## Archive for the ‘Curve Fitting’ tag

## Partial Derivatives for Residuals of the Gaussian Function

I needed to get the partial derivatives for the residuals of the Gaussian Function this week. This is needed for a curve fit I’ll use later. I completely forgot about Maxima, which can do this automatically — so I did it by hand (Maxima is like Maple, but it’s free). I’ve included my work in this post for future reference. If you want a quick refresh on calculus or a step-by-step for this particular function, enjoy :D. The math below is rendered with MathJax.

The Gaussian Function is given by …

$$ f(x) = ae^{-\frac{(x-b)^2}{2c^2}} $$

*a*,*b*,*c*are the curve parameters with respect to which we differentiate the residual function*e*is Euler’s number

Given a set of coordinates I’d like to fit (x_{i}, y_{i}), i ∈ [1, m], the residuals are given by …

$$ r_i = y_i – ae^{-\frac{(x_i-b)^2}{2c^2}} $$

We want to get …

$$ \frac{\partial{r}}{\partial{a}}, \frac{\partial{r}}{\partial{b}}, \frac{\partial{r}}{\partial{c}} $$