Is it true that if $X(t)$ is an Ito process and $p(t)$ is non-stochastic, then the ordinary chain rule applies, that is, $$d(X(t)p(t)) = dX(t)p(t) + X(t)p'(t)dt?$$
2 Answers
Note
Let $X_t$ and $Y_t$ be two Ito processes. By application of Ito's lemma, we have $$d(X_t\,Y_t)=Y_t\,dX_t+X_t\,dY_t+d[X_t\,,\,Y_t]$$ Now, if $Y_t$ is of finite variation, then covariation $[X_t\,,\,Y_t]=0$. Let $p(t)$ be differentiable with continuous derivative, thus it has finite variation.In other words $$d[p(t),X(t)]=0$$
- 11,342
This can be obtained directly from Ito's product rule:
$$ d(X(t)Y(t)) = X(t)dY(t) + Y(t)dX(t) + dX(t)dY(t) $$
For illustration, assume your $dX(t)$ and $dp(t)$ has form:
$$ d(X_t) = \mu_1dt + \sigma_1dW_t \\ d(Y_t) = \mu_2dt $$
Since your $p(t)$ is non-stochastic, it only has derivative w.r.t time and thus the final term is :
$$ dX(t)dY(t) = \mu_1dt(\mu_2dt) + \sigma_1dW_t(\mu_2dt) = 0 $$ as a result of cross term and quadratic variation of time.
One other way to understand the product rule is to use the two-dimensional Ito formula and let $f(t,x,y) = xy$.
- 91
-
Could you please provide a reference where to look for this particular Itô's product rule. – Luisa Estrada May 16 '21 at 14:30
-
@leplata See for example Jacod&Shiryaev, Limit Theorems for Stochastic Processes (2003), Theorems I.4.47 and I.4.49. – Maximilian Janisch Apr 28 '22 at 11:13
-
The statement of the Ito-integration by parts formula is also rigorously given in Eberlein&Kallsen, Mathematical Finance (2019), Theorem 3.15. The informal notation used by the poster of this answer is explained in the same book at „Physicist‘s corner 3.17“. – Maximilian Janisch Apr 28 '22 at 11:15