FERMILAB-PUB-19-457-T IFIC/19-37 Sterile Neutrinos, Vacuum and

Gabriela Barenboim1 and Christopher T. Hill2 1Departament de F´ısica Te`orica and IFIC, Universitat de Val`encia-CSIC,E-46100, Burjassot, Spain. 2Theoretical Department, Fermi National Accelerator Laboratory, P. O. Box 500, Batavia, IL 60510, USA We construct an effective field theory (EFT) model that describes matter field interactions with Schwarzschild mini-black-holes (SBH’s), treated as a scalar field, B0(x). Fermion interactions with SBH’s require a complex spurion field, θij , which we interpret as the EFT description of “holographic information,” which is correlated with the SBH as a composite system. We consider Hawking’s virtual black hole vacuum (VBH) as a Higgs phase, hB0i = V . Integrating sterile neutrino loops, the information field θij is promoted to a dynamical field, necessarily developing a tachyonic instability and acquiring a VEV of order the Planck scale. For N sterile neutrinos this breaks the vacuum 1 to SU(N) × U(1)/SO(N) with N degenerate Majorana masses, and 2 N(N + 1) Nambu-Goldstone neutrino-Majorons. The model suggests many scalars fields, corresponding to all fermion bilinears, may exist bound nonperturbatively by .

PACS numbers:

I. INTRODUCTION and φj with a real Schwarzschild black hole B(x)? By “no-hair” reasoning any interaction, such as φ1 + φ1 → In the present paper we will discuss the issue of the B must have the same physical rate, or same quantum “black hole information paradox” in the context of an amplitude, as any other interaction such as φ1 + φ2 → effective quantum field theory (EFT). Classical EFT is B. A naive “no-hair” theorem would say these must be a powerful tool for summarizing low energy processes indistinguishable processes. where the detailed short-distance behavior of a system However, it is not possible to mathematically write is integrated out. Examples of its application to black down interactions that would have this universality. holes can be found in [1] and references therein, where Once we commit to a certain amplitude for φi + φj → B effective field theories are world-line actions and carry then we will generally not have an equivalent amplitude 0 0 0 0 local operators that represent known cases of the break- for some other process φi + φj → B where (φi, φj) is a down of classical no hair theorems. We also implicitly general linear superposition of the (φi, φj). This is only rely on the intuition of Dvali and Gomez in their picture possible in special cases, such as if we maintain the sym- of threshold quantum black holes [2]. metry U(1) and restrict ourselves to U(1) transforma- We assume there is a quantum limit for black holes, tions of the fields, in which case the process φ1 + φ1 → B near the threshold of production or collisions in scat- is forbidden. tering processes, or in coherent processes such as Bose- Stemming from this it is conventional wisdom that the Einstein condensation. These processes can exist in field global symmetry is broken by . Then theoretic descriptions even through the underlying ob- different configurations of the two scalars may have dif- jects are complicated many body states, e.g., Rubidium ferent amplitudes to interact with B(x). However, given atoms with Z = 37 can be described as pointlike objects, that both processes φ1 + φ2 → B and φ1 + φ1 → B must yet form a quantum condensate which can be described exist, since gravity is “flavor blind,” what rule dictates arXiv:1909.01956v4 [hep-th] 22 Apr 2020 as the vacuum value of a field. Hadronic boundstate pro- their amplitudes? The breaking of a symmetry is actu- duction can be described by pointlike interactions despite ally expressed by the form of the interaction. Breaking the short distance complexity of a hadron. a symmetry actually requires that we supply more infor- Presently we will decribe a black hole by a field, B(x), mation, not less. which can create or destroy a black hole of mass M at One might argue that the quantum effective field the- the event x and satisfies a free field equation of motion. ory does not exist, and the S-matrix does not exist, at We ignore the warped external geometry at large dis- Planckian high energies. However, given a mini black tances compared to the black hole hole, we can consider an idealized low energy experiment of B(x) (following [1]). Fundamental problems must then where the resolvable distance scales are large. We make be faced in constructing the interactions of B(x) with an arbitrary initial combination of the two φi’s and allow matter. them to collide with the mini Schwarzschild black hole, We consider the simplest case of a pair of scalars which B, resulting in a final state black hole B0. We take pains possess a global SO(2) = U(1) symmetry. How do we to observe the exclusive process where no other particles write down a pointlike interaction of a pair of scalars φi appear in the initial or final states. Our question is, what 2

ij 0 is the effective interaction gθ φiφjBB ? What then dic- We distinguish three possible logical cases for a dynam- tates the couplings θij? Theoretically or experimentally ical θij. (I) Information is explicitly lost; (II) Information determining θij may be problematic, but we cannot deny is conserved but does not propagate; (III) Information is its existence. conserved and is carried by the black hole. Quantum mechanics conflicts with no-hair theorems. Case (I) evidently implies an explicit, arbitrary fixed If we believe in Dirac’s formulation of the Hilbert space value of the spurion permeating all of space with a fixed of superpositionable states, an S-matrix and Wilsonian orientation in the group space, hence it breaks SU(N) × effective field theory (which we do) then we must allow U(1) transformation, and the associated Noether current for the possibility that B(x) carries information. To us is violated. It seems that this case makes no sense fun- the resolution is that the information contained in the damentally, since there is no procedure to specify θij(x), ij 0 θ φiφj initial state is transfered to the B state, and even as a random variable, however it may arise spon- resides on it’s horizon. Essentially the black hole B0 has taneoeuly through the formation of a condensate of un- acquired “information hair” and is now defined by a field derlying black holes. Note that just naively summing in- 0 ij αβ i j B = θ B(x) . This acquired information supplements dices, Σij νανβB0(x), implies a particular basis choice any other information that was already present on B, and therefore a particular choice of θij(x) ∼ Σnmδinδjm, though not probed in the experiment. a matrix with all elements = 1. This is logically equiva- In Section II we will reiterate these issues in more detail lent to case I and seems unreasonable unless it arises by for a pair of scalar fields φi, i = (1, 2). spontaneous symmetry breaking. To connect to real-world fields and make these issues Case (II) implies that we introduce a random field vari- more concrete we consider, in addition to scalars, sterile able θij(x), that has no correlation with the black hole neutrinos, possessing an SU(N) × U(1) global symme- itself and, minimally, has no derivatives. This implies try, and consider how they interact with a Schwarzschild that there is no current associated with θij(x), and in- mini-black hole (SBH) of mass MP lanck. In an effec- formation is not carried by the SBH, yet the neutrino tive field theory, we describe the SBH by a real quantum global current is conserved. We can treat θij(x) as a ran- field, B0(x), whose excitations are minimal mass, tiny dom field, and average over θij(x) configurations through black holes. We can follow the intuition of [2], who view which the black hole and the neutrinos propagate. This the spectroscopy of mini-blackholes as a tower of states is somewhat akin to a “spin-glass” in condensed matter labelled by a quantum number N where N is effectively physics [3]. It simply ends up promoting θij(x) to being the number of self gravitating gravitons in the core of the a propagating dynamical field. BH. Case (III) implies that the information is attached to We know that an s−wave pair of sterile neutrinos can the effective SBH and is dynamically transported by it. fall into the SBH, or scatter close to it. We can write This requires a“conjoined kinetic term” where the θij(x) couplings of the neutrinos to B0 that preserves or violates moves with B0(x) as a composite field, ∼ θij(x)B0(x). the global symmetry. This then raises the question of Here θij(x) is positionally correlated with B0(x). Case how the neutrino number current is carried (or destroyed) (III) is the most sensible to us, and we interpret it tenta- by the black hole? tively as an effective field theory description of the holo- A conserved global-charge current cannot be carried graphic principle [4]. Once we engineer the interaction by a real scalar field alone. To engineer a coupling of a described in Case (III) above, we find that there are po- αβ i j Weyl neutrino pair, ∼  να(x)νβ(x), to B0(x) we need tential dynamical consequences. a complex “information” that is either intrinsic to the As an application, we consider the effect of this on the black hole or is a field that can be attached to the SBH. neutrino vacuum. The vacuum of the underlying B0(x) We will designate this as θij(x). Physically, the neutrino field may be nontrivial. Indeed, we know of three classi- pair falls into the SBH, but never appears to cross the cal and important (effective) scalar fields in nature, the horizon. However, at some point we can no longer distin- Higgs field of the standard model (SM), the σ-meson of guish the neutrino pair and black hole from a pure black QCD chiral dynamics, and the Ginzburg-Landau field of hole, though there may be some information record in a superconductor (an effective description of a Cooper the EFT black hole that is absorbing them. pair). In each of these examples the vacuum is a “Higgs In the effective field theory we describe the information phase.” In our present scenario we consider the possibil- by a “spurion,” a complex field that transforms under ity that B0(x) → V + B where V is a nontrivial vacuum SU(N) identically to the neutrino pair and allows us to expectation value (VEV). tie all indices together in the interaction vertex. This A black hole Higgs phase will, in our model, have im- is co-localized with the SBH and the neutrinos at the plications for the dynamical behavior of information en- interaction vertex. The spurion represents information coded in θij: owing to the effects of neutrino loops ex- on the horizon of the SBH. The EFT is describing a mini- ternal to the horizon (in the EFT) global information black hole and the region immediately surrounding it, becomes a propagating massless field in a Higgs phase of external to the horizon, which includes anything orbiting B0(x). Moreover, this back reaction of the neutrino fields or in the process of infall as seen by the Schwarzschild induces an instability, causing a tachyonic potential for observers and this includes θij(x). θij(x) to develop, and in turn, θij(x) acquires a VEV. 3

This happens in both cases (II) and (III) (though there cannot decribe the process since we certainly expect that are slight differences) and we obtain an effective poten- φ1 + φ2 can collide to make a Schwarzschild black hole tial that leads to a VEV for θij, and sterile neutrinos and should have the same S-matrix. develop Planck scale masses. The SU(N) × U(1) global However, consider the interaction vertex, symmetry is broken to SO(N). 2 ∗2 ig(Φ − Φ )B = −2gφ1φ2B (2)

II. QUANTUM INCONSISTENCY WITH THE This breaks the U(1) symmetry and indeed now describes “NO-HAIR” THEOREM the process h1, 2|Bi = 2g. However, suppose we consider the collision h10, 20|Bi = 2g where:

Consider the “no-hair” theorem of classical black holes. |10i = cos(θ)|1i + sin(θ)|2i This allows for a black-hole, B, to have gauge charges, |20i = − sin(θ)|1i + cos(θ)|2i (3) or quantum numbers. For example, black holes with the quantum numbers of the standard model Higgs boson These are simple superimposed states and can readily be could in principle be produced in collisions, such as a pair produced in a laboratory, such as with neutral K-mesons, of eL +e ¯R → B. However we could equally well or oscillating neutrinos. However, we now see that the ¯ consider a collison cL + tR → B of a lefthanded-charm amplitude is theta dependent: with an anti-righthanded-top quark. These initial states have the same gauge charges, but involve different h10, 20|Bi = 2g cos(2θ) (4) combinations of flavors of initial state fermions. Gravity, being flavor blind, supposedly cannot make a flavored Hence we cannot write a local field description of φi + black hole. Hence, B must be universally coupled to any φj → B that yields an S-matrix, hij|Bi = S that is and all fermion bilinears. constant and independent of i and j. The flavor blindness However, a universal coupling to any and all fermion or “no-hair” theorem, in this sense, is incompatible with bilinears cannot exist in quantum mechanics. In the quantum theory. above example if we allow the latter process then there The only sensible resolution is that the black hole does will be some mixed state collision, such as a Cabibbo ro- have memory of the initial state, with global quantum ij tated charm-strange combination, (cL cos θ + sL sin θ) + numbers. The black hole field is then ∼ B which is a t¯R → B, where we did not Cabibbo rotate the top quark. BH that has quantum flavor information, hence ”hair”. This process will have a θ-dependent rate and thus does The vertex takes the form not respect the flavor-blind universality of gravity. ij To simplify, consider the production of a black hole in gΣijφiφjB (5) a collision of two scalar particles φi , with two global Bij remembers the initial states that made it and the flavors, i = 1, 2. We want to describe the process resulting interaction is SO(2) invariant. Equivalenly, the φi + φj → B in an effective field theory. We could black hole must be a complex field in the interaction describe this process directly by an S-matrix element, Φ2B + hc and the phase transformation Φ → eiθΦ to- hij|Bi where B can be a Schwarzschild black hole if the gether with B → e−2iθB must be a symmetry. This collision of the scalars is s-wave. The existence of the implies the BH carries a global U(1) charge. S-matrix implicitly assumes that |Bi is a quantum state. By insisting on the classical no-hair theorem we would Since gravity is flavor blind, a no-hair theorem would conclude that global charges cannot exist. However, it evidently imply that the S-matrix element, hij|Bi = S, has been recently understood that classically time depen- must be a constant independent of the choice of normal- dent hair exists. For example, reference [1] nicely treats ized in-states for the φi particles, for arbitrary i and j. and reviews the situation. A BH immersed in a time (A weaker statement might be that the probability is dependent scalar field will absorb and reemit this field the same for each such initial state). In a quantum field producing a time dependent, coulombic “halo.” If, for theory description the S-matrix must exist and we can example, the time dependent field is a Nambu-Goldstone introduce a field for the black hole which we designate as boson (eg, axion) then the associated charge is given by B(x), and we also have the two fields φi(x). Our equiv- the chiral current f∂tφ so indeed, the BH has acquired a alent field theory problem is then, how do we write the global chiral charge density. local interaction vertex for the process hij|Bi? √ Moreover, by claiming that the U(1) symmetry is bro- Let us introduce a complex field Φ = (φ1 + iφ2)/ 2 ken by gravity, by itself, in no way resolves this issue. Consider the interaction: Even if global symmetries do not exist at short distances, 1 but quantum mechanics applies at all distances, then we g|Φ|2B = gΣ φ2B (1) 2 i i face these issues. The Dirac superposition of Hilbert space states is the underling conflict here. In this re- This would forbid φ1+φ2 → B i.e.,, i 6= j. It equivalently gard, we should mention that ‘t Hooft has reformulated forces the S-matrix to be diagonal in flavor, hij|Bi = δijS quantum mechanics with an underlying “digital,” cellu- hence h1, 2|Bi = 0. Therefore a U(1) = SO(2) singlet lar automata structure [5] where the Dirac superposition 4 of states is only effective, but not fundamental. Pre- they interact with the B + V field. Our main objective sumably Dirac superposition arises by rapidly fluctuat- presently is to provide a field theory description of phys- ing discrete alternatives (on Planckian time scales). Here ical processes that involve matter interactions with the it is argued that there is no conflict with quantum me- SBH’s. This is immediately related to the issue of infor- chanics at larger distances and time scales. Perhaps this mation loss as described in the introduction. is alternative path to resolving these issues, but reveals In the original view of Hawking, global charges would the radical revision required to achieve consistency. The simply be swallowed by mini-black holes which subse- Dvali-Gomez theory of quantum black holes is more con- quently evaporate, and large violations of global charge servative [2], preserving the familar features of quantum conservation would be expected. This has been imple- mechanics and argues that no UV completion of revi- mented to conjecture, e.g., gravitationally induced vio- sion of quantum theory is necessary. In the Dvali-Gomez lation of B + L in the standard model [16] (see also an theory the no-hair theorem, , etc, are exception, [17]). In the modern prevailing view global emergent only for large N (quanta) black holes that ap- charges are conserved, with the global charge “informa- proach a classical limit. tion” holographically painted onto the horizon of the black hole, to be recovered upon evaporation [4].2 For N Weyl fermions we have the global SU(N)×U(1) ¯α˙ µ iβ III. EFFECTIVE FIELD THEORY kinetic term ψi i∂µσαβ˙ ψ . For the Weyl fermion pair, αβ i j i j  ψαψβ ∼ [ψ ψ ], we require a complex spurion, θij(x), We introduce a real scalar field B0(x) to decribe to tie indices of fermions onto B(x), as Schwarzschild mini-black holes as pointlike “particles” of the minimal mass MP , which become the excitations of B2(x) [ψi(x)ψj(x)]θ (x) 0 + h.c. (8) B0(x). In what follows we will neglect Hawking radia- ij 2 MP tion. We choose the Lagrangian of B0(x) to be For the case of ψ representing neutrinos this is depicted 1 1 2 2 2 L = ∂B0∂B0 − M B0 + JB0 + Λ (6) in Fig.(3). Note, we include here B0 (x) since the fermion 2 2 pair is colliding with an existing black hole in the initial 3 Hawking proposed a virtual black hole vacuum (VBH) state and producing one in the final state. In order to [6], in which the vacuum is viewed as consisting of Eu- make an SU(N) invariant interaction we require that θ lies in the symmetric representation of SU(N) of dimen- clidean (instanton-like) loops of tiny black holes, appear- 1 −1 sion 2 N(N + 1). ing and disappearing on time scales of order MP (for a review, see [7]).1 This is connected in the literature Naturally, θij(x) refers to the associated horizon infor- to various ideas in AdS hologaphy, gravitational instan- mation of B0(x). However, in the context of field theory tons and string theory [8][9][10][11]. The consequences we must face the issue of how to treat the spurion dy- of topological instanton fluctuations at the Plank scale, namically. In the following we consider two possibilities. associated with anomalous neutrino currents, have also In case (II) we argue that the spurion is simply a ran- been considered recently [12]. Other authors have dis- dom complex valued variable at the point of interaction. cussed possible gravitational effects in neutrino physics We therefore average path integrals over θij(x). In this [13, 14]. sense, the theory is analogous to a “spin-glass” in which For a simple model of a VBH, we have added a source spins propagate through a random potential, and the par- tition function is averaged over the potentials [3]. This term, J. The vacuum value of B0 is therefore shifted and 2 promotes θij(x) to a physical field. Since gravity loses we obtain the field B0 = B + V where V = J/M in a Higgs phase: all memory of the information, there is no further au- tocorrelation between θij(x) and itself, e.g., no mass or †ij 1 1 J 2 kinetic terms, such as ∼ θij(x)θ (x). Since there is no L → ∂B∂B − M 2B2 − + Λ (7) 2 2 2M 2 current associated with θij(x). we interpret this as “lost information.” We can always choose the source term to cancel an anti- In case (III), following [4] we implement conservation of deSitter Λ = J 2/2M. When par- information, i.e., the holographic principle which implies ticles of the standard model propagate through a VBH a conserved global current that involves the information.

