Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

What is the solution to this ODE (and SDE)?

  1. Apr 18, 2014 #1
    I'm trying to analyze the following Ito stochastic differential equation:

    $$dX_t = \|X_t\|dW_t$$

    where [itex]X_t, dX_t, W_t, dW_t \in \mathbb{R}^n[/itex]. Here, [itex]dW_t[/itex] is the standard Wiener process and [itex]\|\bullet\|[/itex] is the [itex]L^2[/itex] norm. I'm not sure if this has an analytical solution, but I am hoping to at least find an analytical expression for the expected value [itex]E[X_t][/itex].

    In order to gain intuition for this problem, I'm considering the following ordinary differential equation:

    $$\dot{z}(t) =\|z(t)\|b(t)$$

    where [itex]z(t), b(t) \in \mathbb{R}^n [/itex] and everything is completely deterministic. Does anyone know the analytical solution to this second equation, and under what conditions it exists?
  2. jcsd
  3. Apr 18, 2014 #2


    User Avatar
    Homework Helper

    It is most easily solved in spherical polar coordinates where [itex]\|z\| = r[/itex] so that [itex]z = r\mathbf{e}_r[/itex] and [itex]\mathbf{b} = b_r\mathbf{e}_r + \mathbf{b}_n[/itex] where [itex]\mathbf{b}_n \cdot \mathbf{e}_r = 0[/itex]. We then have [tex]
    \dot r \mathbf{e}_r + r \dot{\mathbf{e}_r} = r(b_r\mathbf{e}_r + \mathbf{b}_n).[/tex] Since [itex]\mathbf{e}_r[/itex] and [itex]\dot{\mathbf{e}_r}[/itex] are orthogonal we have [tex]
    \dot r = rb_r(t), \\
    \dot{\mathbf{e}_r} = \mathbf{b}_n.
    [/tex] The radial and angular components thus decouple and the radial component has solution [tex]
    r(t) = r(0) \exp\left( \int_0^t b_r(s)\,ds\right).
    [/tex] The angular component represents the motion of a point on the unit [itex](n-1)[/itex]-sphere. If [itex]\mathbf{b}[/itex] is continuous then a solution should exist, but finding a co-ordinate representation of it valid for all time may be impossible, as one can see from the angular components in [itex]\mathbb{R}^3[/itex], [tex]
    \dot \theta = b_\theta(t), \\
    \dot \phi = \frac{b_\phi(t)}{\sin \theta(t)}
    [/tex] with solution [tex]
    \theta(t) = \theta(0) + \int_0^t b_\theta(s)\,ds, \\
    \phi(t) = \phi(0) + \int_0^t \frac{b_\phi(s)}{\sin \theta(s)}\,ds,[/tex] which becomes invalid if ever [itex]\theta(t) < 0[/itex] or [itex]\theta(t) > \pi[/itex].
  4. Apr 18, 2014 #3
    Thanks a lot for the answer. Can you explain why [tex]
    \theta(t) = \theta(0) + \int_0^t b_\theta(s)\,ds, \\
    \phi(t) = \phi(0) + \int_0^t \frac{b_\phi(s)}{\sin \theta(s)}\,ds,[/tex] becomes invalid if
    ever [itex]\theta(t) < 0[/itex] or [itex]\theta(t) > \pi[/itex]?
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook