I am working through the proof of Itō's formula contained in the book "Continuous Martingales and Brownian Motion" by Revuz and Yor and am stuck at a point in the proof.

Theorem (Itō's formula). Let $X = (X^1, \ldots, X^d)$ be a continuous vector semimartingale and $F \in C^2(\mathbb{R}^d, \mathbb{R})$; then, $F(X)$ is a continuous semimartingale and $$ F(X_t) = F(X_0) + \sum_i \int_0^t \frac{\partial F}{\partial x_i}(X_s)dX_s^i + \frac{1}{2}\sum_{i,j}\int_0^t\frac{\partial^2 F}{\partial x_i\partial x_j}(X_s)d\langle X^i, X^j\rangle_s. $$

The equation immediately shows that $F(X)$ is a continuous semimartingale. Through stopping and approximation it suffices to prove the equation in the case where $F(\cdot)$ is a polynomial. Therefore it is enough to show the following claim, since the equation is true for constant functions.

Claim. If the equality holds for the polynomial $F$, then it holds for all polynomials $$G(x_1, \ldots, x_d) := x_{i_0}F(x_1, \ldots, x_d), \;\; 1 \leq i_0 \leq d$$

Using the integration by parts formula, one obtains $$ G(Y_t) = G(Y_0) + \int_0^t Y_s^{i_0}dF(Y)_s +\int_0^t F(Y_s)dY_s^{i_0} + \langle Y^{i_0}, F(Y)\rangle_t$$

This is now the point where I am not sure how to proceed exactly. Clearly, I want to "replace" the $dF(Y)_s$ in the first integral using Itō's formula for $F$ (which we assumed to hold) to obtain

\begin{align} G(Y_t) = G(Y_0) & + \sum_{i=1}^d \int_0^t Y_s^{i_0}\partial_i F(Y_s) dY_s^i + \int_0^t F(Y_s) dY_s^{i_0} \\ & + \frac{1}{2}\sum_{i,j}^d \int_0^t Y_s^{i_0}\partial_{i,j}^2F(Y_s)d\langle Y^i, Y^j\rangle_s + \langle Y^{i_0}, F(Y) \rangle_t \end{align}

My question now is, how to justify this in a rigourous manner or which intermediate steps I am missing, since it is not obvious to me why one can just make this formal replacement.

Thanks for any input.


I think what you are missing is the "composition property" (I do not remember the name exactly) of stochastic integrals (this is proposition IV.2.4 in Revuz-Yor, this is the case $L^2$ but it can be later be generalized), if $$ V_t = \int_0^t X_s dY_s $$ where $$ Y_t = \int_0^t W_s dZ_s $$ then $$ V_t = \int_0^t X_s W_s dZ_s$$ This can be reinterpreted in the differential notation as $$ dV_s = X_s W_s dZ_s$$ In your case $Y = F(X) = \text{Ito's formula}$ and $G = V$

  • $\begingroup$ Thanks, that is exactly what I have been looking for! $\endgroup$ – caligula Jun 27 '13 at 13:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.