CCSE Members of the Electrodynamics Development Team

Revathi Jambunathan

Weiqun Zhang

Andrew Myers

Ann Almgren

Andy Nonaka

Don Willcox

Zhi Jackie Yao

Accelerator Modeling with WarpX

WarpX is an accelerator modeling code for the exascale being developed as part of DOE's Exascale Computing Program. CCSE researchers contribute to the core AMReX-based infrastructure for WarpX, and the implementation of sophisticated mesh refinement algorithms. Ongoing work includes extension of WarpX to work efficiently on hybrid CPU/GPU platforms.

The official ECP project page is here. The WarpX project is also a member of the NERSC NESAP-for-ECP program.

The project is led by Jean-Luc Vay of LBL's Applied Physics and Accelerator Technologies Division.

More information about the WarpX software can be found at

Astrophysical Plasma

In a new project which builds on the wave propagation methodology used in accelerator modeling, we are extending the capabilities of WarpX to model pulsar magnetosphere and relativistic magnetic reconnection relevant to various astrophysical phenomena.

Even though relativistic magnetic reconnection has been widely studied in the literature, the key mechanisms that govern the transfer of energy from the magnetic field to kinetic energy of accelerated charged plasma species are not well understood. Modeling the Harris-sheet magnetic reconnection problem using the ab-initio particle-in-cell approach will allow for the study of microscale kinetic behavior and its effect on the macroscale acceleration, reconnection rate, and plasmoid instabilities.

Pulsar magnetospheres and magnetic reconnection have never been studied with an AMR code; using this capability for modeling Maxwell's equations requires special care at the coarse-fine grid interface. The use of AMR will greatly reduce the computational cost since the length-scale of the magnetic reconnection system increases by nearly two orders of magnitude away from the current-sheet. The disparity in length-scales in further amplified for the pulsar magnetosphere simulation, where the difference between the smallest length-scale and radius of the neutron star is six orders of magnitude for a realistics pulsar.

Traditionally, to reduce the computational cost of modeling pulsars, the magnetic field energy of the system has been scaled down such that the length-scale difference is artificially decreased to two-orders of magnitude. However, decreasing the magnetic field strength directly affects the important pair-production processes that affects the prediction of gamma ray emission and current-sheet dynamics.

The extended WarpX code, with dynamic mesh refinement to capture the evolving current sheet of the magnetosphere, will be used to study the effect of energy-scaling on the gamma-ray emission and Poynting flux predictions for a rotating oblique pulsar.

For more information, contact Revathi Jambunathan.


Microelectronics research is rapidly gaining prominence as a strategically important direction in DOE and Berkeley Lab. See this document this document for more details. There are a number of DOE initiatives targeting the development of next-generation microelectronic devices, particularly in the fields of new memory and data devices, interconnects, innovative materials, and 5G components.

However, currently available modeling capabilities cannot effectively capture the multiphysics aspect of emerging microelectronics. Furthermore, increased spatial resolution demands require a shift toward even larger-scale simulations. As a result, emerging post-CMOS technologies often rely on trial-and-error development strategies due to the lack of adequate simulation tools. There is an ever-increasing need for higher-fidelity electromagnetic (EM) simulations via higher spatiotemporal resolution and/or improved coupling that can seamlessly incorporate new physics into algorithms for widely-used, standard models.

We are part of the two DOE-funded Microelectronics-CoDesign programs at Berkeley Lab (Click here for the full list of awards) . We are addressing address the need for enhanced modeling for more realistic devices by developing an algorithmically flexible capability that is performant on manycore/GPU-based supercomputers. See the ARTEMIS page for more information of the code package we are developing for microelectronics and quantum chip modeling.

For more information, contact Zhi Jackie Yao, Revathi Jambunathan, or Andy Nonaka.