Dimension reduction for random marked sets with complementary variables is investigated. The models in question are random tessellations and numerical solution of stochastic differential equations.
Sliced inverse regression and some other dimension reduction techniques are developer. Applications in materials research are demonstrated.