Spacetime as a Whole Alireza Jamali∗ Senior Researcher Natural Philosophy Department, Hermite Foundation† [email protected] July 17, 2022 Abstract Following a suggestion put forward by ’t Hooft, that ‘black holes should be treated just like atoms’, the notion of ‘wave-black hole duality’ is proposed, which requires having a proper energy-momentum tensor of spacetime itself. Such a tensor is then found as a consequence of ‘principle of minimum gravitational potential’; a principle that corrects the Schwarzschild metric and predicts extra periods in orbits of the plan- ets. In search of the equation that governs changes of observables of spacetime, a novel Hamiltonian dynamics of a Pseudo-Riemannian manifold based on a vector Hamilto- nian is adumbrated. The new Hamiltonian dynamics is then seen to be characterized by a new ‘tensor bracket’ which enables one to finally find the analogue of Heisenberg equation for a ‘tensor observable’ of spacetime. Keywords— Wave-black hole duality, nonmetricity, corrections to gravitational potential, min- imum gravitational potential, geometry of implicit line elements, corrected Schwarzschild metric, energy-momentum tensor of spacetime, wavefunction of a black hole, Hamiltonian dynamics of a manifold, tensor Poisson bracket Contents 1 Introduction 2 2 Principle of Minimum Gravitational Potential 4 3 Energy-momentum of Spacetime itself 6 ∗Corresponding author †3rd Floor - Block No. 6 - Akbari Alley - After Dardasht Intersection - Janbazané Sharghi - Tehran - Iran 1 4 Wavefunction of a Black hole 7 5 Hamiltonian dynamics of a Manifold 8 6 Conclusion 11 1 Introduction To approach quantum gravity a useful methodological guiding principle advocated by ’tHooft[1]isthat‘Ifwetrytodescribesomethinglikeblackholes,theirbehaviorshouldbeunderstood in the same language as the one we use for other particles; black holes should be treated just like atoms, [...]’1 Followingthismethod,wewouldexpectaquantumapproachtogravitytobeginbyinvestigatingthe status of the first building block of quantum theory of particles, the de Broglie relation, in general relativity(hereafterGR).Ifforthepurposeofquantizinggravityablackholeandaquantumparticle should be treated formally the same, it would be conceivable to formally generalize p = ~k to a µ µ 2-tensor2, and postulate the wave-black hole duality T = ~K , (1) µν µν in which T would be the energy-momentum tensor of the black hole, and K its ‘wavetensor’ µν µν with dimensions [K] = [K ] = Length−3×Time−1 = Frequency×Volume−1 = ‘Frequency density’. (2) µν It is critical not to confuse T which oughts to represent energy of the black hole, i.e. energy- µν momentum tensor of Schwarzschild spacetime3 itself, and the Dirac delta energy-momentum tensor that acts as the source for Schwarzschild solution [2]. We are consequently faced with the (controversial) question of the energy-momentum tensor of the spacetime which is the same as the energy-momentum tensor of the gravitational field according to the Equivalence Principle. As we make no use of any result from the literature on this question, we do not review the previous attempts and refer the interested reader to [3, 4, 5]. It suffices to say that there is still no consensus on a satisfactory tensor representing energy-momentum of spacetime itself. From Electromagnetism (and previous classical field theories) we know that the energy-momentum tensor of a field is constructed from the field strength squared, which is the ‘derivative’ of the potential squared. We also know that in GR the notion of potential is replaced by the metric tensor. Altogether therefore we would expect the field strength tensor of gravity to be constructed from Q = ∇ g ; ρµν ρ µν which is called nonmetricity tensor and has been studied in extensions of general relativity [6]. The obstaclethatwenowfaceisthataslongasthecovariantderivativeistakenwithrespecttotheLevi- Civitaconnectionthatarisesfromthesamemetricg , thistensoriszero. Hadthistensornotbeen µν zero, we could readily construct from it a Lagrangian density and consequently energy-momentum tensor of gravitational field. Things would then probably be similar to electromagnetism and there 1Bold letters by me. 2No spatial index is used in this text. Metric signature (−,+,+,+) is assumed. 3As Schwarzschild metric is the prototype of solutions of Einstein Field Equations and black holes, in this essay we take synonymous ‘spacetime’ and Schwarzschild solution. 2 would be a good chance we could quantize the free field Lagrangian, in a manner similar to the way one does for electromagnetism [7, 8]. If we insist on realizing this expectation we need to use a metric g˜ different from the one that is µν the solution of the Einstein Field Equations. To determine this new metric, note that g˜ ‘plays µν the role’ of the electromagnetic four-potential, therefore applying Hamilton principle to the action ∫ 1 √ S = J Jαβγ −g d4x, (3) J αβγ 2 where 1 1 J := (Q˜ −Q˜ +Q˜ ) = (∇ g˜ −∇ g˜ +∇ g˜ ), (4) µνρ µνρ νµρ ρµν µ νρ ν µρ ρ µν 2 2 inwhichthecovariantderivativeisthesameasinstandardGR(Levi-Civitaconnectionarisingfrom the solution of the Einstein Field Equations), determines the new metric4. In principle therefore we have g˜ determined, which in turn gives the energy-momentum tensor of spacetime via µν 1 δS T = −√ J . µν −gδgµν Application of Hamilton principle to (3) and a gauge fixing yields ∇µ∇ g˜ = 0, (5) µ ρσ in vacuum. In practice, however, solving such equations is complicated and has been studied by Hadamard[9], DeWitt and Brehme[10], and Friedlander[11]. As we expect, the solutions turn out to be expressed in terms of tensorial generalizations of Green functions (bi-tensors). Without attending to solve (5), three key implications can be seen: • (5) is linear in g˜ . Owing to linearity of covariant derivatives, (our minimal extension of) µν general relativity is linear in a special sense. This shows that although GR is nonlinear, in a formally ‘higher level’, it is linear and the linearity may well be employed for quantizing it. • Unlike the case with the Hilbert action[12], for (3) the corresponding Euclidean action is bounded below. • Fromtheperspectiveof(3)beingagroundforquantizationof‘freefieldgravity’,theproblem of non-renormalizability of Hilbert action[13, 14] evaporates. These implications confirm our initial vision that this procedure would be apt to quantizing GR. In view of (3) it is feasible that Hilbert action is but a means to the goal of quantization of GR, not its foundation; yet it is indispensable, as the covariant derivative in (5) requires a given metric, determined independently. Whilefinding a solution to the procedureoutlined above5, poses a momentarypracticalbarrier, wenowshowthatbyembarkingonaroutetoputtheprocedureonafirmphysicalbasis,thisbarrier can be circumvented, facilitating a solution for the case of Schwarzschild spacetime. 4For brevity and without loss of generality, our discourse is limited to vacuum (without source terms). 5By ‘the procedure’ hereafter we mean the one sketched in the beginning of text; the one that starts by the action (3), and is supposed to yield the energy-momentum tensor of spacetime (itself). 3 2 Principle of Minimum Gravitational Potential Having in mind that the essence of the above procedure is having two distinct metrics6, let us begin with the historical-logical precursor of GR, where Einstein7 began: application of special relativity (hereafter SR) to Newtonian gravity. For now we adopt a completely classical and New- tonian perspective towards gravity, and consider a massive particle in a gravitational potential ϕ. Conservation of energy requires Energy of the particle = Energy of the field lost to the particle, viz. 1 m∥v∥2+mϕ = −mϕ, (6) 2 hence ∥v∥2 ≡ −4ϕ, (7) in which the (weak) equivalence principle is assumed8. To apply the second principle of SR9, from (7) we have 1 max∥v∥2 = −minϕ, 4 hence c2 minϕ = − . 4 This is a fundamental necessity that GR must locally respect, along with the principles of SR. This principle of minimum gravitational potential (hereafter MGP) will turn out to be the firm physical ground we were seeking. To see the utility and importance of this principle, apply (7) to the γ factor (of SR) to get10 1 Φ := √ ; (8) 1+ 4ϕ c2 applying MGP now to Newtonian gravity, results in mc2 1 U = − √ , (9) 2 1+ 4ϕ c2 for the gravitational potential energy of a particle with mass m. Taylor approximation results in the corrected potential energy ( ) GMm GM U = − 1+3 , (10) r c2r 6Although (only) similar in words, our approach is fundamentally different from premetric and bimetric theories. For that reason we only refer the interested reader to [6, 15, 16, 17]. 