Moving Mesh Hydrodynamics
UPDATE: The parallel 3D version is now up! It is still not too user friendly, so if you are intrested contact me and I'll be happy to help with it.
RICH (Racah Institute Computational Hydrodynamics) is a Voronoi based moving mesh code that I co-developed along with Almog Yalinewich. Our code is open source and available at github. We are currently finishing the 3D version of the code, which will be stable soon. Since Voronoi based codes are relatively new, it might take some time for a new user to understand how to correctly use them. Feel free to contact me with any question about how to run/install/customize our code.
Pressure Based Load Balancing
Pressure Based Load Balancing
When running large scale simulations it is crucial to maintain a good load balance while trying to minimize communications between processors. In this paper we introduce a new load balancing scheme that minimizes communications. The computational domain is decomposed such that each processor is assigned a Voronoi cell. After each time step, the Voronoi generating points are moved in a manner as to try to maintain an even load balance. This effectively moves a processor's Voronoi cell along with the flow of the workload (i.e. SPH particles or grid cells), thus minimizing the need to transfer data between processors. The movie below shows a demonstration of how the scheme works. The blue crosses represent the location of each processor Voronoi generating point, the green dots are the hydrodynamic particles (distributed exponentially in space) and the red lines show the Voronoi cells. The load balance is defined here as the maximum number of particles in a given cpu over the ideal number.
More Accurate Moving Mesh
Hydrodynamics on a moving Voronoi mesh provides many benefits. However, since this field is still in its infancy,
it is important to achieve a strong understanding of its benefits and pitfalls. We show that current implementations of Voronoi based moving mesh do not correctly take into account the change in the volume of cells. This leads to a lack of convergence when the resolution is high. By correcting the hydrodynamical variables to accurately reflect the change in the cell's volume we allow RICH to achieve second order convergence.
Ignition of detonation in accreted helium envelopes
Are the Largest Asteroids Primoridal?
The last stages of terrestrial planet formation are relatively uncertain. The predominate theory advocates a series of giant impacts between planetary embryos, which naturally leads to terrestrial planets having a slow spin. A competing theory suggests that terrestrial planets formed by forming a semi-collisional disk around them and accreting it, resulting in terrestrial planets having a relatively fast spin. Discriminating between the two theories is hard since in our solar system there are only four terrestrial planets. However, the largest asteroids have undergone the same formation mechanism as the terrestrial planets and can be used to help better understand how planets form. A caveat to this approach is that asteroids undergo mutual collisions inside the Main Belt and thus their spins might be altered during their lifetime. We show that under a few simplifying assumptions the spin distribution of asteroids is a Levy flight (a distribution characterized by a power law tail) and that there were not enough collisions to explain the observed spin rates. Since the largest asteroids spin relatively fast, this suggests that the formation of terrestrial planets involves some semi-collisional accretion.
Histogram of spins for the largest asteroids. The red line is the theoratical expectation from collisions during the lifetime of the solar system.
The Multi-Dimensional Structure of Radiative Shocks
Radiative shocks, behind which gas cools much faster than the dynamical time, play a key role in a range of astrophysical transients, including classical novae, young supernovae (SNe) interacting with circumstellar material (e.g.~SNe IIn), and binary star mergers. These shocks are susceptible to several instabilities (e.g. NTSI). We show that these instabilities reduce the effective temperature of the emitted radiation as well as potentially serve as an acceleration ground for high energy ions. Link to movies section.