A KAUFFMAN BRACKET POLYNOMIAL FOR THE NON-ORIENTABLE THICKENING OF A NON-ORIENTABLE SURFACE3
Knots in ˆ
Σ and their diagrams on Σ can be defined by analogy with the case of
knots in a thickened orientable surface (below, we will refer to the latter case as
the “classical case”). Knots in ˆ
Σ will be considered up to ambient isotopy.
In our case, two diagrams on Σ represent the same knot if and only if they are
connected by a finite sequence of Reidemeister moves R1, R2, R3(which are just the
same as classical ones) and ambient isotopies. indeed, let Σ be represented by poly-
gon of which sides are assumed to be identified (maybe) with a twist and a diagram
is drawn inside the polygon. If we isotope the knot in question inside the direct
product of the polygon and the segment, then we can refer to the corresponding
classical fact. Otherwise, it is necessary to consider a few auxiliary moves. These
moves describe how we push arcs and crossings of the diagrams through the bound-
ary of the polygon. In the case of knots in orientable thickening of non-orientable
surface (see, for example, [2], [1]), some of these moves permute overgoing and un-
dergoing branches at pushing crossing. In our case, ˆ
Σ is the direct product hence
we have no moves permuting overgoing and undergoing branches, i.e., all auxiliary
moves represent ambient isotopies of a diagram. Therefore, in proofs below, we can
restrict ourselves to Reidemeister moves R1, R2, R3.
1.2. Pseudo-classical knots. A knot K⊂ˆ
Σ is called an pseudo-classical if it is
an orientation-preserving path in ˆ
Σ. In other words, a knot in ˆ
Σ is pseudo-classical
if its regular neighborhood in the manifold is homeomorphic to the solid torus.
Clearly, in a non-orientable manifolds not all knots have the property since, in
this case, there are knots whose regular neighborhood is homeomorphic to the solid
Klein bottle. However, the latter kind of knots is out of consideration in the present
paper.
Let Kbe a pseudo-classical oriented knot represented by a diagram D⊂Σ. Since
the knot is pseudo-classical, the diagram (or, more precisely, the projection of K
onto Σ) viewed as a closed directed path on the surface is an orientation-preserving
path. Hence, 2-cabling D2⊂Σ of Dis a diagram of a 2-component link (recall that
2-cabling of a diagram is, informally speaking, the same diagram drawn by doubled-
line with preserving over/under-information in all crossings). Note again that this
is not the case for an arbitrary diagram on Σ. This is because Σ is assumed to be
non-orientable hence there are knot diagrams on the surface, those are orientation-
reversing paths. Hence, their 2-cabling represents not a 2-component link but a
knot which goes along the given knot twice. In the latter case, 2-cabling bounds
the M¨obius strip (going across itself near each crossing of the given diagram) while
2-cabling of a pseudo-classical knot bounds an annulus (having self-intersections).
In the case of an oriented surface, one can speak of left and right components of
2-cabling of an oriented knot diagram. Since Σ is non-orientable, we do not have
the concepts of left and right. However, we will use the terms “left component”
and “right component” of D2keeping in mind that in our situation this is nothing
more than a convenient notation. A choice (which component is chosen as the left
and which is chosen as the right) will be called the labeling of D2. The components
will be denoted by L(D)and R(D), respectively. Below, L(D)and R(D)are
assumed to be oriented so that their orientations are agreed with the orientation of
D. Sometimes we will rename R(D)into L(D)and vice versa, the transformation
will be called the relabeling of D2.
A single crossing of Dbecomes in D2a pattern of 4 crossings (see Fig. 1). There
are two distinguished crossings in each of these patterns: the first one in which