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

I am seeking references for precise statements and rigorous proofs of some facts about the actions of quantum root vectors and $R$-matrices on crystal bases for finite-dimensional representations of quantum groups. I am very new to crystal bases, so I would also appreciate corrections if my questions are not well-formulated. I am putting the questions first, followed by the motivation for those who are curious.


Do you know references (with proofs) for the following statements:

  1. Let $V$ be a finite-dimensional $U_q(\mathfrak{g})$-module. Then the divided powers $E_{\beta}^{(t)}, F_\beta^{(t)}$ of the quantum root vectors have matrix coefficients given by Laurent polynomials in $q$, with respect to the global crystal basis for $V$.

  2. Let $V,W$ be finite-dimensional $U_q(\mathfrak{g})$-modules. Then the matrix coefficients of the $R$-matrix $R_{V,W}$ are Laurent polynomials in $q$, with respect to the tensor product of the global crystal bases for $V,W$.

I believe I have proofs for these statements, but it would be nice to just reference something definitive instead of writing the proofs out myself.


Let $U_q(\mathfrak{g})$ be the quantized enveloping algebra of $\mathfrak{g}$ for $q$ not a root of unity, with generators $E_i,F_i,K_i$ corresponding to the simple roots of $\mathfrak{g}$. Using an action of the braid group of $\mathfrak{g}$ on $U_q(\mathfrak{g})$ one can define quantum root vectors $E_\beta,F_\beta$ for all positive roots $\beta$. (This depends on a choice of decomposition of the longest word of the Weyl group, so assume that we have fixed such a decomposition.)

Let $R_{U,V}$ be the action of the $R$-matrix on $U \otimes V$, (as in Chari-Pressley or Klimyk-Schmudgen, say) so $\tau \circ R_{U,V}$ is the braiding. I would like to make sense of the statement that $R_{V,W} \to \mathrm{id}_{V \otimes W}$ as $q \to 1$ (and hence the braiding tends to the flip as $q \to 1$). This is not trivial because for different $q$'s, the operators $R_{V,W}$ are really operators on different vector spaces. This is where the crystal bases come in.

As I understand it, a crystal basis for a module has the property that the matrix coefficients of the generators $E_i,F_i$ of $U_q(\mathfrak{g})$ (and divided powers of the generators) are given by universal Laurent polynomials in $q$ whose coefficients are independent of $q$. Using this basis we can think of all of the algebras $U_q(\mathfrak{g})$ for various $q$'s acting on the same vector space. The point is that Laurent polynomials are continuous and well-defined at $q=1$, i.e. they are specializable to $q=1$.

Taking the tensor product of the crystal bases for $V$ and $W$, we can think of all of the $R$-matrices for various $q$'s acting on the same space as well, and it makes sense to ask if this family of $R$-matrices is continuous in $q$, and if so, whether it can be extended to $q=1$.

The formula for the action of the $R$-matrix is a big sum of products of operators of the form

$$ \frac{1}{[t]_{q_\beta}!} E_\beta^t \otimes F_\beta^t$$

with coefficients given by Laurent polynomials in $q$. Putting the $q$-factorial under, say, the $E_\beta^t$ term gives the divided power $E_\beta^{(t)}$. If the quantum root vectors and their divided powers act by Laurent polynomials, then the $R$-matrix does as well, and hence everything in sight is continuous in $q$, can be specialized to $q=1$, and it is clear that at $q=1$ the $R$-matrix is just the identity.

share|cite|improve this question
up vote 4 down vote accepted

I never found a precise reference for the statement about the R-matrix, so I ended up writing it up myself. The precise statements and proofs can be found in $\S 4.1$ of my paper with Alex Chirvasitu, Remarks on quantum symmetric algebras, available here.

share|cite|improve this answer

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.