Tuesdays 10:30 - 11:30 | Fridays 11:30 - 12:30
Showing votes from 2016-05-06 12:30 to 2016-05-10 11:30 | Next meeting is Friday Aug 22nd, 11:30 am.
In cosmological $N$-body simulations, the representation of dark matter as discrete "macroparticles" suppresses the growth of structure. This effect occurs even on scales many times larger than the particle spacing; for example, modes above $k_{\rm Nyquist}/4$ inherit a 1 to 3% error in the matter power spectrum at $z=1$. Particle linear theory (PLT) (Marcos et al. 2006) analytically describes this effect for particle lattices and reveals that the continuum growing modes are not the proper lattice growing modes, which excites transient power. We develop initial conditions with the correct growing modes and additionally rescale these modes to account for growth suppression. We also introduce a scheme for computing second-order Lagrangian perturbation theory (2LPT) from direct force calculations without using Fourier transforms. We test rescaling and our 2LPT by comparing the $z=1$ haloes and power spectra with those of an oversampled reference simulation. The combination of rescaling and 2LPT achieves 1% accuracy in the power spectrum down to $k_{\rm Nyquist}$, versus $k_{\rm Nyquist}/4$ without rescaling or $k_{\rm Nyquist}/13$ without 2LPT. Rescaling also corrects a 5 to 10% deficit of haloes below 500 particles, just as 2LPT corrects an equal deficit above 1000 particles. Similarly, the combination of 2LPT and rescaling produces the most accurate (sub-percent) halo two-point correlation functions. We anticipate that our 2LPT will be useful for large simulations where FFTs are expensive and that rescaling will be useful for suites of medium-resolution simulations used, for example, in cosmic emulators and galaxy survey mock catalogs. Code to generate initial conditions is available at https://github.com/lgarrison/zeldovich-PLT.