In this notebook  https://github.com/suberlak/AOS/blob/main/AOS_DM-35922_cocenter_vs_recenter.ipynb   we find that the impact of shift applied at the imageCoCenter step (below 6 pixels) makes numerically no difference to the estimated Zernike coefficients. The magnitude of  co-center has to be increased beyond 9 pixels to make any  difference at all . The figure below shows results of a simulation of corner wavefront sensors. Regardless of donut location, the amount of cocenter shift needs to be increased beyond 9 px to make any difference from the baseline (no shift). 

Work documented in the notebook as well as ticket comments.

To close this out

• is the notebook in AOS analysis notebook repo?
• Can you add an introductory text including author and what versions of the software it runs under. See https://github.com/lsst-ts/ts_analysis_notebooks/blob/develop/aos/vignetting/wepVignettedPerformanceLimits.ipynb

Added an updated notebook (with introductory text and other structure) to https://github.com/lsst-ts/ts_analysis_notebooks/pull/31

Looks good to me

Updates look good

Adding a notebook link after the PR merge https://github.com/lsst-ts/ts_analysis_notebooks/blob/develop/aos/algorithm/imageCoCenterWithLsst.ipynb