Next: Multiplicative Reversible Version of
Up: Reversibility; Estimates for the
Previous: Recursive Construction of the
  Contents
Determining the Rate of Convergence under Reversibility
- Remarks
-
- Notice that (97) yields the following more precise
specification of the convergence estimate (96).
We have
as the column vectors
and
hence also the row vectors
where
form an orthonormal basis in
and thus by the Chauchy-Schwarz inequality
- Consequently,
 |
(98) |
- However, the practical benefit of the estimate (98)
can be limited for several reasons:
- The factor in front of
in (98) does
not depend on the choice of the initial distribution
.
- The derivation of the estimate (98) requires the
Markov chain to be reversible.
- It can be difficult to determine the eigenvalue
if the
number of states is large.
- Therefore in Section 2.3.5 we consider an alternative convergence estimate,
- which depends on the initial distribution
- and does not require the reversibility of the Markov chain.
- Furthermore, in Section 2.3.7 we will derive an upper
bound for the second largest absolute value
among the
eigenvalues of a reversible transition matrix.
Next: Multiplicative Reversible Version of
Up: Reversibility; Estimates for the
Previous: Recursive Construction of the
  Contents
Ursa Pantle
2006-07-20