arXiv:1709.07261v1 [gr-qc] 21 Sep 2017 usqetgnrlztos(e ..tenc eiw[ review nice the e.g. (see generalizations subsequent fdffrn hsclcnet,fo h td fgravita- of range study [ vast the lensing a from tional in contexts, also physical mentioned, but different theorems already of singularity as of only, terms not in application found have [ holds con- theorem were the singularity Hawking define their which and to under Penrose paper ditions Raychaudhuri’s that by surprise inspired should a late it of and as singularities, relation come of was the not existence the It out with work pointed his background. that given himself the a Raychaudhuri in in resulted expansion out cosmological carried equation he Raychaudhuri of that renown analysis kinematic The flows isotropy. im- the and its homogeneity and on principle plications cosmological the assuming without hsclojcs[ objects physical ere ftemte itiuin nhssmnlpaper seminal his in sym- distribution; the [ matter of the artifact an of than that metries more idea nothing the is singularity from a started Motivated Raychaudhuri nature (GR). of cosmology, the Relativity by features General around generic of debate the theory Einstein’s and the met- singularities in in gravitational interested bands of energy got the electronic he in of als, properties While the work- on forcedly itself! was ing Raychaudhuri Kumar structure Amal break- space-time 50’s a early the is, that of Ein- singularity, long down gravitational of a the a harbour and solutions take ric, that met- metric Schwarzschild realize not space-time (FLRW) the Friedmann–Lemaître–Robertson–Walker to did simplest equations, of relativists of field it two stein of point Hercules the community this exception: the even of from an for Relativity, not pillars time General is the view, marks model. that ineluctably ory ∗ 1 † 1 ete rpsdamdlo iedpnetuniverse time-dependent of model a proposed then he ] oeie eerdt steLna–acaduiequation Landau–Raychaudhuri the as to referred Sometimes [email protected] [email protected] ic t neto,teRyhuhr qainadits and equation Raychaudhuri the inception, its Since h perneo iglrte napyia the- physical a in singularities of appearance The ftm-ieadnl uvsi pc-iewt h otgen most the with space-time a th in literature, curves the null in and time first time-like contribut the e of field for orthogonal torsion derive, and generic We tangent completely particles. to a general of in presence leads a the field in torsion curves a close of infinitesimally between vector separation 1 etod aeáia nvriaed ih,Cmu eGual de Campus Minho, do Universidade Matemática, de Centro ie pc-iewt o-aihn oso,w ics t discuss we torsion, non-vanishing with space-time a Given 5 .INTRODUCTION I. ,t rc omto nshrclastro- spherical in formation crack to ], 6 ,adfial osm oefundamen- more some to finally and ], nvriaed iba-U,A.RvsoPi ,14-0 Lis 1049-001 1, Pais Rovisco Av. UL, - Lisboa de Universidade acaduieuto nsaetmswt torsion with space-times in equation Raychaudhuri eatmnod íia nttt ueirTcio-IST, - Técnico Superior Instituto Física, de Departamento 1 o h vlto fthe of evolution the for 2 etoMliicpia eAtoíia-CENTRA, - Astrofísica de Multidisciplinar Centro al Luz Paulo ,2, 1, ∗ 2 n icnoVitagliano Vincenzo and , 4 3 . ]) ]. nrni pno atradtetrinfil [ field torsion the and the between matter coupling of direct spin a intrinsic through environ- gravitational achieved account a catalyzed to in ment, provide effects initially quantum they nontrivial been scheme for simple has relatively models atten- the by the these literature; toward the tion in considered extensively been any along null. field imposed, inde- is tensor still curve the metric space-time with is the of (defined metric derivative connection) covariant the pendent the with is, compatibil- connection that the the that of however, dubbed note, ity field field; tensor tensor a torsion defining be non-vanishing, to connection general further the in and of part equations) anti-symmetric field the independent requiring of to sets approach indepen- two Palatini find an so-called the assuming (using by of connection characterized dent The One is the is theory proposals ECSK while. theory. of a (ECSK) kind this for Einstein–Cartan–Sciama–Kibble of around example now prototypical the been gen- has non-Riemannian of to geometries non-trivial, idea Relativity The a General Einstein field. with eralizing tensor space-time torsion a generic in completely congruence embedded a curves to of construction Raychaudhuri the erature, iseneutosa qain fsaefra(o)equi- (non) [ a thermodynamics for (modified) state spacetime the of librium of equations derivation the as of equations case Einstein the in as issues tal li hois[ gravity Kaluza- of theories dimensional theory Klein higher en- any in and theories, low twistors string considers the that super in that of inevitably fact limit and the ergy naturally to extended appears been torsion has interest this however oso a nfc ese steeeto atrfield matter a Yet of effect spin-torsion. called the usually as where manifold, ECSK seen Lorentzian property, a be this in fact manifold. to in free exception can particular torsion torsion mat- a Lorentzian extra is an a theory as an in general, in in field the seen, recast ter words, be be cannot other cannot field in torsion field tensor, torsion energy-momentum non-null effective ef- a the of imposed, are fects unless assumptions general, comparison further in in particular gravity: nature some of theories unique modified a other with exhibit torsion with ries hoiso rvt ihnnvnsigtrinhave torsion non-vanishing with gravity of Theories lit- the in time first the for generalize, we paper this In acaduieuto o congruence a for equation Raychaudhuri e st eaieaclrto ewe test between acceleration relative a to es ogune eso httepresence the that show We congruence. et otecnrec;i particular, in congruence; the to ffects rctrinfield. torsion eric eeuto o h vlto fthe of evolution the for equation he a,41-5 rg,Portugal Braga, 4710-057 tar, 19 2, .O o fta,gaiainltheo- gravitational that, of top On ]. † o,Portugal boa, 7 – 9 ]. 10 – 18 ]; 2

ECSK theory by no means represents the most generic Having in mind the general affine connection, the com- case of a theory equipped with a torsion field (see for ex- mutator between two vectors can be expressed in terms ample [19, 20]) and, although this circumstance is widely of the torsion-full covariant derivative as recognized, it is intriguing to note that most of the work γ α γ α γ γ α β [u, v] = u αv v αu 2Sαβ u v . (6) done on gravitational theories with non-vanishing torsion ∇ − ∇ − only considers the ECSK setup: for instance, the very im- This last equation and the definition of the Riemann ten- γ portant question of whether the presence of torsion can sor associated with Cαβ, avoid the formation of singularities resulting from gravi- ρ ρ ρ ρ σ ρ σ tational collapse has only been answered for the case of a Rαβγ = ∂βCαγ ∂αC + C Cαγ CασCβγ . (7) − βγ βσ − spin-torsion field [11–13]. In this paper we then start fill- lead to a modified version of the relation between the ing such a crucial gap in the literature: we will focus on curvature and the commutator of two covariant derivative the study of the effects of the most generic torsion field in the case of non vanishing torsion, on the kinematics of test particles and derive the Ray- ρ ρ chaudhuri equation for a congruence of null and time-like Rαβγ wρ = [ α, β] wγ +2Sαβ ρwγ . (8) curves in the spacetime. ∇ ∇ ∇ The Riemann tensor for a non-symmetric connection The paper is organized as follows: in Sec. II we intro- does not have all the usual symmetries. However, from duce the basic definitions and set the conventions that (7) we see that will be used throughout the article; in Sec. III we derive ρ ρ the evolution equation of the separation vector between Rαβγ = Rβαγ , (9) test particles in space-times with torsion, define the kine- − matical quantities of a congruence of curves and derive and, using the symmetries of the contorsion tensor, (4) the Raychaudhuri equation, both for a time-like and null and (5), congruence of curves; finally, we summarize the main re- Rαβγρ = Rαβργ . (10) sults and sort out our conclusions in Sec. IV. − So, the skew symmetry of the Riemann tensor is still ver- ified. To define the Ricci tensor in terms of the Riemann II. CONVENTIONS AND NOTATIONS tensor we will adopt the convention [21, 22] γ Rαβ = Rαγβ , (11) Due to the wide variety of different conventions used in literature, let us start by gently introducing the basic which, using (7), can be expressed in terms of the con- definitions and setting the conventions that will be used nection coefficients as throughout the article. Introduce a covariant derivative, γ γ ρ γ ρ γ Rαβ = ∂γC ∂αC + C C C C . (12) αβ − γβ αβ γρ − γβ αρ β β β σ αU = ∂αU + C U , (1) The Ricci scalar is defined as ∇ ασ R gαβR . constrained to be metric compatible, αgβγ = 0, but = αβ (13) ∇ γ otherwise with a completely generic connection Cαβ. The Note that in the case of a general affine metric-compatible anti-symmetric part of the connection define a tensor connection it is still possible to consider an independent which is called the torsion tensor contraction of the Riemann tensor defining a 2-rank ten- γδ ǫ 1 sor, Rαβ = g gǫβRαγδ ; however a further contraction S γ Cγ = Cγ Cγ . (2) αβ [αβ] αβ βα R R ≡ 2  −  withe the metric results in = , that is the Ricci scalar is unequivocally defined. − Using such definition, it is possible to split the connec- A further important remarke is about the notation we tion into an appropriate combination of the torsion tensor γ will be using to describe the matter sector. For a gen- plus the usual metric Christoffel symbols Γαβ, eral Riemann–Cartan theory of gravity, the matter La- γ γ γ γ γ grangian can couple (non-)minimally to the torsion ten- C =Γ + Sαβ + S αβ Sβ α . (3) αβ αβ − sor which introduce new degrees of freedom in the prob- lem. While the stress-energy tensor is still defined as The sum of the three torsion pieces on the right hand usual as the variation of the matter action with respect side of last equation is frequently dubbed in literature as to the metric, we need to introduce a new object, the in- the contorsion tensor, K γ S γ + Sγ S γ ; αβ αβ αβ β α trinsic hypermomentum, defined as the variation of the using the anti-symmetry of the≡ torsion tensor,− it is an action with respect to the independent connection, easy task to verify straightforwardly the two symmetries of contorsion, ρ 1 δSMatter ∆µν ρ . (14) ≡ √ g δΓµν Kαβγ = Kαγβ , (4) − − Such quantity encapsulates all the information of the mi- croscopic structure of the particle, i.e. intrinsic spin, dila- γ γ K[αβ] = Sαβ . (5) ton charge and intrinsic shear. 3

III. RAYCHAUDHURI EQUATION For infinitesimally close curves, the evolution of the sepa- ration vector along the fiducial curve is entirely described A. The separation vector and its evolution by the tensor field Bαβ. Let us emphasize that in the derivation of (19) and (20) we have not specified the type α Introduced the basic definitions and identities we are of the tangent vector to the fiducial curve, U , hence, this U α now in the position to generalize the Raychaudhuri equa- equations are equally valid for the case of being time- tion for the case of an N-dimensional space-time with like, space-like or light-like, with the fiducial curve being non-null torsion. either a geodesic or not. The notion of separation (sometimes deviation) vector Let us now verify some physical implications of (19) between two infinitesimally close curves is quite intuitive: and (20) in the case of the presence a non-vanishing tor- c n U α define a congruence of curves, not necessarily geodesics, sion tensor. The derivative along of the quantity α such that each curve of the congruence is parameterized reads by an affine parameter λ. Consider a second congruence, D (n U α) α = n aβ +2S U σU αnγ , (21) this time of geodesics, parametrised by an affine parame- dλ β σγα ter t, such that each geodesic intersects a curve of the first congruence at one and only one point of the space-time. where we have defined the acceleration vector appearing for non geodesics fiducial curves as Given two curves in the first congruence, c1 and c2, and a geodesic of the second congruence, γ, let the two points α γ α a U γ U . (22) p and q be the intersection points of γ with, respectively, ≡ ∇ c1 and c2, with c1 (λ0)= γ (t0)= p. Let us now assume The expression (21) represents the failure of the separa- that the point q is in a small enough neighbourhood of tion vector nα and the tangent vector U α to stay orthog- ∂γ onal to each other, that is, if at a given point nα and the point p such that q = γ (t0 + δt) p + ∂t δt. If n ≈ t0 U α are orthogonal to each other, a general non-null tor- is the tangent vector to the geodesic γ in p, then α sion, Sαβγ, or a non-null acceleration, a , will spoil the ∂γ preservation of such orthogonality along the curve. Note n δt = q p (15) that a torsion field, with no further imposed symmetry, ≡ ∂t − t0 will lead to effects parallel to the direction of U α (second

term on the right hand side of (21)), contributing to a gives also a meaningful notion of the separation between relative acceleration between two initially infinitesimally the curves c and c . 1 2 close particles. Let us consider a coordinate neighbourhood that con- The analysis of (21) leads to the conclusion that the tains the points p and q, such that p = xα , q = x′α = tensor B describing the behaviour of the separation xα + nα , with { } { } αβ { } vector will have, for the case of a generic torsion tensor, ∂xα a non-null component tangential to the fiducial curve c. nα = δt , (16) ∂t Without loss of generality, it is then possible to write Bαβ in terms of two components, one orthogonal and α and let U be the tangent vector to the curve c1 (from the other parallel to c: given a projector hαβ onto the here on we will drop the index 1) at p, hypersurface orthogonal to the curve c at a given point, we can write ∂xα U α = . (17) ∂λ Bαβ = B⊥αβ + Bkαβ , (23) In order to find the general expression for the evolution with α of the separation vector, n , we will start by computing γ σ B αβ h h Bγσ , (24) the Lie derivative of n over the tangent vector U and vice ⊥ ≡ α β B Bαβ B αβ . (25) versa. From the definition of the Lie derivative and using kαβ ≡ − ⊥ (16) and (17) we find that In analogy to the General Relativity case, we want to define the kinematical quantities identifying expansion, n U = U n =0 . (18) L L shear, and vorticity - θ, σαβ and ωαβ, respectively - of neighbouring curves of the congruence. These quantities Using (6) and (18) it is possible to derive an equation will only depend on the orthogonal part of the tensor for the change of the separation vector along the fiducial B , B , so that, defining the expansion, shear and curve c, αβ ⊥αβ vorticity as β α α β γ U βn = Bβ n , (19) θ = B , (26) ∇ ⊥γ hαβ where σαβ = B⊥(αβ) γ θ , (27) − h γ α α α γ Bβ = βU +2Sγβ U . (20) ωαβ = B , (28) ∇ ⊥[αβ] 4

2 B⊥αβ can be decomposed into [24] (see also [25, 26] for early single-pole approximation descriptions). A quite interesting result is that single- hαβ pole particles without intrinsic hypermomentum follow, B⊥αβ = γ θ + σαβ + ωαβ . (29) h γ as in general relativity, geodesics of the metric connec- β (g) α tion, v β v = 0, no matter what is the underlying Before we continue and derive the Raychaudhuri equa- theory of∇ gravity. tion let us emphasize that the definitions of the kinemat- Things become much more complicated for particles ical quantities given by (26)-(28) are always valid when- with non-vanishing intrinsic hypermomentum. We will ever the tensor Bαβ is related with the variation of the focus on the case of Riemann-Cartan space-time, since separation vector between curves of a congruence by an in our setup nonmetricity is trivially zero. We will also equation such as (19). A formal proof for this assertion consider particles endowed with only mass and intrinsic can be seen in [23]. spin but we neglect dilaton charge and intrinsic shear for simplifying reason. This means that the hypermo- mentum tensor reduces (for a single-pole particle [24]) ρ ρ B. Raychaudhuri equation for a congruence of to ∆[µν] = τµν v , where τµν is the anti-symmetric spin time-like curves density tensor. In this case the equations of motion for a particle read The results in the previous section are quite general κ κ 2 κ β σ and valid for curves of any kind; however, the procedure v κvα = vαv κ ln m v κ v v στβα + h ∇ − ∇ − m ∇ ∇ that defines the projector αβ strictly depends on the µ β ν σ
+ Sαβ v (mvµ +2v v στνµ)+ specific family of curves considered. Once the projector ∇ is assigned, (29) will give an actual expression in terms 1 ν β µ + Rαβµ v τ ν aα , of the tangent vector and the torsion tensor. 2  ≡ Let us then start by considering the generalized Ray- (32) chaudhuri equation for a congruence of time-like curves. In this case we will impose that the fiducial curve is α σ κ v ατµν 2v[µv v κτσ|ν] =0 , (33) parametrised by the proper time, τ, and, in order to avoid ∇ − ∇ confusion with the general case, we will label its tangent α α Notably, this equation is independent by the specific vector as v , with vαv = 1. The operator projecting choice of the gravitational part in the action. It is impor- onto the hypersurface orthogonal− to vα is given by tant to stress that for a non-spinning particle, τµν van- ishes, the second of equations (33) is trivially satisfied hαβ gαβ + vαvβ , (30) ≡ while the first one recovers the equation of the geodesics β (g) α of the metric, v β v = 0. Note that the rest mass, and fulfills the following conditions that is the projection∇ of the particle 4-momentum on the rest frame, is guaranteed to be constant along the con- h vα , αβ =0 gruence only in the case of non-spinning particle. γ hαhγβ = hαβ , (31) Using equations (30)-(33) in equations (24) and (25) hσ =N 1 , we find σ − ρ ρ σ where N is, again, the dimension of the space-time. B⊥αβ = αvβ +2Sραβv +2Sρασv v vβ + vαaβ ,(34) In order to calculate the two components, orthogonal ∇ ρ σ Bkαβ = 2Sρασv v vβ vαaβ . (35) and parallel, of Bαβ another ingredient is necessary: we − − must find an expression for the trajectories along which For the case of a congruence of time-like curves in a N- free particles move. In a general non-Riemannian man- dimensional space-time (29) can be written as ifold, it is not an easy task to determine such physical curves. The statement usually claimed in the litera- 1 ture that particles follow geodesics, either of the Levi- B⊥αβ = hαβθ + σαβ + ωαβ , (36) N 1 Civita generic connection or of the metric connection, − turns out to be rather unsatisfactory and naïve. Ina where (31) was used. manifold equipped with a completely generic affine con- Taking the variation of the tensor B along the fidu- nection, the particle trajectories can be correctly deter- ⊥αβ cial curve, and remembering the expression (8) for the mined starting from the equations of motion, with the latter obtained themselves from the conservation laws for canonical energy-momentum and hypermomentum. A comprehensive treatment for the calculation of the prop- agation equations of single-pole and pole-dipole particles 2 We warn about a slight different notation in [24]: for example in metric-affine theories of gravity has been developed in the definition of the torsion tensor is there twice our definition. 5

Riemann tensor, we find spin, Refs. [14, 28] consider the simplifying assumption γ γ on the torsion tensor Sαβ = Sαβv , with S(αβ) = 0 and D B⊥αβ γ ρ γ S vα =v γ B αβ = Rγαβρv v + αaβ αβ = 0; in Ref. [12] instead torsion is constrained ⊥ γ γ σ ǫ dτ ∇ ∇ − to be Sαβ = ηαβσǫ v v S , where ηαβσǫ is the com- γ γ ρ ( αv ) ( γ vβ) 2v Sγα ρvβ+ (37) pletely anti-symmetric Levi-Civita tensor. However, in − ∇γ ∇ ρ − γ ∇ +2v γ (Sραβv )+ v γ (vαaβ)+ these special cases the extra symmetries imposed on the γ∇ ρ σ ∇ torsion tensor imply that the second term on the right +2v γ (Sρασv v vβ) , ∇ hand side of (20) is null, reducing the problem to the Shear and vorticity are, respectively, traceless and anti- torsion free case. symmetric, so that, contracting α and β on both sides of previous equations, we obtain C. Raychaudhuri equation for a congruence of null curves Dθ ρ γ α γ α = Rγρv v + αa ( αv ) ( γ v ) dτ − ∇ − ∇ ∇ γ ρ α γ ρ 2v Sγα ρv +2v γ (Sρv ) , (38) Let us now derive the Raychaudhuri equation for a − ∇ ∇ congruence of null curves. As in the previous subsection, α where Sρ Sρα . Taking into account that to avoid any confusion, we will re-label the tangent vector ≡ α α to the fiducial curve and call it k , so that k kα = 0. βα β α ρ α β B αβB = αv ( βv )+2v Sρβ αv + ⊥ ⊥ ∇ ∇ ∇ In this case, unfortunately, if hαβ is the projector onto
β ρ α β α ρ γ +2Sρα v βv +4Sρα Sγβ v v + the hypersurface that is orthogonal to the fiducial null ∇ curve, it cannot be naively definede by (30), since it would +4S vαvγ aβ , αβγ not be orthogonal to kα (it would be h kα = k = 0). (39) αβ β The way out of this problem is through the introduction6 and using (36) we can write (38) as of an auxiliary null vector field, ξα, suche that [29] α Dθ ρ γ 1 2 αβ βα k ξα = 1 , (41) = Rγρv v θ + σαβσ + ωαβω + − dτ − − N 1  ξ ξα . − α =0 (42) β ρ 1 α α α +2Sρα v hβ θ + σβ + ωβ Using (41) and (42) we can now properly introduce a N 1  projector onto the hypersurface orthogonal to both kα α γ − ρ β ρ α α + αa +2v γ (Sρv )+2Sρα v vβa . and ξ as ∇ ∇ (40) hαβ = gαβ + kαξβ + ξαkβ , (43) This equation represents the generalization of the Ray- chaudhuri equation for a time-like congruence of curves satisfying the followinge properties in the presence a generic torsion field. Here, we would h kα =h ξα =0 , like to make a couple of comments: first of all, this equa- αβ αβ σ tion has been obtained using only geometrical arguments, heαhσβ =ehαβ , (44) plus the canonical energy-momentum conservation equa- e ehσ =eN 2 . tion to define the equations of motion of the particles; σ − this means that the result is independent by the spe- As in the time-likee case, the extra ingredient to be cific geometrical theory that we are choosing: once that taken into account is the effective particles trajectories. the theory has been assigned, then the Ricci tensor can An important caveat is here due: particles that are mov- be related with the energy-momentum tensor accordingly ing along null curves are massless particle; the most nat- with the (modified) Einstein field equations. Secondly, it ural candidates in the Standard Model are then photons. in interesting to stress that the extra-force responsible However, the minimal coupling procedure to general- of the acceleration term reported in the first line of (40) ize the electromagnetic field to non-Riemannian environ- is of purely geometric origin, related to the extra cou- ments preserve photons by having a non-vanishing intrin- pling of the intrinsic spin (viz. intrinsic hypermomentum sic hypermomentum; hence, photons follow the geodesics in the most general setup) with post-Riemannian struc- determined by the Christoffel symbols, tures. Note also that our result differs from previous kα (g)kβ =0 . (45) versions of the torsion-full Raychaudhuri equation avail- ∇α able in literature. More concretely, refs. [4, 19, 27] did As a side note, let us stress that some theories, such as the not take properly into account the relation between the Standard Model Extension [30], allow for a non-minimal tensor Bαβ and the evolution of the separation vector be- coupling of the electromagnetic field with geometry (and tween infinitesimally close curves of the congruence, as- eventually other fields), which means a non trivial intrin- suming that the expression of the tensor Bαβ is a priori sic hypermomentum: anyway, in this case the extra oper- the same as in the case of null torsion. Few quite specific ators will also introduce some effective mass for the pho- models for torsion accidentally result anyway in the cor- ton (the simplest case being the explicitly massive Proca rect expression: retracing the properties of the intrinsic field) and a fortiori they could not follow null curves. 6

In the context of the Standard Model, neutrinos de- Tracing Eq. (46) and computing its covariant deriva- serve a separate mention; the recent confirmation of the tive along the fiducial null curve we find phenomenon of neutrino oscillations [31, 32] directly im- ply that neutrinos might be massive [33], avoiding them Dθ α β 1 2 αβ βα = Rαβk k θ + σαβ σ + ωαβω to follow null paths3. In spite of that, let us take into ac- dλ − − N 2  − count the behaviour of the Standard Model massless neu- ˜ α β ρ hβ α α γ ρ trino, viz. a single-pole, massless Dirac particle. Dirac +2Sρα k θ + σβ + ωβ +2k γ (Sρk ) N 2  ∇ particles have a completely anti-symmetric hypermomen- − α γ ρ βγ α α 2 tum, which implies, in the single-pole approximation, a +2 α (S γρk k )+4 h Sαβγ k Sαk ∇ − vanishing intrinsic spin density tensor, τµν = 0 [34, 35].
S αβkµ S kδ hρ S kδ Back to (33), this would mean that the particles follow +4 µ δβα β δρα h − i metric geodesics, as in the non-spinning case, and as for α µ νβ νβ + S µν k h g 4B +6B photons. − kαβ kβα  
It is finally possible to use (41)-(45) to find the or- α e δβ δβ 2Bkδ Bkβα h g , thogonal and tangential components of the tensor Bαβ −  −  that defines the dynamics of the separation vector of the e (50) null congruence of geodesics; a straightforward calcula- which represents the Raychaudhuri equation for a con- tion gives gruence of null geodesics in the presence of a torsion ten- sor field. Here the literature is more sparse: ref. [36] de- rives the Raychaudhuri equation for a null congruence γ γ σ γ σ of curves in the particular case of a completely anti- B⊥αβ = αkβ 2Sαγβk 2Sαγσk k ξβ +2Sγσβkαk ξ ∇ − γ σ ρ− γ γ symmetric torsion, missing in any case the correct defi- +2Sγρσkαk k ξβξ + kαξ γ kβ + kβξ αkγ nition of the Bαβ tensor (although without affecting the γ σ γ σ∇ ∇ 2Sαγσkβk ξ + kαkβξ ξ σkγ final result). − γ σ ∇ γ ρ σ +2Sβγσk k ξα +2Sγσρkαkβk ξ ξ Looking at Eq. (50) it is not clear that the Raychaud- +2k ξ ξσS kγkρ (46) huri equation for a null congruence is independent of the β α σγρ choice of the auxiliary null vector ξα that was introduced to define the projector onto the orthogonal hypersurface B = 2S kρk ξσ k ξρ k +2S kρk ξσ to the congruence, Eq. (43): we have terms that depend kαβ ρσβ α β α ρ αρσ β αβ − ρ σ −γ ∇ ρ σ on the projector h and Bkαβ, which themselves depend 2Sρσγ k kαkβξ ξ +2Sαρσk k ξβ on ξα. However, since the expansion θ is a scalar quan- − ρ σ γ σ γ ρ e 2Sργσk k kαξ ξβ 2ξ ξαkβSσγρk k tity, its rate of variation along the congruence will be − γ σ −ρ 2ξαSβγσk k kαξ ρkβ related to the the rate of variation of θ by − ρ σ − ∇ kαkβξ ξ σkρ (47) − ∇ Dθ e Dθ = kµ θ = kµ∂ θ = kµ∂ θ = kµ (g)θ = . dλ µ µ µ µ dλ Similarly to what was done in (29), we can relate the ∇ ∇ e e e e (51) component B αβ with the expansion, shear and vorticity ⊥ D µ (g) of the congruence of the null curves, where dλ = k µ represents the covariant derivative along the fiducial∇ curve where only the metric connection e is considered. Computing the derivative, we recover the hαβ B⊥αβ = θ + σαβ + ωαβ , (48) usual general relativity expression for the Raychaudhuri N 2 e− equation governing the evolution of the expansion: that in combination with (46) provides an expression for the expansion of the congruence of null geodesics in terms α α Dθ α β 1 2 αβ βα of k , ξ and S , given by = Rαβk k θ + σαβσ + ωαβω , αβγ dλ − − N 2  e e − γ γ (g) γ e e e e e e (52) θ = γ k +2Sγk = k θ . (49) ∇ ∇γ ≡ where the tilded quantities are calculated with the Christoffel connection, and Note that also in this case, as in general relativity,e the scalar expansion θ does not depend on the auxiliary vec- g µ θ = ( )k , tor ξµ chosen to define a proper projection operator to ∇µ (g) γ (g) γ (g) treat the null curves congruence case. σµνe = (µ kν) + ξ k(µ γ kν) + ξ k(ν µ) kγ ∇ ∇ ∇ (53) γ σ (g) (g) µ + kµkν ξ ξ kγ k , e ∇σ − ∇µ (g) γ (g) γ (g) ωµν = [µ kν] + k[µξ γ kν] + ξ k[ν µ] kγ . 3 For the sake of completeness, we want to mention that also in ∇ ∇ ∇ this case there is still room for the (unlikely) scenario with one Notee that, in spite of the explicit appearance of the auxil- out of three neutrino species exactly massless. iary vector ξµ in the two expressions for σµν and ωµν , the

e e 7 total expression (52) does not depend on ξµ, as becomes sor allows the possibility to test models equipped with a evident after calculating the contractions nontrivial Riemannian connection through the study of the motion of test particles. Let us expand a bit on this µν 1 ˜2 2 ˜ 1 (g) ν (g) µ point: the matter source appearing on the right hand side σµν σ = θ + θ + µ k ν k − N 2 N 2 2∇ · ∇ of the equation of motion for the torsion tensor (what is − − e e 1 (g) ν (g)µ usually dubbed hypermomentum, obtained from the vari- + k kν , 2∇µ · ∇ ation of the classical matter action with respect to the νµ 1 (g) ν (g) µ 1 (g) ν (g)µ independent connection) depends on the specific coupling ωµν ω = k k k kν . 