alanrockwood
Member
- Joined
- Oct 11, 2006
- Messages
- 2,185
- Format
- Multi Format
Just a quick comment about polynomials: It is known that polynomials can sometimes show pathological behavior when fitting certain kinds of data. For example, a polynomial might go through every point in a data set but have wild swings between the data points that do not accurately represent the underlying function being fitted. This sort of thing tends to be more common when using high order polynomials, and if I am not mistaken experimental data, which generally contains noise, is likely to accentuate this problem.
Pade functions are another option for parameterizing data. Pade functions often work better than polynomials of comparable complexity in the sense of providing a smoother fit or interpolation as well as providing a better extrapolation outside the fitting range.
For all you non-geeky people, sorry for the geek talk, but for us geeks this sort of thing is fun.
Pade functions are another option for parameterizing data. Pade functions often work better than polynomials of comparable complexity in the sense of providing a smoother fit or interpolation as well as providing a better extrapolation outside the fitting range.
For all you non-geeky people, sorry for the geek talk, but for us geeks this sort of thing is fun.