It’s an interesting question how one can generalize classical continued fractions for real numbers to the p-adic setting. There are many curious p-adic continued fraction algorithms out there. I am going to demonstrate a curious construction of convergents whose limit is a root of in . It is probably an invention of a bicycle, so if you recognize where this result appeared before, let me know!
Theorem
Let be a rational prime such that and let be a root of in . Let be chosen so that and , and define
Then for all , so the sequence converges to . Furthermore, for all , the rational has the smallest possible naive height, i.e.,