Luminous Accretion Disks

Many black holes in our universe are surrounded by very hot gas. When this gas falls in it circularizes and forms an accretion disk. Radiation emitted by this accretion disk can be detected by ground and space based observatories and used to test the laws of general relativity and plasma physics at their extremes. The complexity of such disk-black hole systems warrants the use of highly complex numerical simulations. The most luminous of those accretion disks are especially challenging to simulate since they are very thin and their rotation axis is often misaligned with respect to the black hole spin axis. This requires extremely high resolutions and runtimes.

Luminous Accretion Disks with Toroidal Magnetic Fields

Here you see a general relativistic MHD (GRMHD) simulation of a 65 degrees tilted thin (H/R=0.02) black hole accretion disk at an unprecedented effective resolution (13440x4608x8192) performed with my GRMHD code H-AMR on OLCF Summit. Due to this high resolution I was able to include, for the first time, a toroidal magnetic field, which is more realistic but much harder to resolve than the poloidal magnetic fields used in my previous simulations. This makes it arguably the most expensive GRMHD simulation to date. It ran on 5400 NVIDIA Tesla V100 GPUs on OLCF Summit for several weeks.

I show on the left panel a 3D rendering of the accretion disk and on the right panel the corresponding 2D vertical slice of density along the x-z plane. The black hole spin (a=0.94) axis is oriented along the positive z-axis and 1 R_g is approximately the radius of the black hole. As one can see, the accretion disk tears apart (due to frame dragging of space-time by the spin of the black hole) multiple times into a rapidly precessing inner sub-disk surrounded by a slower precessing outer sub-disk. The precession of this inner sub-disk could perhaps explain mysterious quasi periodic oscillations in the light curves of black holes.

Furthermore, this simulation confirms that an accretion disk threaded with a toroidal magnetic field can align within a few gravitational radii (R_g) with the spin axis of the black hole (seen e.g. around t=70,000 R_g/c), as predicted by Bardeen&Petterson 4 decades ago and recently proven in my paper for thin disks threaded with a poloidal magnetic field. However, the Bardeen-Petterson effect only occurs sporadically and for much smaller radii than predicted analytically.

Luminous Accretion Disks with Poloidal Magnetic Fields

Here we perform a similar simulation to the one above, but include a strong poloidal magnetic field, which is a necessary ingredient to form a jet. As can be seen, the black hole drags the space-time around the disk, shredding this thin disk (H/R=0.03) totally apart. The jet, in red, gets disrupted due to differential precession between inner and outer disk. Jet precession has been invoked some types of quasi-periodic oscillations (QPOs) and has been recently confirmed (see my previous paper). As can be read in my latest paper, the violent collisions between disk and jets might lead to extra dissipation and transfer energy and angular momentum between the disk and jet.

Luminous Accretion Disks with Poloidal Magnetic Fields which are thicker

Here we perform a simulation of a tilted accretion disks threaded with a poloidal magnetic fields which is thicker (H/R=0.1) than the two examples (H/R=0.02-0.03) shown above. As can be seen, the disk almost stops precessing while it slowly aligns with the black hole. This demonstrates that disk tearing is necessary to produce a disk which precesses for the many periods necessary to explain quasi-periodic oscillations and justifies the order(s) of magnitude more expensive simulations described above. Details of this work are described in my paper.

Non-luminous Accretion Disks with Poloidal Magnetic Fields

The animation below shows a GRMHD simulation of a geometrically thick (H/R=0.3) accretion disk tilted by 30 degrees at low (model W-U) and high (model W-R) resolution. As can be seen, the disk barely shows any precession due to the torque of the spinning black hole (a=0.9375). The reason for this is that the disk is expanding in radius, which reduces the amount of torque the spinning black hole can exert on the accretion disk. Expansion in radius is only correctly captured at high resolutions. Details can be found in this paper.

Luminous Accretion Disks with 2T Radiation

Though the previously described simulations have led to ground-breaking new insights, they lacked a consistent treatment of radiation and thermal decoupling between ions and electrons that feeds back on the disk dynamics. Namely, the radiation was approximated by a cooling function derived from an overly simplistic analytical model. This cooling model does not take into account various radiation driven instabilities and does not allow for a self-consistent vertical structure to develop. This makes it very hard to benchmark such simulations against multi-frequency observational data. In addition, a combination of angular momentum cancellation in torn disks and angular momentum extraction by magnetized winds causes those disks to accrete up to ~10 faster than analytic predictions suggested (see image below; made by collaborator Gibwa Musoke). In this scenario, the density of the plasma drops and, due to the short infall timescale, it is expected to decouple into a two-temperature (2T) fluid composed of hot ions and cold electrons..

truncation (1).png

In the first video below you see the first 2-dimensional simulation (performed on a single A100 GPU) of a radiative 2T accretion disk using the newly developed version H-AMR. As can be seen, radiative cooling causes the geometrically thick disk to collapse into a thinner accretion disk, as would be expected theoretically. However, this collapse happens less quickly than in a similar simulation which did not include 2T effects (2nd video below). Within the coming months we will extend this work to 3D.

Previous
Previous

H-AMR: The Next-Gen GRMHD Code

Next
Next

Accretion Disk Microphysics