I am looking and searching for a paper (I think published on JRS or JCRS, but I am not quite sure) that determined based on specific aberration maps of humans, the number of zernike coefficients that could be iteratively fitted and retrieved before getting numerical unstability and obtaining physically impossible values for the retrieved coefficients (the number depended on the specificmap, but was ~100 polynomials I remember, 12-13 radial orders).
Can anyone help me, and remind me the full reference of it? Thx