What is the solution for a one-dimensional random walk in potentials?

Click For Summary
SUMMARY

The discussion centers on solving a one-dimensional random walk in potentials defined by a probability distribution p(x) over the interval [0,1]. The walker starts at x=0.5 with a step size of dx=0.01, and the boundaries are defined such that p(0)=0 and p(1)=1. The challenge lies in calculating the steady-state distribution of the walker's position, which can be modeled as a Markov chain with transition probabilities based on p(x). The solution involves setting up a differential equation and considering the implications of boundary conditions on the walker's movement.

PREREQUISITES
  • Understanding of one-dimensional random walks
  • Familiarity with probability distributions and Markov chains
  • Knowledge of differential equations
  • Concept of steady-state distributions in stochastic processes
NEXT STEPS
  • Study Markov chain models and their applications in random walks
  • Learn about differential equations in the context of stochastic processes
  • Explore the concept of steady-state distributions in probability theory
  • Investigate the implications of boundary conditions on random walks
USEFUL FOR

Researchers in statistical mechanics, mathematicians studying stochastic processes, and anyone interested in modeling random phenomena in one-dimensional systems.

blue2script
Messages
44
Reaction score
0
I have a problem concerning a one-dimensional random walk in potentials. Assume a one-dimensional space [0,1] and a probability distribution p(x). At every point x we have a probability p(x) to go left and 1-p(x) to go right. Assume some smooth distribution of p(x) with boundaries p(0) = 0 and p(1) = 1. Now begin a random walk at x=0.5 with some step-size dx (e.g. d = 0.01) and capture the position of the walker at every time-step t. The boundary constraint assures that the walker remains inside [0,1].

I would assume that after sufficient time steps I get a steady distribution of the position of the walker. This would be equivalent to the probability distribution of finding the walker at some point in the potential.

However, I have yet no idea how to calculate this distribution from some given p(x). I tried to set up a differential equation using the fact that in the steady case the flow from point x to x + dx and back must be zero. However, I would get a pole at x = 0.5 which is pretty useless. I can post the calculation if someone is interested.

I would be glad for every hint how one could solve this problem. Thanks in advance!
Blue2script
 
Physics news on Phys.org
blue2script said:
I have a problem concerning a one-dimensional random walk in potentials. Assume a one-dimensional space [0,1] and a probability distribution p(x). At every point x we have a probability p(x) to go left and 1-p(x) to go right. Assume some smooth distribution of p(x) with boundaries p(0) = 0 and p(1) = 1. Now begin a random walk at x=0.5 with some step-size dx (e.g. d = 0.01) and capture the position of the walker at every time-step t. The boundary constraint assures that the walker remains inside [0,1].

I would assume that after sufficient time steps I get a steady distribution of the position of the walker. This would be equivalent to the probability distribution of finding the walker at some point in the potential.

However, I have yet no idea how to calculate this distribution from some given p(x). I tried to set up a differential equation using the fact that in the steady case the flow from point x to x + dx and back must be zero. However, I would get a pole at x = 0.5 which is pretty useless. I can post the calculation if someone is interested.
I assume a fixed step size ; we can let it tend to zero later.
Firstly, the boundary conditions don't warrant that the walker won't trip at 0 or 1 ( for instance, take d=1/300). If you want that constraint with initial position =1/2, the step size must be 1/2n for some integer n , making the walk discrete.
In such a case, this can be modeled as a Markov chain with 2n+1 states ( x=1/2 & other 2n possible positions ;with transition probabilities p(xn)&c. at each state. The limit of the transition probability matrix yields the long run probabilities.
Finally,letting d->0 gives a continuous time Markov process.

P.S. : Is this a model of any physicsl phenomenon? I ask out of sheer curiosity.
 

Similar threads

  • · Replies 3 ·
Replies
3
Views
790
Replies
29
Views
2K
  • · Replies 6 ·
Replies
6
Views
2K
  • · Replies 18 ·
Replies
18
Views
3K
  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 12 ·
Replies
12
Views
2K
Replies
9
Views
4K
  • · Replies 11 ·
Replies
11
Views
2K
  • · Replies 1 ·
Replies
1
Views
1K
Replies
3
Views
2K