`ldexp(x,n::Integer)` computes `x ⋅ 2^n`. It would be nice if this were extended to `ArbReal` and `ArbFloat`. Right now I'm using `x * ArbFloat(2)^n`.