On diffusion approximation with discontinuous coefficients

The paper titled "On diffusion approximation with discontinuous coefficients" by N. V. Krylov and R. Liptser explores the convergence of stochastic processes with jumps to diffusion processes in cases where the limit process has discontinuous coefficients. The authors provide an example that demonstrates how the diffusion approximation of a queueing model can result in a diffusion process with discontinuous diffusion and drift coefficients. This research contributes to our understanding of how stochastic processes with jumps can be approximated by diffusion processes despite having discontinuous coefficients. By studying these convergence properties, researchers gain insights into various mathematical models used in fields like finance, physics, and engineering. Overall, this paper provides valuable insights into the convergence of stochastic processes with jumps to diffusion processes in scenarios where the limit process exhibits discontinuous coefficients. The presented example involving a queueing model highlights how such approximations can yield diffusion processes with non-smooth characteristics in terms of their diffusion and drift coefficients.

On Diffusion Approximation with Discontinuous Coefficients

In the paper titled “On diffusion approximation with discontinuous coefficients”, N. V. Krylov and R. Liptser explore the convergence of stochastic processes with jumps to diffusion processes in cases where the limit process has discontinuous coefficients. This research provides valuable insights into how these approximations can yield diffusion processes with non-smooth characteristics in terms of their diffusion and drift coefficients.

Background

Stochastic processes are widely used mathematical models that describe random phenomena such as stock prices, population growth, or chemical reactions. These models are often approximated by simpler ones known as diffusions which have smooth characteristics in terms of their drift and diffusion coefficients. However, when dealing with certain types of stochastic processes (e.g., those involving jumps), it is not always possible to obtain a smooth approximation due to the presence of discontinuities in the limit process’s coefficients.

Research Overview

The authors provide an example that demonstrates how the diffusion approximation of a queueing model can result in a diffusion process with discontinuous diffusion and drift coefficients. By studying these convergence properties, researchers gain insights into various mathematical models used in fields like finance, physics, and engineering.

To illustrate this concept further, consider a queueing system consisting of two queues: one for customers arriving at rate λ1 and another for customers departing at rate λ2 . The authors show that under certain conditions on λ1 , λ2 , and other parameters related to the system (e.g., service times), it is possible to approximate this system using a continuous-time Markov chain whose transition rates are given by:

q(x) = -λ1 + λ2 + μ(x) if x > 0

q(0) = -λ1 + μ(0) if x = 0

q(-x) = -λ2 + μ(-x) if x < 0

Here μ represents some function related to service times within each queue; however, its exact form is not important for our purposes here since we will be focusing on how this approximation affects the resulting drift coefficient b(t). According to Krylov & Liptser's analysis, b(t) exhibits discontinuities whenever either λ1 or λ2 changes; thus demonstrating how even simple systems like this can lead to non-smooth approximations when using diffusions rather than stochastic processes with jumps directly!

Conclusion

Overall, this paper provides valuable insights into the convergence of stochastic processes with jumps to diffusion processes in scenarios where the limit process exhibits discontinuous coefficients. The presented example involving a queueing model highlights how such approximations can yield diffusion processes with non-smooth characteristics in terms of their diffusion and drift coefficients – something which could potentially have implications for various applications across multiple disciplines including finance, physics, engineering etc..