Rotation curve. In the disk plane each mass component gives a radial pull. Bulge: Plummer sphere, g = GM r/(r²+a²)^1.5. Disk: Miyamoto-Nagai, g = GM r/(r²+(a+b)²)^1.5. Halo: logarithmic, g = v₀² r/(r_c²+r²). A circular orbit needs v_c² = r g. Without the halo the curve falls off roughly like Kepler; a flat curve at large radius is the classic sign of dark matter.
Epicycles. A star on a slightly non-circular orbit oscillates radially at the epicyclic frequency κ² = r dΩ²/dr + 4Ω², where Ω = v_c/r. In a frame rotating at Ω_p its path is a rosette; Ω - κ/2 is the rate at which an elongated orbit precesses.
Spiral arms. The arms here are a rigidly rotating density wave (Lin-Shu idea): a logarithmic spiral potential Φ_s = -ε v₀² cosψ, ψ = m(φ - Ω_p t) + (m/tan i) ln(r/r₀), with m=2 arms and pitch angle i. Stars pass through the wave rather than being fixed to it, so the arms do not wind up. Gas feels the same potential plus drag towards circular motion, so it piles up along the arm and the dark dust lane sits slightly ahead of the stellar peak. Resonances: corotation where Ω = Ω_p, inner Lindblad where Ω - κ/2 = Ω_p.
Simplifications. Stars are test particles (no self-gravity, no star-star scattering), the wave amplitude is imposed, not self-consistent, the gradient of the amplitude taper is ignored, and the disk is razor-thin with no bar. Particle counts are tiny compared with 10¹⁰–10¹¹ real stars. Masses and scales are Milky-Way-like but illustrative, not a fit.