Virtual-knot problem 60 — Problems on Free Knot Cobordism
Statement
Problems on Free Knot Cobordism: The methods used for proving the fact that the invariant gives an obstruction to the sliceness are not immediately generalized for obtaining lower estimates on the slice genus of free knots. Two reasons are as follows. First, when we define a parity and justified parity for double lines on , we chose an arbitrary curve connecting the two preimages of the point on the double curve. We assert that any two curves connecting these two preimages (and behaving correctly in neighborhoods of the ends) are homotopic. It is true in the case of cobordism of genus zero, but in the case of a surface of an arbitrary genus it is, indeed, not true.
Thus, in order to define a parity for double lines we have to impose some restrictions on the spanning surface: we have to require that the cohomology class dual to the graph was –homologically trivial. The significance of the property for even-valent graphs to be –homologically trivial is closely connected with atoms (for details see Sec. the cited item and ).
Another problem is that the Reeb graph of an arbitrary Morse function (not necessarily corresponding to the disc) is not necessarily a disc. Consider Fig. the cited item.
Thus, starting from a free knot for which, we say, , we (in principle) can turn it by a Morse bifurcation into free two-component link consisting of two free knots and , for which , and then by another Morse bifurcation we can reconstruct this trivial link into the unknot. The invariant is not an obstruction to this, since the sum of and is zero.
In some cases we can overcome these two difficulties for cobordisms (of arbitrary genus).
Let be a surface with boundary . Obviously, the collection of double lines of defines a relative –homology class . This homology class is an obstruction for the surface to be checkerboard-colorable; also, this is an obstruction for well-definedness of even/odd double lines.
Namely, if we look at the definition of an even/odd double line: we see that there is an ambiguity in the choice of path connecting two preimages of a generic point on the double line. For the case of a disc cobordism, the parity of double lines is well defined, because all such curves are homotopic. For the unique obstruction to this well-definedness is the class .
We call a cobordism of genus checkerboard (or atomic/) if the corresponding class vanishes.
The next task (after detecting which -stratum is even and which one is odd) is to distinguish between and . To this end, one should do the same for preimages of points lying on odd -strata, connect them by a generic curve, and count the intersection with even double lines. So, we see that the only obstruction is the relative –homology class generated by even double lines.
We say that a checkerboard cobordism is -atomic if vanishes.
The following theorem holds.
[see ] Assume for a -component framed -graph we have . Then there is no -atomic cobordism spanning the knot of any genus.
In the papers the fourth author constructed a strengthening of the group given in the present work (in the notation our group is ) and the invariants of free knots with values in the classes of conjugate elements from . The idea is as follows. Even chords are further partitioned into chords of different types, it leads to more accurate definition of generators and relations in the group; these constructions are closely connected with iterated parities and the map deleting odd crossings. It seems that all invariants related to the groups also give an obstruction to the sliceness of a knot. Moreover, in the paper the author constructed invariants of virtual knots in which the over/undercrossing structure was taken into account besides the parity of chords. We devote a separate paper to the investigation of a connection of these invariants with cobordisms.
Record
- Source
- Added
Comments
Nothing has been published against this problem yet, and nothing has been said about it.
Sign in with an institutional address to take part in the discussion. Reading every thread stays open to everyone.
Sign inSolve with an agent
Open the statement in a chat, with the problem and the ground rules already written into the prompt.