1 Hawking’s rationale behind considering Euclidean may 2 We are not considering anomalies, as in [12], which more defini- in part have been that the black hole loops are simply finite tively break the global symmetries by violating the conservation action instantonic field configurations, and Hawking radiation of the global current and with an instanton allows a mechanism does not arise in Euclidean space. Given the source J there is no to exchange visible charge with the vacuum. decay of the static VEV, V in our Higgs phase. In any case, the 3 One could extend this to a model of production with vertex ∼ decay width Γ can be significantly less than MP of a physical g[ψψ]θB0 but this is complicated by a large energy dependence Planckian SBH of mass MP [15]. in a form factor which suppresses g(µ) for µ << MP [18]. 5

To do this we must view the SBH as analogous to a very IV. STERILE NEUTRINOS heavy atom, H(x) (such as Uranium) that is bound to a light particle φ(x) (such as a neutron in its nucleus) A. If Information is Lost to make a composite system, φ(x)H(x). The physical properties of the composite state are similar to those of Assume we have N sterile neutrino flavors. This will the unbound H(x), but the location in space-time of the imply an SU(N)×U(1) invariant kinetic term, and we as- light field φ(x) must be correlated with the heavy H(x). sume this is a valid global symmetry at the Planck scale. However, the heavy particle carries the bulk mass of the Consider an s-wave pair of massless right-handed neutri- system. This imples both a conventional kinetic term for nos scattering off of an SBH. Here we have a unique situa- H(x), e.g., 1 ∂H∂H, and a conjoined kinetic term for the 2 tion in the SM that an s-wave combination of two mass- composite system, less right-handed fermions, of flavors i and j can have zero local gauge coupling constant but nonzero global 1 µ flavor. 2 ∂µ(φ(x)H(x))∂ (φ(x)H(x)). (9) 2MP We assume the neutrinos interact with the SBH B0 field as in Fig.(1):

However, the mass term involves only H(x), as 2 M 2H(x)2. i j αβ B0 (x) να(x)νβ(x) θij(x) 2 + h.c. (13) MP In our case H(x) ∼ B0(x) and φ(x) ∼ θij(x). The con- joined kinetic term ∂(θ†B )∂(θB ), and no stand-alone 0 0 where θ (x) is a constant dimensionless spurion (note ∂(θ†)∂(θ) term implies that θ(x) can never escape B (x) ij 0 we given θ dimensions of a mass). This is case(I) alluded unless the system decays through a fermion pair. This to in the Intoduction and will break the conservation of leads to the following Lagrangian: the global currents of the neutrinos explicitly. There is nothing intrinsic to gravity that can dictate the flavor 1 1 2 2 structure or phase of θij in SU(N) × U(1) and we con- L = ∂B0∂B0 − M B + JB0 + Λ 2 2 0 clude that this is not a sensible theory. 1 ∗ij + 2 ∂(θ B0)∂(θijB0) (10) MP B. Random Field as a Spin-Glass The fact that θ is always accompanied by a factor of B (x) implies that it cannot escape the SBH, other than A more reasonable possibility is (II) that θij(x) is a 0 complex random variable. We have the interaction of by a neutrino interaction. In the limit B0 → V the ki- netic term becomes: eq.(13), but θij(x) can have no autocorrelation due to gravity, since the SBH has lost all knowledge of the ij indices, hence no term like µ2θ∗ θij is induced. This is V 2 ij hB θ|L|B θi ∼ L + T r(∂ θ†∂µθ) (11) the “information lost” scenario. 0 0 0 M 2 µ P A given choice of θij(x) describes a particular subpro- cess. However we then have to average over this field. Remarkably, in the VBH vacuum upon replacing B0 → The system is analogous to “spin-glasses” which have V constant, the information field θ freely propagates Hamiltonians that involve random variables (such as the through the medium. We can then absorb a factor of Edwards-Anderson model [3]). 2 2 † µ † µ V/MP into θ to write (V /M )∂µ(θ )∂ (θ) → ∂µθ ∂ θ For spin-glasses the averaging is done over the partition and θ is then canonical. Mainly, there is now a conserved functions and not in the action itself. In our case, we global current that involves both the fermions and the average over the path integrals, and this has the effect of field θ. promoting θij to a random quantum field: The interaction vertex with the fermions then becomes, Z  Z  again, that of eq.(8), in the VBH and suitably renormal- 4 i j αβ 2 2 → Dθ exp i d x θijναν  (B0 /MP ) + h.c. (14) ized β

i j While with a fixed θij(x) the neutrino flavor current con- [ψ (x)ψ (x)]θij(x) + h.c. (12) servation would be violated, it isn’t hard to see that upon averaging over θ matrix elements yield a conserved global µ A The interaction annihilates an information-less SBH B0 current h∂ νT¯ σµνi = 0. from the vacuum, and creates a composite SBH in the The theory is singular, however, since the equation of vacuum with information, θij(x)B0(x). θ is thus created motion of θ would enforce the vanishing of the vertex. (annihilated) by absorbing (producing) a fermion pair, To see this, we pass to the VBH vacuum upon√ replacing always in coincidence with B0 through its VEV. We now B0 → V constant, and absorb the factor of Z = V/MP apply this to sterile neutrinos. into θ to canonically normalize θ. If we then introduce a 6

2 ∗ ij small µ θijθ term in the action, the fermion current is manifestly conserved upon use of the equation of motion of the neutrinos and θ. In the VBH vacuum we have  i j αβ 2 †ij νανβ θij + h.c. − µ θijθ (15)

2 2  where Z = V /MP . The corresponding 4-fermion inter- ν action of eq.(15) is i [νiνj][¯ν ν¯ ] i j (16) ν µ2 j and is singular as µ → 0. Likewise, as µ2 → 0 the equa- i j αβ tion of motion of θ enforces νανβ B0(x) → 0. The key feature is the absence of the mass term for FIG. 1: Low 4-momentum sterile neutrino pair of global fla- θij(x) in the Planck scale effective Lagrangian. There are vors (i, j) disappears into Schwarzschild BH (or the neutrino no derivatives of θij(x) at this stage and no current built exchanges global charge with the SBH in a t-channel). This of θij(x) and the theory is singular at MP . This would be is described by a complex spurion field θij , and absence of an in our opinion, a realization of Hawkings information-lost M 2T r(θ∗θ). hypothesis. We will see below, however, that this situation is un- 1 2 2 † stable and effects of the back-reaction of the neutrinos 2 MP B0 , and there can be no T r(θ θ) term. The effec- will lead to a nonsingular dynamics for θij(x), on scales tive theory for the neutrino pair interaction in the VBH µ < MP and a spontaneous breaking of SU(N). The condensate becomes: singularity at MP share features with a Landau pole. 2 i j αβ B0 i j αβ νανβ θij 2 + h.c. → Zνανβ θij + h.c. (18) MP C. Information is Carried by Black Hole where in the second term we have replaced B0 by the con- 2 2 However, we can locally conserve the information densate and Z = V /M . Moreover, in the condensate the kinetic term becomes, (case III). We introduce a new effective field, Bij = θij(x)B0(x), that is composite and may represent a SBH † † µ with the information of a neutrino pair encoded on its ∂Bij∂Bij → ZT r(∂µθ ∂ θ) (19) horizon holographically. The effective theory for the neu- trino pair interaction with the BH’s becomes: Therefore, if we canonically renormalize θ we can adjust Z = 1 and our theory in the condensate becomes: B (x)Bij†(x) νi (x)νj (x)αβ 0 + h.c. (17) i j αβ † µ α β νανβ θij + h.c. + ηT r(∂µθ ∂ θ) (20) MP How do we view Bij? An analogy was given in Section This provides an insight into what is meant by conserva- III to the isotopes of Uranium. We can freely add or re- tion or loss of information, at least in the EFT: If we con- move neutrons from the Uranium nucleus, and the mass serve information then η = 1; If information is (weakly) is not dramatically changed, nor are the chemical proper- lost then η = 0. Note that with η = 1 there is now a ties. Therefore we can view a Uranium atom as a ground- formal conserved information current state nucleus, U0, which with an additional neutron n(x) ↔ jA =νT ¯ Aσ ν + iT r(θ†T A θ) (21) becomes the effective field U0(x)n(x). Chemically (elec- µ µ ∂ µ tromagnetically) we cannot easily discern which isotope we are dealing with. Yet another analogy might be bugs Information is now dynamically transferred from neutri- that end up flattened on the windshield of a car, that nos to θ and can propagate freely through the condensate. have little effect on the properties of the car, but become part of a conjoined kinetic term with it. Similarly, the effective field Bij(x) is essentially a SBH, D. Back-Reaction of Neutrinos and Spontaneous Symmetry Breaking B0(x) with the information θij(x) on it’s horizon, but there is no experiment we can do to detect θij(x), other than observing neutrinos emitted in Hawking radiation We can compute the action for the system at an en- as the BH decays. The mass of a θij(x) spurion field is ergy scale µ by using the renormalization group. This zero. follows the procedure known as the “block-spin renor- The kinetic term becomes the “conjoined kinetic term” malization group” for treating the Nambu–Jona-Lasinio of eq.(10). The mass term is given by that of B0 alone, model [19], developed in Bardeen-Hill-Lindner [20]. For 7

This is most conveniently rewritten in terms of the phys- ical neutrino mass: 4   2   N 2 2 Nmν MP V = − 2 MP mν + 2 ln 2 + 2 (29) 16π 32π mν

Here have substituted mν for the field θ and we have a renormalization invariant potential as a function of a FIG. 2: The infrared theory is controlled by the neutrino dynamical mν . loops. We now extremalize the potential with respect to mν , equivalent to extremalizing in θˆ. We obtain: 2   2   a fixed choice of θij we integrate out the fermions, de- 16π ∂ 2 2 MP 3 0 = 2 V = −MP + mν ln 2 + (30) scending from a scale MP to a scale µ. This is obtained N ∂mν mν 2 from the fermions loops in Fig.(2) with loop-integrals, The physical solution for the neutrino mass is Z MP d4k m2 = 0.424 M 2 m = 0.651 M (31) 4 . (22) ν P ν P µ (2π) The potential has a runaway for large values of mν >> We start with eq.(20) as the defining action at M . P MP , but this is unphysical since we insist upon the cutoff The result of integrating neutrino loops is an induced MP . action for θij of the form: We remark that this result is sensitive to the sublead- ing log behavior of the loops (constants), which differs ∗ µ 2 ∗ λ ∗ ∗ Z∂µθ ∂ θij − M θ θij − (θ θjkθ θli) (23) from [20]. In that case a large hierarchy is tuned by de- ij µ ij 2 ij kl manding a precisely tuned cancellation between a bare Using the results for Weyl spinors in Hill, Luty and mass term and the loop. Here we have no bare mass Paschos, [21], we immediately obtain: term but we find a solution in a small log limit. Hence, the boundstate field θ necessarily develop a vacuum ex- 1 M 2  1 M 2  P P pectation value (VEV) due to the VBH vacuum. Z = η + 2 ln 2 λ = 2 ln 2 8π µ π µ We note that we have neglected the production ver- 2 1 2 2 tex, ∼ ννθB0/MP , since we are mainly interested in Mµ = − 2 (MP − µ ) (24) 4π neutrino momenta below MP . However, our loop cal- culation informs us that the neutrinos, in the SBH Higgs Anticipating spontaneous symmetry breaking, we denote 2 phase, form a nonzero VEV, hννi ∼ mν M together with the field VEV, θij = δijθ. Then P hθi ∼ MP which may be bootstrapped back to be the ∗ 2 ∗ ∗ 4 θijθij = Nθ θijθjkθklθli = Nθ (25) source term J for the BH condensate itself. The unrenormalized potential is: E. Phenomenology N Nθ4 M 2  V = − (M 2 − µ2)θ2 + ln P (26) un 4π2 P 2π2 µ2 The sterile neutrinos thus obtain a large common Ma- Note that the infrared cut-off on our loops is determined jorana mass, ∼ MP . The N sterile neutrinos, coupled by the neutrino mass. The unrenormalized neutrino mass to gravity, have a global SU(N) × U(1) symmetry which is µ = mν = 2θ, and therefore the potential becomes is now broken to SO(N) θ contains N(N + 1) real de- grees of freedom. The SU(N)/SO(N) breaking implies 4   2   N 2 2 Nθ MP there is one phase and there remain 1 N(N + 1) massless Vun = − MP θ + ln + 2 (27) 2 4π2 2π2 4θ2 Nambu-Goldstone bosons. To renormalize we want the kinetic term to be brought We consider the SM with 3 families including 3 sterile to canonical normalization. neutrinos. The SU(3) symmetry of the νiR is essentially ˆ2 2 an accidental symmetry given only the gravitational cou- The renormalized field VEV is therefore√θ = Zθ . The neutrino mass becomes µ = m = 2θ/ˆ Z = 2g θˆ, plings of the neutrinos. The left-handed doublets, √ ν r ψ couple via the Higgs boson to the right-handed neu- where g = 1/ Z is the renormalized Yukawa coupling. Li r trinos through the Higgs field H as: Note that if η = 0 we see that gr displays the character- ¯ istic Landau pole at µ → MP which is the compositeness yijψLiHνjR (32) condition of the field θ [20]. Thus we have the renormal- ized potential: Generally the yij will break the SU(3). Integrating out the heavy R-neutrinos yields the Weinberg operator, 2 4 ˆ4 2 ! ! gr N 2 ˆ2 gr Nθ MP 1 Vren = − (M )θ + ln + 2 (28) kj ¯ ¯ T 4π2 P 2π2 2 ˆ2 yijy (ψLiH)(ψLkH) + h.c. (33) 4gr θ MP 8

