Black holes (BH), the most extreme objects in the universe, are also the most extreme engines. They power relativistic jets.
In order to understand how relativistic jets are formed, one needs to understand the conditions in the vicinity of the black holes.
These are fed by accreting material (gas) which is heated as it falls into the gravitational potential of the BH. In the hot, ionized gas, small magnetic fields are amplified by magneto-rotational instability (MRI), generating strong magnetic fields.
In addition, the hot gas emits radiation via various processes, such as Bremsstrahlung, synchrotron and (inverse)_Compton.
To study these very complicated systems, since the early 2000’s people start using general-relativistic magneto-hydrodynamic (GR-MHD) codes.
These codes solve the MHD equation in curved space time, as is in the vicinity of the BHs, thereby enable to study the properties of the disk / jet and their relation to the BH properties (such as spin) or the initial conditions – disk structure, magnetic field configuration, etc.
Starting in 2018, we started to build a new GR-MHD code, which we call cuHARM (cuda-HARM). This code is especially optimized to run on graphical processing units (GPUs), making it one of the most efficient (maybe even the most efficient) code of its kind. Using this code, we can study how the properties of the jets in various objects such as GRBs, XRBs or AGNs are associated with the initial conditions.
The main developer of cuHARM is my post-doc, Damien Begue.
Why having our own code ? Well, the first obvious answer is because this makes us independent; even more importantly, codes always have limitations. Since we develop our own code, we are fully aware of them. And even more importantly, this enables us to put as much physics in it as we like.
Recently, we completed a major upgrade to cuHARM, by adding a radiative module to it. For that, we use the Variable Eddington Tensor (VET) method. This means that we solve the radiative transfer equation on a separate, geodesic grid:
While being more complex than the ‘closure scheme’ used by most codes that have radiation, this method accurately captures the angular distribution of the radiation field, thereby enabling to calculate the radiation contribution in regions of intermediate optical depth, which are very relevant to many objects.
For example, this method is required to accurately calculate the beam crossing problem, which cannot be calculated using averaging methods, such as the M1:
A description of cuHARM appears here.
We recently published the full explanation on the radiative module of this code: see
Below, shown a movie of the radiative energy density, given by the 00 component of the radiative stress energy tensor, calculated as the simulatiom evolves over 10,000 gravitational radii (r_g).