In mathematics, Khovanov homology is an oriented link invariant that arises as the homology of a chain complex. It may be regarded as a categorification of the Jones polynomial.
It was developed in the late 1990s by Mikhail Khovanov, then at the University of California, Davis, now at Columbia University.
To any link diagram D representing a link L, we assign the Khovanov bracket [D], a chain complex of graded vector spaces. This is the analogue of the Kauffman bracket in the construction of the Jones polynomial. Next, we normalise [D] by a series of degree shifts (in the graded vector spaces) and height shifts (in the chain complex) to obtain a new chain complex C(D). The homology of this chain complex turns out to be an invariant of L, and its graded Euler characteristic is the Jones polynomial of L.
This definition follows the formalism given in Dror BarNatan’s 2002 paper.
Let {l} denote the degree shift operation on graded vector spaces—that is, the homogeneous component in dimension m is shifted up to dimension m + l.
Similarly, let [s] denote the height shift operation on chain complexes—that is, the rth vector space or module in the complex is shifted along to the (r + s)th place, with all the differential maps being shifted accordingly.
Let V be a graded vector space with one generator q of degree 1, and one generator q^{−1} of degree −1.
Now take an arbitrary diagram D representing a link L. The axioms for the Khovanov bracket are as follows:
In the third of these, F denotes the `flattening’ operation, where a single complex is formed from a double complex by taking direct sums along the diagonals. Also, D_{0} denotes the `0smoothing’ of a chosen crossing in D, and D_{1} denotes the `1smoothing’, analogously to the skein relation for the Kauffman bracket.
Next, we construct the `normalised’ complex C(D) = [D][−n_{−}]{n_{+} − 2n_{−}}, where n_{−} denotes the number of lefthanded crossings in the chosen diagram for D, and n_{+} the number of righthanded crossings.
The Khovanov homology of L is then defined as the homology H(L) of this complex C(D). It turns out that the Khovanov homology is indeed an invariant of L, and does not depend on the choice of diagram. The graded Euler characteristic of H(L) turns out to be the Jones polynomial of L. However, H(L) has been shown to contain more information about L than the Jones polynomial, but the exact details are not yet fully understood.
In 2006 Dror BarNatan developed a computer program to calculate the Khovanov homology (or category) for any knot.
One of the most interesting aspects of Khovanov’s homology is that its exact sequences are formally similar to those arising in the Floer homology of 3manifolds. Moreover, it has been used to produce another proof of a result first demonstrated using gauge theory and its cousins: Jacob Rasmussen’s new proof of a theorem of Kronheimer and Mrowka, formerly known as the Milnor conjecture (see below). Conjecturally, there is a spectral sequence relating Khovanov homology with the knot Floer homology of Peter Ozsváth and Zoltán Szabó (Dunfield et al. 2005). Another spectral sequence (OzsváthSzabó 2005) relates a variant of Khovanov homology with the Heegaard Floer homology of the branched double cover along a knot. A third (Bloom 2009) converges to a variant of the monopole Floer homology of the branched double cover.
Khovanov homology is related to the representation theory of the Lie algebra sl_{2}. Mikhail Khovanov and Lev Rozansky have since defined cohomology theories associated to sl_{n} for all n. In 2003, Catharina Stroppel extended Khovanov homology to an invariant of tangles (a categorified version of ReshetikhinTuraev invariants) which also generalizes to sl_{n} for all n. Paul Seidel and Ivan Smith have constructed a singly graded knot homology theory using Lagrangian intersection Floer homology, which they conjecture to be isomorphic to a singly graded version of Khovanov homology. Ciprian Manolescu has since simplified their construction and shown how to recover the Jones polynomial from the chain complex underlying his version of the SeidelSmith invariant.
At International Congress of Mathematicians in 2006 Mikhail Khovanov provided the following explanation for the relation to knot polynomials from the view point of Khovanov homology. The skein relation for three links
$$L
1
,
L
2
{\displaystyle L_{1},L_{2}}
and
$$L
3
{\displaystyle L_{3}}
is described as
Substituting
$$λ
=
q
n
,
n
≤
0
{\displaystyle \lambda =q^{n},\ n\leq 0}
leads to a link polynomial invariant
$$P
n
(
L
)
∈
Z
[
q
,
q
−
1
]
{\displaystyle P_{n}(L)\in \mathbb {Z} [q,q^{1}]}
, normalized so that for
$$n
>
0
{\displaystyle n>0}
$$
P
0
(
u
n
k
n
o
t
)
=
1
{\displaystyle P_{0}(unknot)=1}
. For
$$n
>
0
{\displaystyle n>0}
$$
P
n
(
L
)
{\displaystyle P_{n}(L)}
$$
s
l
(
n
)
{\displaystyle sl(n)}
and
$$P
0
(
L
)
{\displaystyle P_{0}(L)}
via that of the quantum Lie superalgebra
$$U
q
(
g
l
(
1

1
)
)
{\displaystyle U_{q}(gl(11))}
.
The first application of Khovanov homology was provided by Jacob Rasmussen, who defined the sinvariant using Khovanov homology. This integer valued invariant of a knot gives a bound on the slice genus, and is sufficient to prove the Milnor conjecture.
In 2010 fuzzy socks wholesale, Kronheimer and Mrowka proved that the Khovanov homology detects the unknot. The categorified theory has more information than the noncategorified theory. Although the Khovanov homology detects the unknot, it is not yet known if the Jones polynomial does.