Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

BRS: Some Recent GTR-Related arXiv Eprints

  1. Jun 1, 2010 #1

    Chris Hillman

    User Avatar
    Science Advisor

    I'd like to initiate a new "running thread" in the nonpublic BRS subforum, similar to the one maintained by marcus and some others in "Beyond the Standard Model" public subforum, but focusing on eprints related to gtr. I hope to try to explain for SA/Ms (without the distractions from questions and comments from fringe promoting figures and students lacking sufficient background to follow along) my view on some interesting eprints. I expect that at times I'll probably go back several years (at least), and at other times I'll discuss eprints in the week they appear.
     
    Last edited: Jun 1, 2010
  2. jcsd
  3. Jun 1, 2010 #2

    Chris Hillman

    User Avatar
    Science Advisor

    BRS: "Non-Commutative Geometry (NCG) Inspired Black Holes" (part I)

    Anyone who has followed the gr-qc section of the arXiv in 2010 probably noticed some recent papers by a group of europhysicists which claim to "complete" the so-called "NGG Inspired Black Hole program", in particular:

    "Kerrr" black hole: the Lord of the String
    Anais Smailagic, Euro Spallucci
    Code (Text):

    arxiv.org/abs/1003.3918
     
    (Yes, "Kerrr", meaning "Kerr regularized", as in: lacking any strong curvature singularities.)

    Charged rotating noncommutative black holes
    Leonardo Modesto, Piero Nicolini
    Code (Text):

    arxiv.org/abs/1005.5605
     
    These allegedly are "exact solutions" of the EFE which are "inspired by" noncommutative geometry, a far reaching program initiated and powerfully developed by Fields medalist Alain Connes, as explicated in his landmark book, Non-commutative Geometry. As part of this program, Connes has long expressed the hope that NCG "should" result in the long sought (unique?) quantum theory of gravitation.

    Unfortunately, the description in the abstracts is seriously misleading, and although I think experts will instantly understand the severe limitations of the achievements to date of this "NCG inspired black hole" program, physicists who lack certain background could easily be fooled, I fear, into thinking these papers are more important or decisive than is really the case. At the same time, I do not believe these papers are without merit, or that they are wrong, or anything like that. They just don't accomplish anything like what their authors claim. Rather, my "executive summary" of how I regard the "true situation" with applications of NCG to quantum gravity runs like this:
    • NCG has already resulted in revolutionary achievements in pure mathematics, and can be expected to be developed much further in the new century. Since the motivations of NCG include elementary non-relativistic (!) quantum theory (but much more, on the purely mathematical side), it would not be surprising if ideas from NCG wind up playing an important role in the long sought quantum theory (theories? oh no!) of gravitation.
    • Some time back, rumors flew that Connes would soon announce the general solution of the quantum-corrected EFE, but my understanding is that he has pulled back from such strong claims. Rather, papers by Connes and others in this area have been limited to providing NCG motivation for various guesses about what an "effective field theory" incorporating quantum corrections to the EFE might look like, in the context of gtr. Roughly, there is a general expectation that the long sought quantum theory of gravitation might ultimately result in suprisingly simple "classical approximations" which incorporate corrections which in the context of gtr would for example replace vacuum solutions which model black holes with spacetimes which incorporate (inside the horizon) something which looks, classically, like an "classically impossible" configuration of anistropic fluid matter plus perhaps something like a Lambda contribution.
    • The europhysicist contributions to the "NCG Inspired Black Hole" program are basically inductive guesses about what the simplest possible "effective field theory" "quantum corrections" to Schwarzschild vacuum, Kerr vacuum, and Kerr-Newman electrovacuum might look like. Their guesses result in spacetime models which are not really "black holes" at all (since no event horizon) but can be summarized like this:
      • replace m with m(r) (having a particular not terribly complicated functional form) in the exact solutions in gtr,
      • for large r, they are close mimics of their gtr counterparts,
      • for small r, one finds (in the rotating cases) a coordinate annulus containing a slice of de Sitter lambdavacuum, interpreted as the interior of a kind of spinning "cosmic string", surrounding a region of Minkowski vacuum, and surrounded by a nonvacuum region containg anisotropic fluid and a kind of "variable Lambda" contribution, which interpolates between the innermost regions and outer regions which closely mimic their gtr counterparts.
      These guesses completely ignore important and purely classical considerations involving the (surprising and game changing) effect of infalling radiation and matter, particularly the notion of "mass inflation", and IMO are not terribly plausible guesses after all, and not even strongly supported by NCG motivations.
    • Because the energy-momentum tensor obtained by computing the Einstein tensor of the Lorentzian manifolds proposed as so-called "NCG inspired black holes" corresponds to a mere inductive guess not well motivated by NCG, much less derived from an actual theory, the phrase "exact solution" is IMO entirely inappropriate, since if you call such things "exact solutions", almost any Lorentzian manifold could be called an "exact solution", which would render the EFE vacuous, and I do not think the EFE is vacuous.

    For background, one should really start by at least glancing through the book by Connes (I'm sure there's a much easier introduction--- the first place to look is probably John Baez's This Week archive), and then perhaps glancing at papers such as these:

    Gravity in Non-Commutative Geometry
    A. H. Chamseddine, G. Felder, J. Fröhlich
    Code (Text):

    arxiv.org/abs/hep-th/9209044
     
    (This paper explains how ideas of NCG applied to "nonabelianize" function fields on Lorentzian manifolds leads to the expectation of an "effective classical field" correction resulting from, roughly speaking, the idea that a minimal length scale associated with quantum fluctuations in the geometry should lead to certain kinds of corrections to classical fields, which can be plugged back into the classical context in lieu of actual quantum theory of gravitation.)

    Aspherical gravitational monopoles
    Alain Connes, Thibault Damour, Pierre Fayet
    Code (Text):

    arxiv.org/abs/gr-qc/9611051
     
    (Damour is a leading expert in classical gravitation, and Connes is of course the leader in the NCG programme, so this an important paper for those following applications of NCG to physics.)

    (Non)Commutative Finsler Geometry from String/M--theory
    Sergiu I. Vacaru
    Code (Text):

    arxiv.org/abs/hep-th/0211068
     
    (These ideas can be combined with M-theory and the recent re-emergence of Finsler geometry into both mathematics, following the powerful example of Chern, and physics.)

    Since the book by Connes is rather imposing (but the first chapter is somewhat readable and offers an introduction to the motivating ideas), let me back up and say a few words about NCG generally. You probably noticed that Connes's canonical examples of a noncommutative geometry are
    • the Bohr model for energy levels in atoms
    • the space of Penrose tilings (with the inflation introduced by J. H. Conway)
    Since during my misguided younger years I wrote a diss on Sturmian tilings (aka "generalized Penrose tilings", a notion due to de Bruijn--- the interest in these constructions comes largely from the interplay of local and global properties), it will not come as a shock that I have always been fascinated by NCG. Indeed, I have always been fascinated by a powerful and general notion in mathematics, the Galois connection, which includes the Gelfand correspondence in functional analysis, which should be regarded as the true center of gravity (no pun intended) in NCG. Another application of Galois connections involves how "analytic extensions" of "locally defined holomorphic functions given by power series" gives rise to the important notion of a sheaf, which involves the circle of ideas which comes up whenever one mentions a "unique (or sometimes, nonunique) maximal analytic extension" of a given "local solution" of the EFE.

    (to be cont'd)
     
    Last edited: Jun 1, 2010
  4. Jun 1, 2010 #3

    Chris Hillman

    User Avatar
    Science Advisor

    BRS: Non-Commutative Geometry (NCG) Inspired Black Holes" (part II)

    (cont'd)

    The notion of a Galois connection is very general (see the book by Davies and Priestly, Introduction to Lattices and Order). Given any relation from one set to another set, it provides a dual (order reversing) correspondence between two lattices (in the sense of lattice theory, a subject in modern algebra) and a notion of "closure operation" which is defined algebraically, but in applications where a topology is involved, this closure operation often agrees with the topological closure. The dual lattices are not always interesting, but for relations which are known to be important, quite often they are very interesting.

    For example, in the relation on a finite dimensional inner product space (vector space equipped with a euclidean inner product) in which we say that [itex]\vec{u}, \, \vec{v}[/itex] are related if [itex]\vec{u} \cdot \vec{v} = 0[/itex], instead of two dual lattices we have just one which is self-dual, the "closed objects" in the lattice are the linear subspaces, the closure operation is "linear span", and the duality relation on the "closed objects" is "orthogonal subspaces". For Hilbert spaces, the closure operation is "topological closure of the linear span", and otherwise as for finite dimensional inner product spaces.

    There are many other applications in geometry, analysis, logic, and computer science, and the most interesting applications seem to always be related nicely to category theory approaches. Thus, I think it is reasonable to characterize galois connections as one of the great unifying principles of mathematics.

    In algebraic geometry, a key relation between polynomials on real or complex projective spaces (or affine spaces) and point sets in the underlying space is [itex]f(x) = 0[/itex], i.e. the polynomial is related to the point x in [itex]\mathbb{C}P^n[/itex] when f vanishes on x, i.e. x is a root of f. This leads to the dual lattices studied in algebraic geometry, in which ideals of the ring of polynomials are dual to "algebraic varieties" (the points, curves, and surfaces defined by the vanishing of all the polynomials in a given ideal). (See the wonderful textbook by Cox, Little, and O'Shea, Ideals, Varieties, and Algorithms, one of the ten best books written by humans too date IMO, say what you like about Melville or Tolstoy.) It turns out that the polynomial rings studied in algebraic geometry provide a kind of not very obvious topology called the Zariski topology, in which not every point is closed (so our topological spaces are not Hausdorff spaces), and the non-closed points correspond to some very interesting and important phenomena in the ring involving the "Hilbert generating function". (Same Hilbert known to many SA/Ms for discovering the EFE a few days before Einstein, and for many other pioneering achievements in gtr, but his achievements in pure mathematics are far more important than such comparative trivialities. And see the book by Schenck, Computational Algebraic Geometry, for more about the Hilbert function and the algebraico-geometric information lurking in what it says about nonclosed points. Often, they arise when the Zariski closure of a "singular point" in an algebraic variety includes a line, or something like that.)

    In the theory of commutative rings with a (neccessarly unique) multiplicative unit element 1, maximal and prime ideals play a key role. (It turns out that every maximal ideal is prime, although this is not obvious from the definition. Prime ideals get their name because in algebraic number theory they really do correspond to prime numbers.) A very important class of "tractable" topological spaces are "compact Hausdorff spaces", and a very important example of a commutative ring with unity is the space of real-valued continuous functions on a compact Hausdorff space X, which turns out to be a commutative Banach algebra A with (neccessarly unique) unit 1 and (neccessarily unique) norm |.|. Then the same relation, x is related to f when f(x) = 0, gives rise to a very important duality between closed ideals (of the Banach algebra) and closed sets (in the underlying space), in which the points correspond to prime ideals of the algebra (the "spectrum" of the algebra--- the term really does yield what you think in the case of the Bohr atom!). Furthermore, if we write the set of points on which f vanishes as V_f ("variety" of f), the complements of the V_f is a sub-basis for the topology on our compact Hausdorff space X. The grand result is that we obtain a complete mapping between algebraic concepts concerning the Banach algebra A of real-valued continuous functions on X and topological concepts in X. In this correspondence, it turns out that the fact that all points are closed in X corresponds to the fact that all prime ideals are in fact maximal in A.

    More generally, one can inquire about noncommutative rings with unity (such as result from introducing "a bit o' Heisenberg" into the brew), in which not every prime ideal is maximal, so these rings lead to non-Hausdorff topological spaces in which not all points are closed--- a phenomenon already familiar from algebraic geometry. This leads to a vast programme initiated by Connes in which one systematically examines all the places in mathematics where commutative function rings occur, and generalize to noncommutative function rings (more or less neccessarily) interpretable as incorporating "a bit o' Heisenberg", and then, using the Gelfand duality, obtains "topological spaces" in which not all points are closed, and in which any notion of "distance" (more or less neccessarily) incorporate the ideas of some "minimal length" (the "Planck scale") and "quantum fluctuations" in the geometry (no "points" at all, but incompletely locatable "minimal regions").

