In the first part we consider partial sums
where
the positve integers
are defined by the
expansion
of a real number
which is chosen according to some probability measure
.
Under some regularity conditions imposed on the strictly monotone
function
we prove that the suitably normalized
sum
has an
stable limit distribution
(
) and derive uniform bounds of the approximation
error provided
possesses a strictly positive,
Lipschitz continuous Lebesgue density. Special emphasis is put on
the case
for which
coincides with the
power sum
of the partial
quotients
and
for
(where
)
of the continuous fraction expansion
.
In the second part we present large deviation relations
(in the sense of H. Cramér) for the sequences
and
, where
( resp.
) is the
denominator (resp. denumerator) of the
th approximant
of
. The proofs of the results
are based on a method developed in [1].
[1] | HEINRICH, L. (1996) Mixing properties and central limit
theorem for a class of non-identical piecewise monotonic ![]() |
E-Mail: | heinrich@math.uni-augsburg.de |
Homepage: | www.math.uni-augsburg.de/stochastik/heinrich |