Entity – Hardware-agnostic Particle-in-Cell Code for Plasma Astrophysics. I: Curvilinear Special Relativistic Module
Hayk Hakobyan, Ludwig M. Böss, Yangyang Cai, Alexander Chernoglazov, Alisa Galishnikova, Evgeny A. Gorbunov, Jens F. Mahlmann, Alexander Philippov, Siddhant Solanki, Arno Vanthieghem, Muni Zhou · 2025
Read Entity – Hardware-agnostic Particle-in-Cell Code for Plasma Astrophysics. I: Curvilinear Special Relativistic Module by Hayk Hakobyan, Ludwig M. Böss, Yangyang Cai, Alexander Chernoglazov, Alisa Galishnikova, Evgeny A. Gorbunov, Jens F. Mahlmann, Alexander Philippov, Siddhant Solanki, Arno Vanthieghem, Muni Zhou (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-17710.txt.
Introduction
Hayk Hakobyan and ten co-authors introduce Entity, a free and open code for simulating plasmas in space, where charged particles and fields affect each other. Entity runs on many kinds of graphics and central processors through the Kokkos library, without tying users to one machine.
The focus here is a module that handles curved grids and fast, near light speed motion. Tests on standard plasma cases and on magnetized regions around compact objects check that the code stays stable, saves charge, and runs fast. Entity gives researchers a flexible base for new methods and grid shapes.
Catalog record
| CREATOR | Hayk Hakobyan, Ludwig M. Böss, Yangyang Cai, Alexander Chernoglazov, Alisa Galishnikova, Evgeny A. Gorbunov, Jens F. Mahlmann, Alexander Philippov, Siddhant Solanki, Arno Vanthieghem, Muni Zhou |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 23 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-17710
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Entity – Hardware-agnostic Particle-in-Cell Code for Plasma Astrophysics. I: Curvilinear Special Relativistic Module Hayk Hakobyan, Ludwig M. Böss, Yangyang Cai, Alexander Chernoglazov, Alisa Galishnikova, Evgeny A. Gorbunov, Jens F. Mahlmann, Alexander Philippov, Siddhant Solanki, Arno Vanthieghem, Muni Zhou Abstract Entity is a new-generation, fully open-source particle-in-cell (PIC) code developed to overcome key limitations in astrophysical plasma modeling, particularly the extreme separation of scales and the performance challenges associated with evolving, GPU-centric computing infrastructures. It achieves hardware-agnostic performance portability across various GPU and CPU architectures using the Kokkos library. Crucially, Entity maintains a high standard for usability, clarity, and customizability, offering a robust and easy-to-use framework for developing new algorithms and grid geometries, which allows extensive control without requiring edits to the core source code. This paper details the core general-coordinate special-relativistic module. Entity is the first PIC code designed to solve the Vlasov-Maxwell system in general coordinates, enabling a coordinate-agnostic framework that provides the foundational structure for straightforward extension to arbitrary coordinate geometries. The core methodology achieves numerical stability by solving particle equations of motion in the global orthonormal Cartesian basis, despite using generalized coordinates like Cartesian, axisymmetric spherical, and quasi-spherical grids. Charge conservation is ensured via a specialized current deposition technique using conformal currents. The code exhibits robust scalability and performance portability on major GPU platforms (AMD MI250X, NVIDIA A100, and Intel Max Series), with the 3D particle pusher and the current deposition operating efficiently at about 2 nanoseconds per particle per timestep. Functionality is validated through a comprehensive suite of standard Cartesian plasma tests and the accurate modeling of relativistic magnetospheres in curvilinear axisymmetric geometries. show][email protected] I Introduction Particle-in-cell (PIC) algorithms ( Dawson_1983 ; Hockney.Eastwood_1988 ; Birdsall.Langdon_1991 ; Villasenor.Buneman_1992 ) have been used extensively since the end of the last century to study plasma phenomena ranging from laboratory plasma applications ( Verboncoeur_2005 ; Esarey.etal_2009 ; Macchi.etal_2013 ) , material science ( Ding.etal_2020 ; Ngirmang.etal_2025 ) , space physics ( Lapenta_2012 ; Li.etal_2021 ) , and, ultimately, astrophysics ( Zenitani.Hoshino_2005 ; Spitkovsky2008a ; Sironi.Spitkovsky_2014 ; Guo2014 ; Zhdankin_2017 ; Comisso_2019 ) . Particularly in the past two decades, the astrophysical community has expressed a growing interest in PIC, due to its ability to reconstruct particle distribution functions and make clear predictions on observational appearances of astrophysical objects and environments. Despite the growing complexities and capabilities of modern PIC codes, there are clear bottlenecks in the applicability of the algorithm in astrophysics, especially when modeling large-scale dynamics. Firstly, in astrophysics, the separation of global scales of the systems to plasma-kinetic scales is often extremely large, and resolving both dimensions is – most of the time – infeasible. This bottleneck is somewhat mitigated in some of the codes, which introduce curvilinear (so-called “body-fitted”) grids that enhance the resolution in places where it is most necessary, while sacrificing it at large scales. Secondly, modern computing infrastructures are quickly moving towards more GPU-centric architectures, whereas most of the existing codes are explicitly designed to run on CPUs. And finally, code usability has long been an important barrier-of-entry for many PIC applications, with the existing codebases being greatly outdated, non-modular, and difficult (oftentimes, impossible) to modify and customize to more specific needs. In this work, we introduce the fully open-source particle-in-cell code written in C++ – Entity 1 1 1 Documentation: entity-toolkit.github.io – designed to address all the issues raised above, while still maintaining a high standard for usability, clarity, and customizability. The goal of Entity is to accommodate most of the use-cases of PIC in modern astrophysical research by implementing all of the features which have thus far been used by the community, while also providing a robust and easy-to-use framework for developing new algorithms. Currently, Entity supports a range of different coordinate grids (with a simple capability of adding new ones), including general relativity (details will be published in paper II). Entity also implements advanced field-solver stencils, and high-order particle shape functions (details will be published in paper III) which – due to our coordinate-agnostic design – work for an arbitrary grid geometry. We are also currently developing a set of modules for radiative, quantum-electrodynamic (QED), and hadronic processes, including two-body particle-productions and scatterings, the details for which will be published in future papers. Importantly, Entity is built on an architecture-agnostic framework – Kokkos – which allows for easy performance portability on various GPU and CPU architectures. In this paper, we introduce the basic philosophy behind the design of Entity , primarily focusing on the general-coordinate special-relativistic module. The paper is organized as follows. In Section II , we discuss the PIC algorithm for general curvilinear coordinates. Section introduces the general architectural structure of the code, describing the main data containers and archetypal kernels. In Section we discuss the main API – the problem generators – and describe how it enables the code’s customizability and adaptability to various setups. Performance of the code on three major GPU architectures and its scalability are presented in Section . Finally, in Section we describe a set of standard problem generators included with the code, which help benchmark and test various features of the algorithm. II Particle-in-cell Algorithm In Curvilinear Space Entity solves the Vlasov–Maxwell system of equations in flexible coordinate systems. In this section, we discuss numerical algorithms and implementations in special relativity. Currently, we support one-, two- and three-dimensional simulations in Cartesian coordinates and axisymmetric (2.5D) simulations in spherical coordinate systems, with applications to neutron star and black hole magnetospheres. We decided to avoid implementing 3D spherical geometry in favor of the so-called cubed-sphere numerical grid (currently under development), the details of which will be described in a future publication. II.1 Units and normalizations In this section, we describe the physical unit system we employ in Entity . Note, that internally, Entity uses the so-called code units (described in the next section). These are, ideally, invisible to the end-user, and all the conversions are – most of the time – performed implicitly. Most of the parameters defined in the configuration input file, or settings performed in the problem generator (described in section ), as well as the output, are either in physical or dimensionless units. Physical units in Entity are based on Gaussian units, which we further dimensionlessize using multiple fiducial quantities. First of all, c — the speed of light in vacuum — is set to unity, c=1 . The user then has to provide the extent of the box, e.g., (x_{\rm min},x_{\rm max}) in the 1D Cartesian case, or (r_{\rm min},r_{\rm max}) in 2D spherical geometry. These distances are defined in physical units , the interpretation of which is left to the user. 2 2 2 In spherical geometry, e.g., when modeling a magnetosphere of a neutron star, it is often advisable to use r_{\rm min}=1 , meaning that all the length-scales will be normalized to the radius of the star. Time in these units is then naturally defined in terms of the light-crossing time of 1 unit length: time to cross a 1D Cartesian box is then exactly x_{\rm max}-x_{\rm min} units of time. To further make the quantities dimensionless, we define the following fiducial quantities: n_{0} – fiducial number density, B_{0} – fiducial electric/magnetic field strength, m_{0} – fiducial mass,