    The non-Hausdorff nature of the space of Penrose tilings emerges when one thinks of this as having the character of a sheaf in which "local tilings" can be extended (as in analytic extension in the classical theory of functions) to larger local tilings, but possibly not to a global tiling. These local tilings correspond to local sections, and a global tiling to a global section. There is a theme in mathematics (since the middle of the last century) which says "it all comes down to cohomology", and Conway, Lagarias, and Thurston have provided a complete theory of certain other tiling spaces (such as domino tilings) in which the possible obstructions to finding a tiling in a given tiling space which tiles a region with a certain boundary are cohomological. There has been some work involving C-* algebras (a theory originally inspired by the needs of "geometric quantization") which suggests that something similar may happen in the possible "obstructions" to extending a local patch (local section) to a larger patch. Propp and Wilson have gone in a quite different direction, treating the space of Penrose tilings as a branched manifold, but I think it is more natural to treat tiling spaces as sheaves.

    So, this is all fully rigorous hard analysis (and topology, and algebra, and mathematical logic), and Connes's program of NCG stands as another of the great unifying principles of mathematics (but much, much harder than the Galois connections with which we began). So, the mathematical motivation for expecting NCG to eventually say something profound about quantum gravity is in fact very strong.

    But this doesn't imply that the results claimed for the so-called "NCG Inspired Black Hole" program mean very much! As I already said, so far this program has simply resulted in inductive guesses, motivated by the desire for simplicity (no different from assuming a simplifying symmetry, really!), but in fact only very vaguely (no pun intended on the "vague convergence" which is so important in functional analysis!) motivated by Connes's program of NCG, and certainly not rigorously derived from the long promised noncommutative reformulation of the EFE, which as I understand it, has not yet been found. In fact, physicists seem to pay even less attention to the NCG approach to finding the long sought quantum theory of gravitation than they to to the spin foam approach or quantum cosmology approach, much less the loop quantum gravity approach, much less the M-theory approach! Which suggests that Connes has either not made sufficient progress for applications, or that he has not yet explained his work in language the physicists can understand.

    The Modesto & Nicolini paper refers to "the Newman-Janis algorithm" but a more appropriate name is "Newman-Janis trick" since there is no real procedure here (and Newman & Janis didn't claim to have an algorithm). A readable account of the original Newman-Janis trick can be found in the textbook by D'Inverno; basically, it is a slick way to "derive" the Kerr vacuum from the Schwarzschild vacuum by making an inspired guess, namely modifying a Newman-Penrose tetrad in a certain way.

    Smailagic and Spallucci characterize the approach of Schwarzschild and Kerr & Newman like this:
    I think this is a bit unfair: assuming spherical or axial symmetry when searching for an exact solution to some system of coupled nonlinear PDEs is hardly "ad hoc". Neither is searching for a source-free solution. (In constrast, the "inductive guesses" of these europhysicists are however truly "ad hoc"!)
    First, it is true that when working with linear field equations, it is formally possible (and often desirable) to set up a Green function approach and to "work outward" from a known (spatially compact) source to obtain the "vacuum exterior". But there is nothing wrong with with starting with a vacuum exterior having the expected symmetry and say a fluid solution with the same symmetry and matching the two across the zero pressure surface. That is in fact a reasonable elementary approach in Newtonian gravitation. And in Newtonian gravitation, it turns out that the rotating case is much, much harder, but the same approach does work and results in McLaurin rotating fluid balls--- plus some interesting different families of rotating fluid balls, with an unexpected "phrase transition" as the angular momentum of the ball is increased--- matched to vacuum exteriors which can be expressed only using special functions.

    The EFE is nonlinear, unlike the Poisson equation governing slowly varying Newtonian gravitational fields, so we should expect solving the EFE to be much harder. It turns out that after decades of work the situation in gtr is similar to that in Newtonian gravitation, when following the "outside-inward" strategy. Not surprisingly, finding suitable exact perfect fluid solutions in gtr and matching them to suitable exact vacuum exteriors is even harder in gtr, and this work is still not complete.

    Not surprisingly, following the "inside-outward" approach is also much more difficult in gtr. And it's not hard to see some fundamental reasons why. In Newtonian physics, due to the linearity of the field equation, one can employ the powerful Green function approach; in gtr, one cannot. And physically, the nonlinearity of the EFE ensures that approximations which are useful and valid in Newtonian gravitation simply don't work in gtr. Similarly, thought experiments like "spinning up a fluid ball" which are not too hard to formulate in Newtonian gravitation are very difficult to formulate in gtr, because gtr is a more demanding theory and in particular demands that the modeler provide details of how mass-energy impinges on a nonrotating fluid in order to give the ball some angular momentum. Even defining a physically transparent local coordinate chart suitable for such modeling has proven challenging. And we must expect "memory effects" (the exterior vacuum due to a "spun-up" rotating fluid ball might depend on the history of the spin-up phase).

    Second, we are not talking really about "integration constants" here, but rather about two parameters (in the Kerr vacuum solution) which we need to interpret physically. In is true that one way of doing this is to consider the weak field limits, from which it is fairly easy to rewrite the parameters in terms of two new parameters [itex]m, \, a[/itex] which can be reasonably interpreted as mass and specific angular momentum. But this detour into an approximation is not required! In fact, the parameters m, a in the Kerr solution can be determined by applying the well-motivated formalism (which is however limited to axially symmetric models) of Komar; it is not neccessary to take any weak field limits; rather one computes the Komar mass and Komar spin of the exact Kerr and Kerr-Newman solutions and then chooses the parameters so that these are m and ma respectively.

    So this sentence is really terribly misleading, IMO.

    Rather, in any classical field theory we expect to see field equations having the form
    [tex]
    ({\rm something} \, {\rm computed} \,{\rm from} \, {\rm the} \, {\rm field} ) = {\rm source}
    [/tex]
    In a linear field theory, we expect to be able to apply a Green function approach to work "inside outwards", directly determining the source-free exterior from a given configuration of sources in some compact region. But even there, an "outside inwards" approach assuming some symmetry in the source-free exterior region will probably be easier. And in a nonlinear field theory, we may have little choice but to adopt such an approach if the "inside outwards" approach proves too hard. And in gtr, we have the additional technical challenges that due to the freedom of choice in local coordinate chart, and due to fundamental features of Lorentzian manifolds guaranteeing that "distance in the large" will be a difficult concept, finding mathematically helpful and physically transparent coordinate charts is not so easy. Adding in the difficulty of accounting for details in even defining a legal "initial state" of the source increases the technical difficulties by an order of magnitude.

    As it happens, it is possible to combine the "outside inward, assuming mathematically convenient symmetry" approach with the "inwards outside" approach, to some extent. In particular, for axisymmetric vacuum exteriors, one can try to define boundary values on the axis r=0 (not "zero distance" in the appropriate Ernst-Lewis chart, but something like an "axis of rotation" interrupted by a region corresponding to an event horizon) and work outwards to derive a unique exterior. Meinel and Neugebauer followed this appraoch to obtain the exact vacuum exterior produced by a rigidly rotating thin disk of dust particles (which reduces in a suitable limit to the expected Kerr vacuum, and in another limit to the Minkowski vacuum).

    So this sentence is also terribly misleading, IMO.
     
    Last edited: Jun 2, 2010
  5. Aug 26, 2010 #4

    Chris Hillman

    User Avatar
    Science Advisor

    "Hagihara observers don't exist"?

    In 1008.3553v1, Hioe and Kuebel claim that circular orbits ("stable" or otherwise) in the Schwarzschild exterior only exist for a small range of r just outside r=3m. See especially the top of p. 13 at
    Code (Text):

    arxiv.org/abs/1008.3553
     
    Unfortunately, the claim of Hioe and Kuebel that Hagihara observers don't exist at all--- that is, that timelike geodesics with constant r don't exist at all except in a small range just outside r=3m--- is easily seen to be wrong.

    It would be suspicious in any case because it directly contradicts Chandrasekhar and many other careful authors! But I intend to prove that they are wrong here. I'll give an elementary approach which allows one to analyze arbitrary timelike geodesic motion in any coordinate plane "through the origin". The only point I won't try to prove here is the fact that all timelike geodesic motion in the Schwarzschild vacuum is confined to some such plane (but that is proven in many textbooks, Chandrasekhar's monograph, etc.) But I'll write down a proper time parameterized geodesic sufficiently simple that any SA/M can easily plug it into the geodesic equations and verify by hand that it is indeed a timelike geodesic. Since it obviously corresponds to a circular orbit in the equatorial plane, Hioe and Kuebel are simply incorrect.

    Another comment before I begin: Hioe and Kuebel fail to mention most of a vast quantity of relevant work in the literature; IMO their citations indicate that they simply don't know the literature on the Schwarzschild vacuum well enough to write about this topic. It is in fact quite well known that
    • timelike and null geodesics in the Schwarzschild vacuum yield "planar trajectories"; that is, the trajectories of these spacetime curves, when projected to a hyperslice t=t_0, lie in a coordinate plane "through the origin",
    • Schwarzschild geodesics can be expressed in terms of elliptic functions,
    • some older discussions using effective potentials to analyze, not the Schwarzschild vacuum but a weak-field approximation to it, are deficient,
    • modern textbook discussions, which use effective potentials to analyze the exact Schwarzschild vacuum solution, don't always stress that "stability" refers to in-plane perturbations,
    • the so-called "innermost stable circular orbit" is not stable to small radial inward perturbations (more or less by definition!),
    • the behavior of general timelike and null geodesics has been well known for a long time; see e.g. Chandrasekhar's monograph for pictures illustrating the main points,
    • in particular, it is well known that some trajectories with decreasing r and small "angular momentum over energy" parameter have no turning point but plunge into the hole (generalizes the obvious case of purely radial inward motion, e.g. Lemaitre observers),
    • Chandrasekhar's monograph, while very careful, thorough, and authoritative, does not neccessarily offer the simplest approach to studying timelike geodesics (and the author wasn't trying to meet that goal but had other goals in mind).
    So not only is their main conclusion wrong, those more or less correct claims they make are already well known.

    OK, on to the debunking. To avoid repeating in detail things I've said more than once in the past, I'll only sketch a simple approach using frame fields:

    At each event in the exterior region, boost the frame field modeling static observers
    [tex]
    \begin{array}{rcl}
    \vec{f}_1 & = & \frac{1}{\sqrt{1-2m/r}} \; \partial_t \\
    \vec{f}_2 & = & \sqrt{1-2m/r} \; \partial_r \\
    \vec{f}_3 & = & \frac{1}{r} \; \partial_\theta \\
    \vec{f}_4 & = & \frac{1}{r \, \sin(\theta)} \; \partial_\phi
    \end{array}
    [/tex]
    first in [itex]\partial_r[/itex] direction by [itex]\alpha[/itex] and then in the [itex]\partial_\phi[/itex] direction by [itex]\beta[/itex], with these boost parameters depending only on r. Then with
    [tex]
    p = \sinh(\alpha), \; q = \sinh(\beta)
    [/tex]
    treated as undetermined functions depending only on r, the new frame field is:
    [tex]
    \begin{array}{rcl}
    \vec{e}_1 & = & \sqrt{1+q^2} \; \left(
    \frac{\sqrt{1+p^2}}{\sqrt{1-2m/r}} \; \partial_t
    + p \, \sqrt{1-2m/r} \; \partial_r \right)
    + \frac{q}{r \sin(\theta)} \; \partial_\phi \\
    \vec{e}_2 & = &
    \frac{p}{\sqrt{1-2m/r}} \; \partial_t
    + \sqrt{1+p^2} \, \sqrt{1-2m/r} \; \partial_r \\
    \vec{e}_3 & = & \frac{1}{r} \; \partial_\theta \\
    \vec{e}_4 & = & q \; \left( \frac{\sqrt{1+p^2}}{\sqrt{1-2m/r}}
    \; \partial_t
    + p \, \sqrt{1-2m/r} \; \partial_r \right)
    + \frac{\sqrt{1+q^2}}{r \sin(\theta)} \; \partial_\phi
    \end{array}
    [/tex]
    Now demand that p, q are such that
    [tex]
    \nabla_{\vec{e}_1} \vec{e}_1 = 0
    [/tex]
    in the equatorial plane [itex]\theta=\pi/2[/itex]; to prevent possible misunderstanding, see the sketch below. (By the spherical symmetry, our analysis holds true for any other coordinate plane "through the origin", and since all geodesic motion is planar in that sense, for any timelike geodesics). We find that
    [tex]
    q = J/r, \; \;
    p = -\sqrt{E^2/V-1}, \; \;
    V = (1-2m/r) \; (1 + J^2/r^2)
    [/tex]
    where V is of course the usual effective potential, J is the angular momentum in the equatorial plane of the test particle, and E is its energy (where strictly speaking, one obviously has to say more about where and how these quantities are to be measured, which I won't bother to do.)

    When p is nonzero, p < 0 corresponds to the phase of history (when it exists) when the observer is moving inwards (r is decreasing), and p > 0 for the phase (when it exists) where the observer is moving outwards. Some motions include only outward or only inward motion.

    This frame field is only valid where the chart is valid (on the exterior region), but a similar approach can be adopted for a frame field such as the slow-fall frame, written in Eddington influx chart, where the frame field and local coordinate chart are both valid on the future interior region as well as exterior region. In the case of planar motion which falls inside the event horizon, this allows us to continue the history of our observer.

    The case J = 0 recovers the motion of test particles which are moving purely radially (in particular, the Lemaitre observers who fall in freely and radially "from rest at r=infty"). For small J/E there is no turning point and such observers invariably enter the event horizon. The "shape" of the planar trajectories is in fact determined by J/E alone.

    Those of you who use GRTensorII under Maple can painlessly verify by direct computation that this procedure does indeed yield timelike geodesics representing planar motion in the equatorial plane. Furthermore, the curvature components measured by an inertial observer moving in the equatorial plane are
    [tex]
    \begin{array}{rcl}
    {E\left[\vec{e}_1\right]}_{ab} & = &
    \left[ \begin{array}{ccc}
    \frac{-2m}{r^3} \, \left( 1+\frac{3J^2}{2r^2} \right)
    & 0 & 0 \\
    0 & \frac{m}{r^3} \, \left(1+\frac{3J^2}{r^2} \right)
    & 0 \\
    0 & 0 & \frac{m}{r^3}
    \end{array} \right] \\
    && \\
    {B\left[\vec{e}_1\right]}_{ab} & = &
    \left[ \begin{array}{ccc}
    0 & 0 & 0 \\
    0 & 0 & \frac{-3mJ}{r^4} \, \sqrt{1+\frac{J^2}{r^2}} \\
    0 & \frac{-3mJ}{r^4} \, \sqrt{1+\frac{J^2}{r^2}} & 0
    \end{array} \right]
    \end{array}
    [/tex]
    It is striking that these expressions (which turn out to remain valid for observers who move into the future interior region) do not depend upon E. This generalizes a well known remark of MTW concerning purely radial motion.

    The given frame field is inertial (in the equatorial plane, assuming we have substituted the expressions for p,q I gave above), but the spatial frame is spinning, in general, as we move along an integral curve of [itex]\vec{e}_1[/itex]. Specifically, in the equatorial plane, the spatial frame is spinning with "spin axis" given by the spatial frame vector [itex]\vec{e}_3[/itex]. To fix that, we can introduce a secular spin about [itex]\vec{e}_3[/itex] with a rotation rate which is given by a third undetermined function of r only. Then demanding that the Fermi derivatives vanish gives the desired nonspinning inertial frame. This procedure does not change [itex]\vec{e}_1[/itex], whose integral curves give the world lines of our observers.

    To reiterate, so far I have been discussing the motion of general inertial test particles whose motion is confined to the equatorial plane (or by spherical symmetry, to any other coordinate plane "through the origin").

    Now, when [itex]E^2 = V[/itex], r is obviously constant (since the [itex]\partial_r[/itex] component drops out of [itex]\vec{e}_1[/itex]), so this condition gives the Hagihara observers who move in circular orbits (which are stable against small in-plane perturbations by the usual analysis in all the textbooks). So these observers certainly exist.

    The easiest way to find the frame of a Hagihara observer explicitly is probably to repeat the above but boosting the frame of a static observer only in the [itex]\partial_\phi[/itex] direction. Then we find that for a Hagihara observer orbiting at r=r0 (where r0 > 3m) we must take
    [tex]
    J = \pm \sqrt{ \frac{m \, r_0}{1-3m/r_0}}, \; \;
    E = \sqrt{1-2m/r_0} \, \sqrt{1+\frac{m/r_0^3}{1-3m/r_0}}
    [/tex]
    The frame field becomes (for one choice of sign):
    [tex]
    \begin{array}{rcl}
    \vec{e}_1 & = & \frac{1}{\sqrt{1-3m/r}} \; \partial_t
    - \frac{\sqrt{m/r}}{\sqrt{1-3m/r}}
    \; \frac{1}{r \, \sin(\theta)} \; \partial_\phi \\
    \vec{e}_2 & = & \sqrt{1-2m/r} \; \partial_r \\
    \vec{e}_3 & = & \frac{1}{r} \; \partial_\phi \\
    \vec{e}_4 & = &
    \frac{-\sqrt{m/r^3}}{\sqrt{1-2m/r} \, \sqrt{1-3m/r}}
    \; \partial_t + \frac{\sqrt{1-2m/r}}{\sqrt{1-3m/r}}
    \; \frac{1}{r \, \sin(\theta)} \; \partial_\phi
    \end{array}
    [/tex]
    Again, it must be understood that an integral curve of the timelike unit vector field here, [itex]\vec{e}_1[/itex], is only a timelike geodesic for the case [itex]\theta=\pi/2[/itex]; then we have the following simple expression for the proper time parameterized world line of a Hagihara observer orbiting at r=r0 where r0 > 3m:
    [tex]
    \begin{array}{rcl}
    t & = & t_0 + \frac{s}{\sqrt{1-3m/r_0}} \\
    r & = & r_0 \\
    \theta & = & \pi/2 \\
    \phi & = & \phi_0 - \frac{s \, \sqrt{m/r_0^3}}{\sqrt{1-3m/r0}}
    \end{array}
    [/tex]
    The other integral curves give circular orbits lying in some coordinate plane "parallel to" the equatorial plane, which naturally have a nonzero constant acceleration.

