betaFeedback

WorksPhysics and Astronomy

2-dimensional Metric-Affine Geometries and Dynamical Gravity

Abstract

We study non-Riemannian metric-affine geometries in 2-dimensions. We explicitly show that if the connection of the space is not metric compatible (i.e. the non-metricity is non-vanishing) then the underlying 2-dimensional manifold is not an Einstein manifold, in contrast to the Riemannian case. Furthermore we prove that in general the curvature in such spaces is specified completely by the Ricci tensor which, however, is not symmetric now. An enlarged symmetry of the scalar curvature under vectorial connection shifts is also found. We then apply our results to the study of 2-dimensional Metric-Affine Gravity Theories and discuss some special cases.

Introduction

It is a widely well known mathematical result, and quite trivial to prove, that in 2 dimensions the full Riemannian curvature is completely specified by the Riemannian Ricci scalar. From a physical perspective, in this Riemannian geometry setting, the 2-dimensional Einstein-Hilbert action is topological; it has no local propagating degrees of freedom, and is therefore trivial, the Einstein tensor vanishes identically. This is so because in 2-dimensions, the Riemannian curvature, given its symmetries, has only one independent component and is therefore completely determined by the Ricci tensor through the well-known formula

R~ijkl=R~2(gikgjl−gilgjk)(1)\widetilde{R}_{ijkl} = \frac{\widetilde{R}}{2}(g_{ik}g_{jl}-g_{il}g_{jk}) \tag*{(1)}
R~ij=R~2gij(2)\widetilde{R}_{ij} = \frac{\widetilde{R}}{2}g_{ij} \tag*{(2)}

where the tildes denote Riemannian quantities computed with respect to the torsionless and metric-compatible Levi-Civita connection. In other words, what (2) tells us is that every 2-dimensional Riemannian manifold is an Einstein space.

In order to obtain non-trivial dynamics, an additional dilaton field has to be included leading to the so-called 2-dimensional JT gravity [1, 2].1 Alternatively, one may obtain non-trivial dynamics by enriching the underlying geometry by allowing for torsion and nonmetricity. This is the most well motivated geometric approach to 2-dimensional gravity. Quite surprisingly, not many studies have been done on the matter, with the notable exceptions [4, 5, 6].

In this note we provide some useful novel results related to the geometry of 2-dimensional metric-affine spaces. Given these results we then consider applications to gravity and show how one can have dynamical 2-dim gravity as a consequence of the generalized hh=geometry. In what follows the signature of the 2-dim metric will be ((−1)s,1)((-1)^s,1) where s=0s=0 correspond to a Euclidean and s=1s=1 to Lorentzian manifold respectively. When we study gravity theories we will stick to the Lorentzian signature. The Levi-Civita density is denoted by ϵαβ\epsilon_{\alpha\beta} with ϵ01=−ϵ10=1\epsilon_{01}=-\epsilon_{10}=1 and ϵ00=0=ϵ11\epsilon_{00}=0=\epsilon_{11}. From this we construct the 2-dimensional Levi-Civita tensor

εαβ=∣g∣ϵαβ(3)\varepsilon_{\alpha\beta}=\sqrt{|g|}\epsilon_{\alpha\beta} \tag*{(3)}

and note the 2-dim identity

εαβεμν=gανgβμ−gαμgβν(4)\varepsilon_{\alpha\beta}\varepsilon_{\mu\nu}=g_{\alpha\nu}g_{\beta\mu}-g_{\alpha\mu}g_{\beta\nu} \tag*{(4)}

The Metric-Affine setup

A generic nn-dim manifold endowed with a metric gg and an independent affine connection ∇\nabla, denoted as (M,g,∇)(\mathcal{M},g,\nabla) defines a metric-affine (non-Riemannian) geometry. On this extended manifold, we define the curvature, torsion and non-metricity tensors according to [7]

Rμναβ:=2∂[αΓμ∣ν∣β]+2Γρ[αμΓρ∣ν∣β](5)R^{\mu}{}_{\nu\alpha\beta}:=2\partial_{[\alpha}\Gamma^{\mu}{}_{|\nu|\beta]}+2\Gamma^{\mu}_{\rho[\alpha}\Gamma^{\rho}{}_{|\nu|\beta]} \tag*{(5)}
Sμνλ:=Γλ[μν](6)S_{\mu\nu}{}^{\lambda}:=\Gamma^{\lambda}{}_{[\mu\nu]} \tag*{(6)}
Qαμν:=−∇αgμν(7)Q_{\alpha\mu\nu}:=-\nabla_{\alpha}g_{\mu\nu} \tag*{(7)}

The departure of the affine connection Γλμν\Gamma^{\lambda}{}_{\mu\nu} from the Levi-Civita one defines the distortion tensor [7, 8]

Nλμν:=Γλμν−Γ~λμν=12gαλ(Qμνα+Qναμ−Qαμν)−gαλ(Sαμν+Sανμ−Sμνα)(8)N^{\lambda}{}_{\mu\nu}:=\Gamma^{\lambda}{}_{\mu\nu}-\widetilde{\Gamma}^{\lambda}{}_{\mu\nu}=\frac{1}{2}g^{\alpha\lambda}(Q_{\mu\nu\alpha}+Q_{\nu\alpha\mu}-Q_{\alpha\mu\nu})-g^{\alpha\lambda}(S_{\alpha\mu\nu}+S_{\alpha\nu\mu}-S_{\mu\nu\alpha}) \tag*{(8)}

where Γ~λμν\widetilde{\Gamma}^{\lambda}{}_{\mu\nu} is the usual Levi-Civita connection computed solely by the metric and its first derivatives. From the distortion, torsion and non-metricity can be retrieved according to (see for instance [9])

Sμνα=Nα[μν],Qναμ=2N(αμ)ν(9)S_{\mu\nu\alpha}=N_{\alpha[\mu\nu]},\qquad Q_{\nu\alpha\mu}=2N_{(\alpha\mu)\nu} \tag*{(9)}

Out of these we can construct the associated torsion and nonmetricity vectors by taking the contractions:

Sμ:=Sμλλ(10)S_{\mu}:=S_{\mu\lambda}{}^{\lambda} \tag*{(10)}

and

Qα:=Qαμνgμν,qν=Qαμνgαμ(11)Q_{\alpha}:=Q_{\alpha\mu\nu}g^{\mu\nu},\qquad q_{\nu}=Q_{\alpha\mu\nu}g^{\alpha\mu} \tag*{(11)}

respectively. As for the curvature tensor, we can construct the three contractions

Rνβ:=Rμνμβ,R^αβ:=Rμμαβ,R˘μβ:=Rνμαβgνα(12)R_{\nu\beta}:=R^{\mu}{}_{\nu\mu\beta},\qquad\widehat{R}_{\alpha\beta}:=R^{\mu}{}_{\mu\alpha\beta},\qquad\breve{R}^{\mu}{}_{\beta}:=R^{\nu}{}_{\mu\alpha\beta}g^{\nu\alpha} \tag*{(12)}

The first one is generalized Ricci tensor (which is not symmetric in general), the second is the so-called homothetic curvature and the last one is known as the co-Ricci tensor. Observe that in constructing the Ricci scalar and homothetic curvature no metric is required, whereas in order to define the co-Ricci a metric should be given. The generalized scalar curvature is still uniquely defined, since

R:=gμνRμν=−gμνR˘μν,gμνR^μν=0(13)R:=g^{\mu\nu}R_{\mu\nu}=-g^{\mu\nu}\breve{R}_{\mu\nu},\qquad g^{\mu\nu}\widehat{R}_{\mu\nu}=0 \tag*{(13)}

Using eq. (8) allows to split any quantity into its Riemannian (i.e. computed with respect to the Levi-Civita connection) and non-Riemannian contributions of torsion and nonmetricity. For instance inserting the decomposition (8) into the definition (5) we obtain for the curvature tensor 2

Rμναβ=R~μναβ+2∇~[αNμ∣ν∣β]+2Nμλ[αNλ∣ν∣β](14)R^{\mu}{}_{\nu\alpha\beta}=\widetilde{R}^{\mu}{}_{\nu\alpha\beta}+2\widetilde{\nabla}_{[\alpha}N^{\mu}{}_{|\nu|\beta]}+2N^{\mu}{}_{\lambda[\alpha}N^{\lambda}{}_{|\nu|\beta]} \tag*{(14)}

Taking further contractions we can also find the post-Riemannian expansion of the scalar curvature

R=R~+14QαμνQαμν−12QαμνQμνα−14QμQμ+12Qμqμ+SμναSμνα−2SμναSαμν−4SμSμ+2QαμνSαμν+2Sμ(qμ−Qμ)+∇~μ(qμ−Qμ−4Sμ)(15)\begin{aligned} R={}&\widetilde{R}+\frac{1}{4}Q_{\alpha\mu\nu}Q^{\alpha\mu\nu}-\frac{1}{2}Q_{\alpha\mu\nu}Q^{\mu\nu\alpha}-\frac{1}{4}Q_{\mu}Q^{\mu}+\frac{1}{2}Q_{\mu}q^{\mu}+S_{\mu\nu\alpha}S^{\mu\nu\alpha}\\ &-2S_{\mu\nu\alpha}S^{\alpha\mu\nu}-4S_{\mu}S^{\mu}+2Q_{\alpha\mu\nu}S^{\alpha\mu\nu}+2S_{\mu}(q^{\mu}-Q^{\mu})+\widetilde{\nabla}_{\mu}(q^{\mu}-Q^{\mu}-4S^{\mu}) \tag*{(15)} \end{aligned}

Alternatively, in terms of the distortion, this expansion takes the much more compressed form

R=R~+∇~μ(Nμαβgαβ−Nαμα)+NμαβgαβNγμγ−NαμνNμνα(16)R=\widetilde{R}+\widetilde{\nabla}_{\mu}(N^{\mu\alpha\beta}g_{\alpha\beta}-N^{\alpha\mu}{}_{\alpha})+N_{\mu\alpha\beta}g^{\alpha\beta}N^{\gamma\mu}{}_{\gamma}-N_{\alpha\mu\nu}N^{\mu\nu\alpha} \tag*{(16)}

Finally, an object that we will be dealing with later on is the so-called Palatini tensor [10], defined as the variation of the scalar curvature relative to the affine connection:

Pλμν:=δRδΓμνλ=−∇λ(−ggμν)−g+∇σ(−ggμσ)−gδλν+2(Sλgμν−Sμδλν+gμσSσλν)=δλν(Qμ−12Qμ−2Sμ)+gμν(12Qλ+2Sλ)−(Qλμν+2Sλμν).(17)\begin{aligned} P_{\lambda}^{\mu\nu} := \frac{\delta R}{\delta\Gamma^{\lambda}_{\mu\nu}} = -\frac{\nabla_{\lambda}\left(\sqrt{-g}g^{\mu\nu}\right)}{\sqrt{-g}}+\frac{\nabla_{\sigma}\left(\sqrt{-g}g^{\mu\sigma}\right)}{\sqrt{-g}}\delta^{\nu}_{\lambda}+2\left(S_{\lambda}g^{\mu\nu}-S^{\mu}\delta^{\nu}_{\lambda}+g^{\mu\sigma}S^{\nu}_{\sigma\lambda}\right) \\ &=\delta^{\nu}_{\lambda}\left(Q^{\mu}-\frac{1}{2}Q^{\mu}-2S^{\mu}\right)+g^{\mu\nu}\left(\frac{1}{2}Q_{\lambda}+2S_{\lambda}\right)-\left(Q_{\lambda}^{\mu\nu}+2S_{\lambda}^{\mu\nu}\right). \tag*{(17)} \end{aligned}

Torsion and Non-Riemannian Curvature in 2-Dimensions

In 2 dimensions, even for the generalized non-Riemannian background of Metric-Affine Geometry, many simplifications occur. The affine connection has now 8 independent components, torsion has 2 and non-metricity 6. Of course, the metric still has three components as in the Riemannian case but now the curvature has 4 independent components, the same number as the Ricci tensor. This suggests that, in principle, in this generalized geometric setting, one should be possible to determine the curvature from the Ricci tensor. As we explicitly prove in the following, this is indeed the case. Before proving this, we shall also review some known results for the 2-dimensional torsion, approached from a somewhat different perspective.

2-dimensional Torsion

The very fact that torsion has only 2 components implies that only its trace part is non-vanishing namely, it has the simple form

Sαβγ=2S[αδβ]γ(18)S_{\alpha\beta}^{\gamma}=2S_{[\alpha}\delta_{\beta]}^{\gamma} \tag*{(18)}

with Sα:=Sακ  κS_{\alpha}:=S_{\alpha\kappa}^{\ \ \kappa} being the torsion vector introduced earlier. Of course, even though this is intuitive by counting the degrees of freedom, a more formal proof is required. Let us prove this statement. Since in 2-dimensions any rank-3 totally antisymmetric tensor, or higher rank tensor with 3 indices antisymmetrized, vanishes identically, we have

δ[λγSαβ]κ=0.(19)\delta_{[\lambda}^{\gamma}S_{\alpha\beta]}^{\kappa}=0. \tag*{(19)}

Expanding the latter and using also the antisymmetry property of the torsion tensor, it follows that

δλρSαβκ+δβρSλακ+δαρSβλκ=0(20)\delta_{\lambda}^{\rho}S_{\alpha\beta}^{\kappa}+\delta_{\beta}^{\rho}S_{\lambda\alpha}^{\kappa}+\delta_{\alpha}^{\rho}S_{\beta\lambda}^{\kappa}=0 \tag*{(20)}

Contracting in λ=κ\lambda=\kappa we readily obtain

Sαβγ=2S[αδβ]γ(21)S_{\alpha\beta}^{\gamma}=2S_{[\alpha}\delta_{\beta]}^{\gamma} \tag*{(21)}

which is the stated result. Note that no metric was used to derive this result; it holds true for a purely affine space as well, without the need of a metric. This torsion form implies that in 2-dim the contortion tensor assumes the simple form

Kαμν:=Sμνα−Sανμ−Sαμν=−4S[αgμ]ν(22)K_{\alpha\mu\nu}:=S_{\mu\nu\alpha}-S_{\alpha\nu\mu}-S_{\alpha\mu\nu}=-4S_{[\alpha}g_{\mu]\nu} \tag*{(22)}

Finally, it is trivial to show that for the above form of torsion, the following identities hold true

∇κSij k=−2∇[iSj],∇[iSj]=∂[iSj],SijkSk=0(23)\nabla_{\kappa}S_{ij}^{\ k}=-2\nabla_{[i}S_{j]},\quad\nabla_{[i}S_{j]}=\partial_{[i}S_{j]},\quad S_{ijk}S^{k}=0 \tag*{(23)}

These relations will be found to be useful when we compute the various contractions of the non-Riemannian curvature. All of the above were well established standard results; let us now move on to obtain a new result for the non-Riemannian curvature.

2-DIMENSIONAL CURVATURE

We shall now focus on the curvature tensor of a 2-dimensional metric-affine space. In contrast to the Riemannian case, we now have the following.

Theorem 1. In 2-dim space endowed with a general linear connection ∇\nabla, the curvature tensor is completely determined by the generalized (asymmetric) Ricci tensor. Its coordinate expression reads

R μβγλ=2Rμ[γδβ]λ(24)R^{\lambda}_{\ \mu\beta\gamma}=2R_{\mu[\gamma}\delta^{\lambda}_{\beta]} \tag*{(24)}

Proof. Firstly, let us note that in deriving this expression no metric is required. Namely, the validity of the relation to be proved does not require the introduction of a metric but only that of a linear connection.3 Indeed, starting from the dimension-dependent identity

R μ[αβκδγ]λ=0(25)R^{\kappa}_{\ \mu[\alpha\beta}\delta^{\lambda}_{\gamma]}=0 \tag*{(25)}

expanding, and using also the antisymmetry of the curvature in its last indices it readily follows that

R μαβκδγλ+R μγακδβλ+R μβγκ=0(26)R^{\kappa}_{\ \mu\alpha\beta}\delta^{\lambda}_{\gamma}+R^{\kappa}_{\ \mu\gamma\alpha}\delta^{\lambda}_{\beta}+R^{\kappa}_{\ \mu\beta\gamma}=0 \tag*{(26)}

Contracting the latter in κ=γ\kappa=\gamma, we immediately get

R μαβλ=−2Rμ[αδβ]λ(27)R^{\lambda}_{\ \mu\alpha\beta}=-2R_{\mu[\alpha}\delta^{\lambda}_{\beta]} \tag*{(27)}

as stated.

Therefore we have:

Corollary 1. In a 2-dim non-Riemannian space, the curvature tensor is specified completely by the Ricci tensor.4 This construction does not rely on a metric. Furthermore, the vanishing of the scalar curvature does not necessarily imply that the Ricci and curvature tensors are zero.

Remark. Notice that no use of a metric was made in obtaining the last relation. The above result is also true for purely affine spaces, namely manifolds endowed with only a connection.