2∇µ · ∇ν − 2∇µ · ∇ between matter and torsion itself: different models will (54) e e lead to different field equations for the torsion and hence- Since Eq. (52) is equivalent to Eq. (50), we find that the forth to different solutions. This means that the eventual Raychaudhuri equation for a null congruence is indepen- contributions of torsion to the various kinematical quan- µ dent of the vector field ξ . tities and their evolution will be dependent on the spe- As a final comment, let us address the following im- cific chosen model; the analysis then of the evolution of a portant remark: one could be tempted to naively think matter fluid in a region of space-time will give a chance that, since (52) does not depend explicitly of the torsion to distinguish between different allowed couplings. tensor, then the evolution of the expansion of the curves One of the most important achievements of this paper followed by massless particles does not depend on the is the generalization of Raychaudhuri equation - the presence of a torsion field. This conclusion is not correct equation for the evolution of the expansion of a congru- since although the torsion tensor does not explicitly ap- ence of curves - for the case of time-like and null curves pear in (52), it does affect the geometry of the space-time in an N-dimensional space-time with the most generic in a way described by the (modified) Einstein field equa- torsion field. The study of the evolution of the expansion tions; the metric solution of the Einstein equations will 4 of time-like and null curves in spacetime is obviously be itself affected by the such modification , and so will be important due to its role in defining the evolution of the corresponding (metric) Ricci tensor Rµν appearing in gravitational collapse and possible formation of singular- (52). e ities. Moreover, as was shown in Ref. [36], the expansion of null light-rays also plays a preponderant role on the definition of the throats of (dynamical) wormholes, IV. CONCLUSIONS namely exotic solutions requiring the violation of energy conditions; a still open question is whether or not the In this paper we derived the equation for the evo- degrees of freedom of a completely generic torsion field lution of the separation vector between infinitesimally can somehow avoid the violation of the null energy close curves of a congruence in space-times with non-null condition of whatever matter present at the dynamical generic torsion field, clarifying some of the ambiguities wormhole’s throat. We firmly believe that the study of lingering in the literature about the role of the torsion such possibilities will be of great interest in the near tensor. We concluded that the presence of a torsion field future. leads in general to tangent and orthogonal effects to the congruence, in particular, the presence of a generic tor- sion field contributes to a relative acceleration between test particles. This effects happen either for free-falling or accelerated particles following time-like, null or space- ACKNOWLEDGMENTS like curves. The evolution equation of the separation vector can be further separated and be used to study the kinemat- The Authors wish to thank José P. S. Lemos for early ical quantities that characterize a congruence of curves, discussions on a first version of the paper. We thank namely the expansion, shear and vorticity. We derived, FCT-Portugal for financial support through Project No. for the first time in the literature, how such kinematical PEst - OE/FIS/UI0099/ 2015. PL thanks IDPASC and quantities depend on a completely generic torsion field. FCT-Portugal for financial support through Grant No. Knowing how the kinematical decomposition of the PD/BD/114074/2015. VV is supported by the FCT- geodesics congruence is influenced by the torsion ten- Portugal grant SFRH/BPD/77678/2011.

4 Just as a workable example, consider the simple Einstein–Cartan tion, rewritten in terms of the Christoffel connection eliminating case: one of the equations of motion describes algebraically the the torsion tensor, relates the (Christoffel) Einstein tensor to a torsion tensor as a function of the spin tensor; the other equa- combined version of the stress-energy tensor, that now takes into account also spin density terms [10]. 8