7And others, most notably Nordström. 8We have set ϕ(∞)=0 everywhere in our discourse. This eliminates the arbitrariness due to addition of a constant to ϕ. Also, higher-order special-relativistic corrections to kinetic energy have no bearing on the argument and are thus dropped. 9Known as the Principle of Constancy of Velocity of Light. 10Forbrevity,weuseasubstitutionintheγ factor,butclearlyitisnotnecessaryfortheγ factortobegiven tomodelMGP.Onecansimplyarguethat wearelookingforthesimplestfunction withaboundeddomain. In the elementary functions there are only two such functions: square root, and logarithm. Logarithm is excluded simply because it does not allow for the minimum itself to occur. 4 which is known –at least– since Wells[18] to account for the Perihelion precession of Mercury11. Wells found the new term by comparison with the effective potential of GR and interpreted the result in terms of the notion of effective (field) theories, but he was not able to determine higher- order terms, and instead envisaged of ‘performing precise experiments’ to find the new terms. We now see that MGP uniquely determines all terms without reference even to GR, let alone new experiments. BeforeconsideringstatusofMGPinGR,first,observethatthefirsttermof(9),i.e. −mc2/2,which results when r → ∞, agrees (expectedly) with the effective potential arising from Schwarzschild geodesics of GR[22], ( ) c2 GM l2 l2GM UGR = m − + − , eff 2 r 2r2 r3c2 but we expect this energy (energy of a particle in absence of gravity) to be mc2, not half of it. We see that ‘the 2-factor problem’ is only partially solved by GR, in that GR gets the observable value right; partially because the effective potential is still half mc2. So the question of where has the other half of mc2 gone? Calls for an explanation, for which hypotheses non fingo12. Second, notice that using the definition of gravitational potential, (9) is c2 1 ϕ˜= − √ , 2 1+ 4ϕ c2 pointing to the transformational consequences of MGP. If we are to find such transformations, it is crucial to first find the geometry arising from the Φ factor (8). A simple-minded expectation would be dϵ2 = c4dm2+4c2dUdm, in which we have generalized the definition of gravitational potential ϕ = U/m to a local one dU ϕ := . dm Using the above definition of ϕ and (7), we expect this metric to yield E = γmc2, but it wrongly gives E = mc2/γ. We therefore propose ( ) 4 c4dm3 = dϵ2 dm+ dU , (11) c2 which does yield E = γmc2 upon the application of (7). Unfortunately however, (11) does not conformtoanyknowngeometryasitslineelementdϵisgivenimplicitly. Withoutproperknowledge 11The first occurrence of such correction that I found is in [19], but Wells, as far as I know and could see, was the first to focus on this correction and show it matches GR (in its Newtonian limit) for the perihelion precession problem. There are some corrections based on Einstein-Infeld-Hoffmann equations[20, 21] 3G(m +m ) 1 2 , rc2 which, even if we overlook its problem for a single particle, and let m = m , gives a wrong correction. 1 2 Weinberg[3] considered corrections of higher orders in terms of three coefficients upon which an a priori relation is projected, i.e. g =1+2αϕ+2(αγ−β)ϕ2, 00 which need not be the case. 12As the results of my investigations in this direction are not yet conclusive enough to be presented. 5 of the novel geometry arising from this line element it is not possible to find the transformations. Sadly, the transformations are not yet fully derived to report here13. It is rather astonishing how deep and powerful MGP is. Not only it gives us the effective potential of GR, but also it locally corrects GR itself, for the effective potential of GR has only four terms, while the correction from MGP adds infinitely many new terms. On the mathematical physics side, owing to MGP, (11) offers an entire novel domain of research: geometry of implicit line elements. 3 Energy-momentum of Spacetime itself Proceeding to status of MGP in GR, although the conceptual framework of GR is deeper and able of making other predictions that (9) cannot make, according to (7) violating MGP (locally) is equivalent to (local) violation of the second principle of SR. As such MGP must be locally respected by GR, which is however evidently not the case, unveiling a friction between SR and GR. If we apply MGP to the Schwarzschild metric, a new metric √ 4ϕ dr2 ds′2 = − 1+ c2dt2+ √ +r2dΩ2, (12) c2 1+4ϕ/c2 is found, to which the Schwarzschild metric is an approximation. Having two metrics which was the essence of our initial procedure, we can now show the relation of MGP and energy-momentum of spacetime. To that end, discard O(ϕ3) in (12), leaving ( ) ( ) r r2 r r2 −1 ds2 = − 1− s − s c2dt2+ 1− s − s dr2+r2dΩ2, (13) r 2r2 r 2r2 which has the same form as the Reissner–Nordström metric, ( ) ( ) r r2 r r2 −1 ds2 = − 1− s + Q c2dt2+ 1− s + Q dr2+r2dΩ2. r r2 r r2 Recall that the r−2 term in Reissner–Nordström metric is due to the energy-momentum tensor of the electromagnetic field[23], q2 TEM = ; tˆtˆ 2r4 meaningthatther−2 termin(13)resultswhenonesolvesEinsteinFieldEquationsforaspherically symmetric body with energy-momentum tensor 1 GM 1 T = − ( )2 = − (∇ϕ)2; (14) tˆtˆ 4πG r2 4πG 13Allthiswasforwhenonebeginsbyalinearapproachdϵ2 =c4dm2+4c2dUdm. Atotallydifferentvision is entailed when one considers another possibility: thinking of c2 1 ϕ˜=− √ , 2 1+ 4ϕ c2 as a functional J[ϕ] on the function space of ϕs. This approach is enormously difficult, beginning right away from a nonlinear functional. 6 but this is (twice) the Newtonian limit of energy density of the gravitational field! This observation suggests the interesting result that corrections of MGP to Schwarzschild metric account for the energy of the spacetime itself, i.e. the tensor g¯ , µν ( ) ( ) r2 r2 −1 ds¯2 = g¯ dxµdxν = − 1− s +O(r−3) c2dt2+ 1− s +O(r−3) dr2+r2dΩ2, (15) µν 2r2 2r2 yields the energy-momentum tensor of spacetime when substituted in Einstein Field Equations. The energy-momentum tensor of spacetime T is thus explicitly given by µν c4 T = G¯ . (16) µν µν 16πG It is important to notice that, first, this equation is not the Einstein Field Equation, differing by a factor of 2; second, ∇µT ̸= 0, which will be important later. µν Before applying our results to (1), it is important to mention that the corrected metric (12) is not without predictions. The geodesics of (12) are given by ( ) ( )√ du 2 c2κ2 c2 4GMu = − u2+ 1− , (17) dφ l2 l2 c2 where u = 1/r(φ) and κ,l are constants of integrations. Unlike the geodesic equation of GR which is solved by Weierstrass ℘ function[24], (17) is not possible to be solved even in terms of ellip- tic functions14. We can however readily show the mere existence of at least two new periods of the solution –assuming its existence–. The importance of new periods is in that as Schwarzschild geodesics are given in terms of elliptic functions which are doubly-periodic[25], and since the Ke- plerian geodesics are singly-periodic, perihelion precession of Mercury is the result of the second period. Since Jacobi [25, 26] we know that elliptic functions are the most general multiply periodic func- tions possible in a single variable. On the other hand, elliptic functions can solve at most second order differential equations, while (17) is of fourth order15. This shows that the solution of (17) has at least four periods16. 4 Wavefunction of a Black hole Having adduced that there is no rational obstacle to soundness of our proposal of wave-black hole duality, it is time to consider wavefunction of a black hole (spacetime) Ψ = e−ΩiKµνgµν, (18) 14Algebraically (17) is y4 =x5+x+z, which is called a tetragonal curve, while the GR curve is y2 =x3+x+b, called an elliptic curve. 15Study of this class of equations is still going on; see [27]. Computer programs and numerical approaches exist to solve these equations and find the numerical value of the periods, but it seems almost impossible to me to see the physical meaning of new periods without having a clear analytical vision beforehand; it is like giving someone the mere numerical value of the perihelion and expecting them to infer from it what physicalphenomenaitissignalling. Forthisreasonitisnotofphysicalvaluetoconsidersolvingtheequation numerically. 16This is only a heuristic argument. A rigorous one would be based on Abelian functions. 7 where f 1 Planck frequency P Ω = = = , l3 t l3 Planck volume P P P is chosen so as to make argument of the exponential dimensionless. This enables us to turn metric and energy-momentum tensors into operators (observables)17 { T → Tˆ (f) = i~Ω∂ f, µν µν gµν (19) gµν → gˆµν(f) = gµνf, where ∂ ∂ := ; gµν ∂gµν implying the uncertainty relation [ ] gˆµν,Tˆ = i~Ωδµδν , (20) ρσ (ρ σ) which is proved similarly to the quantum mechanical case, i.e. for a test function Φ Tˆ gˆµνΦ = Tˆ (gµνΦ) = i~Ω∂ (gµνΦ) = i~Ωδµδν Φ+gµν(i~Ω∂ Φ) = i~Ωδµδν Φ+gˆµνTˆ Φ. ρσ ρσ gρσ (ρ σ) gρσ (ρ σ) ρσ Although slightly similar, (20) is critically different from the usual commutation relation of ‘canon- ical quantum gravity’[13] [ ] hˆab(x),pˆ (y) = i~δaδb δ(x,y), (21) cd (c d) in which metric and energy-momentum tensor are understood to be fields, functions of space, analogous to [ϕa(x),π (y)] = i~δaδ(x,y); b b whereasin(20),metricandenergy-momentumtensorareunderstoodasnotfields,butmere‘tensor- coordinates’ of a new ‘phase space’ –which will be described soon–, analogously to [qˆa,pˆ ] = i~δa. b b The physical significance of this difference is that while (21) means one cannot simultaneously measure precisely the metric and energy-momentum of a single point of space(time), for (20) uncer- taintyisinmeasuringsimultaneouslythemetricandenergy-momentumofaparticularblackholein toto. Whiletheconstantin(21)isa‘purelyquantum-mechanical’one, Ωisaquantum-gravitational constant. 5 Hamiltonian dynamics of a Manifold The foremost arising question is now that of equation governing changes18 of observables. Any such equation must be in terms of dynamics of spacetime manifold, as we are viewing the manifold of spacetime ‘as a whole’; in accord with what (1) and the notion of wavefunction of spacetime 17Technically plane waves are not physically legitimate as they are not normalizable, but this issue can be rigorously avoided in orthodox quantum mechanics; see[28], and similarly here. This therefore is safe as a mere tool to arrive at definitions of operators, and their commutation relations, as we shall see now. 18We do not say ‘evolution’ since conventionally by evolution, change in time is meant, but the resulting structure is ‘timeless’. 8 indicate. To this purpose Hamiltonian approach is better suited in that it is timeless at the level of canonical coordinates (phase space). We accordingly postulate that dynamics of spacetime (as a pseudo-Riemannian manifold) is completely determined by its metric g and its energy-momentum µν tensor T , which are ‘independent’ of one another, thereby requiring all dynamics of the manifold µν to be expressed solely in terms of g and T . This postulate resembles that of ADM formulation µν µν of GR[23, 29], although key differences will be seen. Indeed the coherent structure and relative successes of ADM theory are strong motivations and support for this postulate. An important question now arises as to our expectation from such dynamics, knowing that both ‘canonical ten- sors’ are given by GR and (15). This implies that the conceptual order of this new sought-after Hamiltonian dynamics is the ‘reverse’ of the familiar Hamiltonian dynamics: one here should be- gin with canonical variables and arrive at the Hamiltonian, contrary to the familiar situation in which one begins with the Hamiltonian and arrives at the canonical coordinates (as the solutions of Hamilton equations). This is all naturally anticipated since we do not have even a candidate for the Hamiltonian of a pseudo-Riemannian manifold. Mathematically the postulate is that to a pseudo-Riemannian manifold (M,g) of dimension n we associate a 2n2-dimensional ‘phase space’ P spanned by (g,T). To find the equations governing dynamics of M we must adopt a brief heuristic approach, in which it is assumed that wisdom of the reader indulges us patiently by their understanding of the inevitable investigative nature of any discourse that aims to go beyond current established knowledge. Our task is essentially