Light fermions in quantum gravity

We study the impact of quantum gravity, formulated as a quantum field theory of the metric, on chiral symmetry in a fermionic matter sector. We specifically address the question as to whether metric fluctuations can induce chiral symmetry breaking and bound state formation. Our results based on the functional Renormalization Group indicate that chiral symmetry is left intact even at strong gravitational coupling. In particular, we find that asymptotically safe quantum gravity where the gravitational couplings approach a non-Gaussian fixed point generically admits universes with light fermions. Our results thus further support quantum gravity theories built on fluctuations of the metric field such as the asymptotic-safety scenario. A study of chiral symmetry breaking through gravitational quantum effects may serve as a significant benchmark test also for other quantum gravity scenarios, since a completely broken chiral symmetry at the Planck scale would not be in accordance with the observation of light fermions in our universe. We demonstrate that this elementary observation already imposes constraints on a generic UV completion of gravity.

We study the impact of quantum gravity, formulated as a quantum field theory of the metric, on chiral symmetry in a fermionic matter sector. We specifically address the question as to whether metric fluctuations can induce chiral symmetry breaking and bound state formation. Our results based on the functional Renormalization Group indicate that chiral symmetry is left intact even at strong gravitational coupling. In particular, we find that asymptotically safe quantum gravity where the gravitational couplings approach a non-Gaußian fixed point generically admits universes with light fermions. Our results thus further support quantum gravity theories built on fluctuations of the metric field such as the asymptotic-safety scenario. A study of chiral symmetry breaking through gravitational quantum effects may serve as a significant benchmark test also for other quantum gravity scenarios, since a completely broken chiral symmetry at the Planck scale would not be in accordance with the observation of light fermions in our universe. We demonstrate that this elementary observation already imposes constraints on a generic UV completion of gravity.

I. INTRODUCTION
Any phenomenologically relevant theory of quantum gravity has to satisfy a number of physical requirements. In addition to internal or mathematical consistency, observations demand that such a theory has to provide for the existence of a semi-classical limit. Another phenomenological requirement is the existence of light matter (compared with the Planck scale). In many studies of quantum gravity, matter is either ignored, or treated rather as a kinematic degree of freedom. For instance in string compactifications, information about the existence of light matter is drawn from anomaly-cancelation arguments.
As soon as matter and interactions are taken into account dynamically, the existence of light matter is far from being self-evident: From a bottom-up viewpoint, the regime of quantum gravity is naturally defined as the domain where gravity fluctuations become as relevant as matter fluctuations. If gravity becomes even strongly interacting, its dynamical influence on the matter sector may even be similar to strong matter correlations as induced by the other forces of particle physics.
This question becomes particularly paradigmatic in the case of fermions. In standard particle physics scenarios, fermions are light because their mass is protected by chiral symmetry. So far, the mass of all fermionic matter content in the universe is associated with the phenomenon of chiral symmetry breaking. This mass generation is a consequence of strong correlations among fermions. In the electroweak sector, these strong correlations are provided by the Higgs sector, whereas gluoninduced interactions are responsible for mass generation in the strong interactions.
Gravity at first sight seems to have some similarity to these particle physics examples, as it is similar to Yang-Mills theories in many respects. Therefore, the present work is devoted to a first investigation whether gravitational fluctuations can induce chiral symmetry breaking in a chiral fermion sector. If such a mechanism exists, any theory of quantum gravity would have to control it or evade it in a natural way. Since such a mechanism would be active at or above the Planck scale, it would naturally force fermions to have masses on the order of the Planck scale in contrast to observation.
In our work, gravitational fluctuations are parameterized as fluctuations of the metric field. This is at least an effective description below or even near the Planck scale. In the context of Weinberg's asymptotic safety scenario for quantum gravity [1], this can even be a consistent description up to arbitrarily short distances. As asymptotic safety is built on the existence of a non-Gaußian fixed point in the gravitational couplings, i.e., an interacting UV limit, the interplay between the gravitational sector and the chiral fermion sector is of particular interest.
Evidence for the existence of such a fixed point has been provided in different approaches [2][3][4][5][6]. Also causal dynamical triangulations (see e.g. [7] for a review) may be interpreted as providing evidence for this scenario. In particular the functional Renormalization Group (RG), following the pioneering work of Reuter [8] has facilitated numerous studies supporting asymptotic safety in gravity , for reviews see [35][36][37][38]. First steps in the investigation of the compatibility of asymptotically safe quantum gravity with quantized matter have been performed in [39,40]. Here the backreaction of fermionic and bosonic matter onto the gravitational fixed-point properties were investigated. The requirement of the existence of a physically admissible fixed point with a positive value for the Newton coupling then imposes constraints on the matter content of the universe. Most importantly, the matter content of the standard model of particle physics is compatible with asymptotically safe quantum gravity within the investigated truncation [39]. (For work on the effect of the gravitational fixed point on gauge theories see [41][42][43][44]. Studies dealing with a possible solution of the triviality problem in the Higgs sector through the coupling to gravity have been performed in [45][46][47]. ) In this work, we explore the gravity-induced interactions and potentially strong correlations in a chiral fermion sector for the first time using the functional RG. As gravity shares some features with Yang-Mills theories, we are motivated by recent advances in QCD, where chiral symmetry breaking can be understood as the consequence of a critical dynamics among the chiral fermions which is triggered by gluon-induced strong correlations [48,49]; for a successful determination of the critical temperature for chiral symmetry breaking in QCD using the functional RG, see [50][51][52][53]. It is tempting to speculate that gravity might facilitate a similar mechanism in a strong-coupling regime. If so, such a mechanism might exhibit a dependence on control parameters of the theory such as the fermion flavor number N f . In fact, chiral quantum phase transitions as a function of N f have been observed in many systems such as many flavor QCD, QED 3 or the 3-dimensional Thirring model [49,[54][55][56][57][58][59][60][61].
