0016–2663/04/3802–0079c
2004 Plenum Publishing Corporation 79
Functional Analysis and Its Applications, Vol. 38, No. 2, pp. 79–87, 2004Translated from Funktsional
nyi Analiz i Ego Prilozheniya, Vol. 38, No. 2, pp. 1–11, 2004Original Russian Text Copyright
c
by M. O. Avdeeva
On the Statistics of Partial Quotientsof Finite Continued Fractions
∗
M. O. Avdeeva
Received February 16, 2003
Abstract.
We reﬁne the remainder estimate in the asymptotic formula, earlier obtained in a jointpaper with V. A. Bykovskii, for Arnold’s problem about Gauss–Kuzmin statistics.
Key words:
continued fraction, partial quotient, convergent, Gauss–Kuzmin statistics.
Introduction
The following problem was posed by Arnold in [1, p. 17] (see also [2]).“I. Consider the integer points (
p,q
) lying in the positive quarter of the circle of radius
N
, i.e.,such that
p
2
+
q
2
N
2
,
p >
0, and
q >
0. Let us expand each rational number
α
=
p/q
in acontinued fraction (all these fractions are ﬁnite). Let us count the numbers of ones, twos, threes,etc., in the set of elements of all these fractions and ﬁnd the corresponding frequencies, whichdepend on
N
. Now let
N
be very large. Is it true that these numbers will be close to the Gaussianprobabilities
p
k
=1log2log
1 +1
k
(
k
+ 2)
?”This problem is closely related to the metric theory of Diophantine approximations, dynamicalsystems, and many other areas of mathematics.One formally represents the expansion of a rational number
r
in a ﬁnite continued fractionof length
s
=
s
(
r
) with integer part
t
0
= [
r
] and partial quotient
t
1
,...,t
s
(which are positiveintegers),
t
s
2, in the form
r
= [
t
0
,t
1
,...,t
s
]
.
(0.1)Let
H
k
(
N
) be the total number of partial quotients
t
i
(for all indices
i
, 1
i
s
(
r
)) coincidingwith a positive integer
k
(for all fractions
r
=
α
in the problem posed above). Then the number
H
(
N
) =
∞
k
=1
H
k
(
N
)is the sum of lengths of the continued fractions of the above-mentioned rational numbers
r
.In the present paper, we prove that
H
k
(
N
)
H
(
N
)=
p
k
+
O
1log
N
(0.2)for large
N
uniformly with respect to
k
. This reﬁnes the answer given to Arnold’s question in thepaper [3], where the weaker remainder estimate
O
((log
N
)
−
1
/
2
) was obtained.By deﬁnition, we set
A
i
/B
i
= [
t
i
,...,t
s
] (1
i
s
)
,
(0.3)where
A
i
is an integer,
B
i
is a positive integer, and gcd(
A
i
,B
i
) = 1. Let
s
x
(
r
) be the number of indices
i
in the expansion (0.1) for which
B
i
xA
i
. Since we always have 0
< B
i
A
i
, one canassume that
x
∈
(0
,
1]. It readily follows from (0.3) that the diﬀerence (
k
= 1
,
2
,
3
,...
)
s
(
k
)
(
r
) =
s
1
/k
(
r
)
−
s
1
/
(
k
+1)
(
r
) (0.4)
∗
This work was ﬁnancially supported by the RFBR (project No. 04-01-97000).