I am trying to derive the following recursion relation for 2D conformal bootstrap from Di Francesco (page182), however I am stuck. Starting from the operator algebra(note that I have not written the parts in $\bar{z}$ variable) $$ \Phi_1(z)\Phi_2(0) = \sum_p \sum_{\{k\}}C_{12}^{p\{k\}}z^{h_p -h_1-h_2+K}\Phi_p^{\{k\}}(0) \tag{1} $$ where $h_i$ are the scaling dimensions of the fields $\phi_i(z)$. $K= \sum_i k_i$ and ${\{k\}}$ means the collection of $k_i$ indices. The claim is that, $C_{12}^{p\{k\}}=C_{12}^p\beta_{12}^{p\{k\}}$ where $C_{12}^p$ is the coefficient of 3 point correlation function. As a special case of this, the text considers, when $h_1=h_2=h$ acting on vacuum $|0\rangle$ $$ \Phi_1(z)|h\rangle = \sum_p \sum_{\{k\}}C_{12}^p\phi(z)|h_p\rangle \tag{2} $$ where $\phi(z)= \sum_{\{k\}}z^K\beta_{12}^{p\{k\}}L_{k_1}...L_{k_N}$. Next a state is defined, $$|z,h_p\rangle\equiv \phi(z)|h_p\rangle=\sum_{N=0}^\infty Z^N|N,h_p\rangle \tag{3}$$ where $|N,h_p\rangle$ is a descendant state at level $N$ in the verma module. $$L_0|N,h_p\rangle = (h_p+N)|N,h_p\rangle \tag{4} $$ Next, it says that $L_n=\frac{1}{2\pi i}\oint dz z^{n+1}T(z)$ is operated on both sides of (1). On LHS, I could see that it was

$$ L_n\Phi_1(z)|h\rangle = (z^{n+1}\partial_z + (n+1)hz^n)\Phi_1(z)|h\rangle \tag{5}$$. I however could not see the RHS part where the text claims that it is

$$\sum_p \sum_{\{k\}}C_{12}^pL_n|z, h_p\rangle = \sum_p \sum_{\{k\}}C_{12}^p[h_p + h(n-1)z^n +z^{n+1}\partial_z]|z, h_p\rangle \tag{6}$$

And then he finally obtains, $$L_n|N+n,h_p\rangle=(h_p + h(n-1) +N)|N,h_p\rangle \tag{7}$$ I could intuitively see that acting by $L_n$ on $|N+n,h_p\rangle$ will bring it down to $|N,h_p\rangle$ but I am not able to derive the factor in front. I am thinking that there should be a way to derive (7), directly from (4).

  • $\begingroup$ In your equation (5), a $z^n$ is missing, it should be $L_n\Phi_1(z)|h\rangle = (z^{n+1}\partial_z + (n+1)h z^n )\Phi_1(z)|h\rangle $. $\endgroup$ – Antoine Mar 15 '17 at 11:26
  • $\begingroup$ Note that this is a misprint in the book as well, but it is mentioned in the errata. $\endgroup$ – Antoine Mar 15 '17 at 11:27
  • $\begingroup$ @user40085 Thanks for pointing out. Fixed it. You also seem to have come across this particular thing. Have you been able to work it out? $\endgroup$ – levitt Mar 15 '17 at 12:24
  • $\begingroup$ You have another typo in equation (2), the power of z is missing. $\endgroup$ – Antoine Mar 15 '17 at 14:57

Note first that while copying the equations from the book, you forgot several powers of $z$.

You can substitute the corrected equation (2) into your equation (5), and obtain $$L_n \left[ \sum_p \sum_{\{k\}}C_{12}^p z^{h_p - 2h} |z,h_p\rangle \right] = (z^{n+1}\partial_z + (n+1)hz^n) \left[ \sum_p \sum_{\{k\}}C_{12}^p z^{h_p - 2h} |z,h_p\rangle \right] $$ On the LHS, the operator $L_n$ goes through the scalars and hits the state $|z,h_p\rangle$. On the RHS, you just compute the derivative, and obtain $$ \sum_p \sum_{\{k\}}C_{12}^p z^{h_p - 2h} L_n |z,h_p\rangle = \sum_p \sum_{\{k\}}C_{12}^p z^{h_p - 2h} (z^n (h_p - 2h) + z^n h (n+1) + z^{n+1} \partial_z) |z,h_p\rangle $$ Finally, you get $$\boxed{ \sum_p \sum_{\{k\}}C_{12}^p z^{h_p - 2h} L_n |z,h_p\rangle = \sum_p \sum_{\{k\}}C_{12}^p z^{h_p - 2h} (z^n (h_p +h(n-1)) + z^{n+1} \partial_z) |z,h_p\rangle } \tag{6} $$ as claimed in the book.

From this, one deduces the relation $$ L_n |z,h_p\rangle = (z^n (h_p +h(n-1)) + z^{n+1} \partial_z) |z,h_p\rangle \tag{8} $$

Now to prove equation (7), just substitute $$|z,h_p\rangle=\sum_{N=0}^\infty z^N|N,h_p\rangle \tag{3}$$ into (8) : $$ \sum z^N L_n |N,h_p\rangle = (z^n (h_p +h(n-1)) + z^{n+1} \partial_z) \sum z^N|N,h_p\rangle $$ and simplify both sides : $$ \sum z^{n+N} L_n |N+n,h_p\rangle = \sum z^{N+n} (h_p +h(n-1) + N)|N,h_p\rangle $$ Identifying equal powers of $z$, you obtain $$ \boxed{L_n |N+n,h_p\rangle = (h_p +h(n-1) + N)|N,h_p\rangle }$$

| cite | improve this answer | |
  • $\begingroup$ Thanks a lot! I feel stupid for not seeing that step. $\endgroup$ – levitt Mar 15 '17 at 18:56

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.