). This region represents an outer part of the black ring solution. Region can be 1 A smoothly connected with by identifying y = with y = . 2 In , signs ofλ ...λAare++++-,both ∂ −an∞d ∂ aresp∞acelike,butthere exists a timelike Killing vector A2 1 5 ∂t ∂ψ as their linearcombination. This regionrepresentsanergospherewith a limiting surface of stationaritylocated at y = and a horizon at y =1/ν. ∞ In ,signsofλ ...λ are++-++,spacetimeisnon-stationaryandrepresentstheregionbelowthehorizon. 3 1 5 A Curvature singularity is located at y =1/λ. In region , signs of λ ...λ are -++++. This region is asymptotically flat. In the neighbourhood of the 1 5 curvature sinBgularity located at y = 1/λ, both ∂ and ∂ are timelike and thus this region contains closed ∂t ∂ψ timelike curves. In the vicinity of the “flat point” (x,y)=(1,1), ∂ becomes spacelike and the norm of ∂ ∂ψ ∂t approaches -1. This region represents a spacetime of a spinning naked singularity. In region , signs of λ ...λ are +++-+. Note that ∂ is timelike and thus this region admits closed C1 1 5 ∂φ timelike curves. Curvature singularity is located at x = . In region , signs of λ ...λ are +-+++. 2 1 5 ∞ C Curvature singularity is located at x = 1/λ. Regions and are not asymptotically flat and their physical 1 2 C C interpretation is unclear. 5 4 Black ring - algebraic structure Inthissectionweclassifytheblackringsolutionanditsvariousspecialcases. Ourmethodistosolvethenecessity conditions (6) and then check that these solutions indeed represent WANDs by calculating components of the Weyl tensor in an appropriate frame. 4.1 Myers-Perry metric is of type D By setting λ=1 in (8) we obtain the Myers-Perrymetric [7] with a single rotation parameter. It turns out that the second equation in (6) admits two independent solutions 1 νyx y+νx+1 2νy νx 1 y2 1 L = − − R∂ √ν∂ − ∂ + − ∂ .(11) ± (x2 1)( 1+νy) x y t− ψ ± s(x y)(y 1) x x2 1 y − − (cid:18) − (cid:19) − − (cid:18) − (cid:19) When we choose a frame with ℓ L and n L all components of the Weyl tensor with boost weights 2,1,- + − ∼ ∼ 1,-2 vanish and the spacetime is thus of type D. These two vectors were given in Boyer-Lindquist coordinates in [8] and further discussed in App. D in [3]. 4.2 Black ring is of type II on the horizon The transformation F(y) dχ = dψ+ − dy, (12) G(y) p F(y) dv = dt R√λν(1+y) − dy (13) − G(y) p (see [10] for a similar transformation) leads to a metric regular on the horizon y =1/ν F(x) ds2 = (dv+√λνR(1+y)dχ)2 −F(y) R2 dx2 G(x) + F(x) G(y)dχ2 2 F(y)dxdy +F(y)2 + dφ2 . (14) (x y)2 − − − G(x) F(x) − (cid:20) (cid:16) p (cid:17) (cid:18) (cid:19)(cid:21) The second equation in (6) admits a solution ν 1 L=∂ ∂ . v x − λR(1+ν) r One can check that boost order of the Weyl tensor in the frame with ℓ = L is 0 and thus the black ring is of the type II on the horizon. 4.3 Black ring is of type I i In order to solve the first equation in (6) I =ℓbℓcℓ[eCa] [dℓf] =0, (15) eadf bc we denote ℓa =(α,β,γ,δ,ǫ) (16) and from (15) we obtain a set of fourth order polynomial equations in α...ǫ. An additional second order equationfollows fromℓ ℓa =0. Since components of the Weyl tensor are quite complicated, it is not surprising a thattheseequationsarealsocomplicated. However,onecanpicksomeofthemwhicharerelativelysimple. Let us start with the static case. 6 4.3.1 The static case The static case can be obtained by setting ν =0. Two particularly simple equations 3αβδǫλ λ2 1 (x y)3 I = − − =0, (17) txφψ 2R2( 1+λx)2( 1+λy)3 (cid:0) (cid:1) − − 3βγδǫλ λ2 1 (x y)3 I = − − =0 (18) xyφψ −2R2( 1+λx)2( 1+λy)3 (cid:0) (cid:1) − − imply that unless λ equals to -1,0 or 1, at least one component of the WAND (16) vanishes. More detailed analysis of equations (15) shows that the only non-trivial solution is ǫ=0 and R2γ ( 1+λy)2[γ(λx 1)+β(1 λy)] α2 = − − − , (19) (x y)2( 1+λx)(y2 1) − − − β (1 λx) γ(x2 1)+β(1 y2) δ2 = − − − , (20) (y2 1)(x2 1)2 (cid:2) (cid:3) − − with β,γ satisfying the quadratic equation λ x2 1 ( 1+λx)2γ2+λ y2 1 ( 1+λy)( 1+λx)β2 (21) − − − − − λ(cid:0)βγ λ(cid:1)(x+y)[λx( 1+xy)(cid:0)+(1 (cid:1)x2)]+(x y)2(2 λ2)+2( 1+λx)+2xy(1 λy) =0. − − − − − − − (cid:8) (cid:9) Note that a vectoras well as its arbitrarymultiple representthe same WAND. We canthus set one component of (16) equal to a given number. Let us set γ = γ . Now we can solve (21) for β. It has in general two real 0 roots, however,only one of them leads to positive values of α2 and δ2 as given by (19) and (20). Let us denote this rootbyβ andcorrespondingsquarerootsofrighthandsidesof(19)and(20)asα andδ . Note alsothat 0 0 0 β = β ,γ = γ satisfy(21)aswellasβ =β ,γ =γ andthatthischangedoesnotaffectvaluesofα andδ . 0 0 0 0 0 0 − − We thus arrive at four distinct WANDs (α ,β ,γ , δ ,0), (α , β , γ , δ ,0). The static black ring is thus 0 0 0 0 0 0 0 0 ofprincipaltype I.Furthermore,ifwe chooseafram±e with ℓ=(−α ,β−,γ ,±δ ,0)andn=(α ,β ,γ , δ ,0),we 0 0 0 0 0 0 0 0 − can see that all components with boost weight 2 and -2 vanish and thus the type is (1,1)=I . i 4.3.2 The stationary case Linear combination 1 (x 1)(x+1)( 1+νx)I (y 1)(1+λ)( 1+νy)I =0 (22) tψyφ tψxφ − − − λ − − leads to αδ αγ√λν(x y)2( 1+λx)+ǫλR y2 1 [β(1 λy)+γ(λx 1)] − − − − − (cid:8) + ǫνλR (1+y) βy(y 1)( 1+λy)(cid:0)+γ (cid:1)2xy+x2+y ( 1+λx) =0. (23) − − − − (cid:2) (cid:0) (cid:1) (cid:3)(cid:9) Assuming α=0andδ =0, (23)is alinearequationfor αandbysubstituting the resultinI =0 weobtain xyφψ 6 6 the quadratic equation for β and γ δǫ β2λ y2 1 ( 1+λy)( 1+νy)( 1+λx) γ2λ x2 1 ( 1+λx)2( 1+νx) − − − − − − − − − n +βγ 2(cid:0)(ν λ)(cid:1)(x y)2+λ2(1 λy)[x2(x+y)+x y(cid:0)]+2λ(1(cid:1) λx)2+2λxy( 1+λy) (24) − − − − − − h +2λ2νxy(1 λx)+λν(1 λy)[3x2y(1 λx)+x3( 1+λy) 3x+y]+2λν(x y) =0. − − − − − − io 7 20 –2 18 16 –4 14 12 y y –6 10 8 –8 6 4 –10 2 –1 –0.5 0 0.5 1 –1 –0.5 0 0.5 1 x x Figure 2: A numerical example for λ = 3/5, ν = 1/3: solutions of equations (23)–(26) are real everywhere in the region and only to the left from the indicated curve in regions , . 