Let
X=(X1,…,Xn) be a continuous semimartingale and let
b:Rn−1→R be a continuous function such that the process
bX=b(X1,…,Xn−1) is a semimartingale. Setting
C={(x1,…,xn)∈Rn|xn<b(x1,…,xn−1)} and
D={(x1,…,xn)∈Rn|xn>b(x1,…,xn−1)} suppose that a continuous function
F:Rn→R is given such that
F is
Ci1,…,in on
ˉC and
F is
Ci1,…,in on
ˉD where each
ik equals 1 or 2 depending on whether
Xk is of bound variation or not. Then the following change-of-variable formula holds:
F(Xt)=F(X0)+\Bigsumni=1\Bigintt012\(∂F∂xi(X1s,…,Xns+)+∂F∂xi(X1s,…,Xns−)\)dXis+\Bigsumni,j=1\Bigintt012\(∂2F∂xi∂xj(X1s,…,Xns+)+∂2F∂xi∂xj(X1s,…,Xns−)\)d⟨Xi,Xj⟩s+\Bigintt012\(∂F∂xn(X1s,…,Xns+)−∂F∂xn(X1s,…,Xns−)\)I(Xns=bXs)dℓbs(X)
where ℓbs(X) is the local time of X on the surface b given by:
ℓbs(X)=P−limϵ↓012ϵ\Bigints0I(−ϵ<Xnr−bXr<ϵ)d⟨Xn−bX,Xn−bX⟩r
and dℓbs(X) refers to the integration with respect to s↦ℓbs(X). The analogues formula extends to general semimartingales X and bX as well. A version of the same formula under weaker conditions on F is derived for the semimartingale (t,Xt,St)t≥0 where (Xt)t≥0 is an Itô diffusion and (St)t≥0 is its running maximum.