MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

This question concerns the HOMFLY-PT category, closely related to Hecke algebras. (See here for example.)

The minimal idempotents of this category are indexed by pairs $(\lambda_+, \lambda_-)$ of Young diagrams. (The sizes of the diagrams are arbitrary and need not be the same. The diagram $\lambda_+$ corresponds to upward oriented strands, while $\lambda_-$ corresponds to downward oriented strands.) Consequently one can define numerical invariants of oriented links whose components are labeled by pairs of Young diagrams. This is the "colored" HOMFLY-PT polynomial.

Of fundamental importance in this subject are the invariants $S_{\lambda_+\lambda_-,\mu_+\mu_-}$ of the Hopf link with its components labeled by pairs of Young diagrams (i.e. idempotents) $(\lambda_+, \lambda_-)$ and $(\mu_+, \mu_-)$. In TQFT language, this is the "S-matrix" of the theory.

My Question:

Has the S-matrix for the HOMFLY-PT category been calculated and published? If not, are partial results in this direction known?

I am aware of this paper by Morton and Lukac, which does the case where $\lambda_-$ and $\mu_-$ are both empty (i.e. all strands oriented the same direction). This paper by Morton and Hadji is also related. Are there other relevant papers that I have missed?

See also the BMW version of this question here.

share|cite|improve this question

The $S$-matrix is given by \begin{equation} \frac{S_{ij}}{S_{00}}=S_{R_i}(q^{\rho})S_{R_j}(q^{\rho+R_i}) \end{equation} where $S_{R}(x_1,\cdots,x_N)$ is the Schur polynomial with highest weight $R$, $S_{R}(q^{\rho})=S_{R}(q^{\rho_{1}},...,q^{\rho_{N}})$ and $\rho$ is the Weyl vector. Furthermore, the paper by Aganagic and Shakirov propoesed the refinement (categorification) of the $S$-matrix \begin{equation} \frac{S_{ij}}{S_{00}}=M_{R_i}(t^{\rho})M_{R_j}(t^{\rho}q^{R_i}) \end{equation} where $M_{R}(x_1,\cdots,x_N;q,t)$ is the Macdonald polynomial with highest weight $R$ and $M_{R}(t^{\rho}q^{R})=M_{R}(t^{\rho_{1}}q^{R_{1}},...,t^{\rho_{N}}q^{R_{n}};q,t)$. It reduces to the above equation for $q=t$. By using the refined topological vertex, Iqbal and Kozcaz showed that the Khovanov-Rozansky polynomial of the Hopf link is actually proportional to the refined $S$-matrix \begin{equation} KhR_{ij}({\rm Hopf},q,t)\propto M_{R_i}(t^{\rho})M_{R_j}(t^{\rho}q^{R_i}) \end{equation} See Eq.(4.10) and appendix B in the paper.

share|cite|improve this answer
Thanks -- I'll have a look at those papers. – Kevin Walker Aug 19 '12 at 23:05
If I'm understanding correctly, these formulas are for the case where all strands are oriented in the same direction. (i.e. the case where $\mu_-$ and $\lambda_-$ of the question are both empty.) This case is also covered in the Morton and Lukac paper I refer to above. But it's nice to know about other papers discussing other viewpoints, so thanks again for your answer. – Kevin Walker Aug 19 '12 at 23:23
Oh, sorry. I did not understand your question correctly. But, that's all I know. – Satoshi Nawata Aug 19 '12 at 23:35

This isn't a complete answer, but it might make some partial progress. First I'll slightly restate the question. If we decompose $S^3$ into two solid tori and write $C$ and $C^{op}$ for the Homfly skein modules of these tori, then there is a pairing $$\langle -,-\rangle: C\otimes_A C^{op} \to k$$ where $k$ is the base ring (which is isomorphic to the skein module of $S^3$) and $A$ is the Homfly skein algebra of the torus. As you say, both $C$ and $C^{op}$ have bases indexed by pairs of partitions, and if I understand right, the entries of the $S$ matrix are evaluations of this pairing.

One brute-force approach for computing $\langle (\lambda, \mu), (\lambda',\mu')\rangle$ is to "move the left hand side to the right using the action of $A$ and then evaluate." An explicit presentation of $A$ is given here There are generators $P(a,b)$ for $a,b \in \mathbb(Z)$ (which are "the $gcd(a,b)$ power sum on slope $b/a$"), and they satisfy the commutation relations $$[P(x), P(y)] = (q^d-q^{-d}) P(x+y),\quad x,y \in \mathbb Z^2,\quad d = det[x\,\,y]$$ (If $x$ and $y$ are primitive and $d=1$ this is just the skein relation.) The paper also gives explicit formulas for the action of the generators $P(a,b)$ in the basis $(\lambda,\mu)$ that you mentioned, and these formulas actually aren't too bad (e.g. if $a,b \not= 0$ then the entries in the matrix are monomials).

One difficulty in completing the answer is that expressing the basis elements $(\lambda,\mu)$ in terms of power sums isn't too easy. (Morton does have a determinental formulas in terms of complete symmetric functions in one of his papers.) Also, I'm not sure if the evaluations $((\emptyset, \emptyset),(\lambda,\mu)\rangle$ have actually been computed.

(By the way, this algebra $A$ actually has a lot of different realization which are not obviously the same. One other realization is as (a specialization of) the Hall algebra of coherent sheaves on an elliptic curve over a finite field (this is described in the paper above). Also, if you've found these numbers since the question was asked I'd be curious to hear.)

share|cite|improve this answer
Thanks. I'll have a look at the paper you refer to. In answer to your final remark, no, I have not not learned anything new about this since asking the question. – Kevin Walker Feb 11 at 2:36

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.