jitter test in PLL using oscilloscope?

could someone analyse why an oscilloscope with 12.5G bandwidth and 50GS/s can detect the ps level jitter? How it works?

it samples the signal over and over, and eventually builds up a good idea of what the signal looks like.

I think if you use memory feature of your scope, you can observe the jitter over a long period.

by sampling over and over, you get a time scale that is smaller than what you would expect. i.e. a 50 GHz scope should only show you 1/50GHz time steps...but you can do better with the equivalent time sampling...a math trick