    Now the proper time parameterized integral curve I just wrote down is very simple, and anyone can easily plug it into the geodesic equations for the Schwarzschild vacuum, written in the Schwarzschild chart on the exterior region, and see that this is indeed a timelike geodesic. So the claim of Hioe and Kuebel is manifestly wrong: Hagihara observers do exist! Just as all the textbooks say.

    The "orbital frequency" for an observer orbiting at r=r0, as measured by a static observer "at r=infty", is given by
    [tex]
    \frac{d\phi}{dt} = \pm \sqrt{\frac{m}{r_0^3}}
    [/tex]
    (same as Kepler expression); the proper time frequency is however
    [tex]
    \frac{d\phi}{ds} = \pm \sqrt{1-3m/r0} \; \sqrt{\frac{m}{r_0^3}}
    [/tex]
    and then there is an additional problem in deciding how the orbiting observer is to determine that he has just completed "one orbit"--- in fact, what I just wrote is the orbital frequency with time measured by our Hagihara observer but completion of an orbit determined by a distant static observer, which may not be quite what we want. So... some things are just plain tricky! And something funny happens to the vorticity at r=6m, which is the location of the "last stable circular orbit"!

    As already noted for the general case, the given frame is inertial but it is spinning within the equatorial frame. By introducing a secular rotation with a rate which is given by some undetermined function of r only, and requiring that the Fermi derivatives wrt e_1 vanish, we can find the nonspinning inertial frames of the Hagihara observer at r=r0. Then the precession wrt the distant stars agrees with the expression found by Rindler and Perlick.

    An analysis somewhat similar to the above has been given in the research literature, for example by Luminet and Marck 1985. See also the three volume textbook by Hagihara on celestial mechanics.

    How did Hioe and Kuebel fall into error? This appears to be a textbook example of, ironically enough, the very category of error they decry in the introduction: improper reasoning from inappropriate parameterization coupled with a poorly chosen approximation scheme! That is, they express the timelike geodesics (possibly correctly, although I haven't checked) using elliptic functions and a certain parameterization of the "space" of such geodesics. Then they try to take "the Newtonian limit" and that appears to be where they went wrong.

    Their current eprint is based upon a preceding eprint and a paper published in Phy Rev. D, which back in the day was a good place to go for high quality papers on gtr. But recently I've been wondering whether, with the retirement of "the greatest generation" and the apparent failure of (most of) a new generation to learn what their predecessors learned during the Golden Age about gtr, even the best journals may soon no longer be able to field referees competent in classical gtr. I hope that conclusion is premature; certainly it is based upon
    • a so far modest number of papers which IMO would never have been published in a good journal in the old days,
    • a steady stream of arXiv eprints which seem to me to strongly suggest that few young physicists have bothered to attempt to become acquainted with the literature of the Golden Age or to learn what the old masters knew about exact solutions.
    Certainly excellent work on classical gtr continues to published in good journals. But those who know the literature of the Golden Age and also the preceeding decades will probably agree with me that the literature on gtr was a godawful mess until about 1960. And unfortunately too many arXiv eprints on classical gtr are basically "reinventing" the kind of errors many researchers made prior to 1960, which tends to suggest that the physics community is somehow moving backwards in time. To be sure, I can think of younger researchers to whom these negative remarks clearly do not apply, but still, it's frustrating for those of us who love books. I mean, most universities still have physics libraries with journals and things, and there is always Amazon for those who don't mind having their reading habits spied upon.

    For those of you who use Ctensor under Maxima, here is a Ctensor file which you can at least use to verify that the frame field I gave is a frame field on the exterior region of the Schwarzschild vacuum:
    Code (Text):

    /*
    Schwarzschild vacuum; Schwarzschild chart;
    general timelike geodesic motion in equatorial plane when put
        q = J/r
        p = +/- sqrt(EE^2/V-1)
        V = (1-2*m/r)*(1+J^2/r^2)

    */
    load(ctensor);
    cframe_flag: true;
    ratchristof: true;
    ctrgsimp: true;
    /* define the dimension */
    dim: 4;
    /* list the coordinates */
    ct_coords: [t,r,theta,phi];
    /* define background metric */
    lfg: ident(4);
    lfg[1,1]: -1;
    /* Declare the dependent and independent variables */
    constant(m);
    depends([p,q],[r]);
    /* Define the coframe covectors */
    /* only need enter the nonzero components */
    fri: zeromatrix(4,4);
    fri[1,1]: -sqrt(1+p^2)*sqrt(1+q^2)*sqrt(1-2*m/r);
    fri[1,2]:  p*sqrt(1+q^2)/sqrt(1-2*m/r);
    fri[1,4]:  q*r*sin(theta);
    fri[2,1]: -p*sqrt(1-2*m/r);
    fri[2,2]:  sqrt(1+p^2)/sqrt(1-2*m/r);
    fri[3,3]:  r;
    fri[4,1]: -sqrt(1+p^2)*q*sqrt(1-2*m/r);
    fri[4,2]:  p*q/sqrt(1-2*m/r);
    fri[4,4]:  sqrt(1+q^2)*r*sin(theta);
    /* setup the spacetime definition */
    cmetric();
    /* compute a matrix whose rows give frame vectors */
    fr: radcan(factor(expand(fr)));
    /* metric tensor g_(ab) */
    lg: radcan(factor(expand(lg)));
    /* compute g^(ab) */
    ug: radcan(factor(expand(trigsimp(invert(lg)))));
     
    And here is a Ctensor file for the general infalling observer in the Painleve chart, which shows that the expressions found above for the Riemann components remain valid in the interior:
    Code (Text):

    /*
    Schwarzschild vacuum; Painleve chart;
    General infalling observer in equatorial plane
        J   ang momentum of obsvr in eq plane
        E   energy of obsvr
    Both quantities "measured at r=infty"
    Notice that Riemann components do not depend upon E;
    this generalizes remark in MTW about purely radial boosts
    */
    load(ctensor);
    cframe_flag: true;
    ratchristof: true;
    ctrgsimp: true;
    /* define the dimension */
    dim: 4;
    /* list the coordinates */
    ct_coords: [tau,r,theta,phi];
    /* define background metric */
    lfg: ident(4);
    lfg[1,1]: -1;
    /* Declare some constants */
    constant(m);
    constant(J);
    constant(E);
    /* Define the coframe covectors */
    /* only need enter the nonzero components */
    fri: zeromatrix(4,4);
    fri[1,1]: -E;
    fri[1,2]:  (E*sqrt(2*m/r)-sqrt(E^2-(1-2*m/r)*(1+J^2/r^2)))/(1-2*m/r);
    fri[1,4]:  J*sin(theta);
    fri[2,1]:  sqrt(E^2-(1-2*m/r)*(1+J^2/r^2))/sqrt(1+J^2/r^2);
    fri[2,2]:  (E-sqrt(E^2-(1-2*m/r)*(1+J^2/r^2))*sqrt(2*m/r))/sqrt(1+J^2/r^2)/(1-2*m/r);
    fri[3,3]:  r;
    fri[4,1]: -J/r*E/sqrt(1+J^2/r^2);
    fri[4,2]:  J/r*(E*sqrt(2*m/r)-sqrt(E^2-(1-2*m/r)*(1+J^2/r^2)))/sqrt(1+J^2/r^2)/(1-2*m/r);
    fri[4,4]:  sqrt(1+J^2/r^2)*r*sin(theta);
    /* setup the spacetime definition */
    cmetric();
    /* compute a matrix whose rows give frame vectors */
    factor(ratsimp(radcan(fr)));
    /* metric tensor g_(ab) */
    lg;
    /* compute g^(ab) */
    ug: trigsimp(invert(lg));
    christof(false);
    /* Compute fully covariant Riemann components R_(mijk) = riem[i,k,j,m] */
    lriemann(true);
    /* Compute R^(mijk) */
    uriemann(false);
    /* Compute Ricci componets R_(jk) */
    ricci(true);
    /* Compute trace of Ricci tensor */
    tracer;
    /* Compute R^(jk) */
    uricci(false);
    /* Compute and display MIXED Einstein tensor G^a_b */
    /* For (-1,1,1,1) sig Flip sign of top row to get G^(ab) */
    einstein(false);
    cdisplay(ein);
    /* electroriemann tensor */
    expand(matrix([lriem[2,2,1,1], lriem[2,3,1,1],lriem[2,4,1,1]],
    [lriem[3,2,1,1],lriem[3,3,1,1],lriem[3,4,1,1]],
    [lriem[4,2,1,1],lriem[4,3,1,1],lriem[4,4,1,1]]));
    /* magnetoriemann tensor */
    matrix([lriem[2,4,3,1],lriem[2,2,4,1],lriem[2,3,2,1]],
    [lriem[3,4,3,1],lriem[3,2,4,1],lriem[3,3,2,1]],
    [lriem[4,4,3,1],lriem[4,2,4,1],lriem[4,3,2,1]]);
     
    Figure:
    • Sketch of the circular trajectories obtained from the integral curves of [itex]\vec{e}_1[/itex], where sphere indicates some sphere [itex]r=r_0[/itex] where [itex]r_0 > 3m[/itex].
     

    Attached Files:

    Last edited: Aug 27, 2010
  6. Sep 18, 2010 #5

    Chris Hillman

    User Avatar
    Science Advisor

    BRS: Gravitational collapse: more trouble ahead at PF, courtesy of the arXiv?

    I have too many balls in the air as it is (see the SA forum thread on "speed of gravity" and BRS threads on "Those Damnable Paradoxes" and "", plus a possible new thread on "Schaum Outline of Expansion and Vorticity"), but I have been distracted by another recent development in the arXiv; see
    Code (Text):

    http://arxiv.org/abs/1009.2877
     
    Thermal Behaviour of Euclidean Stars
    G. Govender, M. Govender, K. S. Govinder

    Bit o' background: Megan Govender has published with Roy Maartens, who has published with George Ellis and Malcolm MacCallum (certainly not kooks!), and who is in any case known to me as a mainstream gtr researcher, and all three authors have coauthored with various persons (e.g. Naresh Dadich, whom I have exchanged some emails with on the very type of solution to be discussed here) quite a few papers on spherically symmetric gravitational collapse (apparently intended as models of the formation of black holes and/or other compact objects).

    Nonethless, I have the difficult task of trying to clearly explain why I think they are clearly just not trying hard enough to be careful. Even worse, in some of these papers including the one just cited, they almost appear to be inviting readers to draw incorrect conclusions. I don't know whether this is intentional or not, but sooner or later, it seems likely that some gtr-ignorant fringe fan who reads the arXiv (yes, such persons exist) will spot the oddities and start posting complaints in PF that "the truth that BH's have been disproven [sic] is being censored [sic] from the conclusion section of arXiv eprints!" (Which would be funny in way, but not for very long.)

    I'll let curious SA/Ms search the arXiv under the three authors names to chase down their other papers, which seem to all be available as arXiv eprints. Some of the papers cited as references are also in the arXiv.

    The problem with the most recent eprint is their Fig. 3, which is displayed without explanation or comment, but which a naive person is likely to read as evidence that in a "more realistic" model of spherically symmetric gravitational collapse than the OS model, the collapse approaches asympotically a certain radius, as a function of coordinate time in a comoving chart. IOW, naive persons are likely to leap to the conclusion that the authors have "shown that black holes do not exist" [sic].

    I certainly hope the authors didn't mean to leave that impression, because their paper certainly does not show any such thing. One reason is that their model cannot be fairly characterized as "more realistic than the OS collapsing dustball", although at first glance one might get that impression. Another is that the authors appear to have not yet thought hard about the physical significance of what they have computed, e.g. their Fig. 3 should be replotted in terms of proper time of a fluid particle.

