This section is about methods of calculation derivative numerically. Description covers classic central differences, Savitzky-Golay (or Lanczos) filters for noisy data and original smooth differentiators.
- Central Differences
- Savitzky-Golay or Low-noise Lanczos differentiators
- Smooth noise-robust differentiators
The necessity of computing derivative numerically arises when its analytical form is unknown or cannot be derived from limited information available about the function. In the most common case only function’s values are known or can be calculated at any point. This situation requires development of special methods on computing derivative based on function’s values only. Such methods should be capable to give sensible derivative estimation even if data is corrupted by noise.