Ito公式的证明很繁琐,暂时不写证明。完整的证明可以看Karatzas和Shreve在1991年的Brownian motion and stochastic calculus.2nd ed.。
V ∗ = { ( Y ( t ) , t ≥ 0 ) : 实 值 连 续 随 机 过 程 , 适 应 的 ( a d a p t i v e ) , 可 测 的 , 且 P ( ∫ 0 ∞ Y ( t ) 2 d t < ∞ ) = 1 } V^*=\{(Y(t),t \ge 0):实值连续随机过程,适应的(adaptive),可测的,且\mathbb{P}\left(\int_0^{\infty}Y(t)^2\mathrm{d}t<\infty\right)=1\} V∗={(Y(t),t≥0):实值连续随机过程,适应的(adaptive),可测的,且P(∫0∞Y(t)2dt<∞)=1}
定理:设 h ∈ V ∗ h\in V^* h∈V∗, ( g ( t ) , t ≥ 0 ) (g(t),t\ge 0) (g(t),t≥0)是一个适应的过程,且满足 ∀ T > 0 \forall T>0 ∀T>0, ∫ 0 T ∣ g ( t ) ∣ d t < ∞ \int_0^{T}|g(t)|\mathrm{d}t<\infty ∫0T∣g(t)∣dt<∞几乎处处成立。令 X ( t ) : = ∫ 0 t g ( s ) d s + ∫ 0 t h ( s ) d W ( s ) , t ≥ 0 , X(t):=\int_{0}^{t}g(s)\mathrm{d}s+\int_0^{t}h(s)\mathrm{d}W(s),t\ge 0, X(t):=∫0tg(s)ds+∫0th(s)dW(s),t≥0,设 F ∈ C 2 , 1 ( R × R + , R ) F\in C^{2,1}(\mathbb{R}\times\mathbb{R}^+,\mathbb{R}) F∈C2,1(R×R+,R), Y ( t ) = F ( X ( t ) , t ) Y(t)=F(X(t),t) Y(t)=F(X(t),t),则 Y ( t ) = Y ( 0 ) + ∫ 0 t ∂ F ∂ x ( X ( s ) , s ) h ( s ) d W ( s ) + ∫ 0 t ( ∂ F ∂ t ( X ( s ) , s ) + ∂ F ∂ x ( X ( s ) , s ) g ( s ) + 1 2 ∂ 2 F ∂ x 2 ( X ( s ) , s ) h 2 ( s ) ) d s , t ≥ 0. \begin{aligned} Y(t)=&Y(0)+\int_0^t\frac{\partial F}{\partial x}(X(s),s)h(s)\mathrm{d}W(s)\\ &+\int_0^{t}\left(\frac{\partial F}{\partial t}(X(s),s)+\frac{\partial F}{\partial x}(X(s),s)g(s)+\frac{1}{2}\frac{\partial ^2 F}{\partial x^2}(X(s),s)h^2(s)\right)\mathrm{d}s,t\ge 0. \end{aligned} Y(t)=Y(0)+∫0t∂x∂F(X(s),s)h(s)dW(s)+∫0t(∂t∂F(X(s),s)+∂x∂F(X(s),s)g(s)+21∂x2∂2F(X(s),s)h2(s))ds,t≥0.
定理:设h是一个 R d × m \mathbb{R}^{d\times m} Rd×m值过程,其分量都属于 V ∗ V^* V∗, ( g ( t ) , t ≥ 0 ) (g(t),t\ge 0) (g(t),t≥0)是一个 R d \mathbb{R}^d Rd-值适应的过程,且 ∀ T > 0 \forall T>0 ∀T>0, ∫ 0 ∞ ∥ g ( t ) ∥ d t < ∞ \int_0^{\infty}\|g(t)\|\mathrm{d}t<\infty ∫0∞∥g(t)∥dt<∞几乎处处成立。令 X ( t ) : = ∫ 0 t g ( s ) d s + ∫ 0 t h ( s ) d W ( s ) , t ≥ 0 , X(t):=\int_0^tg(s)\mathrm{d}s+\int_0^th(s)\mathrm{d}W(s),t\ge 0, X(t):=∫0tg(s)ds+∫0th(s)dW(s),t≥0,设 W W W是一个 m m m-维布朗运动, F ∈ C 2 , 1 ( R d × R + , R p ) F\in C^{2,1}(\mathbb{R}^d\times \mathbb{R}^+,\mathbb{R}^p) F∈C2,1(Rd×R+,Rp), Y ( t ) = F ( X ( t ) , t ) , t ≥ 0 Y(t)=F(X(t),t),t\ge 0 Y(t)=F(X(t),t),t≥0,则 Y ( t ) = Y ( 0 ) + ∫ 0 t D F ( X ( s ) , s ) h ( s ) d W ( s ) + ∫ 0 t ( ∂ F ∂ t ( X ( s ) , s ) + D F ( X ( s ) , s ) g ( s ) + 1 2 ∑ i , j = 1 d ∂ 2 F ∂ x i ∂ x j ( X ( s ) , s ) ( ∑ l = 1 m h i , l ( s ) h j , l ( s ) ) ) d s , t ≥ 0. \begin{aligned} Y(t)=&Y(0)+\int_0^tDF(X(s),s)h(s)\mathrm{d}W(s)\\ &+\int_0^t\left(\frac{\partial F}{\partial t}(X(s),s)+DF(X(s),s)g(s)+\frac{1}{2}\sum_{i,j=1}^{d}\frac{\partial^2F}{\partial x_i\partial x_j}(X(s),s)\left(\sum_{l=1}^{m}h_{i,l}(s)h_{j,l}(s)\right)\right)\mathrm{d}s,t\ge 0. \end{aligned} Y(t)=Y(0)+∫0tDF(X(s),s)h(s)dW(s)+∫0t(∂t∂F(X(s),s)+DF(X(s),s)g(s)+21i,j=1∑d∂xi∂xj∂2F(X(s),s)(l=1∑mhi,l(s)hj,l(s)))ds,t≥0.其中 D F = ( ∂ x i F j ) 1 ≤ i ≤ d , 1 ≤ j ≤ p DF=(\partial_{x_i}F_j)_{1\le i\le d,1\le j\le p} DF=(∂xiFj)1≤i≤d,1≤j≤p表示 F F F的Jacobian矩阵。