In applied mathematics, construction of an irreducible Markov Chain in the Ising model is the first step in overcoming a computational obstruction encountered when a Markov chain Monte Carlo method is used to get an exact goodness-of-fit test for the finite Ising model.
Contents
- Markov bases
- Construction of an irreducible Markov chain
- Irreducibility in the 1 dimensional Ising model
- Conclusion
- References
The Ising model was used to study magnetic phase transitions at the very beginning, and now it is one of the most famous models of interacting systems.
Markov bases
For every integer vector
A Markov bases for the Ising model is a set
(i) For all
(ii) For any
and
for l = 1,...,k.
The element of
The paper published by P.DIACONIS AND B.STURMFELS in 1998 ‘Algebraic algorithms for sampling from conditional distributions’ shows that a Markov basis can be defined algebraically as in Ising model
Then by the paper published by P.DIACONIS AND B.STURMFELS in 1998, any generating set for the ideal
Construction of an irreducible Markov chain
We cannot get a uniform samples from
A simple swap is defined as
Z denotes the set of sample swaps. Then two configurations
with
for l = 1,...,k
The algorithm can be describe as:
(i) Start with the Markov chain in a configuration
(ii) Select
(iii) Accept
Although the resulting Markov Chain is possible cannot leave initial states, the problem does not arise for the 1 dimensional Ising model which we will introduce in the following. In high dimension we can overcome this problem by using Metropolis-Hastings algorithm in the smallest expanded sample space
Irreducibility in the 1-dimensional Ising model
Before prove of the irreducibility in 1-dimensional Ising model, we present two lemma below:
Lemma 1:The max-singleton configuration of
Lemma 2:For
and
for l = 1,...,k
Since
Conclusion
Even though we just show the irreducibility of the Markov chain based on simple swaps for the 1-dimension Ising model, we can get the same conclusion of 2-dimension or higher dimension Ising model.