.................................................................
Partial separability, graph theory, and global optimization

(joint work with A.~R.~Conn, IBM, Yorktown Heights, NY)

We present a way of exploiting partial separability in particular
global optimization problems. The aim is to reduce significantly the
dimension of the search space. The procedure relies on graph theory
tools such as graph partitioning with node separators.  We illustrate
the idea on a distance geometry problem which arise in the
interpretation of nuclear magnetic resonance data and in the
determination of protein structures.