Model. A particle-mesh (PM) N-body code in comoving coordinates. Dark matter is a set of equal-mass particles; in a flat universe with scale factor a and E(a) = √(Ωm a-3 + 1 − Ωm) each particle obeys dx/da = p / (a3E), dp/da = −∇φ / (aE), with the Poisson equation ∇2φ = (3/2) Ωm δ / a (units H0 = 1). Here: line-mass 1/r gravity (a 2D slice of the universe: infinite parallel strings).
Numerics. Cloud-in-cell (CIC) deposit of particles on a 1282 grid, FFT Poisson solve with a spectral gradient, CIC interpolation of the force back to the particles (the same kernel, so momentum is conserved), kick-drift-kick leapfrog in a with da = 0.01.
Initial conditions. A Gaussian random field with P(k) ∝ kn exp(−(k/kc)2), scaled to the chosen linear rms σ today, started at a = 0.05 with Zel'dovich displacements x = q + D(a) ψ. The linear growth factor is D = (5/2) Ωm E ∫0a da'/(a'E)3 (D(1) = 1).
What you see. The green dashed line in the spectrum plot is the initial power spectrum grown by D2 (linear theory); the yellow line is the simulation. At small k they agree; at large k gravity has moved power from large to small scales (non-linear collapse into filaments and knots) — the gap between the lines is the non-linear growth. Dark energy (smaller Ωm) makes the growth curve flatten after a ≈ 0.6.
Honest simplifications. A toy initial spectrum (no baryon wiggles, no transfer function), a tiny particle number (16384) so haloes are not resolved below a couple of cells, a cubic box of arbitrary size (not tied to Mpc), no baryons or gas, and glowing galaxy sprites are drawn on the densest particles only for illustration. Dark matter is not directly visible; real maps come from lensing and galaxy surveys. The 2D version is a slice of a universe made of parallel strings, so its force law differs from real gravity.
Independent checks (run on load, same core):
| running… |