If a metric is given along with a connection (metric-affine space), then contracting (27) with gμαg^{\mu\alpha}, and bringing all indices down, it follows that

R~κβ=−Rgκβ+Rκβ(28)\widetilde{R}_{\kappa\beta}=-Rg_{\kappa\beta}+R_{\kappa\beta} \tag*{(28)}

which relates the co-Ricci to the Ricci tensor and scalar curvature. The antisymmetric part of the latter gives the relation R[νβ]=R~[νβ]R_{[\nu\beta]}=\widetilde{R}_{[\nu\beta]}. Expression (28) allows to express the curvature in the alternative form

Rαβμν=R~ανgμβ−R~αμgνβ(29)R_{\alpha\beta\mu\nu}=\widetilde{R}_{\alpha\nu}g_{\mu\beta}-\widetilde{R}_{\alpha\mu}g_{\nu\beta} \tag*{(29)}

In addition, (28) tells us that in contrast to the Riemannian case, in the 2-dim non-Riemannian geometry not all spaces are Einstein spaces. In terms of the Ricci tensor and scalar curvature, the curvature tensor reads

Rαβμν=2Rgα[μgν]β−2Rα[μgν]β(30)R_{\alpha\beta\mu\nu}=2Rg_{\alpha[\mu}g_{\nu]\beta}-2R_{\alpha[\mu}g_{\nu]\beta} \tag*{(30)}

Remark. It is important to stress, however, that even when we endow our space with a metric, in contradistinction with the Riemannian case, here the Ricci tensor is not fully described only by the metric and the scalar curvature. Of course, this is easily seen from the very fact that RμνR_{\mu\nu} now also contains an antisymmetric part and therefore cannot be constructed only by the (symmetric) metric.

Therefore, given equations (27),(28) and (30) we have the following:

Corollary 2. If a 2-dimensional metric-affine space is either Ricci flat (Rμν=0R_{\mu\nu}=0) or co-Ricci flat (R~μν=0\widetilde{R}_{\mu\nu}=0) then it is completely flat, in the sense that R μναλ=0R^{\lambda}_{\ \mu\nu\alpha}=0 in either case.

Ricci, co-Ricci and Homothetic Curvature

By substituting the torsion relations (23) into (14) and taking the relevant contractions, it is then easy to see that in the antisymmetric part of the Ricci tensor only non-metricity contributes since

R[ij]=12R^ij(31)R_{[ij]}=\frac{1}{2}\hat{R}_{ij} \tag*{(31)}

for this form of torsion. Alternatively, this relation can also be trivially derived by contracting (27) in λ=μ\lambda=\mu. Continuing, using again the fact that any totally antisymmetric rank-3 tensor (or any higher rank tensor antisymmetrized in three indices) identically vanishes in 2-dimensions, we have the identities

R[αβγ]λ=0(32)R_{[\alpha\beta\gamma]\lambda}=0 \tag*{(32)}

Expanding the antisymmetrization and then contracting with gγλg^{\gamma\lambda} we find

R~[αβ]=R[αβ](33)\tilde{R}_{[\alpha\beta]}=R_{[\alpha\beta]} \tag*{(33)}

then comparing the latter with (31) it follows that all antisymmetric parts of the Ricci type tensors are proportional to one another. Of course this is to be expected since in 2-dimensions any antisymmetric rank-2 tensor is a top-form and as such it is proportional to the antisymmetric Levi-Civita tensor ϵμν\epsilon_{\mu\nu}. The proportionality factor is then fixed by employing the identity ϵμνϵμν=−2\epsilon^{\mu\nu}\epsilon_{\mu\nu}=-2 and we readily find

R~[αβ]=R[αβ]=12R^αβ=12∂[αQβ]=−12(ϵμν∂μQν)ϵαβ(34)\tilde{R}_{[\alpha\beta]}=R_{[\alpha\beta]}=\frac{1}{2}\hat{R}_{\alpha\beta}=\frac{1}{2}\partial_{[\alpha}Q_{\beta]}=-\frac{1}{2}\left(\epsilon^{\mu\nu}\partial_\mu Q_\nu\right)\epsilon_{\alpha\beta} \tag*{(34)}

Confirming that the antisymmetric parts of all Ricci-type tensors are proportional to one another.

2-dimensional Scalar Curvature and Palatini tensor

It is easy to check that in 2-dimensions the torsion contributions exactly cancel out and the Palatini tensor (17) acquires the form

Pαμν=Qα2gμν−Qαμν+gαν(qμ−Qμ2)(35)P_{\alpha\mu\nu}=\frac{Q_\alpha}{2}g_{\mu\nu}-Q_{\alpha\mu\nu}+g_{\alpha\nu}\left(q_\mu-\frac{Q_\mu}{2}\right) \tag*{(35)}

This suggests that in 2-dim the scalar curvature is, probably up to total derivatives, independent of the torsion.5 Indeed, given the form of torsion (18) it is a trivial task to show that in the post-Riemannian expansion of RR the quadratic torsion terms and mixed torsion-non-metricity terms exactly cancel out leaving us with

R=R~+14QαμνQαμν−12QαμνQμνα−14QμQμ+12Qμqμ+∇~μ(qμ−Qμ−4Sμ)(36)R=\tilde{R}+\frac{1}{4}Q_{\alpha\mu\nu}Q^{\alpha\mu\nu}-\frac{1}{2}Q_{\alpha\mu\nu}Q^{\mu\nu\alpha}-\frac{1}{4}Q_\mu Q^\mu+\frac{1}{2}Q_\mu q^\mu+\tilde{\nabla}_\mu\left(q^\mu-Q^\mu-4S^\mu\right) \tag*{(36)}

and consequently, up to total derivatives the scalar curvature is independent of torsion in 2-dim. Consequently the Einstein-Hilbert action at the linear level, does not include any torsion contribution. The last term in (36) that contains a torsion derivative will give contributions only if quadratic and higher order terms of the scalar curvature are considered or non-minimal couplings of the form ΦR\Phi R. In what follows, we shall investigate some of these possibilities in detail.

A NEW SYMMETRY OF SCALAR CURVATURE IN 2-DIM

Under the displacement Γμνλ→Γμνλ+Ξμνλ\Gamma^\lambda_{\mu\nu}\to\Gamma^\lambda_{\mu\nu}+\Xi^\lambda_{\mu\nu} the curvature tensor transforms as [9]

Rμναβ→Rμναβ+2∇[αΞμ∣ν∣β]−2SαβλΞμλ+2Ξμλ[αΞλ∣ν∣β](37)R^\mu{}_{\nu\alpha\beta}\to R^\mu{}_{\nu\alpha\beta}+2\nabla_{[\alpha}\Xi^\mu{}_{|\nu|\beta]}-2S_{\alpha\beta}^{\lambda}\Xi^\mu{}_{\lambda}+2\Xi^\mu{}_{\lambda[\alpha}\Xi^\lambda{}_{|\nu|\beta]} \tag*{(37)}

Consequently, taking a double contraction (with δμαgβγ\delta^\alpha_\mu g^{\beta\gamma}) it follows that the scalar curvature transforms according to

R→R+∇α(Ξ(3)α−Ξ(2)α)+Ξμ(2)Ξ(3)μ−ΞαμνΞμνα+qμΞμ(2)−(Qαβγ+2Sαβγ)Ξαβγ(38)R \to R + \nabla_\alpha\left(\Xi^{\alpha}_{(3)}-\Xi^{\alpha}_{(2)}\right)+\Xi^{(2)}_\mu\Xi^\mu_{(3)}-\Xi_{\alpha\mu\nu}\Xi^{\mu\nu\alpha}+q^\mu\Xi^{(2)}_\mu-\left(Q_{\alpha\beta\gamma}+2S_{\alpha\beta\gamma}\right)\Xi^{\alpha\beta\gamma} \tag*{(38)}

where ∇\nabla, SS and QQ are computed with respect to the initial connection and we have defined the three traces Ξμ(1):=Ξαβμgαβ\Xi^{(1)}_\mu:= \Xi_{\alpha\beta\mu}g^{\alpha\beta}, Ξμ(2):=Ξαμβgαβ\Xi^{(2)}_\mu:= \Xi_{\alpha\mu\beta}g^{\alpha\beta} and Ξμ(3):=Ξμαβgαβ\Xi^{(3)}_\mu:= \Xi_{\mu\alpha\beta}g^{\alpha\beta}. Taking

Γμνλ→Γμνλ+ξλgμν−δνλξμ(39)\Gamma^\lambda_{\mu\nu}\to\Gamma^\lambda_{\mu\nu}+\xi^\lambda g_{\mu\nu}-\delta^\lambda_\nu\xi_\mu \tag*{(39)}

it easily follows for arbitrary dimension nn, that under (39), the density −gR\sqrt{-g}R transforms as follows:

−gR→−gR+2∂α(−gξα)+(n−2)−g(2∇αξα−(n−1)ξαξα−ξαqα)(40)\sqrt{-g}R\to\sqrt{-g}R+2\partial_\alpha\left(\sqrt{-g}\xi^\alpha\right)+(n-2)\sqrt{-g}\left(2\nabla_\alpha\xi^\alpha-(n-1)\xi^\alpha\xi_\alpha-\xi_\alpha q^\alpha\right) \tag*{(40)}

therefore in 2-dimensions the last parenthesis drops out and

−gR→−gR+2∂α(−gξα)(41)\sqrt{-g}R\to\sqrt{-g}R+2\partial_\alpha\left(\sqrt{-g}\xi^\alpha\right) \tag*{(41)}

which means that (up to surface terms) the generalized Einstein Hilbert Lagrangian density remains invariant under (39). This implies that (see [11]) the Γ\Gamma-variation of −gR\sqrt{-g}R produces a tensor PμνλP^\lambda_{\mu\nu} of which the trace in 1st and 3rd indices is the same with the trace in 2nd and 3rd, i.e. P(2)=P(3)P^{(2)}=P^{(3)}. We will elaborate more on this result in the next sections.

When the latter observation is combined with the fact that the scalar curvature is projective invariant (in any dimension), namely invariant under the connection shifts

Γμνλ→Γμνλ+δμλϑν(42)\Gamma^\lambda_{\mu\nu}\to\Gamma^\lambda_{\mu\nu}+\delta^\lambda_\mu\vartheta_\nu \tag*{(42)}

it immediately follows that for n=2n=2 the scalar curvature enjoys a larger symmetry. Indeed, given the fact that the connection transformations (39) and (42) are additive, we arrive to the following.

Theorem 2. In 2-dimensions and up to total derivative terms, the scalar curvature RR is invariant under connection transformations of the form

Γμνλ→Γμνλ+δμλϑν+ξλgμν−δνλξμ(43)\Gamma^\lambda_{\mu\nu}\to\Gamma^\lambda_{\mu\nu}+\delta^\lambda_\mu\vartheta_\nu+\xi^\lambda g_{\mu\nu}-\delta^\lambda_\nu\xi_\mu \tag*{(43)}

where ϑμ\vartheta_\mu is an arbitrary 1-form and ξλ\xi^\lambda an arbitrary vector field. This enlarged symmetry is a specific feature of 2-dimensions.6

Classification of Generalized Einstein Spaces in 2-Dim

2-dimensional Riemann-Cartan Geometry

It is interesting to see what changes once we restrict the non-Riemannian background by demanding also the metricity condition, i.e. imposing vanishing non-metricity. We are then in the realm of Riemann-Cartan geometry [12, 13, 14, 15]. In this case, it follows immediately that the full distortion is completely specified by the torsion vector (along with the metric) and reads

Nμνλ=−2(Sλgμν−δνλSμ)(44)N^\lambda_{\mu\nu}=-2\left(S^\lambda g_{\mu\nu}-\delta^\lambda_\nu S_\mu\right) \tag*{(44)}

With this restriction we have the following.

Theorem 3. Every 2-dimensional manifold with a metric compatible but torsionful connection, is an Einstein space in the sense that

Rμν=R2gμν(45)R_{\mu\nu}=\frac{R}{2}g_{\mu\nu} \tag*{(45)}

with the scalar curvature given by

R=R~−4∇~μSμ(46)R = \widetilde{R} - 4\widetilde{\nabla}_{\mu}S^{\mu} \tag*{(46)}

where R~\widetilde{R} is the Riemannian scalar curvature (Riemannian Ricci scalar), ∇~μ\widetilde{\nabla}_{\mu} the Levi-Civita covariant derivative, and SμS^{\mu} the torsion vector. In other words, every 2-dimensional Riemann-Cartan manifold is an Einstein space.

Proof. Starting from the definition of the curvature tensor and substituting in the form of the distortion (44) it readily follows that

Rμναβ=R~μναβ+4δ[βμ(∇~α]−2Sα])Sν−4gν[β(∇~α]−2Sα])Sμ−8δ[αμgβ]νSγSγ(47)R^{\mu}{}_{\nu\alpha\beta} = \widetilde{R}^{\mu}{}_{\nu\alpha\beta} + 4\delta^{\mu}_{[\beta}(\widetilde{\nabla}_{\alpha]}-2S_{\alpha]})S_{\nu} - 4g_{\nu[\beta}(\widetilde{\nabla}_{\alpha]}-2S_{\alpha]})S^{\mu} - 8\delta^{\mu}_{[\alpha}g_{\beta]\nu}S_{\gamma}S^{\gamma} \tag*{(47)}

Contracting we obtain the associated Ricci tensor7

Rμν=R~μν−2gμν∇~αSα(48)R_{\mu\nu} = \widetilde{R}_{\mu\nu} - 2g_{\mu\nu}\widetilde{\nabla}_{\alpha}S^{\alpha} \tag*{(48)}

Then, given that

R~μν=R~2gμν(49)\widetilde{R}_{\mu\nu} = \frac{\widetilde{R}}{2}g_{\mu\nu} \tag*{(49)}

for the Riemannian part in 2 dimensions, substitution of the latter into (49) gives the stated result. □

2-dimensional Riemann-Cartan-Weyl Geometry

Let us now consider a 2-dimensional Riemann-Cartan-Weyl geometry. Traditionally, a Riemann-Cartan-Weyl geometry is characterised by an arbitrary torsion but with a restricted Weyl-type non-metricity, viz.8

Qαμν=Qα2gμν(50)Q_{\alpha\mu\nu} = \frac{Q_{\alpha}}{2}g_{\mu\nu} \tag*{(50)}

for such geometries. Therefore this is the simplest non-trivial extension of the previous Riemann-Cartan case. Now, as we have already seen in 2-dimensions torsion is purely vectorial. Combining the above two results it follows that in 2-dimensional Weyl-Cartan spaces, the distortion tensor takes the form

Nλμν=δμλ14Qν+δνλwμ−wλgμν(51)N^{\lambda}{}_{\mu\nu} = \delta^{\lambda}_{\mu}\frac{1}{4}Q_{\nu} + \delta^{\lambda}_{\nu}w_{\mu} - w^{\lambda}g_{\mu\nu} \tag*{(51)}

where wμ=Qμ/4+2Sμw_{\mu}=Q_{\mu}/4+2S_{\mu}. Then it follows that:

Theorem 4. Every 2-dimensional Weyl-Cartan-Riemann space is a generalized Einstein space in the sense that

R(μν)=R2gμν(52)R_{(\mu\nu)} = \frac{R}{2}g_{\mu\nu} \tag*{(52)}

is identically valid.

Proof. For the distortion (52) we readily compute the associated Ricci tensor

Rμν=R~μν+12∂[μQν]−(∇~αwα)gμν(53)R_{\mu\nu} = \widetilde{R}_{\mu\nu} + \frac{1}{2}\partial_{[\mu}Q_{\nu]} - (\widetilde{\nabla}_{\alpha}w^{\alpha})g_{\mu\nu} \tag*{(53)}
Rμν=R~μν−2d−1gμν∇~αSα−2(d−2)d−1∇~νSμ+4(d−2)(d−1)2(SμSν−S2gμν)(54)R_{\mu\nu} = \widetilde{R}_{\mu\nu} - \frac{2}{d-1}g_{\mu\nu}\widetilde{\nabla}_{\alpha}S^{\alpha} - \frac{2(d-2)}{d-1}\widetilde{\nabla}_{\nu}S_{\mu} + \frac{4(d-2)}{(d-1)^2}(S_{\mu}S_{\nu}-S^{2}g_{\mu\nu}) \tag*{(54)}

But of course, it goes without saying, that in higher dimensions there are more modes to the torsion tensor apart from its trace. Note that R[μν]=12∂[μQν]=12R^μνR_{[\mu\nu]} = \frac{1}{2}\partial_{[\mu}Q_{\nu]} = \frac{1}{2}\hat{R}_{\mu\nu} as it should in 2-dimensions, in agreement with the results of the previous section. The associated scalar curvature to the above Ricci tensor, reads

R=R~−2∇~αwα(55)R = \widetilde{R} - 2\widetilde{\nabla}_{\alpha}w^{\alpha} \tag*{(55)}

Then given the fact that R~μν=R~2gμν\widetilde{R}_{\mu\nu} = \frac{\widetilde{R}}{2}g_{\mu\nu} it immediately follows that

R(μν)=R2gμν(56)R_{(\mu\nu)} = \frac{R}{2}g_{\mu\nu} \tag*{(56)}

Corollary 3. Any 2-dimensional Weyl-Cartan geometry is a generalized Einstein space.

To gain more intuition on this result, for generic nonmetricity we may separate out the Weyl part and decompose the nonmetricity according to

Qαμν=Qα2+Ωαμν(57)Q_{\alpha\mu\nu} = \frac{Q_{\alpha}}{2} + \Omega_{\alpha\mu\nu} \tag*{(57)}

with Ωαμνgμν=0\Omega_{\alpha\mu\nu}g^{\mu\nu}=0. Then, a simple substitution of the latter into the decomposition (36) yields

R=R~+12ΩαμνΩαμν−12ΩαμνΩμνα+∇~μ(qμ−Qμ−4Sμ)(58)R = \widetilde{R} + \frac{1}{2}\Omega_{\alpha\mu\nu}\Omega^{\alpha\mu\nu} - \frac{1}{2}\Omega_{\alpha\mu\nu}\Omega^{\mu\nu\alpha} + \widetilde{\nabla}_{\mu}(q^{\mu} - Q^{\mu} - 4S^{\mu}) \tag*{(58)}

meaning that the Weyl part contributes only a total derivative term to the scalar curvature and the rest of the quadratic Weyl contributions disappear. Consequently, the variation of the scalar curvature is completely independent of the Weyl vector (as is also from the torsion vector) and takes the form

Pαμν=−Ωαμν+gανΩβγμgβγ(59)P_{\alpha\mu\nu} = -\Omega_{\alpha\mu\nu} + g_{\alpha\nu}\Omega_{\beta\gamma\mu}g^{\beta\gamma} \tag*{(59)}

We see therefore that both the scalar curvature and its variation (Palatini tensor) depend only on the non-Weyl part of nonmetricity. It is quite interesting that the same part is responsible for the non-topological character of the generalized Gauss-Bonnet term [16].

Setting up metric-affine theories in 2-dim

For general dimensions, out of the curvature tensor and its contractions one can form 16 quadratic parity even invariants [17] and the full curvature-quadratic action reads

S=12κ∫dnx−g[R+a1R2+a2RμνRμν+a3RμνRνμ+a4R˘μνR˘μν+a5R˘μνR˘νμ+a6R^μνR^μν+a7RμνR˘μν+a8RμνR˘νμ+a9RμνR^μν+a10R˘μνR^μν+a11RμναβRμναβ+a12RμναβRναμβ+a13RμναβRνμαβ+a14RμναβRανμβ+a15RμναβRμανβ+a16RμναβRαβμν](60)\begin{aligned} S = \frac{1}{2\kappa}\int d^{n}x\sqrt{-g}\big[ R &+ a_{1}R^{2} + a_{2}R_{\mu\nu}R^{\mu\nu} + a_{3}R_{\mu\nu}R^{\nu\mu} + a_{4}\breve{R}_{\mu\nu}\breve{R}^{\mu\nu} + a_{5}\breve{R}_{\mu\nu}\breve{R}^{\nu\mu} + a_{6}\widehat{R}_{\mu\nu}\widehat{R}^{\mu\nu} \\ &+ a_{7}R_{\mu\nu}\breve{R}^{\mu\nu} + a_{8}R_{\mu\nu}\breve{R}^{\nu\mu} + a_{9}R_{\mu\nu}\widehat{R}^{\mu\nu} + a_{10}\breve{R}_{\mu\nu}\widehat{R}^{\mu\nu} + a_{11}R_{\mu\nu\alpha\beta}R^{\mu\nu\alpha\beta} \\ &+ a_{12}R_{\mu\nu\alpha\beta}R^{\nu\alpha\mu\beta} + a_{13}R_{\mu\nu\alpha\beta}R^{\nu\mu\alpha\beta} + a_{14}R_{\mu\nu\alpha\beta}R^{\alpha\nu\mu\beta} + a_{15}R_{\mu\nu\alpha\beta}R^{\mu\alpha\nu\beta} + a_{16}R_{\mu\nu\alpha\beta}R^{\alpha\beta\mu\nu}\big] \tag*{(60)} \end{aligned}

In 2-dimensions, however, this number is sufficiently reduced as one may anticipate. Indeed, the very form of the curvature tensor (30) immediately implies that all quadratic curvature invariants would be linear combinations of R2R^{2}, RμνRμνR_{\mu\nu}R^{\mu\nu} and RμνRνμR_{\mu\nu}R^{\nu\mu}. Therefore, only the first 3 quadratic terms in the first line of (60) are relevant in 2-dimensions, and the rest of the terms will just shift the coefficients a1a_{1}, a2a_{2} and a3a_{3}. Then, after renaming the aforementioned parameters, we may then safely set the rest to zero, i.e. ai=0a_{i}=0, i=4,5,…,16i=4,5,\ldots,16. Therefore, in 2-dimensions the most general action that is quadratic in curvature9 is given by

L(2)=R+a1R2+a2RμνRμν+a3RμνRνμ(61)\mathcal{L}^{(2)} = R + a_{1}R^{2} + a_{2}R_{\mu\nu}R^{\mu\nu} + a_{3}R_{\mu\nu}R^{\nu\mu} \tag*{(61)}

namely, it is specified by three parameters only. Consequently, in 2-dimensions the most general Metric-Affine up to second-order curvature invariants, reads

S(2)[g,Γ]=12κ∫d2x−g[R+a1R2+a2RμνRμν+a3RμνRνμ](62)S^{(2)}[g,\Gamma] = \frac{1}{2\kappa}\int d^2x\sqrt{-g}\left[R+a_1R^2+a_2R_{\mu\nu}R^{\mu\nu}+a_3R_{\mu\nu}R^{\nu\mu}\right] \tag*{(62)}

It should be noted that there will be an undetermined scalar mode in this theory. This is not special in 2-dim, any MAG Theory, in arbitrary dimensions, constructed solely by the curvature tensor (and its various contractions) has this property [7]. This is easily seen from the fact that in any dimensions the curvature tensor is invariant under special projective transformations

Γμνλ↦Γμνλ+δμλ∂νϑ(63)\Gamma^\lambda_{\mu\nu}\mapsto\Gamma^\lambda_{\mu\nu}+\delta^\lambda_\mu\partial_\nu\vartheta \tag*{(63)}

with ϑ\vartheta being an arbitrary scalar function. This invariance then means that there will be an undetermined scalar mode which, however, can be fixed by gauge choice. Varying with respect to the metric and the independent affine connection we derive the field equations

−12gμν(R+a1R2+a2RμνRμν+a3RμνRνμ)+(1+2a1R)R(μν)+a2(RμαRνα+RαμRαν)+2a3Rα(μRν)α=0(64)-\frac{1}{2}g_{\mu\nu}\left(R+a_1R^2+a_2R_{\mu\nu}R^{\mu\nu}+a_3R_{\mu\nu}R^{\nu\mu}\right)+(1+2a_1R)R_{(\mu\nu)} +a_2\left(R_\mu{}^\alpha R_{\nu\alpha}+R_{\alpha\mu}R^\alpha{}_\nu\right)+2a_3R^\alpha{}_{(\mu}R_{\nu)\alpha}=0 \tag*{(64)}

and

(−∇α+2Sα)(−gΩλμαν)−−gΩλμβγSγδν=0(65)\left(-\nabla_\alpha+2S_\alpha\right)\left(\sqrt{-g}\Omega_\lambda{}^{\mu\alpha\nu}\right)-\sqrt{-g}\Omega_\lambda{}^{\mu\beta\gamma}S_{\gamma\delta}{}^\nu=0 \tag*{(65)}

respectively, where

Ωλμαν=∂(R+L(2))∂Rλμαν=(1+2a1R)δλ[αgν]μ+2a2Rμ[νδλα]+2a3δλ[αRν]μ(66)\Omega_\lambda{}^{\mu\alpha\nu}=\frac{\partial\left(R+\mathcal{L}^{(2)}\right)}{\partial R^\lambda{}_{\mu\alpha\nu}}=(1+2a_1R)\delta_\lambda^{[\alpha}g^{\nu]\mu}+2a_2R^{\mu[\nu}\delta_\lambda^{\alpha]}+2a_3\delta_\lambda^{[\alpha}R^{\nu]\mu} \tag*{(66)}

Let us firstly focus on the connection field equations. In 2-dimensions and given the form of the torsion tensor (18), the above connection field equations simplify considerably and take the form

∇α(−gΩλμαν)=0(67)\nabla_\alpha\left(\sqrt{-g}\Omega_\lambda{}^{\mu\alpha\nu}\right)=0 \tag*{(67)}

as can be trivially checked. We see that the terms explicit in torsion have canceled out. Let us observe that Ωλμαν\Omega_\lambda{}^{\mu\alpha\nu} can be written as

Ωλμαν=δλ[αHν]μ(68)\Omega_\lambda{}^{\mu\alpha\nu}=\delta_\lambda^{[\alpha}H^{\nu]\mu} \tag*{(68)}

where

Hνμ=(1+2a1R)gνμ+2a2Rμν+2a3Rνμ(69)H^{\nu\mu}=(1+2a_1R)g^{\nu\mu}+2a_2R^{\mu\nu}+2a_3R^{\nu\mu} \tag*{(69)}

Using this, the above connection field equations become

∇λ(−gHνμ)−δλν∇α(−gHαμ)=0(70)\nabla_\lambda\left(\sqrt{-g}H^{\nu\mu}\right)-\delta^\nu_\lambda\nabla_\alpha\left(\sqrt{-g}H^{\alpha\mu}\right)=0 \tag*{(70)}

Contracting in ν=λ\nu=\lambda it follows that

∇α(−gHαμ)=0(71)\nabla_\alpha\left(\sqrt{-g}H^{\alpha\mu}\right)=0 \tag*{(71)}

which when substituted back to (70) results in

∇λ(−gHνμ)=0(72)\nabla_\lambda\left(\sqrt{-g}H^{\nu\mu}\right)=0 \tag*{(72)}

implying that the tensor density −gHνμ\sqrt{-g}H^{\nu\mu} is covariantly constant on-shell. Now, let us turn our attention to the metric field equations. Taking the trace of (64) we obtain

a1R2+a2RμνRμν+a3RμνRνμ=0(73)a_1R^2+a_2R_{\mu\nu}R^{\mu\nu}+a_3R_{\mu\nu}R^{\nu\mu}=0 \tag*{(73)}

This is an algebraic equation for the traces of RμνR_{\mu\nu}. Indeed, if we denote RμνR_{\mu\nu} by the matrix A the above can be written as

a1(Tr⁡(A))2+a2Tr⁡(AAT)+a3Tr⁡(A2)=0(74)a_1\left(\operatorname{Tr}(A)\right)^2+a_2\operatorname{Tr}(AA^T)+a_3\operatorname{Tr}(A^2)=0 \tag*{(74)}

and may be used to eliminate one of them. Substituting this back to the metric field equations (64), we arrive at the simplified form

−12gμνR+(1+2a1R)R(μν)+a2(RμαRνα+RαμRαν)+2a3Rα(μRν)α=0-\frac{1}{2}g_{\mu\nu}R+(1+2a_1R)R_{(\mu\nu)}+a_2\left(R_\mu{}^\alpha R_{\nu\alpha}+R_{\alpha\mu}R^\alpha{}_\nu\right)+2a_3R^\alpha{}_{(\mu}R_{\nu)\alpha}=0

This is fairly general. Let us consider some special cases.

The case a2=a3=0a_2 = a_3 = 0.

We shall first consider the case where only the scalar curvature is included, that is, we set a2=a3=0a_2 = a_3 = 0. Then the metric and connection field equations reduce to

(1+2a1R)R(μν)−12(R+a1R2)gμν=0(75)(1 + 2a_1 R)R_{(\mu\nu)} - \frac{1}{2}(R + a_1 R^2)g_{\mu\nu} = 0 \tag*{(75)}
∇λ(−g(1+2a1R)gμν)=0(76)\nabla_\lambda\left(\sqrt{-g}(1 + 2a_1 R)g^{\mu\nu}\right) = 0 \tag*{(76)}

respectively. Taking the trace of the above metric field equations, for a1≠0a_1 \ne0 we deduce

R=0(77)R = 0 \tag*{(77)}

which, when substituted back to (75), implies that

R(μν)=0(78)R_{(\mu\nu)} = 0 \tag*{(78)}

Given that R=0R = 0, the connection field equations simplify to

∇λ(−ggμν)=0(79)\nabla_\lambda\left(\sqrt{-g}g^{\mu\nu}\right) = 0 \tag*{(79)}

It is important to note that in 2-dimensions the latter does not imply a vanishing non-metricity but rather that

Qαμν=Qα2gμν(80)Q_{\alpha\mu\nu} = \frac{Q_\alpha}{2}g_{\mu\nu} \tag*{(80)}

namely we have non-metricity of the Weyl-type. Since in 2-dim torsion is always of a purely vectorial nature we have 2 vectors completely specifying the non-Riemannian degrees of freedom. However, notice that since the action we started with consists only of the scalar curvature, enjoys the enlarged symmetry (43). Therefore we can choose ϑμ\vartheta_\mu and ξμ\xi_\mu appropriately so as to obtain vanishing torsion and non-metricity vectors. In these particular gauges we have

Sμ=0 , Qμ=0(81)S_\mu= 0 \ , \ Q_\mu= 0 \tag*{(81)}

and subsequently the full torsion and non-metricity vanish, viz.

Sαμν=0 , Qλμν=0(82)S_{\alpha\mu\nu} = 0 \ , \ Q_{\lambda\mu\nu} = 0 \tag*{(82)}

Then, given the post-Riemannian decomposition (36) of RR it follows that

R~=0(83)\widetilde{R} = 0 \tag*{(83)}

and consequently R~μν=0\widetilde{R}_{\mu\nu} = 0. Therefore there is no dynamics for this Theory. The dynamics arises once we include the rest of the Ricci tensor pieces or by supplementing the action with quadratic torsion and non-metricity invariants which manifestly brake the symmetry of the action under connection transformations of the form (43).

Full quadratic theory/including quadratic torsion and non-metricity terms

Theories built only out of the curvature are by themselves interesting however they are restrictive in some sense. Indeed, in the non-Riemannian setup of MAG, along with the curvature the space possesses also the geometric features of torsion and non-metricity. Additionally, as we saw earlier, Theories that are built solely from the curvature tensor have the special projective symmetry which implies the existence of an undetermined scalar mode. Therefore we must also supplement our gravitational action with invariants built out of torsion and non-metricity. In general (i.e. for any dimension) there exist 11 parity even quadratic invariants in torsion and non-metricity. Of course in 2-dimensions this number is reduced substantially.10To maintain simplicity and also make contact with the existing literature it suffices to consider the following two quadratic terms

SμSμ , QμQμ(84)S_\mu S^\mu\ , \ Q_\mu Q^\mu \tag*{(84)}

Let us note that adding more quadratic invariants to the action apart from the above two does not change the qualitative behavior of the Theory and only shifts the various coefficients by introducing more parameters. So, our extended quadratic Theory is now given by

S(2)[g,Γ]=12κ∫dmx−g[R+a1R2+a2RμνRμν+a3RμνRνμ+b1SμSμ+b2QμQμ](85)\mathcal{S}^{(2)}[g,\Gamma]=\frac{1}{2\kappa}\int d^m x\sqrt{-g}\left[R+a_1R^2+a_2R_{\mu\nu}R^{\mu\nu}+a_3R_{\mu\nu}R^{\nu\mu}+b_1S_\mu S^\mu+b_2Q_\mu Q^\mu\right] \tag*{(85)}

This is the starting point for obtaining non-trivial dynamics.

Special Cases

Let us now focus on the special case a2=a3=0a_2=a_3=0 which was studied in [6]. Here we shall verify the conclusions found there but in a holonomic frame. In this special case the connection field equations read

∇λ(−gHμν)−δλν∇α(−gHαμ)+−g(2b1Sμλν+4b2δλμQν)=0(86)\nabla_\lambda\left(\sqrt{-g}H^{\mu\nu}\right)-\delta^\nu_\lambda\nabla_\alpha\left(\sqrt{-g}H^{\alpha\mu}\right)+\sqrt{-g}\left(2b_1S^\mu{}_{\lambda}{}^\nu+4b_2\delta^\mu_\lambda Q^\nu\right)=0 \tag*{(86)}

where now Hμν=(1+2a1R)gμνH^{\mu\nu}=(1+2a_1R)g^{\mu\nu}. Tracing in λ=ν\lambda=\nu it follows that

∇α(−gHαμ)=−g(b1Sμ+4b2Qμ)(87)\nabla_\alpha\left(\sqrt{-g}H^{\alpha\mu}\right)=\sqrt{-g}\left(b_1S^\mu+4b_2Q^\mu\right) \tag*{(87)}

which when substituting back to the above brings it to the form

1−g∇λ(−gHμν)−b1δλνSμ+8b2δλμQν=0(88)\frac{1}{\sqrt{-g}}\nabla_\lambda\left(\sqrt{-g}H^{\mu\nu}\right)-b_1\delta^\nu_\lambda S^\mu+8b_2\delta^\mu_\lambda Q^\nu=0 \tag*{(88)}

Taking the antisymmetric part of the latter in μ,ν\mu,\nu fixes

b1Sμ=8b2Qμ(89)b_1S_\mu=8b_2Q_\mu \tag*{(89)}

Substituting this into (88) and using the Leibniz rule for the covariant derivative, remembering also the definition of non-metricity we find

−Qλ2(1+2a1R)gμν+2a1(∂λR)gμν+(1+2a1R)Qλμν=8b2δλμQν(90)-\frac{Q_\lambda}{2}(1+2a_1R)g^{\mu\nu}+2a_1(\partial_\lambda R)g^{\mu\nu}+(1+2a_1R)Q_\lambda{}^{\mu\nu}=8b_2\delta^\mu_\lambda Q^\nu \tag*{(90)}

Contracting this one time with gμνg_{\mu\nu} and another in ν=λ\nu=\lambda we obtain

Qμ=a12b2∂μR(91)Q_\mu=\frac{a_1}{2b_2}\partial_\mu R \tag*{(91)}

and

qμ=[12+8b21+2a1R]Qμ=[12+8b21+2a1R]a12b2∂μR=∂μ(a14b2R+2ln⁡(1+2a1R))(92)q_\mu=\left[\frac{1}{2}+\frac{8b_2}{1+2a_1R}\right]Q_\mu=\left[\frac{1}{2}+\frac{8b_2}{1+2a_1R}\right]\frac{a_1}{2b_2}\partial_\mu R=\partial_\mu\left(\frac{a_1}{4b_2}R+2\ln(1+2a_1R)\right) \tag*{(92)}

In addition, combining (89) with (91) it follows that

Sμ=4a1b1∂μR(93)S_\mu=\frac{4a_1}{b_1}\partial_\mu R \tag*{(93)}

Consequently, all three torsion and non-metricity vectors are proportional to one another and powered from the scalar curvature. Note that all three vectors are exact. Using the above forms of these vectors we can exactly solve (90) for non-metricity

Qαμν=a1[14b2−21+2a1R](∂αR)gμν+4a11+2a1Rgα(μ∂ν)R(94)Q_{\alpha\mu\nu}=a_1\left[\frac{1}{4b_2}-\frac{2}{1+2a_1R}\right](\partial_\alpha R)g_{\mu\nu}+\frac{4a_1}{1+2a_1R}g_{\alpha(\mu}\partial_{\nu)}R \tag*{(94)}

given also the pure vectorial type of torsion in 2-dim the full torsion reads

Sμνα=8a1b1gα[ν∂μ]R(95)S_{\mu\nu\alpha}=\frac{8a_1}{b_1}g_{\alpha[\nu}\partial_{\mu]}R \tag*{(95)}

Therefore we see that the entire non-metricity tensors are given in terms of the derivative of the scalar curvature. Let us now find the evolution equation that the latter satisfies. To this end, we now turn our attention to the metric field equations, which in this case take the form

(1+2a1R)Rμν−12(R+a1R2)gμν+b1(SμSν−12SαSαgμν)+b2(QμQν−12QαQαgμν)−2b2gμν1−g∂α(−gQα)=0(96)\left(1+2a_1R\right)R_{\mu\nu}-\frac{1}{2}\left(R+a_1R^2\right)g_{\mu\nu}+b_1\left(S_\mu S_\nu-\frac{1}{2}S_\alpha S^\alpha g_{\mu\nu}\right)+b_2\left(Q_\mu Q_\nu-\frac{1}{2}Q_\alpha Q^\alpha g_{\mu\nu}\right)-2b_2g_{\mu\nu}\frac{1}{\sqrt{-g}}\partial_\alpha\left(\sqrt{-g}Q^\alpha\right)=0 \tag*{(96)}

Taking the trace and using (91) it follows that

R2−2−g∂α(−ggαβ∂βR)=0(97)R^2-\frac{2}{\sqrt{-g}}\partial_\alpha\left(\sqrt{-g}g^{\alpha\beta}\partial_\beta R\right)=0 \tag*{(97)}

This is a dynamical Klein-Gordon-like equation for the scalar curvature, which given also the relations (94) and (95) gives dynamics to torsion and nonmetricity as well. Therefore we have non-trivial dynamics! It is worth stressing that, quite remarkably, all parameters have disappeared from the final equation.

Case α3=−a2\alpha_3=-a_2

Let us now extend our previous considerations and allow for non-vanishing a2a_2 and a3a_3 that obey the relation a3=−a2a_3=-a_2. This choice picks out the homothetic curvature part from the quadratic Ricci terms, namely the tensor HμνH^{\mu\nu} now becomes

Hμν=(1+2a1R)gμν+a2R^μν(98)H^{\mu\nu}=\left(1+2a_1R\right)g^{\mu\nu}+a_2\widehat{R}^{\mu\nu} \tag*{(98)}

Then the connection field equations our formalistically the same with (86) but for this form of HμνH^{\mu\nu}. Then similarly to the previous we first take the trace of the connection field equations in λ=ν\lambda=\nu which gives

∇α(−gHαμ)=−g(b1Sμ+4b2Qμ)(99)\nabla_\alpha\left(\sqrt{-g}H^{\alpha\mu}\right)=\sqrt{-g}\left(b_1S^\mu+4b_2Q^\mu\right) \tag*{(99)}

which again when substituted back to (86) leaves us with

1−g∇λ(−gHνμ)−b1δλνSμ+8b2δλ[μQν]=0(100)\frac{1}{\sqrt{-g}}\nabla_\lambda\left(\sqrt{-g}H^{\nu\mu}\right)-b_1\delta^\nu_\lambda S^\mu+8b_2\delta^{[\mu}_\lambda Q^{\nu]}=0 \tag*{(100)}

Taking the symmetric part (νμ)(\nu\mu) of the latter, expanding the covariant derivative, and using (98) we obtain

−Qλ2(1+2a1R)gμν+2a1(∂λR)gμν+(1+2a1R)Qλμν=b1δλ(μSν)(101)-\frac{Q_\lambda}{2}\left(1+2a_1R\right)g^{\mu\nu}+2a_1\left(\partial_\lambda R\right)g^{\mu\nu}+\left(1+2a_1R\right)Q_\lambda{}^{\mu\nu}=b_1\delta_\lambda^{(\mu}S^{\nu)} \tag*{(101)}

Contracting with gμνg_{\mu\nu} it follows that

Sλ=4a1b1∂λR(102)S_\lambda=\frac{4a_1}{b_1}\partial_\lambda R \tag*{(102)}

Using this, if we now contract (101) in λ=ν\lambda=\nu we get

(1+2a1R)(Qμ−12Qμ)=b1Sμ=4a1∂μR(103)\left(1+2a_1R\right)\left(Q^\mu-\frac{1}{2}Q^\mu\right)=b_1S^\mu=4a_1\partial^\mu R \tag*{(103)}

Furthermore, if we now take the antisymmetric part [μν][\mu\nu] of the connection field equations (100) and subsequently contract in λ=μ\lambda=\mu we obtain

a21−g∂μ(−gR^μν)=2a1∂νR−4b2Qν(104)a_2\frac{1}{\sqrt{-g}}\partial_\mu\left(\sqrt{-g}\widehat{R}^{\mu\nu}\right)=2a_1\partial^\nu R-4b_2Q^\nu \tag*{(104)}

Since R^μν\widehat{R}^{\mu\nu} is antisymmetric in its indices, if we act with another derivative, it follows that

2b21−g∂μ(−gQμ)=a1□R(105)2b_2\frac{1}{\sqrt{-g}}\partial_\mu\left(\sqrt{-g}Q^\mu\right)=a_1\Box R \tag*{(105)}

