- #1

- 60

- 0

I wish to calculate:

- the expectation value of the walker's position after a total of n decisions.

- what happens as n approaches infinite?

You are using an out of date browser. It may not display this or other websites correctly.

You should upgrade or use an alternative browser.

You should upgrade or use an alternative browser.

- Thread starter chrisphd
- Start date

- #1

- 60

- 0

I wish to calculate:

- the expectation value of the walker's position after a total of n decisions.

- what happens as n approaches infinite?

- #2

Borek

Mentor

- 28,674

- 3,166

Writing a program that will simulate the walker so that you can test it and find numerical solution should take an hour. Perhaps two.

At least that's what I did in the past when faced with a related problem. See http://onlinelibrary.wiley.com/doi/10.1002/elan.1140040603/abstract

- #3

- 22,089

- 3,296

How are things that you want to calculate related to the fact walker stops at 0 - seems to me like this is unrelated to the problem. But perhaps I am missing something.

It is not unrelated to the problem at all and could have a profound impact on the problem. For example, in the special case that b=c, the condition will take care that it the end, the walker almost always ends up in 0. If the walker does not stop at 0, then the walker would end up visiting every position an infinite number of time.

Anyway, this is a classical problem with stochastic processes. For example, the situation can be modeled using a Markov chain. Do you know Markov chains?? What would be the transition matrix in this case?

- #4

- 60

- 0

Not sure how I could use Markov chains because the transition matrix would have infinite dimensions.

- #5

- 22,089

- 3,296

Not sure how I could use Markov chains because the transition matrix would have infinite dimensions.

I don't really see that as a problem, it just makes the calculations more difficult.

- #6

- 60

- 0

Do you know the procedure for finding expectation value once I've determined a transition matrix?

- #7

- 22,089

- 3,296

Do you know the procedure for finding expectation value once I've determined a transition matrix?

It's been a long time ago since I've done such a thing. I'll have a look at it and answer later, if nobody else answered.

- #8

Borek

Mentor

- 28,674

- 3,166

It is not unrelated to the problem at all and could have a profound impact on the problem. For example, in the special case that b=c, the condition will take care that it the end, the walker almost always ends up in 0. If the walker does not stop at 0, then the walker would end up visiting every position an infinite number of time.

OK. If so, for n approaching infinity, final position seems to be always zero and expected value of the displacement is -A.

Or am I wrong again?

- #9

- 60

- 0

- #10

chiro

Science Advisor

- 4,790

- 132

You need to setup a transition matrix with an absorbing state at 0 and the other states relevant to where you are and how many steps you move.

For a finite n you can construct a matrix with the right values (it will be very sparse).

In terms of where it should end up at infinite steps, if you can show the system will always end up at the absorbent state then you have shown that the probability that you will eventually end up at 0 is 1 given enough time.

I would suggest apart from the sparse matrix approach that you try and construct a series expansion based on a binomial tree (which was suggested) and then see if you can use this finite summation in terms of n to evaluate your expression.

The summation would be based on difference equation in which you can take that and see if you can get something global for evaluating the expectation in terms of n.

Share: