types.backends.ConvergenceType

types.backends.ConvergenceType()

SDE convergence type for numerical integration.

Types: - ‘strong’: Pathwise/strong convergence * Individual sample paths converge * E[|X_numerical - X_true|] → 0 * Needed for: Filtering, control synthesis, single trajectory accuracy * More expensive to achieve

• ‘weak’: Weak convergence
• Distributions/moments converge
• E[φ(X_numerical)] → E[φ(X_true)] for test functions φ
• Needed for: Monte Carlo, statistics, ensemble behavior
• Easier to achieve (higher order possible)

Order Comparison: - Euler-Maruyama: Strong order 0.5, weak order 1.0 - Milstein: Strong order 1.0 - SRA1: Weak order 2.0