As it is mentioned in the question of the 1st link you posted, it's just a result of numeric precision of R. I didn't know about the Rmpfr package, but it seems very useful. What are the problems you're facing using it?
It'll be better to express these problems in form of a reproducible example. In case you don't know how to make one, here's a great post:
Just out of curiosity, why is the dimension 9137\times2? You've 9137 observations I guess, but why 2, and not 1?