On cyclic and nontransitive probabilities

Abstract: Motivated by classical nontransitivity paradoxes, we call an

n

-tuple

(x_{1}, d o t s, x_{n}) i n [0, 1]^{n}

extit{cyclic} if there exist independent random variables

U_{1}, d o t s, U_{n}

with

P (U_{i} = U_{j}) = 0

for

i o t = j

such that

P (U_{i + 1} > U_{i}) = x_{i}

for

i = 1, d o t s, n - 1

and

P (U_{1} > U_{n}) = x_{n}

. We call the tuple

(x_{1}, d o t s, x_{n})

extit{nontransitive} if it is cyclic and in addition satisfies

x_{i} > 1 / 2

for all

i

. Let

p_{n}

(resp.~

p_{n}^{*}

) denote the probability that a randomly chosen

n

-tuple

(x_{1}, d o t s, x_{n}) i n [0, 1]^{n}

is cyclic (resp.~nontransitive). We determine

p_{3}

and

p_{3}^{*}

exactly, while for

n g e 4

we give upper and lower bounds for

p_{n}

that show that

p_{n}

converges to

1

as

n o i n f t y

. We also determine the distribution of the smallest, middle, and largest elements in a cyclic triple.

Recommendations

Cites work

Cited in

(9)

This page was built for publication: On cyclic and nontransitive probabilities