HYPERS-Global

The HYPERS code was developed in the early 2010's by Yuri Omelchenko as a general-purpose hybrid code for modeling multiscale fusion and space plasmas with significant kinetic ion effects. At the initial stage of its development HYPERS was predominantly used to simulate equilibria and collisions of compact magnetically confined plasmas (tori and plumes) and interactions of plasma flows with magnetized and unmagnetized obstacles.

The HYPERS-Global model describes global interactions of solar wind with the Earth's magnetosphere or unmagnetized planets. The HYPERS code is parallel (MPI-based) and compile-configurable (2D/3D in configuration space, 3D in velocity space). The ion components are simulated as kinetic macro-particles and the electrons are treated as a neutralizing inertialess fluid. The Maxwell and electron fluid equations are numerically solved in the radiation-free (Darwin) limit on a logically mapped (stretched or uniform) mesh. The macro-particles are advanced using a Particle-in-Cell method. Unlike conventional codes, which step numerical solutions in time, HYPERS implements a novel "operating system" for scheduling and executing numerical updates, EMAPS: Event-driven Multi-Agent Planning System. EMAPS combines discrete-event simulation and artificial intelligence to enable efficient updates of temporally disparate computational elements (particles, discretized fields, external models, etc.) on their own, self-adaptive timescales, without forcing their synchronous execution at predetermined time steps.

The HYPERS-Global model allows for multiple ion species in the solar wind plasma and approximates the ionosphere with a perfectly conducting or resistive sphere.

The HYPERS-Global model uses the solar wind ion inertial length (DI) as a scale for normalizing all distances, including mesh resolution. The Earth's dipole strength is scaled by the model based on a user specified (nominal) magnetopause distance (RM) given in terms of this characteristic length. To express spatial scales of interest in RE units, model users should judicially define their own ratio of RM/RE. HYPERS adjusts the rate of local updates in accordance with local physics, which makes simulations CPU efficient. However, the model performance may still be restricted by fast ion cyclotron and whistler time scales in the near-Earth space. A particular choice of domain partitioning may significantly impact parallel model performance. The HYPERS-Global model currently does not include detailed physics of ionosphere, plasmasphere and ring current. Thus, results close to the internal boundary may be non-physical. In general, the correctness of hybrid simulations depends on a judicial choice of computational parameters. Therefore, model users are strongly encouraged to contact the model developer directly before using the HYPERS-Global model in their research.

Physical inputs to the HYPERS-Global model are solar wind plasma and magnetic field conditions at the inflow simulation box boundary. The Earth's magnetic field is approximated by a constant magnetic dipole. Its strength and orientation are specified for the entire time interval simulated. The interplanetary magnetic field (IMF) and solar wind plasma properties are specified at the beginning of a simulation. These properties can be changed in simulation time to model interactions of discontinuities with the Earth's magnetosphere. The user can also specify simulation accuracy criteria, a resistivity model, a uniform or stretched mesh, and a variety of boundary conditions to be used in a simulation.

The HYPERS-Global output is produced in the form of text and binary data. Text files describe simulation run geometry and progress. Binary files for post processing and restarts are produced with specified cadences. The post-processing output contains snapshots of magnetic and electric fields, current densities, particle moments (ion densities, velocities, pressures, temperatures) and randomly selected particle distributions. The HYPERS-Global package provides post-processing tools and a collection of IDL routines that can be used for visualizing output data. Since particle data can be extremely large, a judicious choice must be made before starting a simulation when selecting output cadences and probabilities used for dumping particles for post-processing purposes.