Critical phenomena in gravitational collapse
Critical phenomena in gravitational collapse
Overview
One key prediction of general relativity is that when matter is
sufficiently dense it must keep collapsing until it dissappears from
sight inside a black hole. A process such as the collapse of a massive
star can therefore have two qualitatively different outcomes:
either a black hole forms or it does not.
Consider fine-tuning any one parameter in the initial data, for example the
mass of the collapsing star, right to the threshold of black hole
formation. At the black hole threshold, interesting things
happen:
- Universality: The evolutions of initial data which can
be very different (for example different initial shapes of the
collapsing star), except that they are all fine-tuned to the
threshold, all converge to the same critical solution.
- This solution is periodic in the logarithm of spacetime
scale: it shows discrete or continuous self-similarity.
- Often by fine-tuning to the black hole threshold, the mass of the
black hole can be made arbitrarily small (almost all of the mass gets
blown off). The mass scales as a nontrivial power of distance to the
black hole threshold, and this critical exponent is independent
of the initial data.
Mathematically, these phenomena (universality and scaling) are closely
related to critical phase transitions in thermodynamics. The
critical exponents can be calculated renormalisation methods, but
uniquely in general relativity the action of the renormalisation group
is also a time evolution.
Fine-tuning generic initial data, for generic systems, to the black
hole threshold thus allows one to make arbitrarily small black holes,
and this means arbitrarily large spacetime curvature outside the black
hole. In the limit of perfect fine-tuning, a naked singularity
is created from regular initial data: classical relativity then breaks down (and presumably quantum gravity takes over).
These numerical results have forced us to
refine the cosmic censorship conjecture, which states that
naked singularities cannot be created in collapse from regular initial
data for reasonable systems: we now have to emphasise that this cannot
happen for generic initial data, because we now know that it does
happen for a codimension-1 set of smooth initial data, namely those which have
been fine-tuned to the black hole threshold. (See the
bet between John Preskill, Stephen Hawking and Kip Thorne.)
To read more:
Current research
Beyond spherical symmetry
Black holes have angular momentum as well as mass (but no other
properties), and we would like to know how that scales at the black
hole threshold. On the other hand, angular momentum resists
collapse. This suggests that the mathematical similarity with critical
phase transitions can be extended much further and gives rise to new
predictions: C. Gundlach, Critical collapse with angular momentum:
from critical exponents to scaling
functions,
Phys. Rev. D 65, 64019 (2002).
This has motivated my work with Thomas Baumgarte on axisymmetric fluid
collapse with angular momentum, starting with
Phys. Rev. Lett. 116,
22103 (2016), and continuing.
Next to scalar fields and perfect fluids, electromagnetic fields are also well-motivated and clean sources of curvature. But electromagnetic waves cannot be spherically symmetric, so, as with angular momentum, one must relax the symmetry at least to axisymmetry. I am working on this with Thomas Baumgarte and David Hilditch, starting with Phys. Rev. Lett. 123,
171103 (2019), and continuing.
Vacuum critical collapse
However, the ultimate goal is the formation of naked singularities in vacuum gravitational collapse, that is the implosion of ultr-strong gravitational waves under their own weight. This is unlikely to happen in nature, but puts the question really sharply: does classical general relativity break down for a codimension-one set of initial data, because a naked singularity develops? In vacuum one cannot blame the matter for this, so it is the cleanest system in which to ask this question.
It turns out that vacuum gravity is also the most difficult system in which to investigate the threshold of collapse, and hence potential naked singularities. Again, gravitational waves cannot be spherically symmetric, and one needs to relax at least to axisymmetry. But it turns out that numerical simulations are much harder in axisymmetric vacuum collapse than in axisymmetric collapse with matter, and we are still at an early stage here.
However, the phenoma we see are already quite complex. There are many indications of curvature scaling (and hence at the creation of naked singularities), but there indications both for and against this happening through (discrete) self-similarity, and universality is also currently unclear.
My own current approach is to use time evolutions in null coordinates. These have been used a lot in spherical symmmetry, but not so much beyond. They have a number of advantages:
- One does not need to solve the Einstein constraint equations in order to find initial data.
- Rather, the constraint equations are solved for the full metric at each time step, but in null coordinates this is much easier.
- Imposing boundary conditions is also much easier, and in critical collapse this can be used to shrink the numerical grid to adapt itself to the critical solution, giving rise to an outer boundary that is shrinking faster than the speed of like (and so is future spacelike).
To read more:
- The first numerical investigation of
critical phenomena in non-rotating axisymmetric vacuum collapse is
Abrahams
and Evans (1993), but a number of research groups have been unable to
reproduce those results, until two recent breakthrough papers
- Ledvinka and Khirnov (2021)
- Suarez Fernandez, Renkhoff, Cors Agullo, Bruegmann and Hilditch (2022)
- In Baumgarte, Gundlach and Hilditch (2023) we showed thate there is definitely no universality in vacuum critical collapse.
- In Baumgarte, Bruegmann, Cors, Gundlach, Hilditch, Khirnov, Ledvinka, Renkhoff and Suarez Fernandez (2023) we showed that three different codes agree quantitatively in specific critical collapse simulations, and that the still confusing variety of results is therefore not numerical error!
- Some prelimary results of my null code in spherical symmetry are in Gundlach, Baumgarte and Hilditch (2019). Three papers in axisymmetry are in preparation.
- Quite recently, there has also been interest in this from mathematical relativists working with the methods of PDE theory, for example Rodnianski and Shlapentokh-Rothmann (2019) and
Shlapentokh-Rothmann (2022) on self-similar spacetimes that generate naked singularities from regular initial data in vacuum gravity, or Merle, Raphael, Rodnianski and Szeftel (2019) on self-similar singularities arising from smooth initial data in the spherical Einstein-perfect fluid system, and a proof, in Kehle and Unger (2024) that the threshold solution can be an extremal black hole (in the Einstein-Maxwell-charged Vlasov system in spherical symmtrey), and a conjecture that this is generic.