## Approximation du crochet de certaines semimartingales continues.(French)Zbl 0539.60045

Sémin. probabilités XVIII, 1982/83, Proc., Lect. Notes Math. 1059, 144-147 (1984).
[For the entire collection see Zbl 0527.00020.]
It is known that if X is an $$L^ 2$$-semimartingale on a complete right continuous stochastic basis ($$\Omega$$,$${\mathcal F},P,({\mathcal F}_ t)_{0\leq t\leq 1})$$ then $$\sum_{\sigma}E((X_{t_{i+1}}-X_{t_ i})^ 2| F_{t_ i}|)$$ tends to $$<X,X>_ t$$ in $$L^ 1$$ if the norm of the division $$\sigma(0=t_ 0<t_ 1<...<t_ n=t)$$ converges to 0. The author gives the following technical generalization of this fact:
Theorem. Suppose that $$X_ t=W_ t+\int^{t}_{0}H_ sds$$ where W is a Wiener process and H an $$L^ 2$$-bounded predictable process. Suppose also that there exists another $$L^ 2$$-bounded predictable process H’ such that $$\epsilon_ s(t)=E[((W_ s-W_ t)(s-t)^{-1}(H_ s-H_ t- H'\!_ t(W_ s-W_ t))| {\mathcal F}_ t]$$ uniformly tends to 0 in $$L^ 1$$ when s converges to t. Then $L^ 1-\lim_{n}(n/u)(\sum^{n- 1}_{i=0}E[(X_{u(i+1)/n}-X_{ui/n})^ 2| {\mathcal F}_{ui/n}])=\int^{u}_{0}H^ 2_ sds+\int^{u}_{0}H'\!_ sds.$
Reviewer: G.Zbaganu

### MSC:

 60G44 Martingales with continuous parameter 60G48 Generalizations of martingales

### Keywords:

predictable process

Zbl 0527.00020
Full Text: