Doubly stochastic measures, topology, and latticework hairpins (Q2638637): Difference between revisions

The authors show that the topology on the set of doubly stochastic measures (dsm's) studied by \textit{J. E. L. Peck} [Michigan Math. J. 6, 217-220 (1959; Zbl 0090.268)] is equivalent to the topologies on dsm's induced by the \(L^ p\)-norms (p\(\geq 1)\) on the copulas associated with dsm's. (If \(\mu\) is a dsm on \(I^ 2\), then the associated copula is \(C(x,y)=\mu ([0,x]\times [0,y]).)\) If g is an increasing homeomorphism on I with \(g(x)<x\) for x in (0,1), then the union of the graphs of g and \(g^{-1}\) is called a hairpin. A polygonal hairpin is generated by a g which is piecewise linear. A lattice work hairpin is a polygonal hairpin with an additional restriction. The authors show that the dsm's concentrated on polygonal hairpins are dense in the set of dsm's concentrated on hairpins and that the dsm's concentrated on lattice work hairpins are nowhere dense in the set of dsm's concentrated on hairpins.