with
cube edges, ``hit points''.
on each edge (see figure 4).
.
In this case a pair of hits on
edges that share a common Low vertex will be connected by a boundary curve
within the face.
Thus we call this choice ``Knit Low''.
Similarly, the choice that makes the Highs nonadjacent (and thus the Lows
adjacent)
is termed
``Knit High'' (see the squares in figures 4 and 5).
These are the only two choices for face F, since
if there is a path
between the two Highs
then there cannot be a path between between the two Lows in
,
as it
would have to cross
(and conversely). In two dimensions
these will be our only two choices. As we shall see we will make a similar
choice for 3 and higher dimensions. In this case we will regard
them as adjacent if we determine that they are path connected through
the interior of a cube sharing face F, however, as we shall
see when we discuss
critical 4-hit faces,
no topological generality will be lost if we assume the path is through
F
We now explain the sense in which edge-connectivity is inconsistent.
.
Proof: If f interpolatesthen -f interpolates
since
implies that
. Since
implies
, then
.
![]()
Proof:
If we reverse the signs of all the readings, the High vertices of -f are the Low vertices of f (and vice versa). The Low vertices of the 4-hit face F are adjacent in. Thus making these same two vertices non-adjacent as Highs of
yields
(compare figure 4 with figure 5).