3 1 2 A A A Now the equationℓaℓ =0 is linear in δ2 and using (23) we can express δ2 in terms of β,γ,ǫ. Substituting this a expression in I =0 gives xφyψ γ ( λx+1) β (1 λy) γ3ν ( 1+λx)( 1+λy)2(x y)2 ǫ2βλ y2 1 3( 1+νy)3( 1+λy) − − − − − − − − − − (cid:2) +ǫ2γ( 1+λx) y2(cid:3)n 1 2( 1+νy)2 νx(1 λy)(x 2y)+λ(1(cid:0) νy)+(cid:1) y2(ν λ) =0, (25) − − − − − − − (cid:0) (cid:1) (cid:2) (cid:3)o which is linear in ǫ2. Solving this equation for ǫ2 and substituting to ℓ ℓa =0 leads to a δ2 = β3λ ( 1+λx) y2 1 ( 1+λy)( 1+νy)+βγ2λ ( 1+λx)2 x2 1 ( 1+νx) − − − − − − − γβ2n( 1+λx) λν((cid:0)1 xy)(cid:1)[y(1 λx)+x(1 λy)] ν(x y)2+λ(1(cid:0) λy)(x(cid:1)2 1)+λ(1 λx)(y2 1) − − − − − − − − − − − / ( 1+νx)2 x2h 1 2 γ( 1+λx) ν(x y)2( 1+λy)+λ(y2 1)(1 νy) io − − − − − − − n (cid:0) (cid:1)+hβλ y2 1 ((cid:2)1+λy)( 1+νy) . (cid:3) (26) − − − (cid:0) (cid:1) io Nowwecanexplicitly expressalignednulldirections. Sinceifℓisaligned,thenalsoanymultiple ofℓisaligned, we can set β =1 without loss of generality. Then we solve the quadratic equation (24) for γ and express δ and ǫ from (26) and (25) and finally α from (23). Since the first condition in (6) is equivalence we have thus obtained expressions for WANDs. In the static limit ν 0 the solution corresponds to (19)–(21). Note, however, that for some values of λ,ν,x,y (see Figure → 2) the value of ǫ2 as determined by (25) is negative and thus (23)–(26) do not lead to real WANDs at these points. Thus either spacetime is of type G there or there exists another solution of (15) at these points. This indicates that the corresponding region in the vicinity of the axis x=0 is physically distinct from the rest. References [1] R. Milson, A. Coley, V. Pravda, A. Pravdov´a, Alignment and algebraically special tensors in Lorentzian geometry, gr-qc/0401010. [2] A.Coley,R.Milson,V.Pravda,A.Pravdov´a,Classification oftheWeyltensorinhigher dimensions,Class. Quantum Grav. 21, L35 (2004). [3] V. Pravda,A. Pravdov´a,A. A. Coley,R. Milson, Bianchi identities in higher dimensions, Class.Quantum Grav. 21, 2873 (2004). 8 [4] A. Coley,R. Milson, V. Pravda,A. Pravdov´a,Vanishing scalar invariant spacetimes in higher dimensions, Class. Quantum Grav. 21, 5519 (2004). [5] A. Ashtekar, S. Fairhurst, B. Krishnan, Isolated horizons: Hamiltonian evolution and the first law, Phys. Rev. D, 62, 104025 (2000); A. Ashtekar et all, Generic Isolated Horizons and Their Applications, Phys. Rev. Lett. 85, 3564 (2000). [6] J. Lewandowski, T. Pawlowski, Quasi-local rotating black holes in higher dimension: geometry, gr-qc/0410146(2004). [7] R. C. Myers, M. J. Perry, Black holes in higher dimensional space-times, Ann. Phys. 172, 304 (1986). [8] V. Frolov, D. Stojkovi´c, Particle and light motion in space-time of a five-dimensional rotating black hole, Phys. Rev. D 68, 064011 (2003); V. Frolov and I. Novikov, Black hole Physics: Basic concepts and new developments, Kluwer, Dodrecht (1998). [9] H. Stephani, D. Kramer, M. MacCallum, C. Hoenselaers, E. Herlt, Exact Solutions to Einstein’s Field Equations, Second Edition, Cambridge University Press, Cambridge (2003). [10] R. Emparan, H. S. Reall, A rotating black ring solution in five dimensions, Phys. Rev. Lett. 88, 101101 (2002). [11] H. Elvang, R. Emparan, Black rings, supertubes, and a stringy resolution of black hole non-uniqueness, JHEP 0311, 035 (2003). 9