If we now assume a typical (large) value for the ykj ∼ 1 that the dynamics is more subtle. Owing to the local in eq.(33) we see that the scale of the induced Majorana gauge field, the Bij field essentially describes a Reissner- mass terms of the observable L-neutrinos, with vweak ∼ Nordstrom (RN) black hole. This acquires a slight mass 2 19 −6 (175 GeV), is ∼ vweak/(10 GeV) ∼ 3 × 10 eV, which enhancement of order αMP above the mass of the SBH is rather small. According to [22], the best fit to neutrino MP . This mass enhancement is associated with the in- 2 2 −5 2 data implies we require ∆m12 ∼ mν ∼ 7×10 eV which formation field θij, i.e., in the VBH the field θ freely −2 implies mν ∼ 0.8 × 10 eV. Our results suggest a scale propagates, but carries the charge of the RN hole and 2 † of observable neutrino masses that is small by roughly a acquires the small RN mass term αMP T r(θ θ). This factor of ∼ 3 × 10−3. means that the tachyonic instability is blocked for small 2 2 2 Our result for the Majorana mass scale depends only V/MP . However, if an effective coupling g = V /MP 2 upon MP and is rather immutable. However this is the exceeds a critical value gc the field θ acquires a VEV and Planck mass at extremely high energies (of order MP ). the gauge symmetry is then spontaneously broken. It should be noted that a number of authors have argued A general picture that is emerging here, upon includ- for significant renormalization effects of the Planck scale, ing local gauging, offers a new non-perturbative bind- 16 and that MP ∼ 10 GeV may be reality in D = 4 [23]. ing mechanism for fermions to produce scalar fields. In Of course, with extra dimensions MP can be significantly turn, this suggests a large system of composite Higgs modified, but our set up requires D = 4 and would other- bosons. Moreover, a near critical value of the coupling, wise have to be re-explored if D 6= 4. However, neutrinos g ∼ V/M implies deep scalar boundstates with (nearly) with the Type I seesaw may be uniquely probing grav- vanishing masses. Perhaps a more refined theory might ity at MP and offer credence to a signficantly reduced lead to a conformal window with a low mass scale for the Planck mass at high energies. di- boundstate, θ. In our crude approximations 1 this would be a coupling tuned arbitrarily near critical- The 2 N(N + 1) = 6 Nambu-Goldstone bosons (Ma- jorons) will have decay constant f ∼ MP but poten- ity g ≈ gc. Since there are 1176 Weyl bilinears in the tials governed by the explicit SU(3) symmetry breaking standard model [25], there may exist a large number of Weinberg operator. Schematically ∼ m4 cos((φ/f) + χ), composite scalars in nature that are marginally subcrit- where χ is a CP phase and may range from m ∼ mν to ical boundstates of elementary fermions due to gravity, m ∼ vweak depending upon the details of eq.(33). This with masses that extend down to the Higgs mass scale 125 potentially offers a number of cosmological possibilities, GeV. In this picture, “scalar democracy,” the standard from late time phase transitions, dark energy, to provid- model Higgs is composed of tt¯ , and its nearest neigh- ing an inflaton [24]. Discussion of these is beyond the bor ¯bb would be expected at a mass scale ∼ 5.5 TeV scope of the present paper. and accessible to an upgraded LHC [25]. In this scheme, the Higgs boson(s) of the standard model are essentually mini-black holes [26] V. CONCLUSIONS

Acknowledgments We have given an effective field theory treatment of information loss or conservation in the dynamics of a mini-black hole interacting with fermions. In particu- We thank participants of Simplicity-III at the lar, we have focused upon sterile neutrinos with a global Perimeter Institute, Waterloo, Canada, for discus- SU(N) × U(1) symmetry. sions. GB acknowledges support from the MEC and Our present model illustrates how “information” might FEDER (EC) Grants SEV-2014-0398, FIS2015-72245- be described in analogy to an induced effective random EXP, and FPA2017-845438 and the Generalitat Va- field θ. The weak information loss of global charge would lenciana under grant PROMETEOII/2017/033. She forbid gravitationally induced auto-correlation of θ, i.e., also acknowledges partial support from the European no kinetic, mass or interaction terms. However, consis- Union FP10ITN ELUSIVES (H2020-MSCAITN-2015- 674896) and INVISIBLES-PLUS (H2020-MSCA-RISE- tency with the holographic view promotes θij to a local field and θ “piggy-backs” on a Schwarzshild black hole. 2015-690575). CTH acknowledges the Fermi Research θ becomes dynamical in the black hole condensate. Alliance, LLC under Contract No. DE-AC02-07CH11359 The absence of the θ mass term for sterile neutrinos with the U.S. Department of Energy, Office of Science, has an immediate and dramatic physical consequence. Office of High Energy Physics. This implies that an instability driven by sterile fermion loops will always lead to a condensate of θ ∼ MP . In our case the instability is provided by the neutrino loops external to the black holes, and θ becomes a sterile νν boundstate. Our present paper is introductory, but let us men- tion a future application. In a subsequent paper we ex- tend these results to locally charged fermions. We find 9

[1] L. K. Wong, A. Davis and R. Gregory, Phys. Rev. D 100, Int. J. Mod. Phys. A 16, 2399 (2001). no.2, 024010 (2019) doi:10.1103/PhysRevD.100.024010 [17] D. Stojkovic, F. C. Adams and G. D. Starkman, Int. J. [arXiv:1903.07080 [hep-th]]. Mod. Phys. D 14, 2293 (2005). [2] G. Dvali and C. Gomez, Fortsch. Phys. 61, 742 (2013); [18] T. Banks and W. Fischler, “A Model for high-energy Phys. Lett. B 719, 419 (2013); JCAP 1401, 023 (2014). scattering in quantum gravity,” hep-th/9906038. [3] S. F. Edwards, P. W. Anderson, ”Theory of spin glasses”, D. M. Eardley and S. B. Giddings, Phys. Rev. D 66, Journal of Physics F, (5): 965974 (1975); 044011 (2002) D. Sherrington, S. Kirkpatrick, Phys. Rev, Letters, 35 [19] Y. Nambu and G. Jona-Lasinio, Phys. Rev. 122, 345 (26): 17921796, (1975). (1961). [4] G. ’t Hooft, Conf. Proc. C 930308, 284 (1993); [20] W. A. Bardeen, C. T. Hill and M. Lindner, Phys. Rev. L. Susskind, J. Math. Phys. 36, 6377 (1995); R. Bousso, D 41 (1990) 1647; Rev. Mod. Phys. 74, 825 (2002). W. A. Bardeen and C. T. Hill, Adv. Ser. Direct. High [5] G. ’t Hooft, “The Cellular Automaton Interpretation Energy Phys. 10, 649 (1992); of Quantum Mechanics. A View on the Quantum Na- For references to Higgs as tt¯ see: C. T. Hill and E. H. Sim- ture of our Universe, Compulsory or Impossible?,” mons, Phys. Rept. 381, 235 (2003); Erratum: [Phys. [arXiv:1405.1548 [quant-ph]]; Found. Phys. 43, 597-614 Rept. 390, 553 (2004)]. (2013) doi:10.1007/s10701-013-9702-z [arXiv:1205.4107 [21] C. T. Hill, M. A. Luty and E. A. Paschos, Phys. Rev. D [quant-ph]]. 43, 3011 (1991). [6] S. W. Hawking, Phys. Rev. D 53, 3099 (1996). [22] M. Tanabashi, et al.,(Particle Data Group), Phys. Rev. [7] G. Amelino-Camelia, Living Rev. Rel. 16, 5 (2013). D98, 030001 (2018). [8] A. Chamblin, R. Emparan, C. V. Johnson and R. C. My- [23] D. F. Litim, Phil. Trans. Roy. Soc. Lond. A 369, 2759 ers, Phys. Rev. D 59, 064010 (1999). (2011); [9] L. J. Garay, Phys. Rev. D 58, 124015 (1998); Phys. Rev. D. F. Litim, R. Percacci and L. Rachwal, Phys. Lett. B Lett. 80, 2508 (1998). 710, 472 (2012); [10] G. W. Gibbons, M. B. Green and M. J. Perry, Phys. Lett. I. Antoniadis and S. P. Patil, Eur. Phys. J. C 75, 182 B 370, 37 (1996). (2015). [11] C. J. Isham, gr-qc/9510063. [24] C. T. Hill, D. N. Schramm and J. N. Fry, Comments [12] G. Dvali and L. Funcke, Phys. Rev. D 93, no. 11, 113002 Nucl. Part. Phys. 19, no. 1, 25 (1989); (2016). J. A. Frieman, C. T. Hill, A. Stebbins and I. Waga, Phys. [13] G. Barenboim, JHEP 0903 (2009) 102; Rev. Lett. 75, 2077 (1995). Phys. Rev. D 82 (2010) 093014. [25] C. T. Hill, P. A. N. Machado, A. E. Thomsen and [14] F. Benatti and R. Floreanini, Phys. Rev. D 64, 085015 J. Turner, Phys. Rev. D 100, no. 1, 015015 (2019); Phys. (2001); JHEP 0002, 032 (2000). Rev. D 100, no. 1, 015051 (2019). [15] D. N. Page, Phys. Rev. D 13, 198 (1976) and Phys. Rev. [26] C. T. Hill, “Composite Higgs Bosons and Mini Black D 14, 3260 (1976). Holes,” [arXiv:2002.11547 [hep-ph]]. [16] F. C. Adams, G. L. Kane, M. Mbonye and M. J. Perry,