we find the distribution of real-life passwords obeys a similar law: For a password dataset, the rank r of a password and its frequency fr follow the equation: