I'm aware of several related stack questions, but my case is a bit different because I assume that the system of SDEs is multiplied by 1-dimensional increment $dW$.

Suppose $\sigma$ is an $n \times n$ matrix and $dW$ is a 1-dimensional Wiener increment. The system of Stratonovitch SDEs is: $$dX = \sigma X \circ dW.$$ I'm unsure how to derive the correction term to convert the system to the Ito form. So far, I think the Ito system of SDEs should be $$dX = \sigma X dW + \frac{1}{2}\sigma^2 Xdt.$$


1 Answer 1


This is true.

In the one-dimensional case the transformation between the two types of integrals is, as we know, $$ \underbrace{\int_0^tY_s\circ\,dW_s}_{\text{Stratonovich}}=\underbrace{\int_0^tY_s\,dW_s}_{\text{Ito}}+\tfrac{1}{2}\langle W,Y\rangle_t\,. $$ In particular,

  • the quadratic covariation of any process $Z$ with a Stratonovich integral is the same as the quadratic covariation of $Z$ with the corresponding Ito integral.

In your case $$ Y_i(t)=(\sigma X(t))_i=\sum_{j=1}^n\sigma_{ij}X_j(t)\,,\quad\langle W,Y_i\rangle_t=\sum_{j=1}^n\sigma_{ij}\langle W,X_j\rangle_t\,. $$ Using the above bullet point and bilinearity of the covariation, \begin{align} \langle W,X_j\rangle_t&=\left\langle W,\int_0^.\sum_{k=1}^n\sigma_{jk}X_k(s)\circ\,dW_s\right\rangle=\sum_{k=1}^n\sigma_{jk}\left\langle W,\int_0^.X_k(s)\,dW_s\right\rangle_t\\[2mm] &=\sum_{k=1}^n\sigma_{jk}\int_0^tX_k(s)\,ds\,. \end{align} This shows \begin{align} dX_i(t)=\sum_{j=1}^n\sigma_{ij} X_j(t)\circ dW_t=\sum_{j=1}^n\sigma_{ij} X_j(t)\,dW_t+\frac12\sum_{j,k=1}^n\sigma_{ij}\,\sigma_{jk}X_k(t)\,dt\,. \end{align} In shorter matrix/vector notation this is the formula you wrote at the end of OP.

  • $\begingroup$ Thank you very much for the answer and details on the derivation. Could you recommend a good textbook source where I could learn more about these basics? $\endgroup$
    – MonteNero
    Commented Apr 23 at 17:45
  • 1
    $\begingroup$ A few other of my MSE answers are dealing with Stratonovich in detail as well. Typical references are the books by Shreve, Karatzas-Shreve, Protter, Oksendal. The most however I learned from an inexpensive book by Hui-Hsiung Kuo: Introduction to Stochastic Integration. This answer shows why. $\endgroup$
    – Kurt G.
    Commented Apr 23 at 17:52

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .