Open antonysigma opened 2 hours ago
Note to self: I notice a slight variation between the Matlab code and the Python code. Not sure if they are relevant to the singular matrix issue.
Currently, the scaling factor is computed as
const vec scaling_factor = (iota(1.0, precision.num_basis + 1) * 3 - 2) * params.NA *
(precision.min_wavelength / wavelength);
which follows the original literature and the code in MicroscPSF-py
python code. The Matlab code is slightly different: the NA is normalized by the max NA parameter as params.NA / maxNA
.
When a lower NA objective lens is specified, the adaptive linear solver implementation [1] reports the following warning:
Microscope parameters:
[1] Sanderson and Curtin 2020, an adaptive solver for systems of linear equations. https://arma.sourceforge.net/armadillo_solver_2020.pdf