Lagrange polynomial interpolation
Naive idea: Interpolate all points at once
- \(\forall t\in[t_0,t_{N-1}]\;,\;\;\displaystyle p(t)=\sum_{i=0}^{N-1} \alpha_i(t)\;p_i\)
\(\forall i\in[0,N-1]\;\; p(t_i)=p_i\)
- Degree of polynomial : \(N-1\)
- - Known solution: Lagrange polynomial
-
- \(\displaystyle p(t)=\sum_{i=0}^{N-1} \alpha_i(t)\;p_i\)
\(\displaystyle \alpha_i(t) = \prod_{k=0, k\neq i}^{N-1} \frac{t-t_k}{t_i-t_k}\)
- - Explanation: By construction \(\alpha_i(t_i)=1\) and \(\alpha_i(t_k)=0\)
- (+) Interpolate all points
- (-) Large oscillations between samples for large degree.
- (-) Non local influence