TL;DR
This paper investigates the reliability of simulated halo mass accretion histories by testing their convergence to self-similarity in scale-free simulations. The study finds that the Rockstar halo finder provides more robust and self-similar mass accretion histories compared to COMPASO, especially for larger halos and later simulation times. This work establishes a benchmark for evaluating the accuracy of halo finders and simulations, crucial for cosmological studies.
What Problem Does This Paper Address?
Understanding the formation and evolution of cosmic structures, particularly dark matter halos, is fundamental to cosmology. A key aspect of this is the "mass accretion history" (MAH) of these halos – how they grow over time by merging with smaller structures or accreting diffuse matter. These histories are vital for interpreting observations of galaxy clusters, probing dark matter properties, and constraining cosmological parameters. However, the accuracy of these histories derived from simulations is paramount. A significant challenge is ensuring that these simulated histories are physically realistic and not artifacts of the simulation's resolution or the algorithms used to identify halos (halo finders). This paper addresses this by using a "scale-free" simulation, a simplified but powerful theoretical tool, to rigorously test the self-similarity of halo MAHs. In a scale-free universe, physical properties should exhibit self-similarity, meaning they can be rescaled to look the same across different scales and times. Deviations from this self-similarity in simulated MAHs indicate a lack of physical accuracy, pointing to issues with the simulation or halo finder.
Methods
The core of this study relies on a scale-free N-body simulation generated using the ABACUS code. Scale-free cosmologies are characterized by a power-law spectrum of initial fluctuations and an expansion law where all dimensionless physical properties should be self-similar. The authors define a "rescaled mass accretion history" ($\Gamma$) that is normalized by a characteristic non-linear mass scale ($M_{NL}$) derived from the simulation's initial conditions. This rescaling allows them to compare MAHs across different halo masses and simulation times.
The key innovation is the use of self-similarity as a benchmark for evaluating the robustness of halo MAHs. They define convergence to self-similarity as the absence of temporal evolution in the rescaled MAH. To quantify this, they employ a two-step convergence criterion:
- $\Delta_1$: Measures the variation in the rescaled MAH within a sliding window of 5 consecutive snapshots. If this variation is below a certain percentage (e.g., 10%, 5%, 2%, 1%), the region is considered potentially converged.
- $\Delta_2$: Compares the MAH of individual snapshots within the converged window to the average MAH of that window. If this deviation is also below the threshold, and at least three consecutive snapshots meet this criterion, the region is deemed converged to self-similarity.
They compare two popular halo finders: ROCKSTAR and COMPASO, applying their convergence criteria to the MAHs derived from each.
Datasets
The study utilizes a single, high-resolution (N=4096$^3$) N-body simulation of a scale-free cosmology with a spectral index $n=-1.5$. This simulation is a re-use of data from previous studies. No new datasets were introduced.
Key Insights and Ideas
- Self-similarity as a universal benchmark: The paper leverages the theoretical property of self-similarity in scale-free cosmologies as a powerful, model-independent tool to assess the physical accuracy of halo finders and simulations.
- Halo finder performance matters: The choice of halo finder significantly impacts the robustness and self-similarity of derived mass accretion histories. Rockstar consistently outperforms COMPASO in this regard.
- Convergence improves with simulation evolution: As the simulation progresses to later times (larger scale factors), the regions of self-similarity become larger and deeper, indicating that structures have had more time to settle into physically consistent states.
- Particle count is a key factor: The number of particles within a halo is a critical determinant of convergence. While Rockstar can achieve high convergence with as few as ~30 particles at late times, COMPASO requires significantly more (~70 particles for 5% convergence).
- Scale-free simulations provide upper bounds: While scale-free simulations are simplified, the robustness limits found here can serve as an upper bound for the performance of halo finders and simulations in more complex, realistic cosmologies like ACDM.
Key Results
- ROCKSTAR: Demonstrates superior self-similarity. It achieves 5% convergence for halos with as few as ~70 particles and 2% convergence with ~30 particles at late simulation times. Convergence is robust across a wide range of halo particle counts (~10$^2$ to 10$^5$) and simulation times.
- COMPASO: Shows weaker convergence. It requires at least 10% convergence for halos between ~10$^2$ and 10$^4$ particles. The convergence is less stable, particularly for high-mass halos at late times, where it exhibits a noticeable drop in MAH.
- Simulation Evolution: The regions of convergence to self-similarity expand significantly as the simulation evolves, particularly after the scale factor quadruples from the time of non-linear structure formation.
- Particle Count Bounds: For Rockstar, convergence is observed for halos with particle counts ranging from ~20 to ~2x10$^4$ at late times. For COMPASO, this range is narrower, approximately 5x10$^2$ to 6x10$^3$ for 2% convergence.
Limitations and Future Work
- Scale-Free vs. ACDM: The study uses a scale-free cosmology, which is a simplification. While the results provide upper bounds and insights, applying them directly to ACDM simulations requires caution, especially at low redshifts where deviations from EdS become significant. Future work could explore the dependence of convergence on the initial power spectrum and compare results across different cosmological models.
- Halo Finder Specificity: The findings are specific to the tested halo finders (ROCKSTAR and COMPASO). Further investigation into other halo finders and their performance is warranted.
- Physical Accuracy: The self-similarity test primarily assesses the robustness and consistency of the simulation and halo finder. It does not definitively prove physical accuracy, as a perfectly self-similar simulation might still be unphysical if the underlying scale-free model doesn't match reality. Future work should combine scale-free validation with non-scale-free validation and comparisons to real observational data.
- Computational Cost: The paper notes that COMPASO has a significant advantage in computational cost and speed, suggesting scenarios where its sub-optimal convergence might be acceptable. Further research could explore hybrid approaches or optimized halo-finding techniques.
Summary updated