Martingale difference arrays and stochastic integrals (Q1064610): Difference between revisions

The author considers two martingale difference arrays \((X_{ni})\), \((Y_{ni})\) and their stopped partial sum processes \((S_ n(t))_{t\geq 0},\quad (T_ n(t))_{t\geq 0}\). If their common distribution converges to that of \((S_ t,T_ t)_{t\geq 0}\) then conditions are given for the weak convergence of the stopped partial sum process of the MDA (\(\phi\) (\(\sum^{i-1}_{k=1}X_{nk})\cdot Y_{ni})_{n,i\in {\mathbb{N}}}\) to the stochastic integral \(\int^{t}_{0}\phi (S)dT\), for some function \(\phi\) : \({\mathbb{R}}\to {\mathbb{R}}.\) This result can be applied to important special cases: A functional limit theorem for diffusion approximations [cf. \textit{T. Lindvall}, J. Appl. Probab. 9, 445-450 (1972; Zbl 0238.60063)] is obtained; for likelihood ratio martingales a representation of the limiting process as an exponential martingale is given [cf. \textit{A. R. Swensen}, Conditions for contigenity of probability measures under an asymptotic negligibility condition. Ph. D. Thesis, Univ. Calif., Berkeley (1980)].