Markov chains on graphs and Brownian motion (Q5939312): Difference between revisions

The authors formulate the model of a symmetric random walk with fixed step length along the edges of a graph, combined with a natural probabilistic rule for jumping over a vertex and choosing there a new edge (which can be the old one) for continuation of the walk. Then they show that by rescaling time and space (as in the classical random walk) and letting the step length tend to zero there is weak convergence to Brownian motion on the edges of the graph, the particle (loosely speaking) on arrival at a vertex choosing with equal probabilities each of the edges meeting there for continuation.