No label defined (Q3973813)

[For the entire collection see Zbl 0721.00025.] This is an introduction to and an announcement of the results of \textit{M. Emery}'s above reviewed paper (Zbl 0753.60045). The emphasis here is more on ``chaotic'' representation than in Emery's paper. If \(X_ t\) is a square integrable martingale, one can define multiple integrals. \[ I_ n(f)=\int_{s_ 1>s_ 2>\ldots>s_ n}f(s_ 1,s_ 2,\ldots,s_ n)dX_{s_ 1}\ldots dX_{s_ n} \] for every \(n\geq 1\). The set of all such \(I_ n(f)\) is called the \(n\)-th chaos, \(W_ n\). If the sum of the \(W_ n\) is all of \(L^ 2\), we say \(X_ t\) has the chaotic representation property (CRP). Emery's paper shows that a martingale discovered by Azema, \(X_ t=\text{sgn}(B_ t)\sqrt{t-g_ t}\), has the CRP. (\(B_ t\) is standard Brownian motion and \(g_ t\) is the last zero before time \(t\)).