finding counterparts of Hamilton equations19. Let us ask for the counterpart of the (call it ‘first’) Hamilton equation dp ∂H = − . (22) dt ∂q This equation, when its right-hand-side is let equal to zero, yields the conservation of energy. As the counterpart of conservation of energy in GR is ∇µT = 0, µν we look for an equation that yields ∇µT = 0 whenever its one side vanishes. Straightforwardly µν one is led to ∇µT = −∂ Hσ, (23) µν gνσ in which vector Hamiltonian Hσ is yet to be defined, as explained above. Observe that [H] = [Hσ] = Energy×Length−4, (24) which is energy ‘density’ per spacetime, manifesting that the proposal does not distinguish between space and time even in its measurement units (dimensions). It should be mentioned that for a proper energy-momentum of spacetime (16), the left-hand-side of (23) is not zero. The task of ensuring that the total energy of spacetime as a whole is conserved, is undertaken by another equation, to which we shall soon arrive. Assuming (23), immediately the analogue of the second Hamilton equation is expected to be ∇µg = ∂ Hσ, µν Tνσ 19The word counterpart is vital. This cannot be a ‘generalization’ of classical Hamiltonian dynamics in any sense. There is no methodologically continuous way to get from metric, which knows no coordinates, to coordinates, let alone a total (proper) time derivative. Therefore we cannot expect this new analytical mechanics of manifolds to be a generalization proper; it would be a similar but different structure from the symplectic structure; as we shall see. 9 but the left-hand-side is zero by the metric-compatibility condition of a pseudo-Rimannian mani- fold20. Thus ∂ Hσ = 0. (25) Tνσ Equations(23)and(25)therefore, definethevectorHamiltonianofapseudo-Riemannianmanifold. The notion of a vectorial Hamiltonian is not totally alien to quantum gravity and unexpected. In exercising the Hamiltonian constraints –which lead to Wheeler-DeWitt equation– one is practically using a vector Hamiltonian Ha [13, 23]; the ‘momentum (spatial) constraint’. This newly proposed structure now allows us to put the fundamental commutation relation of quantum gravity(20) on a tenable ‘classical’ origin, by proving the analogue of Liouville’s theorem in this new phase space P: ( ) ∑∑ ∂ Hσ div vH = (∂ ,∂ ) Tνσ = (∂ ∂ Hσ −∂ ∂ Hσ) = 0. (26) gνσ Tνσ −∂ Hσ gνσ Tνσ Tνσ gνσ gνσ ν σ The above equation21 guides us to define a bracket for an arbitrary vector V, and the vector Hamiltonian H, both functions on P, by ∑( ) {V,H} µ := (∂ Hµ)(∂ Vµ)−(∂ Vµ)(∂ Hµ) . (27) ν gλν Tλν gλν Tλν λ Similarly for an arbitrary 2-tensor M on P, and the vector Hamiltonian H, we define ∑( ) {M,H} ν := (∂ Mκν)(∂ Hν)−(∂ Hν)(∂ Mκν) λµ gλµ Tλµ gλµ Tλµ (28) κ This new bracket enables us to write (23) and (25) as ∇µM = {M,H} (29) µν ν in which M can be metric or energy-momentum tensor22. But we can take H as given by (23) µ and (25), and postulate that (29) determines changes of any 2-tensor M on P. For a vector V this postulate would be ∇µV = {V,H}, µ implying ∇µH = {H,H} = 0, (30) µ which is the promised conservation of Hamiltonian (conservation of total energy of spacetime). This equation determines changes of the vector Hamiltonian itself. To sum up, in this new dynamics, 1. One begins by solving Einstein Field Equations to find metric tensor; first canonical tensor, 2. Then the second canonical tensor, energy-momentum tensor, is found via the procedure, or (16), 3. One then finds vector Hamiltonian using (23) and (25), Finally 20Evidently, should one desire to retain complete formal symmetry and similarity with classical Hamilto- nian mechanics, a geometry with non-metricity shall be assumed. 21Technically (26) requires a generalization of Clairaut’s theorem. To save a huge space, the proof, which requires some theoretical development, will be presented elsewhere. 22ItisimplicitlyunderstoodthatEinsteinSummationConventiondoesnotholdfortheindexforwhichΣ is explicitly written. Apart from that, this notation is consistent with index notation of tensors, and indices canbecontractedaccordingtoEinsteinsummationconvention. Itisherethereforeassumedthatcontracted indices are not written anymore, i.e. {M,H} :={M,H} θ, and so on. ν νθ 10