arXiv subjects

Jonathan Thornburg

Publications and source records attributed to Jonathan Thornburg.

24 records · Page 2Linked to original sources

arXiv2001

A New Transition between Discrete and Continuous Self-Similarity in Critical Gravitational Collapse

We analyze a bifurcation phenomenon associated with critical gravitational collapse in a family of self-gravitating SU(2) $σ$-models. As the dimensionless coupling constant decreases, the critical solution changes from discretely self-similar (DSS) to continuously self-similar (CSS). Numerical results provide evidence for a bifurcation which is analogous to a heteroclinic loop bifurcation in dynamical systems, where two fixed points (CSS) collide with a limit cycle (DSS) in phase space as the coupling constant tends to a critical value.

gr-qc↗

arXiv2000

A Multiple-Grid-Patch Evolution Scheme for 3-D Black Hole Excision

When using black hole excision to numerically evolve a fully generic black hole spacetime, most 3-D 3+1 codes use an $xyz$-topology (spatial) grid. In such a grid, an $r = \constant$ excision surface must be approximated by an irregular and non-smooth "staircase-shaped" excision grid boundary, which may introduce numerical instabilities into the evolution. In this paper I describe an alternate scheme, which uses multiple grid patches, each with topology $\{r \times ({\rm angular coordinates})\}$, to cover the slice outside the $r = \constant$ excision surface. The excision grid boundary is now smooth, so the evolution should be less prone to instabilities. With 4th order finite differencing, this code evolves Kerr initial data to ${\sim} 60M$ using the ADM equations; I'm currently implementing the BSSN equations in it in the hope that this will improve the stability.

gr-qc↗

arXiv2000

Type II Critical Collapse of a Self-Gravitating Nonlinear $σ$-Model

We report on the existence and phenomenology of type II critical collapse within the one-parameter family of SU(2) $σ$-models coupled to gravity. Numerical investigations in spherical symmetry show discretely self-similar (DSS) behavior at the threshold of black hole formation for values of the dimensionless coupling constant $\ccbeta$ ranging from 0.2 to 100; at 0.18 we see small deviations from DSS. While the echoing period $Δ$ of the critical solution rises sharply towards the lower limit of this range, the characteristic mass scaling has a critical exponent $γ$ which is almost independent of $\ccbeta$, asymptoting to $0.1185 \pm 0.0005$ at large $\ccbeta$. We also find critical scaling of the scalar curvature for near-critical initial data. Our numerical results are based on an outgoing-null-cone formulation of the Einstein-matter equations, specialized to spherical symmetry. Our numerically computed initial-data critical parameters $p^*$ show 2nd order convergence with the grid resolution, and after compensating for this variation in $p^*$, our individual evolutions are uniformly 2nd order convergent even very close to criticality.

gr-qc↗

arXiv1999

A 3+1 Computational Scheme for Dynamic Spherically Symmetric Black Hole Spacetimes -- II: Time Evolution

This is the second in a series of papers describing a 3+1 computational scheme for the numerical simulation of dynamic black hole spacetimes. We discuss the numerical time-evolution of a given black-hole-containing initial data slice in spherical symmetry. We avoid singularities via the "black-hole exclusion" or "horizon boundary condition" technique, where the slices meet the black hole's singularity, but on each slice a spatial neighbourhood of the singularity is excluded from the domain of the numerical computations. After first discussing some of the key design choices which arise with the black hole exclusion technique, we then give a detailed description of our numerical evolution scheme for spherically symmetric scalar field evolution, assuming that a black hole is already present on the initial slice. We use a free evolution, with Eddington-Finkelstein-like coordinates and the inner boundary placed at a fixed coordinate radius well inside the horizon. Our numerical scheme is based on the method of lines (MOL), where spacetime PDEs are first finite differenced in space only, yielding a system of coupled ODEs for the time evolution of the field variables along the spatial-grid-point world lines. These ODEs are then time-integrated by standard methods. We use 4th order finite differencing in both space and time, with 5 and/or 6 point spatial molecules (off-centered near the grid boundaries), and a Runge-Kutta time integrator. The spatial grid is smoothly nonuniform, but not adaptive. We present numerical black hole + scalar field evolutions showing that this scheme is stable, can evolve "forever" (we have gone to t > 4000m), and is very accurate. At a resolution Delta_r/r = 3% near the horizon, typical errors in g_ij(K_ij) at t=100m are <= 1e-5(3e-7), and the energy constraint is < 3e-5.

gr-qc↗

arXiv1998

A 3+1 Computational Scheme for Dynamic Spherically Symmetric Black Hole Spacetimes -- I: Initial Data

When using the black hole exclusion (horizon boundary condition) technique, $K$ is usually nonzero and spatially variable, so none of the special cases of York's conformal-decomposition algorithm apply, and the full 4-vector nonlinear York equations must be solved numerically. We discuss the construction of dynamic black hole initial data slices using this technique: We perturb a known black hole slice via some Ansatz, apply the York decomposition (using another Ansatz for the inner boundary conditions) to project the perturbed field variables back into the constraint hypersurface, and finally optionally apply a numerical 3-coordinate transformation to (eg) restore an areal radial coordinate. In comparison to other initial data algorithms, the key advantage of this algorithm is its flexibility: $K$ is unrestricted, allowing the use of whatever slicing is most suitable for (say) a time evolution. We have implemented this algorithm for the spherically symmetric scalar field system. We present numerical results for a number of Eddington- Finkelstein--like initial data slices containing black holes surrounded by scalar field shells. Using 4th order finite differencing with resolutions of $Δr/r \approx 0.02$ (0.01) near the (Gaussian) perturbations, the numerically computed energy and momentum constraints for the final slices are $\ltsim 10^{-8}$ ($10^{-9}$) and $\ltsim 10^{-9}$ ($10^{-10}$) in magnitude. Finally, we briefly discuss the errors incurred when interpolating data from one grid to another, as in numerical coordinate transformation or horizon finding. We show that for the usual moving-local-interpolation schemes, even for smooth functions the interpolation error is not smooth.

gr-qc↗

arXiv1996

Finding Apparent Horizons in Numerical Relativity

This paper presents a detailed discussion of the ``Newton's method'' algorithm for finding apparent horizons in 3+1 numerical relativity. We describe a method for computing the Jacobian matrix of the finite differenced $H(h)$ function by symbolically differentiating the finite difference equations, giving the Jacobian elements directly in terms of the finite difference molecule coefficients used in computing $H(h)$. Assuming the finite differencing scheme commutes with linearization, we show how the Jacobian elements may be computed by first linearizing the continuum $H(h)$ equations, then finite differencing the linearized (continuum) equations. We find this symbolic differentiation method of computing the $H(h)$ Jacobian to be {\em much} more efficient than the usual numerical perturbation method, and also much easier to implement than is commonly thought. When solving the discrete $H(h) = 0$ equations, we find that Newton's method generally converges very rapidly. However, if the initial guess for the horizon position contains significant high-spatial-frequency error components, Newton's method has a small (poor) radius of convergence. This is {\em not} an artifact of insufficient resolution in the finite difference grid; rather, it appears to be caused by a strong nonlinearity in the continuum $H(h)$ function for high-spatial-frequency error components in $h$. Robust variants of Newton's method can boost the radius of convergence by O(1) factors, but the underlying nonlinearity remains, and appears to worsen rapidly with increasing initial-guess-error spatial frequency. Using 4th~order finite differencing, we find typical accuracies for computed horizon positions in the $10^{-5}$ range for $Δθ= \frac{π/2}{50}$.

gr-qc↗