Our study of the gravitationally-stimulated chiral dynamics is based on a truncation of the full quantum effective action that concentrates on a Fierz-complete basis of chiral fermionic four-point functions in the point-like limit. Again, this is motivated by analogous studies in other theories, where such an ansatz provides both for an intuitive as well as quantitatively meaningful approach to chiral symmetry breaking.
As our main result, we do not find any indications for gravitationally-stimulated chiral symmetry breaking within this ansatz. Whereas the Gaußian fermion matter fixed point turns into an interacting non-Gaußian one, the universality properties of this fixed point receive rather small modifications in the asymptotic-safety scenario if fermionic degrees of freedom dominate the matter sector. As a general pattern, gravitational binding which would favor chiral symmetry breaking is compensated by gravitational contributions to anomalous scaling of the fermion interactions. Within this minimal truncation, we can therefore conclude that asymptotic safety is well compatible with the existence (and observation) of light fermions despite an interacting UV sector which stimulates fermion self-interactions. This paper is structured as follows: We will introduce the functional RG as a tool for our investigation in Sect. II, and introduce the system that we study in Sect. III. Results concerning the asymptotic-safety scenario as well as for general classes of effective theories of quantum gravity are presented in Sect. IV. Finally, we conclude in Sect. V. Technical details can be found in appendix V.

II. FUNCTIONAL RG
The functional RG facilitates the non-perturbative evaluation of full correlation functions. Within the formulation used here, we focus on the scale-dependent effective average action which is the generating functional of 1PI correlators that include all fluctuations from the ultraviolet (UV) down to the infrared (IR) scale k. At k = 0, Γ k coincides with the standard effective action Γ = Γ k=0 . The scale dependence of the effective average action is governed by the Wetterich equation [62] Here, ∂ t = k ∂ k , Γ (2) k is the second functional derivative of Γ k with respect to the fields, and R k is an IR regulator function. Accordingly the right-hand side of the Wetterich equation depends on the full (field-dependent) regularized propagator Γ , which is matrixvalued in field space. The supertrace STr contains a trace over the spectrum of the full propagator in all appropriate indices (i.e. on a flat background in the absence of classical background fields it translates into a momentum integral and a trace over Lorentz and internal indices). For Grassmann-valued fields the supertrace involves an additional negative sign. For reviews on the functional RG and the Wetterich equation see [63][64][65][66][67][68].
Since quantum fluctuations generate all possible operators compatible with the symmetries and the field content of the microscopic action, the effective (average) action lives in theory space which is spanned by all these operators. Expanding Γ k = n g n (k)O n into the infinite sum of all operators O with running couplings g n (k) allows to rewrite the Wetterich equation as an infinite tower of coupled differential equations. In practice, a truncation of theory space to a smaller subspace is necessary. In the case of asymptotically safe quantum gravity, numerous studies have shown a high degree of stability of truncations of the Einstein-Hilbert type and beyond. The results appear to converge under generalizations of the truncation in various ways, as well as under a change of the regularization scheme, thus providing strong support for the existence of the NGFP in full theory space [8-19, 21-31, 34].
The perturbative non-unitarity of theories involving higher derivative operators does not directly apply here: Indeed unitarity has to be reinvestigated within the nonperturbative setting and can remain intact within the asymptotic-safety scenario (see e.g. corresponding discussions in [15,27,36,37]).
In the following, we investigate the compatibility of the asymptotic-safety scenario with the existence of light fermionic matter. More generally, our results can be applied to generic effective theories of quantum gravity formulated in terms of fluctuations of the metric field and their interplay with chiral fermions.

III. CHIRAL FERMIONS IN QUANTUM EINSTEIN GRAVITY
In the case of QCD-like theories, many studies based on functional methods suggest that chiral symmetry is broken for gauge couplings larger than a critical value [48,49,[69][70][71]. In direct analogy it is tempting to expect that there exists a critical value for the Newton coupling at which metric fluctuations break chiral symmetry. This would agree with the picture that gravity is always attractive and thus should support fermionic binding phenomena.
We investigate this scenario in a specific fermionic system with a chiral SU(N f ) L × SU(N f ) R symmetry. We parameterize this system by an action of the form These fermionic self-interactions form a complete basis of four-fermion operators in the pointlike limit which are invariant under the chiral symmetry. The parentheses indicate expressions with fully contracted Dirac indices. The γ matrices are understood to live in curved spacetime, being related to their flat-space cousins γ a by the vielbein: γ µ = e µ a γ a . Flavor indices are denoted by Latin letters i, j, ... and run from 1 to N f . The covariant derivative ∇ µ is given by where ω ab µ denotes the spin connection. The latter can be determined in terms of the Christoffel connection by demanding that ∇ µ e µ a = 0. All other non-derivative SU(N f ) L × SU(N f ) R symmetric four-fermion operators, e.g. a flavor non-singlet scalar-pseudo-scalar interaction, can be transformed into some combination of the above ones by a Fierz transformation. Including all of the basis operators implies that we cover all possible channels for chiral symmetry breaking in the point-like, i.e. momentum-independent, limit. This is important, as gravity might pick one specific channel to induce the breaking of chiral symmetry. This ansatz of operators in the chiral sector is strongly motivated by similar lines of reasoning in QCD-like theories or other strongly-correlated fermionic systems. There, the ansatz is capable of describing the approach to chiral symmetry breaking qualitatively as well as quantitatively. Of course, gravity may choose to break chiral symmetry in a fashion differing from Yang-Mills theory; potential further mechanisms will be briefly outlined below.
In d > 2 dimensional spacetime, four-fermion interactions are perturbatively non-renormalizable. In an RG language this translates into the fact that these couplings are irrelevant at the Gaußian fixed point. Accordingly they have to be set to zero at the UV scale in a fundamental theory in a perturbative setting. Even if being zero initially, such couplings are generated by interactions in the flow towards the IR, for instance, in the context of QCD or also when coupled to gravity. The flow of such fermionic self-interactions can then provide indications for chiral symmetry breaking. Beyond perturbation theory, non-Gaußian fixed points may exist which could allow to construct a non-perturbatively renormalizable (asymptotically safe) theory with non-vanishing four-fermion interactions. In 2 < d < 4 dimensions, for instance, the Gross-Neveu model provides for a simple and well-understood example of asymptotic safety [72].
At this point, we can already discuss the relation between chiral symmetry breaking and the fixed-point structure of the four-fermion couplingsλ ± [49]. For this, we introduce the dimensionless renormalized couplings λ ± and the fermionic anomalous dimension η ψ : Due to the one-loop form of the Wetterich equation, the β functions for λ ± have the generic form Herein the first term arises from dimensional (and anomalous) scaling. The quadratic contributions follow from a purely fermionic two-vertex diagram (cf. diagram (2c) in Fig. 2 below). A tadpole contribution ∼ dλ ± may also exist, as well as a λ ± -independent part ∼ e which results from the coupling to other fields, for instance, arising from the covariant derivative in the kinetic term in Eq. (2). The numerical values for a, b and c depend on the regulator, the contributions d and e will also depend on further couplings. Specific representations will be given below. Fixing all other couplings, the β function of a given fermionic coupling β λ± = ∂ t λ ± as a function of λ ± is a parabola with two fixed points λ * ± where β λ± (λ * ± ) = 0. The coupled system of two fermionic β functions then admits 2 2 fixed points which need not necessarily all be real.
In order to illustrate the relevance of this fermionic fixed-point structure, let us concentrate on the λ + channel and perform a Fierz transformation to the standard scalar-pseudo-scalar channel, where (ψ i ψ j ) 2 ≡ψ i ψ jψj ψ i , and similarly for the pseudoscalar channel. Equation (7) is an exact Fierz identity if the couplings satisfy The fixed-point structure in the (V + A) channel hence implies a corresponding fixed-point structure in the standard scalar-pseudo-scalar channel, where chiral symmetry breaking is expected to be visible. In Fig. 1, the β function β λσ = ∂ t λ σ for this chiral channel is sketched for vanishing gravitational coupling. The two crossings of the parabola with the λ σ axis indicate the two fixed points. The Gaußian fixed point at λ * σ = 0 is IR attractive whereas the non-Gaußian fixed point λ * σ = λ σ,cr > 0 is UV attractive (arrows indicate the flow towards In the absence of further interactions, the parabola-type β function exhibits two fixed points, the Gaußian fixed point at λ * σ = 0 and a non-Gaußian fixed point at λ * σ = λσ,cr > 0. Arrows indicate the RG flow towards the IR. Further interactions can shift the parabola (dashed line) and lead to an annihilation of the two fixed points (dotdashed line). This annihilation is an indicator for an approach to chiral criticality. Such a scenario typically occurs in QCD-like theories [49,[73][74][75].
the IR). If the system is in the attractive domain of the Gaußian fixed point, fermionic correlations die out quickly during the RG flow and the system remains in the chirally symmetric phase.
If the system starts to the right of the non-Gaußian fixed point λ σ > λ σ,cr , the coupling runs towards very large values and actually diverges at a finite RG scale. Whereas the point-like fermionic truncation breaks down here, a transition to a bosonized description with a boson being related to a fermion bilinear identifies the chiral coupling with the mass term of a chiral bosonic field λ σ ∼ 1/m 2 φ . The divergence of the fermionic coupling therefore indicates that the bosonic mass term drops below zero which is a signature for symmetry breaking. It characterizes the onset of the effective chiral potential to acquire a Mexican-hat form. In NJL-type models, the non-Gaußian fixed point is identical to the critical value of the coupling λ σ,cr required to put the system into the chiral symmetry broken phase.
A new feature comes about by coupling the fermionic system to another interaction, e.g. a gauge theory. Then, an originally weakly coupled fermion sector near the Gaußian fixed point can dynamically be driven to chiral criticality. For this, the Gaußian fixed point needs to annihilate with the non-Gaußian fixed point, such that the corresponding β function for λ σ becomes completely negative, as indicated by the dashed and dotdashed line in Fig. 1. Fixed-point annihilation thus is an indicator for an approach to chiral criticality. This scenario is indeed realized, e.g. in QCD-like and other theories [49,[73][74][75]. With regard to the structure of Eq. (6), the last two terms together with the gravitational contribution to η ψ , in principle, have the potential to destabilize the fermionic fixed points and drive the fermionic system to criticality. Whether or not this happens is a quantitative question that has to be addressed by integrating out gravitational fluctuations in a given quantum gravity theory.
For the gravitational part we work in the background field formalism [76], where the full metric is split according to where this split does not imply that we consider only small fluctuations around e.g. a flat background. Within the FRG approach we have access to physics also in the fully non-perturbative regime. This formalism, being highly useful in non-abelian gauge theories (see, e.g. [65,66]), is mandatory in gravity, since the background metric allows for a meaningful notion of "highmomentum" and "low-momentum" modes as implied by the spectrum of the background covariant Laplacian. The desired feature of background independence naively seems spoiled in this way; however, the background formalism turns out to be merely a technical tool, leaving physical results unaffected and thus independent of the background [35].
Following standard approximations we work within a single-metric truncation, i.e. we set g µν =ḡ µν after evaluating Γ (2) k . As suggested in recent studies [31,77] qualitative results in the Einstein-Hilbert sector are not affected by this approximation.
On a general (curved) spacetime our truncation then reads: where the Einstein-Hilbert term and the gauge-fixing term are given by: with Herein,κ = (32πG N ) − 1 2 is related to the bare Newton constant G N . We denote the cosmological constant byλ without any subscript. It should not be confused with the four-fermion couplingsλ ± .
A ghost sector corresponding to Faddeev-Popov gauge fixing is implicitely understood here.
For the vielbein we work in the symmetric vielbein gauge [78,79] such that O(4) ghosts do not occur. This gauge also allows to reexpress vielbein fluctuations purely in terms of metric fluctuations.
Details on the second functional derivative of the effective action can be found in appendix V.
The fermionic self-interactions ∼λ ± do not directly contribute to the pure gravity flow. Technically this is, because no one-loop diagram containing a fermionic fourpoint vertex can be formed that has only gravitons on external legs. Hence, the Einstein-Hilbert sector receives contributions only from the minimally coupled kinetic fermion term as determined in [39], where the approximation Z ψ = 1 has been used.
Our new task here is to compute the gravitationally stimulated flow of the fermion interactionsλ ± . For this, a flat-background calculation, settingḡ µν = δ µν , is fully sufficient, and technically favorable.
In the following, we ignore a non-trivial running of the fermion kinetic term by setting Z ψ = 1. In Yang-Mills theory, this is justified in the Landau gauge α → 0, where this flow of Z ψ vanishes in a similar truncation [74]. In gravity, however, the flow of Z ψ does receive nontrivial contributions even in the Landau-deWitt gauge ρ → α → 0. This marks a first difference between gravity and Yang-Mills theory in this context. We keep track of this difference by maintaining the dependence of the flow on the fermion anomalous dimension η ψ . Whereas η ψ = 0 in the present approximation, we will later treat η ψ as a free parameter to explore possible consequences of this difference between Yang-Mills theory and gravity.
We decompose the metric fluctuations h µν into a transverse traceless tensor, a transverse vector, a scalar, and the trace part. We then specialize to Landau deWitt gauge, ρ → α → 0, which implies that only the transverse traceless tensor h TT µν and the trace mode h =ḡ µν h µν can contribute to the flow of the fermionic couplings (see also [29]).
Splitting Γ (2) k +R k = P k +F k , where all field-dependent terms enter the fluctuation matrix F k , we may now expand the right-hand side of the flow equation as follows: where the derivative∂ t in the second line by definition acts only on the k dependence of the regulator,∂ t = ∂ t R k δ δR k . Since each factor of F k contains a coupling to external fields, this expansion simply corresponds to an expansion in the number of vertices. As we are interested in the flow of the four-fermion coupling, we can neglect terms with more than four vertices. The contributing terms are then given by the diagrams in Fig. 2.
Diagrams (2c), (3a), (4a), and (4b) occur when Yang-Mills theory is coupled to fermions minimally. The additional diagrams can be traced back to the fact that the volume element containing √ g generates additional graviton-fermion-couplings. Also the covariant derivative in the kinetic term generates not only one-but also two-graviton fermion couplings. An additional triangular diagram, built from a two-fermion-two-graviton vertex and two vertices coupling the graviton to one external Contributions to the running of the four-fermion couplings, sorted according to the number of vertices they contain. The diagrams containing curly lines receive contributions only from the trace mode, the diagrams with spiraling lines exist only for the TT mode. The functional RG equation receives contributions from these diagram types with all internal lines and all vertices denoting full regularized propagators and vertices, respectively. The right-hand side of the flow is given by the∂t derivative of these diagrams, yielding corresponding regulator insertions in the internal propagators, c.f. Eq. (14). and one internal (anti)fermion, vanishes in the Landau-deWitt gauge. The former vertex exists only for the transverse traceless, whereas the latter couples only to the trace mode. Since the metric propagator is diagonal in these modes for all choices of gauge parameters α and ρ, a non-vanishing diagram of this type cannot be constructed.
For the following discussion, we introduce the dimensionless renormalized gravitational couplings and the corresponding anomalous dimension for the metric Considering metric fluctuations within a general effective theory of quantum gravity, these coupling constants can acquire a wide range of values near the Planck scale. By contrast, these couplings are tightly constrained in the asymptotic-safety scenario in the deep UV by the properties of a UV fixed point. In fact, these fixed-point values are a result of the formalism and not a parameter of the theory. In this aspect, asymptotic safety is distinct from many other approaches to quantum gravity, since here the microscopic action is a prediction and not an assumption of the theory. The anomalous dimension of the background metric yields η N = −2 as a necessary requirement for the existence of the non-Gaußian fixed point [8]. The fixed point values G * , λ * of the gravitational couplings depend on the regularization scheme and also on the matter content. Within the regularization scheme and gauge choice used in [39], the fixed-point values given in Tab. I were obtained that will be used in our analysis for illustration. In the following, we investigate the influence of metric fluctuations on the chiral fermion sector taking both types of scenarios for quantum gravity into account.

IV. FLOW OF THE FERMION SECTOR
Let us now discuss the flow in the fermion sector as induced by the various diagrams sketched in Fig. 2.
First we observe a cancelation between the two box diagrams (4a) and (4b). This is not specific to gravity, but occurs whenever a Yukawa-type fermion-antifermionφ vertex with a scalar field φ without flavor (or internalsymmetry) and/or Dirac indices, exists in the theory. As these box diagrams involve only the trace mode, this cancelation mechanism between ladder and crossed-ladder topologies is at work here. A similar mechanism is active in a non-chiral Yukawa coupling of the type φψ i ψ i .
This marks an important difference to Yang-Mills in-teractions, as corresponding box diagrams are the only contribution that generate the four-fermion interaction even if they are set to zero initially. Since gravity gives rise to a larger number of vertices from a minimallycoupled kinetic fermion term, the diagram (2a) in figure  2, being absent in Yang-Mills theory, will create this interaction here. Therefore the β functions, as given by Eq. (6) contain similar types of terms as in Yang-Mills theory. Schematically, the β functions for the dimensionful couplingsλ ± for a general regulator R k (p 2 ) are given by The fermionic contribution indicated by the fermion-loop term corresponds to the diagram class (2c) in Fig. 2 and has first been calculated in [74]. The regulator-dependent dimensionful threshold functions are defined by where n = n TT + n f + n h − 1. In the above notation, we have already used regulators of the type R k = Γ (2) r k (y), with a dimensionless regulator shape function r k (y) and y = p 2 /k 2 . Specializing to linear regulators of the type the fermionic flow equations for the dimensionless couplings λ ± , can be determined explicitly Herein, the single terms correspond to the diagrams in fig. 2 in the following sequence: The first terms are the dimensional scaling terms of λ ± . The first term in each square bracket corresponds to the transverse traceless tadpole (1a), the second to the conformal tadpole (1b). The third term in the square brackets that enters the two beta functions with a different sign is represented by the two-vertex diagram (2a) with internal metric propagators only. The mixed two-vertex diagram (2b) results in the fourth term in square brackets. Finally the three-vertex diagram (3a) corresponds to the last term in square brackets. The fermion-loop contributions (2c) are represented in the two differing last terms; they agree with [74]. We find four pairs of (real) non-Gaußian fixed points for λ ± as a function of (G, λ, N f , η N , η ψ ). For gravity approaching a non-Gaußian fixed point, G * = 0, the fermionic Gaußian fixed point is shifted and also becomes non-Gaußian. If all four fixed points persist also beyond this truncation, each one defines a UV universality class of the fermionic matter sector. The fixed points can quantitatively be classified by their number of relevant directions and the corresponding critical exponents.
The universal critical exponents can be read off from the linearized form of the β functions for general couplings g i in the vicinity of the fixed point g j * , where the stability matrix B ij is defined by Eq. (21) is solved by Herein the critical exponents {θ} = −spect(B ij ) are minus the eigenvalues of the stability matrix and V n are the (right) eigenvectors of B ij . The scale k 0 is a reference scale and the C n are constants of integration.
In order for the flow to hit a fixed point in the UV all C n pertaining to irrelevant directions with θ n < 0 have to be set to zero. By contrast, the C n for relevant directions with θ n > 0 are free physical parameters that determine the long range physics. As a consequence a non-Gaußian fixed point can be used to construct a predictive fundamental theory if it has a finite number of relevant directions.
In the absence of gravity, the Gaußian fixed point has two irrelevant directions both with critical exponent θ Gauß = −2, corresponding to the standard powercounting canonical dimension of the fermion interactions in d = 4 dimensions. The other three non-Gaußian fixed points all have at least one relevant direction with critical exponent θ = 2, as can be proven on general grounds [74]. Two of these fixed points have an additional irrelevant direction with negative critical exponent. The last fixed point has another relevant direction with positive critical exponent. Let us now discuss the effect exerted by metric fluctuations onto the chiral fixed-point structure.

A. Asymptotically safe quantum gravity
Let us analyze the fermionic flow in the asymptoticsafety scenario near the gravitational UV fixed point where η N = −2 and G and λ approach their fixed point values G * , λ * as a function of the number of fermions N f as determined in [39], cf. Tab. I. We also set η ψ = 0 here, as it is consistent with our truncation.
As one of our main results, the fermionic fixed-point structure persists under the inclusion of metric fluctuations. Since the four-fermion couplings do not couple back into the flow of the Einstein-Hilbert sector, the stability matrix has a 2 × 2 block of zeros off the diagonal. Therefore the gravitational and fermionic critical exponents are determined by the eigenvalues in the Einstein-Hilbert sector and the fermionic subsector separately. Accordingly the gravitational critical exponents are given by the well-known exponents with positive real parts in the Einstein-Hilbert sector (see [38] for an overview of typical values without the effect of the minimally coupled fermions) and the two real critical exponents from the fermionic subsector.
Again we have four different fermionic fixed points at our disposal each defining its own matter universality class, which have either two, one or no relevant directions. The dependence of the critical exponents on N f at each of the four fixed points is shown in Fig. 3. These critical exponents are determined by inserting the fixed point values of the gravitational couplings taken from [39] into the fermionic part of the stability matrix. As these fixed point values are determined within a slightly differ-ent regularization scheme, this scheme-dependent error adds to the systematic error of our truncation. Nevertheless, the general chiral fixed point structure is rather insensitive to variations of the non-universal input. The critical exponents approach the limiting values of the purely fermionic system for large N f . This is due to the following mechanism being at work here: As shown in [39], the backreaction of a minimally coupled fermion sector onto the Einstein-Hilbert sector shifts λ * to increasingly negative values as a function of N f . In the propagators, a negative value for λ acts similarly to a mass term for the metric. This suppresses the contribution from metric fluctuations to β λ± for large N f . This decoupling mechanism induced by an increasingly negative cosmological constant ensures that the properties of the matter sector will not be strongly altered by metric fluctuations.
This chiral fixed-point structure is illustrated in Fig. 4 It should be emphasized that the decoupling mechanism due to a negative cosmological constant is only active in theories with a dominant number of fermionic degrees of freedom, as it is the case for the standard model. Since minimally coupled scalars shift the fixedpoint value for the cosmological constant towards λ * → 1 2 [39], a larger number of scalars even results in an enhancement of metric fluctuations. As a consequence, even at the shifted Gaußian fixed point the fermionic system can develop strong correlations, since the fixed point values for λ ± can then become quite large (cf Fig. 7  below). Accordingly, in theories with a supersymmetric matter content and low-scale supersymmetry breaking, such a decoupling mechanism might not occur or only in a much weaker fashion. Supersymmetric theories with a quantum gravity embedding may thus have to satisfy stronger constraints as far as the initial conditions of their RG flow are concerned.

B. General effective quantum gravity theories
Let us broaden our viewpoint a bit by considering a wider range of effective theories of quantum gravity. Consider an unknown UV completion of quantum gravity, which may differ strongly from a description in a QFT framework, e.g., by introducing a physical discreteness scale for spacetime, possibly also including the matter sector in a unified manner. Still, effective metric degrees of freedom may be expected to become relevant at a scale k 0 close to or below the Planck scale (see also [80]). The underlying microscopic theory may effectively reduce to a quantum field theory for matter and effective metric degrees of freedom with gravity rapidly becoming semiclassical towards the IR. Without any knowledge about the underlying theory, the effective quantum field theory may be at any point in theory space. For scales k < k 0 accordingly our present framework becomes applicable even for these different approaches to quantum gravity. Within this framework, we thus allow for any value of the chiral couplings and admissible values of the gravitational couplings (i.e., a positive Newton constant, and λ < 1/2 in order to avoid the potentially artificial propagator poles). The anomalous dimensions η N and η ψ , representing inessential couplings, will be determined by the effective field theory. Nevertheless, in our truncated framework, we allow them to acquire a priori unknown values of O(1).
Following the RG flow further towards the IR, the fermionic sector should approach the Gaußian fixed point in standard scenarios in order to be weakly correlated on scales where metric fluctuations become unimportant, and gauge boson and matter fluctuations start to dominate the picture. This requires that the initial values of the four-fermion couplings, which can (in principle) be determined from the underlying microscopic theory, have to lie within the basin of attraction of the Gaußian fixed point. This presents a non-trivial requirement that has to be fulfilled by any theory of quantum gravity.
As a specific example, we depict the basin of attraction for η N = 0, η ψ = 0, N f = 6, G = 0.1 and λ = 0.1 in Fig For generic values of G and λ the system can be altered considerably. As a first observation, the parabolas characterizing the fermionic β functions broaden as a function of G for a fixed value of λ, as shown in Fig. 6. In this example, we plot the β function for λ + for fixed N f = 2, η N = −2, η ψ = 0 and λ = 0. We set λ − on the shifted Gaußian fixed point value.
A particularly strong effect can be observed for positive values of λ. Here, the contribution from the metric sector is further enhanced for λ > 0. Indeed the β functions show the well-known divergence for λ = 1 2 . Whereas the divergence is likely to be an artifact of the simple Einstein-Hilbert truncation in the IR, the response of the gravitational propagators naturally suggests an enhancement of metric fluctuations for positive cosmological constant λ.
As an example, we plot the value of λ + at the shifted We observe that the Gaußian fixed point can be shifted to considerably larger values of (λ − , λ + ). This implies that the system will be strongly-interacting in this sector even at the fixed point which corresponds to the Gaußian fixed point in the absence of gravity. This may exert a strong influence on the physics properties of this shifted Gaußian universality class. As an example for such a significant deformation, we find that a negative anomalous dimension can in principle induce a crucial change in the fixed point structure. In particular the shifted Gaußian fixed point "collides" with a fixed point with one relevant direction at a negative value for η ψ crit (see Fig. 8). At this point, one critical exponent of each fixed-point approaches zero. For values η ψ < η ψ crit the two fixed points move apart again. Assuming that in the regime of strong gravity fluctuations these induce a large negative value for η ψ , the flow towards the semiclassical infrared would then cross η ψ crit . Accordingly points that lie in what at a first glance seems to be the basin of attraction for the shifted Gaußian fixed point can leave this region during the flow. Precise constraints for such quantum gravity theories and the associated chiral sector then depend on the dynamical details of the full flow.

C. Metric vs. gauge boson fluctuations
At first sight, the absence of gravity-stimulated chiral symmetry breaking seems surprising. Since gravity leads to attractive forces between matter, it is plausible to expect that binding phenomena are enhanced upon the inclusion of gravity. It is therefore instructive to confront our results with the chiral symmetry breaking mechanism in QCD-like theories.
Several technical differences have already become obvious: (i) the fermion anomalous dimension η ψ does not vanish in gravity in the Landau-DeWitt gauge (even though we have set it to zero in the present approximation). (ii) ladder and crossed-ladder box diagrams ((4a) and (4b) in Fig. 2) cancel in gravity, but play an important role in the approach to chiral symmetry breaking in QCD.
Further differences can be read off from the above analysis. For the fixed-point annihilation to occur (cf. Fig. 1) the terms ∼ G 2 in the fermionic flows ((2a) in Fig. 2) have to dominate. In the case of gravity, they are however outweighed by the tadpole terms ∼ Gλ ± .
In more physical terms, the ∼ G 2 terms describe the attractive nature of gravity, whereas the tadpole terms ∼ Gλ ± play the role of a gravity contribution to the (anomalous) scaling of the fermion couplings, ∂ t λ ± = +2(1 + η ψ + . . . G)λ ± + . . . . Quite generally, the fixed point structure in the fermionic flows for d > 2 arises from a balancing between dimensional scaling ∂ t λ ± ∼ λ ± and fermion fluctuations ∼ λ 2 ± . Whereas gauge-field fluctuations support the fermionic fluctuation channels, metric fluctuations also take a strong influence on the anomalous dimensional scaling which counteracts the general attractive effect of gravity.
This viewpoint is further supported by other technical observations: whereas gravity is channel blind with respect to the scaling terms, i.e., ∂ t λ i ∼ Gλ i , gauge boson fluctuations with coupling g also give rise to terms ∂ t λ i ∼ g 2 λ j with i = j that rather act like the above mentioned fluctuation terms. Finally, we should mention that there are further examples that fluctuations of attractive forces do not necessarily support binding phenomena: e.g., an effective flavored chiral Yukawa interaction in QCD-like theories contributes via box diagrams with a sign opposite to that of gauge bosons to the fermionic flow [48].
Still, it should be kept in mind that the present analysis is carried out in a restricted truncation of the effective action that -though meaningful for QCD-like theories -might not be sufficient for gravity. Further operators which could potentially be relevant for the interplay between gravity and a chiral fermion sector are discussed below.

V. CONCLUSIONS
We have investigated the quantum interplay between chiral fermions and metric fluctuations in quantum gravity. In contrast to QCD-like systems, where gluon fluctuations can induce strong fermionic correlations leading to chiral symmetry breaking, metric fluctuations do not support this mechanism in an analogous framework. Our result thus indicates that the existence and observations of light fermions is well compatible with a regime, e.g., near or above the Planck scale, where quantum gravity effects in the form of sizeable metric fluctuations set in.
More specifically, light fermions are compatible with the asymptotic-safety scenario of quantum gravity which provides a UV completion of quantum gravity within quantum field theory. In particular, we observe a decoupling mechanism for metric fluctuations: As known from [39], the dominant effect of fermionic fluctuations is a shift of the gravitational non-Gaußian fixed point towards increasingly negative values of the dimensionless renormalized cosmological constant for larger numbers of flavors N f . As a consequence, metric fluctuations decouple from the matter sector as such a cosmological constant acts like a mass term in the propagators of the metric modes and thus suppresses metric fluctuations. Apart from a slight shift of the Gaußian matter fixed point to small but non-vanishing values of the fermion interactions, the universality class of the Gaußian fixed point which is supposed to describe the fermionic matter content of the universe is left rather unaffected. In particular the critical exponents remain very close to the Gaußian values.
Such a compatibility scenario between matter and asymptotically safe gravity holds at least as long as fermions remain the dominant degrees of freedom. For larger numbers of bosonic modes, in particular for a supersymmetric matter content, the whole system may behave differently.
Our results are also applicable to scenarios where quantized metric fluctuations are considered as part of an effective quantum field theory at an intermediate scale below or near the Planck scale. Here the underlying UV completion of gravity, which may use different degrees of freedom for gravity than the metric, or even leave the local QFT framework completely, will determine the initial conditions for the RG flow at this intermediate scale.
Also from this more general viewpoint, we do not find any indications for gravity-stimulated chiral symmetry breaking. Still, our results can be used to decompose the accessible theory space into those branches where the chiral sector remains symmetric and other branches where the chiral sector becomes critical and typically generates heavy fermion masses. The distribution of these branches in theory space depends on the gravitational couplings. In particular, the universality properties of the shifted Gaußian matter fixed point can substantially vary. This analysis provides general constraints on the Planck scale behavior of any microscopic theory of quantum gravity: the existence and observation of light fermions potentially excludes those branches of theory space where the chiral sector is critical at the Planck scale.
Of course, a variety of further aspects could modify our results quantitatively and qualitatively, as our analysis is performed in a limited and comparatively small hypersurface in theory space. Still, the interplay between sizeable metric fluctuations and chiral symmetry has the potential to provide relevant phenomenological constraints for any theory of quantum gravity also in a more complete and quantitatively reliable investigation. Let us try to list some issues that need to be addressed in further studies: Whereas we have used flat-space calculations mainly as a technical tool, the full theory of quantum gravity will predict an (effective) manifold as its solution to the equations of motion. This resulting background may take influence on the chiral status of the matter sector itself, as it may screen or enhance fermionic long-range fluctuations that lead to chiral criticality. Screening mechanisms for chiral symmetry breaking of this type have already been studied in various chiral models, such as two-and three-dimensional (gauged) Thirring models [81][82][83] or the four-dimensional gauged NJL model [84]. Depend-ing on its sign, curvature can act like an IR cutoff that screens the critical IR fluctuations.
Also, if the question of chiral symmetry breaking and restoration is considered in a cosmological context, the thermal evolution of the universe may play an essential role. Broken chiral symmetry could be restored during reheating, if the corresponding temperature is sufficiently high compared with the scale of critical fermion dynamics -independently of whether it is stimulated by metric fluctuations or another mechanism.
Let us also discuss the possibility of gravity-stimulated chiral symmetry breaking which would strongly differ from the scenario arising from QCD-like theories. As gravity supports a richer structure of operators, chiral criticality could be triggered by operators that are typical for gravity, but do not occur for other theories.
Up to canonical dimension four and five, only explicitly symmetry breaking terms (as e.g. d 4 x √ g Rψ i ψ i ) exist. At dimension six (for two-fermion terms) and eight (for four-fermion-terms) we encounter a variety of new terms that are not forbidden by explicit chiral symmetry breaking, for instance dim 6: dim 8: At higher dimensionalities the number of terms increases considerably, as then also e.g. contractions involving the Riemann tensor will be possible. Furthermore couplings involving / ∇ or higher powers of the curvature are possible. Distinguishing between the background and the fluctuation metric leads to an even larger "zoo" of possible operators.
Several comments are in order here: The effect of metric fluctuations implies that none of these couplings will have a Gaußian fixed point, as the antifermion-fermiontwo-graviton vertex generically generates these couplings even if they are set to zero. The corresponding diagrams are indicated in Fig. 9.
These couplings raise several issues: In order for the asymptotic-safety scenario to work, the complete system of gravitational couplings, four-fermion-couplings and mixed couplings such as the above ones has to admit a Non-Gaußian fixed point (even though this does not necessarily require g i * = 0 for all possible couplings). As the above couplings couple non-trivially into the flow of the Einstein-Hilbert action and the four-fermion couplings, they may change our findings in this sector.
Since in the truncation that we have studied here, chiral symmetry breaking is avoided as the gravitational contribution to the anomalous scaling of the Here we have not drawn external graviton lines/couplings to a non-trivial background curvature; these will be generated by taking derivatives of the above diagrams with respect to the desired metric structure. The results then correspond to dimension 6 and 8 operators of the type listed in Eqs. (24) and (25). The upper self-energy diagram on flat space also contributes to the fermion anomalous dimension η ψ . Similar diagrams occur at higher order in the fermion field.
fermionic couplings outweighs the contribution that triggers bound-state formation, we expect that in particular the dimension-6 non-minimal kinetic terms in Eq. (24) represent an interestion extension of our truncation specific to gravity. This is because these generate further contributions ∼ G 2 and do not contribute to the anomalous scaling. As we expect that these couplings are non-zero at any UV fixed point, they constitute a nonvanishing contribution to the β functions of the fourfermion couplings. In other words, these dimension-6 terms genuine to gravity have the potential to act structurally identical in the fermionic flows, as the fermionic self-interaction terms considered so far.
Another question that has remained unaddressed so far, is the question of gravity-induced symmetry-breaking patterns. In the present work, we have imposed a rather standard chiral SU(N f ) L × SU(N f ) R (with additional U(1) factors of particle number and axial symmetry), and implicitly assumed its breaking in a QCD-like fashion, i.e. to a remaining mesonic SU(N f ) symmetry. Other breaking patterns are conceivable, including an originally larger symmetry that may break to the standard chiral symmetry upon large metric fluctuations. Our work merely represents a first step in this direction. In principle, it seems worthwhile to think not only about gravitationally-stimulated symmetry breaking and corresponding condensates, but also about corresponding bound states or excitations on top of condensates. If a gravitationally-stimulated symmetry breaking transition with a remnant standard chiral symmetry occurred near the Planck scale, stable bound states (analogously to hadrons in QCD) may have remained and (if equipped with the right quantum numbers) could contribute to the dark matter in the universe. Furthermore, in analogy to recent ideas in QCD, where a quarkyonic phase with confinement but intact chiral symmetry supports a spectrum of bound states, bound states may form that correspond to bosonized operators, e.g. of the form Eq. (25). These might form at a scale where quantum gravity is strongly interacting, and may then become massive at the much lower scale of chiral symmetry breaking. Supporting a stable bound state over such a large range of scales requires, of course, a highly non-trivial interplay between gravity and matter.
In this notation,ψ and ψ are the constant background fields, whereas the momentum-dependent fluctuation fields are distinguished by carrying an appropriate argument. This allows to evaluate the following vertices: where the momentum is always the momentum of the incoming graviton. The corresponding vertices with the TT mode vanish, as the first term in (A12) contains only the trace mode, the second term vanishes by transversality for constant external fermion fields and the last term vanishes as the contraction γ µ [γ ν , γ κ ]D κ h TT µν = 0. The second variation of the kinetic fermion term with respect to the metric contains only a TT contribution, as the trace contribution is always of the formψ i h / Dhψ i , which can be rewritten as a total derivative for constant external fermions.