Wikipedia has a definition of abstract polyhedra based on partially ordered sets. However, it seems more natural to try to define them more geometrically, building up 1 dimension at a time:
Each face is "completely surrounded" by faces one dimension lower. Each face participates exactly twice, once on each of its "sides", in the surrounding of a face one dimension higher.
Are the poset and geometric definitions of abstract polyhedra equivalent?
The tricky part, unresolved, is defining "surrounded" and (possibly) "sides". In 2D to define a left and right side of a line segment, it needs to be a directed line segment. But then in 3D, the concept of "sides" of a line segment disappears. In 3D, we can order the vertices around a 2D face, then define sides of it by the left- or right-hand rule.
A regular 3D cube would have two 3D faces: its interior and its exterior, the latter having infinite volume, though volume is not a concept in abstract polyhedra.
Calculate the poset graphs for some well-known polyhedra. For the larger ones, text representation of the graph is better.
EmoticonEmoticon