How does NumPy implement fast aggregation functions like sum and mean?

Aggregation functions are implemented as optimized C loops with optional parallelization.

Functions like sum, mean, and std operate using low-level C implementations that iterate over contiguous memory blocks. Some builds use multi-threaded BLAS or SIMD optimizations. NumPy also fuses operations to reduce intermediate allocations.