Free Binary Decision Diagrams (FBDDs) or read-once branching
  programs are a data structure for Boolean functions.  They can
  efficiently be manipulated if only FBDDs respecting a fixed graph
  ordering are considered. However, the size of such  FBDDs may
  strongly depend on the chosen graph ordering.  In this paper it is
  shown that the existence of polynomial time approximation schemes
  for optimizing graph orderings or for minimizing FBDDs implies
  NP=P, and so such algorithms are quite unlikely to exist.  The same
  holds for the related problem of computing minimal size FBDDs that 
  are consistent with a given set of examples. The latter result 
  implies that size bounded FBDDs are not PAC-learnable unless NP=RP.
Free Binary Decision Diagrams (FBDDs) are a data structure
for the representation and manipulation of Boolean functions.
Efficient algorithms for most of the important operations are known if
only FBDDs respecting a fixed graph ordering are considered. However,
the size of such an FBDD may strongly depend on the chosen graph
ordering and efficient algorithms for computing good or optimal graph
orderings are not known.  In this paper it is shown that the existence
of polynomial time approximation schemes for optimizing graph
orderings or for minimizing FBDDs implies NP=ZPP or NP=P,
respectively, and so such algorithms are quite unlikely to exist.