N-body Galaxy Simulation Dataset for Chaos Analysis

This dataset contains the numerical simulation outputs presented in “The exponential growth of infinitesimal perturbations in the long-term evolution of simulated galaxies” by Asano & Portegies Zwart (arXiv:2604.12053). The simulations investigate the chaotic nature of self-gravitating N-body galaxy models by comparing reference runs and perturbed runs that differ only by an infinitesimal perturbation in the initial conditions. Each simulation consists of a pair (or ensemble) of runs: Reference run: r_xx_0 Perturbed runs: r_xx_i (i = 1, 2, ...) The perturbed runs are generated by applying a small displacement to a single particle in the initial condition of the reference run, leading to exponential divergence in phase space. List of simulations Run (i=1...3) Number of particles $N$ softening length $\epsilon$ (pc) r_00_0/r_00_i $1\times10^7$ 50.0 r_19_0/r_19_i $1.25\times10^6$ 141.4 r_20_0/r_20_i $2.5\times10^6$ 100.0 r_21_0/r_21_i $5\times10^6$ 70.7 r_xx_0 and r_xx_i (i ≥ 1) represent the reference and perturbed runs, respectively. Stellar particle data only. Data are stored in compressed NumPy binary format (.npz). Arrays have shape $T \times N \times 6$, where $T$ and $N$ are the number of snapshots and particles; the last axis corresponds to 6D phase-space coordinates $(x,y,z,v_x,v_y,v_z)$. Snapshot interval: $1\, \mathrm{kpc}/ (100 \mathrm{km\, s^{−1}}) \approx 9.78 \, \mathrm{Myr}$. Positions and velocities are given in dimensionless N-body units (Hénon 1971). Particle ordering is identical between paired runs, enabling direct comparison. Code examples https://github.com/tetsuroasano/ChaosTASPZ26