    The basic idea in this kind of solution is to match an interior (imperfect) fluid region, which models a ball of collapsing ordinary matter, to an exterior Vaidya null dust outflux region, which models the exterior of the collapsing ball together with outflowing massless radiation.

    Adopting a suitable polar spherical type chart (the choice made by the authors may not in fact be the best such choice): in the exterior, wrt a suitable frame field, the Einstein tensor has the form appropriate for a null dust
    [tex]
    G^{ab} = 8 \pi \;
    \left[ \begin{array}{cccc}
    \nu & \nu & 0 & 0 \\
    \nu & \nu & 0 & 0 \\
    0 & 0 & 0 & 0 \\
    0 & 0 & 0 & 0
    \end{array} \right]
    [/tex]
    where [itex]\nu[/itex], the energy density of the massless radiation, is a function of t,r only. In the interior, the Einstein tensor has the form appropriate for an imperfect fluid with radially outward heat flux, which might look like
    [tex]
    G^{ab} = 8 \pi \;
    \left[ \begin{array}{cccc}
    \rho & q & 0 & 0 \\
    q & p- \zeta \, \theta - 2 \eta \, \sigma & 0 & 0 \\
    0 & 0 & p - \zeta \, \theta + \eta \, \sigma & 0 \\
    0 & 0 & 0 & p - \zeta \, \theta + \eta \, \sigma
    \end{array} \right]
    =
    \left[ \begin{array}{cccc}
    \rho & q & 0 & 0 \\
    q & a & 0 & 0 \\
    0 & 0 & b & 0 \\
    0 & 0 & 0 & b
    \end{array} \right]
    [/tex]
    where [itex]\rho[/itex] is the mass-energy density of the fluid, p is the isotropic component of the pressure, q is the radial heat flux, [itex]\zeta[/itex] is the bulk viscosity of the fluid, [itex]\theta[/itex] is the expansion scalar of the congruence of world lines of the fluid elements, [itex]\sigma[/itex] is a scalar controlling the anisotropy of the pressure, [itex]\eta[/itex] is the dynamic viscosity of the fluid, and these are all functions of t,r only. One might need to add stress terms; see MTW equation (22.16d). Note that [itex]\rho, p, q, \eta, \zeta[/itex] should all be non-negative, and that we have specified no equation of state (EOS). And note that coordinate components won't look so simple!

    The authors's aim is apparently to find physically reasonable exact solutions and apply the formalism of thermodynamics on curved spacetimes (see MTW section 22.2) to derive conclusions about temperatures &c, and perhaps even to find a more realistic model of complete gravitational collapse than the OS dust ball (such a claim could be strengthened by showing that the OS dust ball arises as a special case of a wider family of exact solutions).

    In order to find exact solutions, it is traditional to play around by making various physically unmotivated but mathematically convenient assumptions and hoping that something nice happens. For example, Naidu and Govender gr-qc/0510013 made a simple Ansatz which leads to a model in which the "radius" of collapsing ball decreases linearly, and physical quantities diverge as the "radius" approaches zero in a physically unreasonable fashion. Other authors have assumed that the world lines of the fluid particles are timelike geodesics--- this happens for any dust solution but seems suspect in the context of the radial transport of energy and momentum.

    In the eprint under discussion, the authors make a different Ansatz, most simply stated as this: for spherically symmetric collapse, the vorticity vanishes, so a unique family of hypersurfaces everywhere orthogonal to the world lines of the fluid elements exists; assume that these are all locally flat (vanishing Riemann tensor of the hyperslices, which are Riemannian three-manifolds). The authors refer to an imperfect fluid ball which satisfies this condition as a "euclidean star". Because the resulting field equations are still hard to solve, the authors then make some further physically unmotivated assumptions which eventually leads to their solution. But one must always remember: "garbage in, garbage out". In this case, I believe their assumptions are physically so unrealistic that their model is actually much less realistic than the simple OS dust ball. But it might still have value as a simple model of the collapse of a ball of ordinary matter to a quasi-equilibrium state supported by internal pressure (and maybe other internal stresses), e.g. as a model of the formation of a brown dwarf (not a black hole).

    It should also be noted that decades of numerical simulations of supernovae, employing contributions from all relevant physical theories, have supported the notion of the formation of black holes from a collapsing supernova core, and these and other simulations suggest that the most significant modifications to predictions from an OS dust ball may come from the influence of nonzero rotation.

    The next thing to do in studying the question of what if anything their solution means physically would be, I think, to check for unphysical negative terms in the Einstein tensor, then for other unphysical behavior, then to reexpress time evolution in terms of the proper time of observers riding with the fluid, if neccessary changing coordinates or appealing to numerical solutions for various choices of their constants. Then, study the effect of different choices for how these affect the physical reasonableness or lack of same. And figure out which, if any, well known solutions are included as a special case. Assuming one has shown that the apparent asympotic approach to some surface radius is not due to using coordinate time in place of proper time of observers riding on the surface, explore the relation of the final state with well known static spherically symmetric perfect fluid solutions--- or, if pressure anisotropies persist, with phenomenological guesses at what an analog in gtr of Newtonian models of spherical bodies made of elastic material might look like.

    (Maybe I am being too harsh, but IMO the authors should already have done all these things.)

    In the end, I am confident that one of two things will happen:
    • where reasonable (if anywhere), their solution will turn out to model a fluid ball which radiates away some energy, contracts, and cools, eventually achieving as a quasi-equilibrium state a slowly cooling ball supported against its gravitational self-attraction by its internal pressure, with radius much larger than the Buchdahl radius, perhaps modeling a "brown dwarf",
    • their solution will be shown to be unrealistic, period (e.g. some quantity which should be non-negative turns out to always be negative).
    (Buchdahl radius: 9/8 times the Schwarzschild radius, the minimal physically permissible radius of a static perfect fluid ball, independent of EOS.)

    Again, I hope it is mere carelessness which has led the authors to have left the impression, in this and some other papers, that they think they might have thrown cold water on the notion that black holes exist in Nature. But in this and other papers they refer to "horizon-free collapse", which certainly seems to be a very odd remark in case they intend only to model such phenomena as contracting and cooling balls supported by internal pressure.
     
    Last edited: Sep 18, 2010
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook