Multimartingales, spectral measures and stochastic integration (Q1324843)

The theory of random functions depending on several parameters and having some martingale-like properties, including the stochastic integration with respect to such functions, was developed in papers of \textit{R. Cairoli} and \textit{J. B. Walsh} [Acta Math. 134, 111-183 (1975; Zbl 0334.60026)], \textit{E. Wong} and \textit{M. Zakai} [Z. Wahrscheinlichkeitstheorie Verw. Geb. 29, 109-122 (1974; Zbl 0282.60030) and Ann. Probab. 4, 570-586 (1976; Zbl 0359.60053)] and \textit{I. I. Gikhman} [Proc. School Semin. Theory Random Processes, Vilnius 1975, Pt. 1, 35-69 (1975); Theory Probab. Math. Stat. 15, 21-29 (1978), translation from Teor. Veroyatn. Mat. Stat. 15, 21-30 (1976; Zbl 0399.60040); Russ. Math. Surv. 37, No. 6, 1-29 (1982), translation from Usp. Mat. Nauk 37, No. 6 (228), 3-28 (1982; Zbl 0509.60043)] and in a number of papers by these and other authors. We investigate random measures on subsets of \({\mathbb{R}}^ k_ +=[0,\infty)^ k\) instead of random point functions. These measures have some martingale-like property and are called multimartingale measures. For \(k=2\) the notion of multimartingale measure actually coincides with the notion of bimartingale in the last cited paper. The set of square integrable multimartingale measures appears to be the Hilbert space \(M^ 2\). Here we describe the integration of previsible functions with respect to square integrable multimartingale measures. The starting point is spectral measures on some algebras with projections in \(M^ 2\) as their values. The idea to use spectral measures in stochastic integration theory was exposed in the paper of \textit{Z. Huang} [Z. Wahrscheinlichkeitstheorie Verw. Geb. 66, 25-40 (1984; Zbl 0519.60054)].