1 Introduction
A knot is a piecewise linear simple closed curve without intersection in (or ). A link consists of such finite simple closed curves in (or ). Each curve is called a component of the link. In fact, a knot is a link with one component. If there is an orientation preserving piecewise linear homeomorphism which preserves the orientation such that , then the link is equivalent to the link . The unknot is the knot equivalent to a circle. How to distinguish nonequivalent links is an essential problem in Knot Theory. We need to find some invariants of links in order to distinguish the distinct links.
The history of knots can be traced back to the beginning of human being tying knots with ropes. Knot Theory, that is a mathematical study of knots, is probably recognized first by Gauss. He introduced the linking number in the 19th century [
19]. Because of the mystery of the problem itself, it is widely concerned by many mathematicians in fields of Topology, Algebra, Geometry, Graph theory Combinatorics, etc. It is used to study related topics by biologists and physicists et al. [
71,
55]. In this paper, we review some results on the Jones polynomial and the Khovanov homology of alternating links and quasi-alternating links. Moreover, a long pass link is introduced and several problems are given. The properties of other polynomial invariants can be referred to the review articles [
13,
57].
2 Preliminaries
A link projection is obtained by projecting a link to the plane. If there are a finite number of double crossings in a projection and the over-crossing line and the under-crossing line transverse at each crossing in the plane, then the projection is called a diagram of the link. If the over and under crossing at a crossing are not considered, a 4-regular planar graph is obtained, which is called the
shadow of the diagram. A
path is an alternating sequence of vertices and edges. A graph is connected if there is a path for any two distinct vertices. A vertex
of a graph
is called a cut vertex if
is the partition of
and
such that
and
have just the common vertex
. Here,
is the subgraph of
whose vertex set is the set of vertices incident with an edge in
and whose edge set is
for
A diagram is reduced if its shadow doesn't contain a cut vertex. Reidermeister and Alexander and Briggs independently introduced three Reidermeister moves (Fig.1) and obtained the following result in 1926 [
63,
2]. Then a diagram is
connected (or non-split) if it isn't equivalent to the disjoint union of two diagrams.
Theorem 1 If two diagrams differ by a finite sequence of Reidermeister moves, then they are equivalent.
Jones discovered the famous polynomial invariant−the Jones polynomial in 1984 [
28], who won the Fields Prize for this work in 1990. Kauffman introduced the bracket polynomial and then gave a relatively simple expression of the Jones polynomial in 1987 [
31]. Let
be a diagram of a link
. Each crossing can be resolved by using 0-revolution or 1-revolution (Fig.2). A state of
is obtained by resolving each crossing. Let
be the collection of all states of
and let
be the number of crossings in
. Obviously,
has
states. Let
and
denote the number of
-revolutions and loops in
, respectively. Then
is the number of
-revolutions in
where
. Set
. The bracket polynomial is as follows:
Definition 1 [
31,
72] Let
be a diagram of an oriented link
. Set
to be the writhe of
and set
. Then the Jones polynomial of
is as follows:
The breadth is the difference between the maximal degree of and the minimal degree . Suppose that for . If for each , then is gapless.
Khovanov proposed a cohomological invariant of a link – Khovanov homology [
33], which is a categorification of the Jones polynomial. The Khovanov homology encodes more information about knots than the Jones polynomial. It detects the unknot, the trefoil knot and the figure-eight knot etc. [
4,
5,
6,
9,
23,
35,
41,
42,
46,
81,
82]. Rasmussen defined a knot invariant
on the Khovanov homology by using Lee’s spectral sequence [
62].
provides a lower bound for the smooth slice genus of
. Rasmussen gave a combinatorial proof of the Milnor conjecture which is proved by using Gauge Theory [
48,
50]. We recall the Khovanov homology over
and the readers are referred to [
7,
32,
33,
39,
74,
77,
84].
Let be a diagram of an oriented link with crossings . A state is obtained by resolving each crossing of where the th element denotes the revolution of the crossing in . Regard every state as a vertex. Two vertices and are adjacent if and only if they differ in exactly one place. Set , . Then . We associate each cycle to a graded vector space where . There is a vector space for each such that
Here, and are the number of positive crossings and negative crossings in , respectively. is the number of 1's and is the number of cycles in . Let
where . Let where
and let such that
Define the differential
where ,
Here, is determined by the parity of the number of 1's before in . is determined by either or on the cycles appearing in the changing disk. is the identity on cycles not entering the changing disk.
Definition 2 [
33] The Khovanov homology (or cohomology) of an orientable link
is the homology of a bi-graded chain complex
where and are given above for .
Definition 3 [
34,
45] If its nontrivial homology groups lie on two adjacent diagonals, then the link
is Khovanov homologically thin (or H-thin). If Khovanov homology is completely determined by its Jones polynomial and its signature
, then
is Khovanov holomogically
-thin.
3 Alternating links
The alternating links give a classical class of links. They play an important role in Knot Theory. In this section, we review some classical results on the alternating links. Some related computations are referred to [
17,
25,
27,
30,
38,
47,
52,
53,
61,
85].
Definition 4 [
70] Let
be a diagram. If one passes the over-crossings and under-crossings alternately while moving along each component of
, then
is an alternating diagram. If a link has an alternating diagram, then it is alternating.
The research on alternating links can be traced back to the work of Tait and others in the end of the 19th century. Tait proposed an important conjecture about alternating knots in 1898.
Conjecture 1 [
70] A connected reduced alternating diagram has minimal crossing number for the knot.
Kauffman, Murasugi and Thistlethwaite obtained the following results and then independently proved Tait's conjecture.
Theorem 2 [
31,
54]
The number of crossings in a connected reduced alternating diagram of is a topological invariant for a link .
Theorem 3 [
54]
Two connected reduced alternating diagrams of an alternating link have the same number of crossings. Theorem 4 [
72]
If a link has a connected reduced alternating diagram of crossings, then: the breadth of is exactly ;
is an alternating polynomial;
the coefficients of the terms of of maximal and minimal degree are both ;
if is prime, and is not a torus link, then is gapless.
Wu obtained the following result by analyzing the black and white regions in the shaded shadow of a link diagram.
Theorem 5 [
26]
Let be a connected reduced diagram with crossings. is its state. If there exists a region satisfying the following conditions: and each adjacent region have precisely one common edge;
both the black mark and white mark occur on the boundary of in the coloring of the shadow,
then
where the revolution of each crossing in is different from that in .
This deduces the following result.
Corollary 1 A prime connected reduced alternating diagram of a link is a diagram with the minimum crossings of .
In the topic of the Khovanov homology of an alternating link, Khovanov obtained the Khovanov homology of
over
as follows [
33].
Theorem 6 The Khovanov homology of torus links for are given by
Bar-Natan computed the Khovanov homology over of all prime knots with crossings up to 11 [
7,
8]. Bar-Natan, Garoufalidis and Khovanov proposed the following conjectures based on computations.
Conjecture 2 [
7] For any prime alternating knot
there exists an even integer
and a polynomial
in
and
with only non-negative coefficients such that
Here, is the graded Poincaré polynomial of the complex in the variable over , is the signature of and contains only power of .
Conjecture 3 [
18] The following result holds for each alternating knot
.
where , is the signature of .
Conjecture 4 [
7,
18] For each alternating knots
, the support of the Khovanov homology lies in the lines
Lee obtained that the Khovanov homology of an alternating knot was determined by its Jones polynomial and signature and then proved Conjectures 2−4 over .
Theorem 7 [
39]
For any oriented connected alternating link , the support of lies in and for some with ,
Theorem 8 [
39]
Let be an oriented connected alternating link with components .
Then where , is the linking number of and .
4 Quasi-alternating links
Ozsváth and Szabó introduced a quasi-alternating link [
56]. The unknot and an alternating link are quasi-alternating links, which shows that a quasi-alternating link is a generalization of an alternating link. In this section we recall some results related to the Jones polynomial, the Khovanov homology and differentiation of quasi-alternating links.
Definition 5 The set of quasi-alternating links is the smallest set of links which satisfies the following properties:
• the unknot is in ;
• the set is closed under the following operation. Let be any link which has a diagram with a crossing satisfying the following properties:
1. both resolutions ;
2.
Then .
Manolescu and Ozsváth proved the following result [
45].
Theorem 9 Each quasi-alternating link is Khovanov-thin over .
Manolescu [
44], Baldwin [
3], Greene [
21], Champanerkar and Kofman [
10] obtained quasi-alternating knots up to 10 crossings. Champanerkar and Kofman [
10] proved the result as follows.
Theorem 10 If and are any quasi-alternating knot diagrams, then is quasi-alternating.
Some infinite families of links are considered later. According to Conway representation, each rational tangle associates a continued fraction such that
where
,
,
for
. Let
be a crossing of a link diagram
. If we replace
by
of
for
, then
extends . If each rational tangle in a product extends
, then the product of rational tangles extends the crossing
. Each Montesinos link
has a diagram which consists of
rational tangle
and
half-twists for
and
. Here,
associate
for
[
1,
16]. Then there are the following results.
Theorem 11 [
21]
The pretzel link for ,
,
is quasi-alternating if and only if (1) ;
(2) ;
(3) , and or ;
(4) , and or .
The result holds on permuting on the parameters and .
Theorem 12 [
10]
Set to be a quasi-alternating diagram at some crossing .
Let be the diagram obtained from by a rational tangle extending .
Then is quasi-alternating. Theorem 13 [
10]
The pretzel link is quasi-alternating for ,
,
each and .
The conclusion hold for all permutations of and and for reflections of all these pretzel links. Theorem 14 [
60]
Set to be a quasi-alternating diagram at some crossing .
Let be the diagram obtained from by a product of rational tangles that extends .
Then is quasi-alternating. Theorem 15 [
60]
The Montesinos link is quasi-alternating if it satisfies one of the following conditions: (1) ;
(2) ;
(3) for some , ;
(4) for some , .
These conclusions contain the results related to quasi-alternating Montesinos links [
80]. In addition, Chbili and Qazaqzeh considered the Jones polynomial of a quasi-alternating link and then obtained the following result [
15].
Theorem 16 Let be a quasi-alternating link whose diagram is quasi-alternating at . Let be a diagram obtained from by replacing by a product of rational tangles extending . If is gapless, then is gapless where is a diagram of .
Chbili and Kaur considered the diagram from a quasi-alternating diagram by replacing the crossing by a reduced alternating tangle that extends . They obtained the conclusion as follows.
Theorem 17 Set to be a quasi-alternating link diagram at a crossing . Let be the diagram obtained from by replacing by a reduced alternating tangle extending . Then is quasi-alternating at every crossing of
5 Long pass links
A long pass link is introduced in this section. It is different from the notation of the quasi-alternating link. Related problems are provided.
Definition 6 [
78] Let
be a link diagram. Assign a letter to each crossing and denote two adjacent crossings
and
by an unordered pair
. Then an embedding presentation of
is obtained. Here,
. If
is an overcrossing at
, then
; otherwise
. Each crossing has an anti-clockwise rotation.
Definition 7 [
78] Let
be a path in a link diagram
. If
,
,
for any
and
, then
is removable. If
,
,
for each
, then
is a pass of length
. If there doesn’t exist a pass
of
such that
, then
is maximal.
If a diagram has a removable pass, then it is equivalent to a diagram without any removable pass. Now consider the related operations.
Definition 8 [
78] Let
be a link diagram and let
be edges for
. If
, then
is the union of
and
. Accordingly,
and
are obtained from
by adding a vertex
on
.
and
is a subdivision of
.
The union implies to remove a crossing . The subdivision implies to add a crossing . Next consider the operations on passes.
Definition 9 [
78] Set
to be a link diagram. Suppose that
is a pass of length
(
, and
is adjacent to
,
is on the boundary of the face
for each
where
. Remove the pass
, that is to remove
and to add
for each
. Then a figure denoted by
is obtained. Denote the face containing
of
by
in
. Add a pass
, that is to add a crossing
on
(
,
) in
such that
is a pass of length
(
. Then a link diagram
is obtained.
Definition 10 [
78] A pass replacement is an operation that one removes a pass
, adds a pass
and then obtains a link diagram
based on a link diagram
where
are given above. If
and
for each
, then
isn't different from
; otherwise
is different from
. If
, then the replacement is a short pass replacement. If
and
are different and
, then the replacement is an equal pass replacement. If
, then the replacement is a long pass replacement.
The pass replacement is always a replacement between different passes later in the paper.
Definition 11 Let be a reduced link diagram. If there isn't a short pass replacement in , then is a diagram without a short pass. If there is neither a short pass replacement nor an equal pass replacement, then is a long pass diagram. If a link (or knot) has a long pass diagram, then it is a long pass link (or knot).
An alternating diagram is obvious to be a diagram without a short pass. The non-alternating knot
in the Rolfsen knot table is a long pass knot. Since it is Khovanov holomogically thick [
64], it isn't quasi-alternating by Theorem 9. So a diagram without an equal pass is a generalization of an alternating diagram and it is different from the notation of the quasi-alternating diagram. Obviously, the Thistlethwaite's unknot [
36] is a diagram without a short pass (Fig.3), which shows that there exists a diagram without a short pass is equivalent to the unknot. Is a long pass link equivalent to the unlink? Several problems are given as follows.
Conjecture 5 Let be a long pass diagram of a link with () crossings. Then is its diagram with the minimum crossings.
If is alternating, this conjecture holds. is a non-alternating link with the minimum crossings satisfying the conjecture.
Problem 1 Characterize the properties of a quasi-alternating and a long pass link, especially diagrams of the unknot.
Problem 2 Study the Jones polynomials of quasi-alternating knots and long pass knots and then study whether the Jones polynomial detects the unknot.
Jones proposed an important conjecture that the Jones polynomial detects the unknot [
29]. The conjecture holds for an alternating knot and certain knots and a knot with crossings less than 25. Some results and a detailed introduction are given in [
31,
54,
72,
11,
37,
24,
67,
75,
76,
83].
Problem 3 Find an effective method to determine the Khovanov homology of some infinite families of links and study the ability of the Khovanov homology to detect links.
It is hard to determine the Khovanov homology of a link. It can be determined for a link with crossings less than 50 and certain links [
65,
12,
20,
40,
43,
51,
49,
58,
59,
66,
68,
73,
69]. It is worth finding methods to determine the Khovanov homology of a general link. The Khovanov homology isn't a complete invariant of a link [
79]. Mathematicians have been paying attention to the following important problem:
Find an effective complete invariant of a link, especially a knot.