Tags: OregonIons/mjolnir
Tags
image: fit optimizations Minor optimizations; the first uses a fixed path for the einstein summation (unfortunately not a great win, but a win nonetheless). The second optimization regards the `scale` parameter. Originally this was intended as a sort of `max beam intensity` parameter, as can be seen in the initial guess for it (array max - array min). However, in the fitting function I had also divided by the normalization for unit integral over all space of the Gaussian, making it into an integrated power. This meant that the initial guess was always a wild underestimate, which often caused the fit to fail due to maximum number of function evaluations. Removing this scaling prefactor improves the robustness of the fitting.