DiracToDiracApproximation

Python interface for deterministic Dirac mixture reduction.

This class reduces a set of M discrete Dirac components in ℝᴺ to L optimized deterministic Dirac components.

The reduction preserves the global structure of the input mixture while minimizing a distance metric computed in the C++ backend.

Available Variants

• approximate_double (single-threaded)
• approximate_thread_double (multi-threaded)
• approximate_function_double (custom weight functions)

Notes

• Input arrays must be NumPy arrays (float64).
• Shapes must match (M x N) and (L x N).
• The output array is modified in-place.