'Mean Aversion' in Stochastic Differential Equations

Click For Summary

Discussion Overview

The discussion revolves around the exploration of stochastic differential equations (SDEs) that do not revert to a mean, contrasting with models like the Ornstein-Uhlenbeck process. Participants are interested in identifying or deriving SDEs that can represent distributions with multiple peaks, such as bimodal or trimodal distributions.

Discussion Character

  • Exploratory, Technical explanation, Debate/contested

Main Points Raised

  • One participant questions the existence of SDEs that avoid mean reversion, seeking examples or derivations.
  • Another participant suggests the possibility of a bimodal or multi-peaked distribution as a potential model.
  • A participant proposes geometric or arithmetic Brownian motion as an example of an SDE that does not revert to a mean, noting its application in stock modeling.
  • There is a suggestion to modify geometric Brownian motion by using a bimodally distributed Wiener process instead of the standard normally distributed one.
  • Some participants express uncertainty about the existence of Wiener processes other than the standard Brownian motion.

Areas of Agreement / Disagreement

Participants do not reach a consensus on the existence of SDEs that represent bimodal or trimodal distributions, and there is uncertainty regarding the modification of Wiener processes.

Contextual Notes

Limitations include the lack of examples for non-mean-reverting SDEs and the dependence on the definitions of distributions and processes discussed.

Apogee
Messages
45
Reaction score
1
I had a brief question regarding SDEs. Typically, I've seen models like the Ornstein-Uhlenbeck process that generally revert back to the mean over time. However, I've been trying to find a stochastic differential equation/process that avoids the mean, such as a sharp increase followed by a sharp decrease. What examples are like this and/or how would one derive something like this?
 
Physics news on Phys.org
Precisely. Is there any SDE/SP that represents a bimodal/trimodal/etc. distribution?
 
Actually none come to my mind right now, although I'm sure there are.

One example of perhaps an SDE that does not fall to a mean is geometric (or arithmetic even) Brownian motion, which is basically a superposition of Brownian motion on top of geometric growth, often used to model stocks.
 
Could I just take geometric Brownian motion and instead of a Wiener process that is normally or lognormally distributed, use a Wiener process that is bimodally distributed?
 
I am not familiar with any Wiener process than the regular Brownian motion...which is normally distributed...o.o
 
Hahaha. Then just a random process that is bimodally distributed.
 

Similar threads

  • · Replies 5 ·
Replies
5
Views
2K
  • · Replies 5 ·
Replies
5
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 19 ·
Replies
19
Views
4K
  • · Replies 7 ·
Replies
7
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 31 ·
2
Replies
31
Views
4K