Line point picking is the selection of a uniformly distributed point on a unit line segment
. This point
divides the line into line segments
of length
and
. If a set of points are thus picked at random, the
resulting distribution of lengths has a uniform
distribution on
. Similarly, separating the
two pieces after each break, the larger piece has uniform
distribution on
(with mean 3/4), and the smaller piece has uniform
distribution on
(with mean 1/4).
The probability that the line segments resulting from cutting at two independently and uniformly sampled points on a unit line segment determine a triangle is given by 1/4.
The probability density function and distribution function for the ratio of small to large pieces are given by
|
(1)
| |||
|
(2)
|
for . The raw moments
are therefore
|
(3)
|
where
and
is the digamma function. The first few are therefore
|
(4)
| |||
|
(5)
| |||
|
(6)
| |||
|
(7)
|
(OEIS A115388 and A115389). The central moments are therefore
|
(8)
|
where
and
is a Pochhammer symbol. The first few are therefore
|
(9)
| |||
|
(10)
| |||
|
(11)
|
This therefore gives mean, variance, skewness, and kurtosis excess of
|
(12)
| |||
|
(13)
| |||
|
(14)
| |||
|
(15)
|
The mean can be computed directly from
|
(16)
| |||
|
(17)
| |||
|
(18)
|
The probability density function and distribution function for the ratio of large to small pieces are given by
|
(19)
| |||
|
(20)
|
for . Paradoxical though it may be, this distribution
has infinite mean and other moments. The reason for this
is that a theoretical bone can be cut extremely close to one end, thus giving huge
ratio of largest to smallest pieces, whereas there is some limit for a real physical
bone. Conditioning the cut to be uniformly distributed on
, where
, ensures that each piece has length
at least
. The conditional mean
is then given by
|
(21)
|