Why is preffered to use squared visibility instead of visibility in long-baseline interferometry, please?
The main thing is Signal to Noise ratio and also a bit of statistical stuff. It increases the SNR (Signal to Noise Ratio) which is of course good. It also provides noise averaging on similar note. Since noise is generally random, squared visibility allows to remove a significant portion of it. Going to the statistical side, you can assume a Guassian distribution and therefore simplify noise treatment. Squared visibility lets you have a larger brightness range by getting all the intensities and also is more accurate. Additionally, its easier to fix astromephric issues, systematic errors, or instrumental effects.