Conjugate sets of Gaussian random fields (Q1088293)

The author considers Gaussian random fields (G.r.f.) (X(x); \(x\in {\mathbb{R}}^ d)\) determined by their structure function r(t), i.e. \(E[(X(x)-X(y))^ 2]=r(| x-y|)\) \((x,y\in {\mathbb{R}}^ d)\). Let \(E\subset {\mathbb{R}}^ d\) and let \(R_ r(x,y| E)\) be the conditional covariance function given E. Let \({\mathcal F}_ X(x| E)=\{y\in {\mathbb{R}}^ d\); \(R_ r(x,y| E)=0\}\) be the maximal conjugate set of \(x\) relative to \(E\). For finite sets \(E\) the author gives an affirmative answer to the following problems under certain reasonable conditions: 1) Let \((X_ 1,r_ 1(t))\) be another G.r.f. Suppose that \({\mathcal F}_ X(x| E)\subset {\mathcal F}_{X_ 1}(x| E)\) holds for certain pairs \(\{\) x,E\(\}\). Is it true that \(r_ 1=r ?\) 2) Suppose that \({\mathcal F}_ X(tx| tE)=t{\mathcal F}_ X(x| E)\) \((t>0)\) holds for certain pairs \(\{x,E\}\). Is it true that \(r(t)=t^{\alpha}\) for some \(\alpha\in]0,2] ?\) These results generalize the author's work published in Lect. Notes Math. 1021, 252-256 (1983; Zbl 0516.60043) and the author's joint work with \textit{A. Noda} published in Nagoya Math. J. 85, 251-268 (1982; Zbl 0446.60036).