Expand description
SIMD-accelerated rolling window operations
This module provides high-performance rolling window calculations using:
- Portable SIMD via the
widecrate (works on stable Rust) - Algorithmic optimizations (deque-based min/max, Welford’s algorithm for std)
- Smart thresholds to avoid SIMD overhead on small arrays
Functions§
- clip
- Clip values (auto-selects SIMD or scalar)
- diff
- Diff operation (auto-selects SIMD or scalar)
- pct_
change - Percentage change (auto-selects SIMD or scalar)
- rolling_
max_ deque - Deque-based rolling maximum - O(n) amortized complexity
- rolling_
mean - Rolling mean (auto-selects SIMD or scalar)
- rolling_
min_ deque - Deque-based rolling minimum - O(n) amortized complexity
- rolling_
std - Rolling std (auto-selects Welford algorithm, always optimal)
- rolling_
std_ welford - Welford’s algorithm for numerically stable rolling standard deviation
- rolling_
sum - Rolling sum (auto-selects SIMD or scalar)