I have a question about the proof of proposition $3.3.6(3)$ in "Tensor Categories" by Etingof et al..

This part states that for $A$, transitive unital $\mathbb Z_+$-ring, there is a unique character taking non-negative values on the basis elements.

The proof uses the fact that if $\chi$ is a character, and $f$ is a vector with entries $\chi(Y)$ for basis elements $Y$, then $f$ is an eigenvector of the matrix of left multiplication by $\sum_{X\in I}X$, where $I$ is a set of basis elements.

But I don't see why this holds. As I understand we want an equation: $$ \sum_{X,Y\in I} \chi(Y) XY = \lambda \sum_{Z\in I} \chi(Z) Z , $$ to hold, so $\sum_{X,Y\in I} \chi(Y) c_{XY}^Z = \lambda \chi(Z)$. But I do not understand what to do next.

In bibliographical notes there is a link to the paper https://arxiv.org/pdf/math/0203060.pdf (Lemma 8.3). Which uses more or less the same fact with $X$ instead of $\sum_{X\in I} X$ as a matrix, which is even more mysterious to me.

Can someone please clarify this for me?


Hm, I probably figured it out myself, but I won't delete the question since I think the formulation in the proof is slightly misguiding.

Use $b_i$ to denote basis elements ($i \in I$), a take element $y = \sum y_i b_i$ and denote $\chi_i = \chi(b_i)$. We have $\chi(y) = \sum y_i \chi_i$. But if the multiplication by $x$ -- any element of the ring is given by matrix $A_{ij}$ we have $xy = \sum b_iA_{ij}y_j$. Now take character of both sides to obtain: $$ \chi(x) \sum \chi_j y_j = \sum \chi_i A_{ij} y_j \ , $$ since this holds we have $\chi(x) \chi_i = \sum \chi_i A_{ij}$, so the row vector $\chi_i$ is indeed the eigenvector of $A_{ij}$. Now the rest of the proof works the same since the can use the Frobenius-Perron theorem.


Hrm, I can’t seem to get it to work without assuming that the ring is based...

Let’s just do the version without the sum from the paper. So you want: $$\sum_{Y\in I} \chi(Y^*) c_{Z X^*}^{Y^*} = \sum_{Y\in I} \chi(Y) c_{XY}^Z = \lambda \chi(Z)$$

Now take $\lambda = \chi(X)=\chi(X^*)$.

  • $\begingroup$ Maybe I’m missing a trick where you just take a transpose of a matrix rather than using star. $\endgroup$ – Noah Snyder Nov 20 '17 at 14:05
  • $\begingroup$ What I find strange is that they do precisely this calculation later for FPdim in fusion ring. So it seems to be strange if the statement in the paper as I first understood it is true for unbaised rings... $\endgroup$ – DerLoewe Nov 20 '17 at 14:17

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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