[1] A. K. Raychaudhuri, “Relativistic Cosmology I”, Phys. [21] V. Vitagliano, “The role of nonmetricity in metric-affine Rev. 98, 1123 (1955). theories of gravity”, Class. Quant. Grav. 31, 045006 [2] R. Penrose, “Gravitational Collapse and Space-Time Sin- (2014). gularities”, Phys. Rev. Lett. 14, 57 (1965). [22] V. Vitagliano, “On the dynamical content of MAGs,” J. [3] S. Hawking and G. F. R. Ellis, The Large Scale Structure Phys. Conf. Ser. 600, no. 1, 012043 (2015). of Space-Time (Cambridge University Press, Cambridge, [23] E. Poisson, A Relativist’s Toolkit: The Mathematics 1973). of Black-Hole Mechanics (Cambridge University Press, [4] S. Kar and S. Sengupta, “The Raychaudhuri equations: Cambridge, 2004). A brief review”, Pramana 69, 49 (2007). [24] D. Puetzfeld and Y. N. Obukhov, “Propagation equa- [5] K. Tomita, “Gravitational lens effect of wall-like objects tions for deformable test bodies with microstructure in and its cosmological implications”, Prog. Theor. Phys. extended theories of gravity,” Phys. Rev. D 76, 084025 85, 57 (1991). (2007) Erratum: [Phys. Rev. D 79, 069902 (2009)]. [6] L. Herrera, “Cracking of self-gravitating compact ob- [25] P. B. Yasskin and W. R. Stoeger, S.J., “Propagation jects”, Phys. Lett. A 165, 206 (1992). Equations for Test Bodies With Spin and Rotation in [7] T. Jacobson, “Thermodynamics of spacetime: The Ein- Theories of Gravity With Torsion,” Phys. Rev. D 21, stein equation of state”, Phys. Rev. Lett. 75, 1260 (1995). 2081 (1980). [8] C. Eling, R. Guedens and T. Jacobson, “Nonequilib- [26] K. Nomura, T. Shirafuji and K. Hayashi, “Semiclassi- rium thermodynamics of spacetime”, Phys. Rev. Lett. cal particles with arbitrary spin in the Riemann-Cartan 96, 121301 (2006). space-time,” Prog. Theor. Phys. 87, 1275 (1992). [9] G. Chirco and S. Liberati, “Non-equilibrium Thermody- [27] M. I. Wanas and M. A. Bakry, “Effect of spin-torsion in- namics of Spacetime: The Role of Gravitational Dissipa- teraction on Raychaudhuri equation”, Int. J. Mod. Phys. tion”, Phys. Rev. D 81, 024016 (2010). A 24, 5025 (2009). [10] F. W. Hehl, P. von der Heyde, G. D. Kerlick and J. M. [28] G. Esposito, “Mathematical structures of space-time”, Nester, “General relativity with spin and torsion: Foun- Fortsch. Phys. 40, 1 (1992). dations and prospects”, Rev. Mod. Phys. 48, 393 (1976). [29] B. Carter, “The general theory of the mechanical, electro- [11] A. Trautman, “Spin and torsion may avert gravitational magnetic and thermodynamic properties of black holes”, singularities.”, Nature Phys. Sci. 242, 7 (1973). in General Relativity: an Einstein Centenary Survey, [12] J. Stewart and P. Hájíček, “Can spin avert singulari- ed. S.W. Hawking and W. Israel (Cambridge University ties?”, Nature Phys. Sci. 244, 96 (1973). Press, Cambridge, England, 1979). [13] W. Kopczýnski, “An anisotropic universe with torsion.”, [30] D. Colladay and V. A. Kostelecky, “Lorentz violating ex- Phys. Lett. A 43A, 63 (1973). tension of the standard model,” Phys. Rev. D 58, 116002 [14] J. Tafel, “A non-singular homogeneous universe with tor- (1998). sion”, Phys. Lett. A 45, 4 (1973). [31] Y. Fukuda et al. [Super-Kamiokande Collaboration], [15] T. P. Sotiriou and S. Liberati, “Metric-affine f(R) theories “Evidence for oscillation of atmospheric neutrinos”, of gravity,” Annals Phys. 322, 935 (2007). Phys. Rev. Lett. 81, 1562 (1998). [16] V. Vitagliano, T. P. Sotiriou and S. Liberati, “The [32] Q. R. Ahmad et al. [SNO Collaboration], “Direct ev- dynamics of generalized Palatini Theories of Gravity,” idence for neutrino flavor transformation from neutral Phys. Rev. D 82, 084007 (2010). current interactions in the Sudbury Neutrino Observa- [17] V. Vitagliano, T. P. Sotiriou and S. Liberati, “The dy- tory”, Phys. Rev. Lett. 89, 011301 (2002). namics of metric-affine gravity,” Annals Phys. 326, 1259 [33] S. M. Bilenky and S. T. Petcov, “Massive Neutrinos and (2011) Erratum: [Annals Phys. 329, 186 (2013)]. Neutrino Oscillations,” Rev. Mod. Phys. 59, 671 (1987) [18] L. T. Santana, M. O. Calvão, R. R. R. Reis and B. B. Sif- Erratum: [Rev. Mod. Phys. 61, 169 (1989)] Erratum: fert, “How does light move in a generic metric-affine back- [Rev. Mod. Phys. 60, 575 (1988)]. ground?,” Phys. Rev. D 95, 061501 (2017). [34] M. Blagojević and F. W. Hehl, chapter 14 in Gauge The- [19] Y. Cai, S. Capozziello, M. De Laurentis and E. N. Sari- ories of Gravitation : A Reader with Commentaries, (Im- dakis, “f (T ) teleparallel gravity and cosmology”, Rept. perial College Press, London, 2013). Prog. Phys. 79, 106901 (2016). [35] M. Vasilic and M. Vojinovic, “Zero-size objects in [20] Y. Mao, M. Tegmark, A. H. Guth, and S. Cabi, “Con- Riemann-Cartan spacetime,” JHEP 0808, 104 (2008). straining torsion with Gravity Probe B”, Phys. Rev. D [36] D. Hochberg and M. Visser, “Dynamic wormholes, anti- 76, 104029 (2007). trapped surfaces, and energy conditions”, Phys. Rev. D 58, 044021 (1998).