Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Ginnungagap -- a massively parallel cosmological initial conditions generator

Sergey Pilipenko, Gustavo Yepes, Stefan Gottlöber, Steffen Knollmann

TL;DR

Ginnungagap addresses the need for massively parallel, open-source cosmological initial-condition generation with support for zoom-in and resolution-refinement. It constructs ICs by generating a Gaussian white-noise field, convolving it with the transfer function to obtain density and velocity fields, and applying the Zel'dovich approximation to place particles, with plans for improved perturbation theory. The code demonstrates robust halo-property consistency across resolutions for both Milky Way–scale and cluster-mass halos, enabling reliable high-resolution resimulations and constrained realizations. Its modular workflow, automated configuration tooling, and open-source license make Ginnungagap a practical backbone for large-scale cosmological simulations across evolving hardware.

Abstract

Ginnungagap is a fully parallel (MPI+OpenMP) code designed to generate cosmological initial conditions for simulations involving very large numbers of particles. It operates in several modes, including the creation of initial conditions with either uniform or spatially varying resolution (for "zoom-in" simulations). The initial conditions can be fully random or derived by extending the resolution of existing ones while preserving the large-scale structures. Ginnungagap is open source and modular, consisting of a collection of independent tools that can be used for a variety of tasks. In this paper, we describe the main features of Ginnungagap and present test results for different types of simulations prepared with it.

Ginnungagap -- a massively parallel cosmological initial conditions generator

TL;DR

Ginnungagap addresses the need for massively parallel, open-source cosmological initial-condition generation with support for zoom-in and resolution-refinement. It constructs ICs by generating a Gaussian white-noise field, convolving it with the transfer function to obtain density and velocity fields, and applying the Zel'dovich approximation to place particles, with plans for improved perturbation theory. The code demonstrates robust halo-property consistency across resolutions for both Milky Way–scale and cluster-mass halos, enabling reliable high-resolution resimulations and constrained realizations. Its modular workflow, automated configuration tooling, and open-source license make Ginnungagap a practical backbone for large-scale cosmological simulations across evolving hardware.

Abstract

Ginnungagap is a fully parallel (MPI+OpenMP) code designed to generate cosmological initial conditions for simulations involving very large numbers of particles. It operates in several modes, including the creation of initial conditions with either uniform or spatially varying resolution (for "zoom-in" simulations). The initial conditions can be fully random or derived by extending the resolution of existing ones while preserving the large-scale structures. Ginnungagap is open source and modular, consisting of a collection of independent tools that can be used for a variety of tasks. In this paper, we describe the main features of Ginnungagap and present test results for different types of simulations prepared with it.

Paper Structure

This paper contains 13 sections, 15 equations, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. ICs Preparation Paradigm
  3. Computing the density and velocity fields
  4. Creating the particle phase-space representation
  5. Increasing or reducing the resolution
  6. Zoomed initial conditions
  7. Managing complex workflows
  8. Code validation tests
  9. Milky Way–mass halo in a 100 Mpc/$h$ box
  10. Cluster-mass halos in a 1 Gpc$/h$ box
  11. Summary and conclusions
  12. Acknowledgements
  13. Increasing the Resolution of the White Noise Field

Figures (4)

  • Figure 1: Left panel: masses of the MW and M31 analogues (squares and circles, respectively) in the simulations with increasing resolution. Right panel: distances of these two halos from their positions in the highest resolution ($8192^3$) simulation. Dashed line shows the ICs grid cell size $L_{box}/N_{1D}$.
  • Figure 2: Top: mass accretion histories of the two MW-mass halos in zoomed simulations with effective resolutions $1024^3$ (solid lines), $2048^3$ (dashed lines), $4096^3$ (dotted lines) and $8192^3$ (dot-dashed lines) particles. Bottom: ratios of the main progenitor mass with respect to that in the highest resolution simulation.
  • Figure 3: Projected DM density of a single cluster simulated with the zoom-in technique with the resolutions, equivalent to (from top to bottom) $3840^3$, $7680^3$ and $15360^3$ particles in the box. The black bar on the left of the plots corresponds to 1 Mpc/h scale
  • Figure 4: Probability distribution functions measured with 324 cluster-mass halos. Left panel: mass ratio between $7680^3$ and $3840^3$ simulations. Right panel: distance between halo canters in $7680^3$ and $3840^3$ simulations.