We additionally assume that the state space
is partially ordered and has a
maximal element
and a minimal element
, i.e., there is a relation on such that
(a)
(b)
and
(c)
and
(d)
Furthermore, we impose the condition
that the update function
is monotonously nondecreasing with respect to the partial order
, i.e., for arbitrary
such that
we have
(91)
Let the innovations
be identical
with probability ,
i.e., we merely consider a single sequence
of independent and
-uniformly distributed random variables and define
.
For arbitrary
and
the
Markov chain
is recursively defined by
Sometimes the update function
is not
monotonously nondecreasing but nonincreasing with respect to the
partial order , i.e., for arbitrary
such that
we have
(98)
In this case the following cross-over technique turns out to
be useful.
Based on the update function
we
construct a new nondecreasing update function
which is given as
(99)
This function has the desired property as by (98) and
(99) we obtain for arbitrary
such
that
i.e.,
is nondecreasing if
is nonincreasing.
Let now
be an update function with
respect to the irreducible and aperiodic transition matrix
with ergodic limit distribution
.
Then the map
defined by
(99) is a valid update function with respect to the
irreducible and aperiodic two-step transition matrix
and it has the same
ergodic limit distribution
.
In the same way that was used to prove Theorem 3.24
one can show that the coupling time
is finite with
probability , i.e.,
and
for all
if
is
nonincreasing.