SciPost Submission Page
The nonrational limit of Dseries minimal models
by Sylvain Ribault
This is not the latest submitted version.
This Submission thread is now published as
Submission summary
Authors (as registered SciPost users):  Sylvain Ribault 
Submission information  

Preprint Link:  https://arxiv.org/abs/1909.10784v3 (pdf) 
Date submitted:  20200602 02:00 
Submitted by:  Ribault, Sylvain 
Submitted to:  SciPost Physics Core 
Ontological classification  

Academic field:  Physics 
Specialties: 

Approach:  Theoretical 
Abstract
We study the limit of Dseries minimal models when the central charge tends to a generic irrational value $c\in (\infty, 1)$. We find that the limit theory's diagonal threepoint structure constant differs from that of Liouville theory by a distribution factor, which is given by a divergent Verlinde formula. Nevertheless, correlation functions that involve both nondiagonal and diagonal fields are smooth functions of the diagonal fields' conformal dimensions. The limit theory is a nontrivial example of a nondiagonal, nonrational, solved twodimensional conformal field theory.
Author comments upon resubmission
I am grateful for the reviewer's work and interest. On the function σ(P) and on the "hard to accept" conclusion, I realize that the article is not clear and explicit enough. The resubmitted version addresses these shortcomings.
List of changes
1. I have written more details on the sign factor σ(P). To begin with, I have explained how the shift equations (2.26) are deduced from ref. [5], and why they do not depend on $r_i,s_i$ in our case. Then, I have written explicit expressions (2.29) and (2.30) for the solutions of the shift equations. Moreover, in order to better show how the signs appear in fourpoint functions, I have added the diagrams (2.32).
2. In order to clarify the statement that "the diagonal and mixed fourpoint functions cannot belong to the same CFT", I have added a few lines of explanation after this statement, at the very end of Section 4. I have also added explanations at the end of Section 5.1.
Current status:
Reports on this Submission
Anonymous Report 1 on 2020614 (Invited Report)
 Cite as: Anonymous, Report on arXiv:1909.10784v3, delivered 20200614, doi: 10.21468/SciPost.Report.1762
Report
Again, I would like to thank the author for their changes. The comments on the limit process require more thought, but they present a point of view that I think is definitely acceptable.
On the issue of the sign $\sigma(P)$, I am sorry that I was not more explicit as I think that the author has not understand my concerns. I was not worried by the shift equations themselves, whether they depend on $(r_2,s_2)$ and $(r_3,s_3)$, only in the solutions and the actual values of $\sigma(P_1)$.
The sign $\sigma(P)$ has to depend on $(r_2,s_2)$ and $(r_3,s_3)$ since, for example, $C_{P,(2,1/2),(4,1/2)} =  C_{P,(4,1/2),(2,1/2)}$ from the standard result for the three point structure constant of fields with integer spin,
$C_{abc} = (1)^{S_c+S_a+S_b} C_{acb}$ where $S_a$ is the spin of the field $a$, [see e.g. eqn (2.2.48) in arxiv:1406.4290] and the facts that the spin of the field $V_P$ is $0$, and of $V_{(r,s)}$ is $rs$.
Hence, if we replace $\sigma(P)$ in eqn (2.25) by the more general notation
$\sigma_{P_1,(r_2,s_2),(r_3,s_3)}$, it has to be the case that
$\sigma_{P_1,(2,1/2),(4,1/2)} =  \sigma_{P_1,(4,1/2),(2,1/2)}$.
I would like to know what values the author has given for $\sigma(P_1)=\sigma_{P_1,(r_2,s_2),(r_3,s_3)}$ to understand if there is any way to undo the nonanalyticity by field redefinitions, splitting the fields into two sets, etc, or (as is obviously suggested) there is none.
I really think that the author should provide the values of $\sigma_{P_1,(r_2,s_2),(r_3,s_3)}$ so that one could check the numerical calculations.
I am sorry that such a seemingly small point should hold up publication, but it seems essential to me to allow readers to reproduce the calculations and to decide for themselves on the possibility of an analytic solution or not.
Author: Sylvain Ribault on 20200615 [id 854]
(in reply to Report 1 on 20200614)In Eq. (2.25) for the threepoint structure constant, in addition to $\sigma(P)$, there should be a prefactor $(1)^{r_2s_3}$. This prefactor obeys $(1)^{r_2s_3} = (1)^{r_2s_2+r_3s_3} (1)^{r_3s_2}$, because $r_1+r_2\in 2\mathbb{Z}$ and $s_2+s_3\in\mathbb{Z}$. So it leads to the expected behaviour of the threepoint structure constant when exchanging the fields $2$ and $3$.
Does this answer the question? Please object if it does not. If I receive no objection within a few days, I plan to submit a revised version with the additional prefactor.
Author: Sylvain Ribault on 20200624 [id 864]
(in reply to Sylvain Ribault on 20200615 [id 854])Yes, this sign factor is described in the notebook Correlators.ipynb in the second text cell, and implemented by the line "product = (1)**sum(fields[i].indices[0]*fields[i+1].indices[1] for i in range(3))" in the method three_shift() of the class FourPoint.
Anonymous on 20200623 [id 863]
(in reply to Sylvain Ribault on 20200615 [id 854])Just to confirm, is this the value of the structure constants used in the numerical checks?