Let us now focus on the metric field equations which read

−12(R+a1R2+a22R^αβR^αβ)gμν+(1+2a1R)R(μν)+a2R^μαR^νβgαβ+b1(SμSν−12SαSαgμν)+b2(QμQν−12QαQαgμν)−2b2gμν1−g∂α(−gQα)=0-\frac{1}{2}\left(R+a_{1}R^{2}+\frac{a_{2}}{2}\hat{R}_{\alpha\beta}\hat{R}^{\alpha\beta}\right)g_{\mu\nu}+(1+2a_{1}R)R_{(\mu\nu)}+a_{2}\hat{R}_{\mu\alpha}\hat{R}_{\nu\beta}g^{\alpha\beta} +b_{1}\left(S_{\mu}S_{\nu}-\frac{1}{2}S_{\alpha}S^{\alpha}g_{\mu\nu}\right)+b_{2}\left(Q_{\mu}Q_{\nu}-\frac{1}{2}Q_{\alpha}Q^{\alpha}g_{\mu\nu}\right)-2b_{2}g_{\mu\nu}\frac{1}{\sqrt{-g}}\partial_{\alpha}\left(\sqrt{-g}Q^{\alpha}\right)=0

Taking the trace we obtain

a1R2+a22R^αβR^αβ=4b21−g∂α(−gQα)(106)a_{1}R^{2}+\frac{a_{2}}{2}\hat{R}_{\alpha\beta}\hat{R}^{\alpha\beta}=4b_{2}\frac{1}{\sqrt{-g}}\partial_{\alpha}\left(\sqrt{-g}Q^{\alpha}\right) \tag*{(106)}

which in view of (105) can be written as

2□gR=R2+a22a1R^αβR^αβ(107)2\Box_{g}R=R^{2}+\frac{a_{2}}{2a_{1}}\hat{R}_{\alpha\beta}\hat{R}^{\alpha\beta} \tag*{(107)}

We see therefore that the collection of (104) and (107) describes the propagation of a scalar (RR) and a vector (QμQ^{\mu}) coupled with each other. Given the solutions of RR and QμQ^{\mu} one can then compute the full torsion and non-metricity tensors. This result is quite interesting; however, it goes beyond the purpose of this study to further investigate the solutions of this Theory. The purpose was to show how the inclusion of torsion and nonmetricity, provides non-trivial dynamics for 2-dimensional gravity.

Conclusions

We have studied the structure of 2-dimensional Metric-Affine Geometries consisting of curvature, torsion and non-metricity. We have explicitly proved that in such geometries, in contrast to the Riemannian case, the Riemann tensor is not completely determined by the scalar curvature, but rather by the Ricci tensor which is now asymmetric. Furthermore, we have found a new symmetry for the generalized scalar curvature in 2-dimensions, the invariance under the connection transformations (43).

Moving on to the physical applications, we used the above results to study 2-dimensional Metric-Affine Theories in vacuum. Considering quadratic Yang-Mills type Theories in curvature, torsion and nonmetricity, we explicitly showed how the extended geometry offers the possibility to have propagating degrees of freedom. In particular, we presented two examples where torsion and nonmetricity are propagating in 2-dimensions.

It would be interesting to study the integrability and black hole solutions of our general theory (85), compared to the Poincare case [18]. Needless to say, the theoretical framework we developed here can be applied to study the whole plethora of 2-dim gravity applications such as holography [19, 20, 21], string theory [22, 23, 24], cosmology [25, 26, 27] and so on.

Acknowledgments

This work was supported by the Istituto Nazionale di Fisica Nucleare (INFN), Sezioni di Napoli, Iniziative Specifiche SKY.

References

  1. [1]C. Teitelboim. Gravitation and hamiltonian structure in two space-time dimensions. Physics Letters B, 126:41–45, 1983.
  2. [2]R. Jackiw. Lower dimensional gravity. Nuclear Physics B, 252:343–356, 1985.DOI
  3. [3]D. Grumiller, W. Kummer, and D. V. Vassilevich. Dilaton gravity in two-dimensions. Phys. Rept., 369:327–430, 2002.
  4. [4]M. O. Katanaev and I. V. Volovich. Two-Dimensional Gravity with Dynamical Torsion and Strings. Annals Phys., 197:1, 1990.DOI
  5. [5]Eckehard W. Mielke, Frank Gronwald, Yuri N. Obukhov, Romualdo Tresguerres, and Friedrich W. Hehl. Towards complete integrability of two-dimensional Poincare gauge gravity. Phys. Rev. D, 48:3648–3662, 1993.pubmed.ncbi.nlm.nih.gov/10016639
  6. [6]Yuri N. Obukhov. Two-dimensional metric affine gravity. Phys. Rev. D, 69:064009, 2004.DOI
  7. [7]J A Schouten. Ricci-calculus. An introduction to tensor analysis and its geometrical applications, 1954.
  8. [8]Friedrich W Hehl, J Dermott McCrea, Eckehard W Mielke, and Yuval Ne’eman. Metric-affine gauge theory of gravity: field equations, noether identities, world spinors, and breaking of dilation invariance. Physics Reports, 258(1-2):1–171, 1995.DOI
  9. [9]Damianos Iosifidis. Metric-affine gravity and cosmology/aspects of torsion and non-metricity in gravity theories. arXiv preprint arXiv:1902.09643, 2019.arxiv.org/abs/1902.09643
  10. [10]Friedrich W. Hehl and G. David Kerlick. Metric-affine variational principles in general relativity. I. Riemannian space-time. Gen. Rel. Grav., 9(8):691–710, 1978.DOI
  11. [11]Damianos Iosifidis. Linear transformations on affine-connections. Classical and Quantum Gravity, 2020.DOI
  12. [12]Élie Cartan. Sur une généralisation de la notion de courbure de Riemann et les espaces à torsion. Comptes rendus de l’Académie des sciences, 174:593–595, 1922.
  13. [13]Élie Cartan. Sur les variétés à connexion affine et la théorie de la relativité généralisée (première partie). Annales scientifiques de l’École Normale Supérieure, 40:325–412, 1923.
  14. [14]Élie Cartan. Sur les variétés à connexion affine et la théorie de la relativité généralisée (première partie) (suite). Annales scientifiques de l’École Normale Supérieure, 41:1–25, 1924.
  15. [15]Élie Cartan. Sur les variétés à connexion affine et la théorie de la relativité généralisée (deuxième partie). Annales scientifiques de l’École Normale Supérieure, 42:17–88, 1925.DOI
  16. [16]O. V. Babourova and B. N. Frolov. Gauss–bonnet type identity in weyl-cartan space. International Journal of Modern Physics A, 12:3665, 1997.DOI
  17. [17]Dmitri Vassiliev. Quadratic metric affine gravity. Annalen Phys., 14:231–252, 2005.
  18. [18]Eckehard W. Mielke, Frank Gronwald, Yuri N. Obukhov, Romualdo Tresguerres, and Friedrich W. Hehl. Towards complete integrability of two-dimensional poincare gauge gravity. Physical Review D, 48:3648–3662, 1993.pubmed.ncbi.nlm.nih.gov/10016639
  19. [19]Ahmed Almheiri and Joseph Polchinski. Models of AdS2 backreaction and holography. Journal of High Energy Physics, 2015(11):014, 2015.
  20. [20]Juan Maldacena, Douglas Stanford, and Zhenbin Yang. Conformal symmetry and its breaking in two dimensional nearly anti-de-Sitter space. Progress of Theoretical and Experimental Physics, 2016(12):12C104, 2016.arxiv.org/abs/1606.01857
  21. [21]Thomas G. Mertens and Gustavo J. Turiaci. Solvable models of quantum black holes: A review on jackiw–teitelboim gravity. Living Reviews in Relativity, 26, 2023.DOI
  22. [22]Philippe Di Francesco, Paul H. Ginsparg, and Jean Zinn-Justin. 2d gravity and random matrices. Physics Reports, 254(1–2):1–133, 1995.
  23. [23]Jacques Distler and Hikaru Kawai. Conformal field theory and 2d quantum gravity. Nuclear Physics B, 321:509–527, 1989.
  24. [24]Francois David. Conformal field theories coupled to 2d gravity in the conformal gauge. Modern Physics Letters A, 3:1651–1656, 1988.DOI
  25. [25]Ritabrata Bhattacharya, K. Narayan, and Partha Paul. Cosmological singularities and 2-dimensional dilaton gravity. Journal of High Energy Physics, 2020(8):062, 2020.arxiv.org/abs/2006.09470
  26. [26]Francisco D. Mazzitelli and Jorge G. Russo. Dilaton quantum cosmology in two dimensions. Physical Review D, 47:4490–4495, 1993.pubmed.ncbi.nlm.nih.gov/10015448
  27. [27]Sukanta Bose and Sayan Kar. Exact solutions in two-dimensional string cosmology with back reaction. Phys. Rev. D, 56:R4444–R4448, 1997.DOI

Paper details

vlnc:W_rvraupkn
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents