The discussion centers on the utility of custom routines for computing the 2D inverse Fourier transform, with an offer to share the code if there is interest. Concerns are raised about the built-in functions fft2 and ifft2 in MATLAB, questioning their effectiveness and accuracy in performing Fourier transforms. The critique highlights that these functions do not provide the expected output of a single variable Fourier transform, which should yield a function evaluated at a specific point k. The conversation emphasizes skepticism towards MATLAB's built-in functionalities, suggesting that they may not always meet user needs, as illustrated by the preference for manually calculating the Hilbert transform instead of relying on MATLAB's version.