Direct Collapse to Supermassive Black Hole Seeds: Comparing the AMR and SPH Approaches

TL;DR

This study compares the AMR and SPH simulation methods in modeling direct baryonic collapse within dark matter halos for supermassive black hole formation, highlighting differences in resolution, timing, and computational resources.

Contribution

It provides a detailed comparison of AMR and SPH approaches in simulating direct collapse black hole formation, revealing differences in resolution, collapse timing, and computational efficiency.

Findings

SPH models collapse earlier due to higher gravitational resolution.

AMR develops slightly higher baryonic resolution during halo growth.

Differences in resolution affect collapse timing and properties of host halos.

Abstract

We provide detailed comparison between the AMR code Enzo-2.4 and the SPH/N- body code GADGET-3 in the context of isolated or cosmological direct baryonic collapse within dark matter (DM) halos to form supermassive black holes. Gas flow is examined by following evolution of basic parameters of accretion flows. Both codes show an overall agreement in the general features of the collapse, however, many subtle differences exist. For isolated models, the codes increase their spatial and mass resolutions at different pace, which leads to substantially earlier collapse in SPH than in AMR cases due to higher gravitational resolution in GADGET-3. In cosmological runs, the AMR develops a slightly higher baryonic resolution than SPH during halo growth via cold accretion permeated by mergers. Still, both codes agree in the buildup of DM and baryonic structures. However, with the onset of collapse, this difference in mass and spatial resolution is amplified, so evolution of SPH models begins to lag behind. Such a delay can have effect on formation/destruction rate of H2 due to UV background, and on basic properties of host halos. Finally, isolated non-cosmological models in spinning halos, with spin parameter {\lambda} ~ 0.01 - 0.07, show delayed collapse for greater {\lambda}, but pace of this increase is faster for AMR. Within our simulation setup, GADGET-3 requires significantly larger computational resources than Enzo- 2.4 during collapse, and needs similar resources, during the pre-collapse, cosmological structure formation phase. Yet it benefits from substantially higher gravitational force and hydrodynamic resolutions, except at the end of collapse.