1: \chapter{Boundary Insertion Operators}
2: \label{ch:bio}
3:
4: To complete the description of the BCFT on the full Riemann Surface,
5: we need to discuss how the $N_0$ distinct copies of the theory, each
6: one living on a different cylindrical end \cyl{k}, interact along the
7: underlying polytope $|P_{T_l}|\,\rightarrow\,M$. In \cite{Carfora2},
8: a remarkable intuition proposed that this interaction could be
9: mediated by boundary conditions changing operators, whose presence is
10: predicted by boundary conformal field theory
11: \cite{Cardy1,Recknagel:1998ih}(see the general theory presented in the
12: previous chapter).
13:
14: In this chapter starting from this statement, we slightly modify it.
15: We describe interaction on the ribbon graph introducing, beside
16: ordinary boundary condition changing operators, a new class of
17: operators which mediate the change in boundary conditions which
18: actually take place non locally on the boundary shared by two adjacent
19: cylindrical ends. As a matter of fact, while in ordinary boundary
20: conformal field theory a jump in boundary conditions on the boundary
21: is explained by the local action of boundary operators, this feature
22: does not fit completely our model, both because, after the glueing
23: process we do not deal with a true boundary, but with a ``separation
24: edge'' between the two adjacent cylindrical ends, and because we do
25: not have a jump in boundary conditions taking place in a precise point
26: of this edge (which we will go on calling boundary, for the sake of
27: simplicity) but with two different boundary conditions which, in the
28: adjacency limit, coexist on this shared boundary. From these
29: considerations, it follows the description of the dynamical glueing of
30: the $N_0(T)$ copies of the bosonic BCFT which we present in this
31: chapter.
32:
33:
34:
35: \section{Boundary Insertion Operators}
36:
37: In the background defined in section \ref{sec:bo}, it may happen that
38: a boundary condition changes along the real line. In radial
39: quantization, this situation is explained with the presence of a
40: vacuum which is no longer invariant under the action of
41: $L_{-1}^{(H)}$. In \cite{Cardy1} it was proposed that such states were
42: obtained by the action of a boundary operator on the true vacuum,
43: \ie{} it may happen that boundary operators induce transitions on
44: boundary conditions along the boundary. These boundary condition
45: changing operators are associated with vectors in the Hilbert space
46: depending on both the boundary conditions, and they cannot be obtained
47: from bulk fields through a bulk to boundary OPE.
48:
49: In this section we will propose a slight modification of this picture,
50: introducing a particular class of boundary condition changing
51: operators which, living on the boundary shared between two adjacent
52: polytopes and carrying an irreducible action of the chiral algebra,
53: will mediate the change in boundary conditions between pairwise
54: adjacent cylindrical ends.
55: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
56: \begin{figure}[!t]
57: \begin{center}
58: \includegraphics[width=.4\textwidth]{immagini/Bio_action}
59: \caption{The insertion of boundary operators on the shared
60: boundary between pairwise adjacent polytope allows to define
61: the superposition of different boundary conditions\cite{Carfora2}}
62: \label{BIO}
63: \end{center}
64: \end{figure}
65: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
66:
67: To fully understand this statement, let us first analyze the geometry
68: on the neighborhood of a shared ribbon graph edge. Ribbon graph's
69: strips represent the uniformization of the singular structure of the
70: RRT in the neighborhood of the 1-skeleton of the original
71: triangulation. The ribbon graph arises as a consequence of the
72: presence of a locally uniformizing coordinate $\zeta(k)$, which
73: provides a counterclockwise orientation in the 2-cells of the original
74: triangulation: such an orientation gives rise to a cyclic ordering on
75: the set of the half edges $\left\{
76: \rho^1(k)^\pm\right\}_{k=0}^{N_1(T)}$ incident on the vertexes
77: $\left\{ \rho^0(k)^\pm\right\}_{k=0}^{N_2(T)}$, stating a 1-to-1
78: correspondence between the 1-skeleton of the original triangulation
79: and the set o trivalent metric ribbon graphs. In this connection, let
80: us consider two adjacent cylindrical ends \cyl{p} and \cyl{q}: these
81: are dual to the adjacent cells $\rho^2(p)$ and $\rho^2(q)$. The
82: cylindrical ends are glued to the oriented boundaries
83: $\partial\Gamma_p$ and $\partial\Gamma_q$ of the ribbon graph, which
84: are denoted respectively by $\partial\Gamma_p$ and $\partial\Gamma_q$.
85: Let us consider the oriented strip associated with the edge
86: $\rho^1(p,q)$ of the triangulation and its uniformized neighborhood
87: $\left(U_{\rho^1(p,q)},z(p,q)\right)$ and let us take in
88: $\left(U_{\rho^1(p,q)},z(p,q)\right)$ two infinitesimally neighboring
89: points $z_1 \,=\, x_1 \,+\ i y_1$ and $z_2 \,=\, x_2 \,+\ i y_2$ such
90: that $x_1 \,=\, x_2$. Thus, if $y_1 \,\rightarrow\, 0$ we are
91: approaching the intersection between the boundary $\partial\Gamma_p$
92: of the ribbon graph and the oriented edge $\rho^1(p,q)$, while when
93: $y_2 \,\rightarrow\, 0$ we are approaching the intersection between
94: the boundary $\partial\Gamma_q$ of the ribbon graph and the oriented
95: edge $\rho^1(q,p)$. From a BCFT point of view, this leads to the
96: conclusion that, in the adjacency limit described above, the
97: ``effective boundary'' between two adjacent cylindrical end is unique
98: and shares the two different boundary conditions associated to
99: $S^{(-)}_{\varepsilon(p)}$ and $S^{(-)}_{\varepsilon(q)}$.
100:
101: In this connection, it is no longer correct to argue the presence of
102: vacuum state which is not invariant under translations along the
103: boundary. The shared boundary is obtained glueing two loops, each of
104: them being part of a domain, on which we have defined a self-contained
105: BCFT, with its associated Hilbert space and vacuum state invariant
106: under translation along the boundary loop.
107:
108: Thus, in order the glueing process to take place coherentely, we have
109: to require that each state of the Hilbert space melt without breaking
110: the conformal and chiral symmetry of the model. This leads to the
111: definition of Boundary Insertion Operators.
112:
113: A CFT is consistently defined on each cylindrical end once we have
114: imposed constraints in \eqref{eq:cont_rq}. In particular, specializing
115: to the Virasoro fields, on \cyl{p} inner boundary we have
116: \begin{equation}
117: \label{eq:nmp}
118: \zeta(p)^2\, T(\zeta(p))\,=\,
119: \bgz(p)^2\bar{T}(\bgz(p))|_{|\zeta(p)|=1}
120: \end{equation}
121: while in the inner boundary of \cyl{q} it holds:
122: \begin{equation}
123: \label{eq:nmq}
124: \zeta(q)^2\, T(\zeta(q))\,=\,
125: \bgz(q)^2\bar{T}(\bgz(q))|_{|\zeta(q)|=1}
126: \end{equation}
127:
128: This conditions allows to combine, on each cylindrical end, $T$ and
129: $\overline{T}$ in a unique object (well defined conformally mapping
130: the cylinder into a semi annular domain in the UHP), thus allowing to
131: associate to each cylindrical end a single copy of the Virasoro
132: algebra (the same obviously holds for the other chiral fields, see eq.
133: \eqref{eq:open_modes}).
134:
135: In this connection, we can pursue further this last construction, and
136: implement a non symmetry-breaking glueing of two adjacent cylindrical
137: end by associating, to each pairwise adjacent couple of them, a unique
138: copy of both Virasoro and chiral algebras. To achieve this, let us
139: consider the neighborhood of the $(p,q)$ edge defined in
140: \eqref{eq:strip} and uniformized with the complex coordinate $z(p,q)$.
141: Conformal mappings between $z(p,q)$ and the coordinates uniformizing
142: each cylindrical end are define as:
143: \begin{equation}
144: \zeta(p) \,=\, e^{\frac{2 \pi i}{L(p)}(L(p) - \hat{L}(p,q) + z(p,q))}
145: \qquad
146: \zeta(q) \,=\, e^{\frac{2 \pi i}{L(q)}(L(q) - \hat{L}(q,p) + z(q,p))}
147: \end{equation}
148: where it holds $z(q,p) = - z(p,q)$. We can easily express the
149: holomorphic and antiholomorphic components of the Virasoro fields
150: defined on each cylindrical end in term of the strip coordinate. For
151: the holomorphic components we have:
152: \begin{subequations}
153: \begin{gather}
154: T_{(p)} (z(p,q)) \,=\, T_{(p)} (\zeta(p))
155: \left(
156: \frac{d\,z(p,q)}{d\,\zeta(p)}
157: \right)^{-2}
158: \\
159: T_{(q)} (z(q,p)) \,=\, T_{(q)} (\zeta(q))
160: \left(
161: \frac{d\,z(q,p)}{d\,\zeta(q)}
162: \right)^{-2}.
163: \end{gather}
164: \end{subequations}
165: Analogous relations hold for the antiholomorphic sector.
166:
167: Rescaling $z(q,p) = - z(p,q)$, then we perform the glueing asking for
168: the following relations to hold:
169: \begin{subequations}
170: \begin{gather}
171: T_{(p)} (z(p,q)) \,=\, T_{(q)} (z(p,q))\,\vert_{y(p,q) = 0} \\
172: \overline{T}_{(p)} (\bz(p,q)) \,=\,
173: \overline{T}_{(q)} (\bz(p,q))\,\vert_{y(p,q) = 0}
174: \end{gather}
175: \end{subequations}
176:
177: These relations allows to associate to each pairwise adjacent couples
178: of theories a single copy of the Virasoro algebra. As a matter of
179: fact, these glueing conditions allows to define a unique holomorphic
180: component of the stress energy-tensor as:
181: \begin{equation}
182: \label{eq:Tpq}
183: T_{(p,q)}\,=\,
184: \begin{cases}
185: T_{(p)} (z(p,q)) & \text{if} z(p,q) \,\cup\,\zeta(p) \,\neq\,0\\
186: T_{(q)} (z(q,p)) & \text{if} z(q,p) \,\cup\,\zeta(q) \,\neq\,0
187: \end{cases}
188: \end{equation}
189: The same holds for the antiholomorphic sector, allowing to define a
190: unique $\overline{T}(\bz(p,q))$. Moreover, $T_{(p,q)}$ and
191: $\overline{T}(\bz(p,q))$ are not independent, because of relations
192: \eqref{eq:nmp} \eqref{eq:nmq}. Thus, to each pairwise adjacent couples
193: of cylindrical ends, we can associate a unique copy of the Virasoro
194: algebra. This can be defined considering a small integration contour
195: crossing the $(p,q)$ boundary, like that at the rhs of fig. \ref{fig:int_bound}
196: \begin{figure}[!t]
197: \centering
198: \includegraphics[width=.75\textwidth]{immagini/int_bound}
199: \caption{A small integration contour intersecting the $(p,q)$ edge
200: of the ribbon graph}
201: \label{fig:int_bound}
202: \end{figure}
203:
204: Thanks to the continuity condition along the boundary, we can actually
205: write:
206: \begin{multline}
207: \label{eq:Lnpq}
208: L_n^{(p,q)} \,=\, \frac{1}{2 \pi i}\oint_{C(p,q)} z(p,q)^{n + 1}
209: T_{(p,q)} (z(p,q)) \,=\,\\
210: \frac{1}{2 \pi i}\oint_{C(p)} d z(p,q)\,z(p,q)^{n + 1} T_p
211: (z(p,q)) \,+\,\\ \frac{1}{2 \pi i}\oint_{C(q)} d z(q,p) \,z(q,p)^{n +
212: 1} T_q (z(q,p))
213: \end{multline}
214:
215: A similar reasoning can be performed for the other chiral fields of
216: the model, but performing a particular attention at the gluing maps
217: $\Omega$ in \eqref{eq:1al}. As a matter of fact, we must keep in
218: account the fact that this glueing process must relate relate the two
219: glueing authomorphisms $\Omega_{(p)}$ and $\Omega_{(q)}$ associated to
220: the BCFTs defined respectively on \cyl{p} and \cyl{q}.
221:
222: With the above remarks, we can associate to each pairwise coupled
223: BCFTs a unique Hilbert space carrying a representation of the
224: $u(1)_{(p,q)}$ algebra. Consequently the associated collection of
225: boundary operators will have a primary $u(1)$ (and Virasoro) index
226: $\lambda(p,q)$. We ask the pair of polytope indexes $(p,q)$ to be
227: ordered, assuming that this means that the operator in question lives
228: on the ``$p$-side'' of the boundary of the strip connecting \cyl{p}
229: with \cyl{q}. Moreover, these operators carry the information about
230: the conformal boundary condition over $S^{(-)}_{\varepsilon(p)}$, thus
231: they are labeled by another index $A(p)$:
232: \begin{equation}
233: \psi^{A(p)}_{\lambda(p,q)}(x(p,q)),
234: \end{equation}
235: where $x(p,q)\,=\,\Re z(p,q)$.
236:
237: In the same way, we can define with
238: \begin{equation}
239: \label{eq:bf_1p}
240: \psi^{B(q)}_{\lambda(q,p)}(x(q,p))
241: \end{equation}
242: (with $x(q,p)\,=\,\Re z(q,p)$) the associated operator, transformed
243: under the conformal mapping $z(p,q)\,=\,-\,z(q,p)$. According to the
244: above assumptions, operators in \eqref{eq:bf_1p} live on the $q$- side
245: of the $(p,q)$ edge, thus carrying the information about the boundary
246: condition on $S^{(-)}_{\varepsilon(q)}$.
247:
248: However, the above distinction between $p$-side fields and $q$-side
249: ones is merely artificial, since the spectrum of boundary operators
250: associated to a given edge is unique. The natural consequence of this
251: is that each of them must carry the information about the boundary
252: conditions adopted on both $S^{(-)}_{\varepsilon(p)}$ and
253: $S^{(-)}_{\varepsilon(q)}$.
254:
255: Thus the complete labeling of Boundary Insertion Operators (BIO) which
256: arises from the BCFT defined on the $(p,q)$-edge is
257: \begin{equation}
258: \label{eq:bio_spectrum}
259: \psi_{\lambda(p,q)}^{B(q) A(p)}(x(p,q));
260: \end{equation}
261:
262: In this connection, these objects naturally mediate the change in
263: boundary conditions between pairwise adjacent boundaries. The labels
264: ordering, indicates the ``direction'' in which BIO acts, with the
265: understanding that:
266: \begin{equation}
267: \label{eq:conformal_bio}
268: \psi^{\lambda(q,p)}_{A(p) B(q)}(x(q,p))
269: \,=\,
270: \psi_{\lambda(p,q)}^{B(q) A(p)}(x(p,q));
271: \end{equation}
272:
273: At this level of the discussion, the action of BIO is still purely
274: formal. Since the switch in boundary conditions (from $A(p)$ to
275: $B(q)$ and viceversa) will act both on the glueing authomorphism and
276: on the Wilson line/brane position (see remarks before equation
277: \eqref{eq:bs_newnot}), we can relate their action to an authomorphism
278: of the algebra relating the two glueing maps $\Omega_{(p)}$ and
279: $\Omega_{(q)}$ (see equation \eqref{eq:1al}). Moreover, BIOs are
280: chiral primary operators of the $U(1)$ algebra: their expression
281: suggests to adopt a description analogue to that one introduced in
282: \cite{Behrend:1999bn} for usual boundary operators in Rational
283: Conformal Field Theory (RCFT). They can be represented as Chiral
284: Vertex Operators (CVO) which, in every point of the boundary, map the
285: coupling between states defined by the boundary condition $A(p)$ and
286: states belonging to the $\lambda(p,q)$ representation of the chiral
287: algebra to the spectrum of states which defines the boundary condition
288: $B(q)$. We can introduce a pictorial description of the interaction
289: as represented by the (\underline{non local}, because the switch
290: between boundary condition take place along the full boundary)
291: interaction vertex:
292:
293: \vspace{1cm}
294: %\psgrid[subgriddiv=1,griddots=10,gridlabels=7pt](0,-3)(15,3)%
295: \psline[linecolor=black]{c-}(5,-1)(6.5,-1)
296: \psline[linecolor=black]{c-}(6.5,-1)(8,-1)
297: \psline[linecolor=black]{c-}(6.5,-1)(6.5,.5)
298:
299: \rput[br](5,-1){$A(p)$}
300: \rput[bl](8,-1){$B(q)$}
301: \rput[bl](6,1.1){$\lambda(p,q)$}
302: \rput[tl](5.9,-.8){$\psi_{\lambda(p,q)}^{B(q) A(p)}$}
303:
304: \vspace{1.7cm}
305:
306: This kind of interaction is equivalent to the usual chiral vertex
307: which represent the map from the fusion of two representation of the
308: (bulk) Virasoro algebra $\lambda(p)$ and $\lambda'(p)$ to the
309: representation $\lambda''(p)$.
310:
311: The analogy stated with the usual CVO formalism is purely formal,
312: since the ``fusion'' process described above involve different objects
313: like boundary states (which are superposition of Ishibashi states) and
314: irreducible representations of the chiral algebra. However, this
315: analogy allows to specify the conformal properties of these operators.
316: It is possible, a priori, to assign a multiplicity index to BIOs,
317: $t\,=\,1,\ldots,\,N_{\lambda(p,q)}^{A(p)\,B(q)}$ for each
318: $\psi^{\lambda(p,q)}_{A(p) B(q)}(x)$ which takes into account the
319: degeneration which can occur in the map process described before.
320: Moreover, BIO are primary operators of the boundary chiral algebra,
321: thus their conformal dimension coincide with the highest weight of the
322: $V_\lambda(p,q)$ module of the $U(1)$ (Virasoro) algebra:
323: \begin{equation}
324: H(p,q) \,=\, \frac{1}{2} \lambda(p,q)^2
325: \end{equation}
326:
327: BIOs live on the ribbon graph, thus their interactions are guided by
328: the trivalent structure of $\Gamma$. A careful and exhaustive
329: analysis, which has been entirely reported in appendix
330: \ref{sec:biobcft}, shows that this structure allows to define all
331: fundamental coefficients weighting the self-interaction of boundary
332: insertion operators. Two points functions are well defined on the
333: edges of the graph, while OPE coefficients naturally defines the
334: fusion of different boundary insertion operators interacting in
335: $N_2(T)$ tri-valent vertexes of $\Gamma$. Moreover, we have shown that
336: thanks to the trivalent structure of the ribbon graph, we can define a
337: set of sewing constraint these coefficients must satisfy. Remarkably,
338: these constraints are perfectly analogues to the sewing constraint
339: introduced in \cite{Lewellen:1991tb} for boundary conditions changing
340: operator, thus enforcing the analogy between these latters and BIO,
341: which merge dynamically the $N_0(T)$ copies of the BCFT which are
342: defined on $M_\partial$.
343:
344:
345: \section{Investigations at the T-duality self dual radius}
346: The above description can be pursued further if we explore the
347: rational limit of the compactified boson BCFT, \ie{} the limit
348: \begin{equation}
349: \label{eq:sdr}
350: \Omega(k)\,\doteq\,\Omega(k)_{s.d.}\,=\,\frac{R(k)_{s.d.}}{l(k)}\,=\,\sqrt{2}
351: \end{equation}
352: where $\Omega(k)_{s.d.}$ is the self-dual radius, \ie{} the fixed
353: point of $T$-duality transformation
354: $\Omega\rightarrow\frac{1}{\Omega}$. If we substitute
355: $\Omega\,=\,\sqrt{2}$ into \eqref{eq:momentum}, we obtain
356: $\lambda_{s.d} \,=\, \frac{ \left( \mu\,+\,\nu \right)}{\sqrt{2}}$ and
357: $ \overline{\lambda}_{s.d}\,=\,\frac{\left( \mu\,-\,\nu
358: \right)}{\sqrt{2}}$, thus we are falling exactly in the situation
359: described in section \ref{sec:amplitude}. The holomorphic and
360: antiholomorphic conformal dimensions of primary fields are
361: respectively $ h_{s.d} \,=\, \left( \frac{\mu\,+\,\nu}{2} \right)^2 $
362: and $ \overline{h}_{s.d} \,=\, \left( \frac{\mu\,-\,\nu}{2}
363: \right)^2$, \ie{} they are the square of an integer or semi-integer
364: number. In full generality, we can write
365: $h(p)\,=\,j(p)^2,\,j(p)\in\,\frac{1}{2}\mathbb{Z}$. This lead to the
366: presence of a null state at the level $2j+1$ in the correspondent
367: $U(1)$ module, which consequently decompose into into an infinite set
368: of Virasoro ones. Consequences of this can be understood considering
369: the torus partition function associated the compactified boson with
370: $\Omega\,=\,\Omega_{s.d.}$ \footnote{In this discussion we will omit
371: the polytope index: it will be restored when we will deal again whit
372: the interaction of the $N_0$ copies of the BCFT on the ribbon
373: graph}:
374: \begin{equation}
375: \label{eq:sd_partfunc}
376: Z(\sqrt{2}) \,=\,
377: \frac{1}{\left|\eta(\tau)\right|^2}
378: \sum_{\mu,\,\nu\,\in\,\mathbb{Z}}
379: q^{\frac{1}{4}(\mu\,+\,\nu)^2}
380: \overline{q}^{\frac{1}{4}(\mu\,-\,\nu)^2}
381: \end{equation}
382:
383: If we redefine $m\,=\,\mu\,+\,\nu$ and $n\,=\,\mu\,-\,\nu$ such that
384: $m\,-\,n\,=\,0\,\text{mod}\,2$, \eqref{eq:sd_partfunc} reads
385: \begin{equation}
386: Z(\sqrt{2}) \,=\,
387: \frac{1}{\left|\eta(\tau)\right|^2}
388: \sum_{\substack{m,\,n\,\in\,\mathbb{Z} \\ m\,-\,n\,=\,0\,\text{mod}2}}
389: q^{\frac{1}{4}m^2}
390: \overline{q}^{\frac{1}{4}n^2}.
391: \end{equation}
392:
393: We can introduce the extended characters\cite{DiFrancesco}:
394: \begin{subequations}
395: \label{eq:extchar}
396: \begin{align}
397: \label{eq:extchar1}
398: C_0(\tau)
399: & \,=\,
400: \frac{1}{\eta(\tau)}
401: \sum_{m \text{even}}q^{\frac{1}{4}m^2} \,=\,
402: \frac{1}{\eta(\tau)}
403: \sum_{m \in \mathbb{Z}}q^{m^2} \,=\,
404: \frac{\theta_3(2\tau)}{\eta(\tau)} \\
405: \label{eq:extchar2}
406: C_1(\tau)
407: & \,=\,
408: \frac{1}{\eta(\tau)}
409: \sum_{m \text{odd}}q^{\frac{1}{4}m^2} \,=\,
410: \frac{1}{\eta(\tau)}
411: \sum_{m \in \mathbb{Z}}q^{(m\,+\,\frac{1}{2})^2} \,=\,
412: \frac{\theta_2(2\tau)}{\eta(\tau)}
413: \end{align}
414: \end{subequations}
415:
416: where $\theta_i(\tau)$ is the $i$-th Jacobi Theta function. The
417: partition function now reads:
418: \begin{equation}
419: \label{eq:extpf}
420: Z(\sqrt{2}) \,=\,
421: |C_0(\tau)|^2 \,+\,
422: |C_1(\tau)|^2.
423: \end{equation}
424:
425: We have actually reorganized the infinite Virasoro modules into a
426: finite number of extended blocks which transforms linearly into each
427: others by modular transformation. The extended $\mathcal{S}^{ext}$ matrix,
428: which encodes the modular data via the relation
429: $C_i(-\frac{1}{\tau})\,=\,\mathcal{S}^{ext}_{ij}C_j(\tau)$, reads\cite{DiFrancesco}:
430: \begin{equation}
431: \label{eq:sext}
432: \mathcal{S}^{ext} \,=\,
433: \frac{1}{\sqrt{2}}
434: \begin{pmatrix}
435: 1 & 1 \\
436: 1 & -1
437: \end{pmatrix}
438: \end{equation}
439:
440: These extended blocks are identified with the two irreducible
441: representations of the affine algebra $\hat{su}(2)_{k=1}$, namely the
442: vacuum representation $\mathcal{H}_0$, generated from a state
443: transforming in the singlet of the horizontal $su(2)$ algebra, and the
444: representation $\mathcal{H}_\frac{1}{2}$, for which the highest weight
445: state transform into the fundamental representation of the horizontal
446: $su(2)$ algebra. Thus, when
447: $\Omega(k)\,=\,\Omega(k)_{s.d.}$, the BCFT defined over each cylinder
448: is equivalent to an $\hat{\mathfrak{su}}(2)_{k=1}$-WZW model: it is an
449: extended non-minimal model generating a Rational Conformal Field
450: Theory (RCFT), a theory whose possibly infinite Verma modules can be
451: reorganized into a finite number of extended blocks. These are
452: irreducible representations of an extended symmetry algebra playing
453: the role of chiral algebra for the underlining CFT.
454:
455: In this connection, the set of labels associated to the irreducible
456: representations of the chiral algebra is finite,
457: $\mathcal{Y} = \left\{
458: 0,\,\frac{1}{2}\right\}$, and the Hilbert space is written as:
459: \begin{equation}
460: \label{eq:su2wzw_hilbspace}
461: \mathcal{H}_{bulk} \,=\,
462: \left(
463: \mathcal{H}_{0}^{\hat{su}(2)}
464: \,\otimes\,
465: \overline{\mathcal{H}}_{0}^{\hat{su}(2)}
466: \right)
467: \,\oplus\,
468: \left(
469: \mathcal{H}_{\frac{1}{2}}^{\hat{su}(2)}
470: \,\otimes\,
471: \overline{\mathcal{H}}_{\frac{1}{2}}^{\hat{su}(2)}
472: \right).
473: \end{equation}
474:
475: Since they will be useful in the following discussion, let us summarize
476: the main features of the conformal field theory associated to the
477: $\hat{\mathfrak{su}}(2)_{k=1}$ WZW model. $\hat{\mathfrak{su}}(2)_{1}$ (bulk)
478: primary fields are\cite{DiFrancesco}:
479: \begin{subequations}
480: \label{su2p}
481: \begin{align}
482: j=0 \,\rightarrow\, & \phi_{(0,0),(0,0)}(z,\bz) \,=\,
483: \mathbb{I}\\
484: j=\frac{1}{2} \,\rightarrow\, &
485: \begin{cases}
486: \phi_{(\frac{1}{2},\frac{1}{2}),(\frac{1}{2},\frac{1}{2})}(z,\bz)
487: \,=\, e^{\frac{i}{\sqrt{2}}X(z)}e^{\frac{i}{\sqrt{2}}X(\bz)}\\
488: \phi_{(\frac{1}{2},-\frac{1}{2}),(\frac{1}{2},-\frac{1}{2})}(z,\bz)
489: \,=\, e^{-\frac{i}{\sqrt{2}}X(z)}e^{-\frac{i}{\sqrt{2}}X(\bz)}
490: \end{cases}
491: \end{align}
492: \end{subequations}
493:
494:
495: The associated boundary theory is defined by a set of glueing
496: conditions on the boundary (see formula \eqref{eq:1al}). Let us
497: remember that, once we are given a glueing authomorphism, we can
498: associate to each irreducible representation of the chiral algebra an
499: Ishibashi state\cite{Ishibashi:1988kg},\ie{} a coherent state solution
500: of the constraint \eqref{eq:cont_bs}. In this connection, let us
501: denote the two Ishibashi states which are associated with the
502: $\mathcal{H}_0^{\hat{su}(2)}$ and the
503: $\mathcal{H}_\frac{1}{2}^{\hat{su}(2)}$ modules respectively with with
504: $\vert 0 \Rangle$ and $\vert \frac{1}{2} \Rangle$. A remarkable
505: property defined by Cardy in \cite{Cardy1} states that the set of
506: boundary conditions of a rational boundary conformal field theory are
507: labelled exactly by the modules of the chiral algebra, while the
508: correspondent boundary states are given by\cite{Cardy1}:
509: \begin{equation}
510: \label{eq:cardy_bs}
511: \vert\vert A(p)\Rangle \,=
512: \sum_{l\,\in\,\mathcal{Y}(p)}
513: \frac{\mathcal{S}^{ext}_{A l}}{\sqrt{\mathcal{S}^{ext}_{0 l}}}
514: |l(p)\Rangle
515: \end{equation}
516: In the case of $\hat{su}(2)_{k=1}$-WZW model, we can write
517: $\mathcal{Y}\,\equiv\,\mathcal{A}\,=\,\left\{
518: 0,\,\frac{1}{2}\right\}$, while formula \eqref{eq:cardy_bs}
519: specializes to:
520: \begin{equation}
521: \label{eq:cbs}
522: \Vert J \Rangle \,=\,
523: 2^{- \frac{1}{4}} \,\vert 0 \Rangle \,+\,
524: (-1)^{2J}\,2^{- \frac{1}{4}}\,\vert \frac{1}{2} \Rangle,
525: \quad\text{with} J = 0, \frac{1}{2}.
526: \end{equation}
527:
528: These properties allow for a complete definition of BIO of a rational
529: conformal field theory. As a matter of fact, since both boundary states
530: and boundary operators are identified by primary labels of the
531: extended chiral algebra, in analogy to what happens for boundary
532: conditions changing operators of rational minimal models, we can
533: define the glueing process as the fusion between the representations
534: associated to the two adjacent boundary states and the one BIO
535: carries. Thus, let us consider the $(p,q)$ edge $\rho^1(p,q)$. If
536: $\Vert J(p) \Rangle$ and $\Vert J(q) \Rangle$ are the boundary
537: conditions shared by $\rho^1(p,q)$, BIOs on $\rho^1(p,q)$ are defined
538: as:
539: \begin{equation}
540: \label{eq:bio}
541: \psi_{j(p,q)}^{J(p)\,J(q)} (x(p,q)) \,=\,
542: \mathcal{N}_{J(p)\,j(p,q)}^{J(q)}\,\psi_{j(p,q)}(x(p,q))
543: \end{equation}
544: \ie{} they are the $\hat{\mathfrak{su}}(2)_{k=1}$ primary fields
545: weighted by hte fusion rule ${N}_{J(p)\,j(p,q)}^{J(q)}$. The latters
546: are given by a combination of the $\mathcal{S}^{ext}$ matrix entries
547: via the Verlinde formula
548: \begin{equation}
549: \label{eq:verlfor}
550: \mathcal{N}_{J(p)\,j(p,q)}^{J(q)} \,=\,
551: \sum_{l\in\mathcal{Y}}
552: \frac{S^{ext}_{J(p)\,l}\,S^{ext}_{j(p,q)\,l}\,
553: \overline{S}^{ext}_{l\,J(q)}}{S^{ext}_{0\,l}}
554: \end{equation}
555:
556: As it stands, this construction, valid for the
557: $\hat{\mathfrak{su}}(2)_{k=1}$ model, does not apply to the
558: compactified boson at the self-dual radius $\Omega_{s.d}$. As a matter
559: of fact, even if the Hilbert space of the two models is the same, the
560: boundary theory associated to the latter one is quite more
561: complicated. It will lead to the definition of BIOs of the
562: compactified boson at the self dual radius as truly marginal
563: deformations of the operators in \eqref{eq:bio} by the action of
564: $SO(3)$ elements.
565:
566: We will dedicate the remaining part of this section to the
567: demonstration of this statement.
568:
569: Due to the Sugawara construction, each highest weight representation
570: of $\hat{\mathfrak{su}}(2)_{k=1}$ is also an irreducible representation of the
571: $c=1$ Virasoro algebra, whose generators commute with the horizontal
572: $su(2)$ ones. Thus, we can decompose each irreducible representation
573: of $\hat{\mathfrak{su}}(2)_{k=1}$ entering in \eqref{eq:su2wzw_hilbspace} with
574: respect to $su(2)\,\otimes\,{Vir}$:
575: \begin{equation}
576: \label{eq:su2dec}
577: \mathcal{H}_{j}^{\hat{\mathfrak{su}}(2)}
578: \,=\,
579: \sum_{n=0}^\infty
580: \mathcal{V}_{(n\,+\,j)^2}
581: \,\otimes\,
582: \mathcal{H}_{n\,+\,j}^{su(2)},
583: \end{equation}
584: where $\mathcal{V}_h$ is the conformal weight $h$ irreducible
585: representation of the Virasoro algebra, while
586: $\mathcal{H}_{j}^{\hat{\mathfrak{su}}(2)}$ (resp. $\mathcal{H}_{j}^{su(2)}$) is
587: the $(2j\,+\,1)$-dimensional representation of $\hat{su}(2)$
588: (resp. $su(2)$). The bulk Hilbert space then decompose as:
589: \begin{equation}
590: \label{eq:su2bulk_dec}
591: \mathcal{H}_{bulk} \,=\,
592: \sum_{\substack{n,\overline{n}\,=\,0\\n\,+\,\overline{n}\,\text{even}}}%
593: ^\infty
594: \left(
595: \mathcal{V}_{\frac{n^2}{4}}
596: \,\otimes\,
597: \overline{\mathcal{V}}_{\frac{\overline{n}^2}{4}}
598: \right)
599: \,\otimes\,
600: \left(
601: \mathcal{H}_{\frac{n}{2}}^{su(2)}
602: \,\otimes\,
603: \overline{\mathcal{H}}_{\frac{\overline{n}}{2}}^{su(2)}
604: \right).
605: \end{equation}
606:
607:
608: As we outlined before, the representations of the Virasoro algebra
609: which occur in the decomposition are all degenerates. Due to this, the
610: set of Virasoro primary fields is larger than in the case of generic
611: compactification radius. The decomposition \eqref{eq:su2bulk_dec}
612: shows that, for fixed $(m,\,n)$, the $U(1)_L\,\times\,U(1)_R$
613: representation of momenta $\lambda_{s.d.}$ and
614: $\overline{\lambda}_{s.d.}$ contains the Virasoro representation
615: provided that both $|m|,|n|\,\leq\,j$ and $j\,-\,m$ and $j\,-\,n$ are
616: integers. Conversely, if $m$ and $n$ satisfies these constraints, the
617: representation $ \mathcal{V}_{\frac{n^2}{4}} \,\otimes\,
618: \overline{\mathcal{V}}_{\frac{\overline{n}^2}{4}}$ of the Virasoro
619: algebra enter exactly once into the representation of
620: $U(1)\,\times\,U(1)$. Thus the Virasoro primary fields can be labelled
621: by the triple of indexes $(j;\,m,\,n)$, with the constraint
622: $|m|,\,|n|\,\leq\,j$. This is the celebrated \textsc{discrete serie of
623: states}\cite{Kac:1978ge,Klebanov:1991hx,Callan:1994ub,Kristjansson:2004ny}
624: $\psi_{j,m} \overline{\psi}_{j,n}$. The associated Ishibashi states
625: are defined in literature as $\vert
626: j,\,m,\,n\Rangle$\cite{Callan:1994ub}.
627:
628: The arising of the discrete serie of states is strictly related to the
629: deformation of an open bosonic string theory obtained by adding
630: suitable boundary operators to the action. As a matter of fact,
631: authors of \cite{Callan:1994ub} showed that the discrete spectrum
632: generated by the free boson once it is compactified at the $T$-duality
633: self dual radius can be equivalently explained with an open string
634: model in which one end of string is subjected to Dirichlet boundary
635: conditions while, on the other end, the boundary condition is
636: dynamically defined by adding to the action the following boundary term:
637: \begin{equation}
638: \label{eq:bt_kp}
639: S_b \,=\, \int d t \frac{1}{2}
640: \left(
641: g e^{i \frac{X(t,0)}{\sqrt{2}}} +
642: \overline{g} e^{- i \frac{\overline{X}(t,0)}{\sqrt{2}}}
643: \right)
644: \end{equation}
645: They showed that the discrete serie of states cited above saturates
646: the dynamic of the boundary problem, since Neumann boundary states has
647: null momentum, and the period of the boundary momenta is such that it
648: injects momenta which are integral multiples of $\frac{1}{\sqrt{2}}$,
649: namely the values carried by the discrete states\cite{Callan:1993mw}.
650:
651: This effects is part of a wider connection in which, once we are given
652: a boundary conformal field theory, we can associate it different
653: models considering fluctuations in boundary condensate, where the
654: condensate is defined by a boundary term added to the bulk action:
655: \begin{equation}
656: S = S_{bulk} + g\int dx \psi(x)
657: \end{equation}
658: If the operator $\psi(x)$ is truly marginal (for what this means, see
659: appendix \ref{ch:bcft_deform}), the deformation does not take the
660: model away from the renormalization group fixed
661: point\cite{Recknagel:1998ih}. Thus the bulk theory remains unvaried,
662: and the perturbation effects involve only a redefinition of the
663: boundary conditions, thus affecting boundary states and the boundary
664: operators.
665:
666: For the sake of completeness, we have included in appendix
667: \ref{ch:bcft_deform} a comprehensive introduction to the fundamental
668: concepts and techniques in marginal deformations of a boundary
669: conformal field theory. These techniques will be used extensively in
670: the following.
671:
672: In the connection of the compactified boson, the presence of the
673: enhanced affine symmetry at special values of the compactification
674: radius coincide with the presence of new massless \textsc{open} string
675: states which can be used to deform the theory\cite{Green:1995ga}.
676: When $\Omega(k) = \sqrt{2}$, the closed affine algebra generators can
677: be represented in term of the boson field via the Frenkel-Kac-Siegel
678: construction of the affine algebra. In the closed string channel, the
679: left moving currents are:
680: \begin{equation}
681: H(\zeta(p)) \,=\, \partial X(\zeta(p)), \qquad
682: E^\pm \,=\, \ordprod{e^{\pm i \sqrt{2} X(\zeta(p))}}
683: \end{equation}
684: and the same construction holds for the antiholomorphic sector.
685: The $\hat{su}(2)_1$ currents modes, $J^a_m$ and $\bar{J}^a_m$
686: close 2 copies of the affine algebra:
687: \begin{equation}
688: \label{eq:algeb}
689: [J^a_m,J^a_n] \,=\,
690: \sum_c i f_{abc} J^c_{m+n} \,+\, k m \delta_{a,b} \delta_{m+n,0}
691: \end{equation}
692:
693: The vertex operators for the massless closed string states are given
694: in term of the currents above as:
695: \begin{gather}
696: \label{eq:massl_vertex}
697: V^a_P(p) \,=\,
698: J^a(\zeta(p)) \bar{\partial} \bar{X}(\bar{\zeta}(p))
699: e^{i P X(\zeta(p)) + \bar{X}(\bar{\zeta}(p))}\\
700: \bar{V}^a_P(p) \,=\,
701: \bar{J}^a(\bar{\zeta}(p)) \partial X(\zeta(p))
702: e^{i P X(\zeta(p)) + \bar{X}(\bar{\zeta}(p))}
703: \end{gather}
704: where $P\,=\,p_L \,+\, p_R$ is the total center of mass momentum of
705: the closed string. The vertex operators for the new open string
706: scalar states can be written in the closed string channel as:
707: \begin{multline}
708: S^a_P \,=\, J^a(\zeta(p)) e^{i P X(\zeta(p))} \\\,\equiv\,
709: \frac{1}{2}\left[
710: J^a(\zeta(p)) \,-\,
711: \Omega(p)(\bar{J}^a(\bar{\zeta}(p))) e^{i P X(\zeta(p))}
712: \right]\vert_{|\zeta|=1,\,|\zeta|=\q},
713: \label{eq:open_vert}
714: \end{multline}
715: where $\Omega(p)$ is the glueing authomorphism on the boundary (see
716: equation \eqref{eq:1al}).
717:
718: The occurrence of extra massless open string states in equation
719: \eqref{eq:open_vert}, indicates that, at this special point in
720: compactification moduli space, also the chiral algebra of the boundary
721: theory is enhanced. The associated currents $\mathbf{J}^a(\zeta(p))$
722: (defined as in equation \eqref{eq:1al}) are truly marginal operators
723: (see appendix \ref{ch:bcft_deform}) and can actually be combined to
724: deform the original theory with a boundary action of the form:
725: \begin{equation}
726: \label{def_action}
727: S_B \,=\, \int d x(p,q) \sum_a g_a \mathbf{J}^a(\zeta(p))|_{|\zeta|=\q}
728: \end{equation}
729:
730: In this connection, deformations in equation \eqref{eq:bt_kp} are a
731: particular case of \eqref{def_action}, in which the deformation
732: involves only generators associated to the simple roots of $su(2)$.
733:
734: Since chiral marginal deformation are truly marginal
735: \cite{Recknagel:1998ih} the deformed model will change only for a
736: redefinition of boundary conditions (thus boundary states and boundary
737: operators).
738:
739: In particular, in \cite{Green:1995ga,Recknagel:1998ih} it had been
740: showed that the boundary condition induced by the presence of an
741: action boundary term like that in \eqref{def_action} is represented by
742: a boundary state defined as\cite{Green:1995ga,Recknagel:1998ih}
743: \begin{equation}
744: \label{eq:bs_schom}
745: \Vert g \Rangle
746: \,=\,
747: g \, \Vert N(0) \Rangle_{s.d.}
748: \qquad \text{with} \qquad
749: g \,=\, e^{\sum_a i g_a J^a_0}
750: \end{equation}
751: where $J^a_0$ are the horizontal $SU(2)$ algebra generators.
752: Thus, According to formula \eqref{eq:bs_schom} boundary states are actually
753: a rotation of a ``generator'' boundary state, which is that associated
754: to Neumann boundary conditions with null Wilson line. This
755: construction cover the full moduli space of boundary states. Naively,
756: one can expect to obtain a second branch in the moduli space of
757: boundary states via the same construction on a ``Dirchlet-like''
758: unperturbed boundary state $\Vert D(0) \Rangle_{s.d.}$. However,
759: Dirichlet boundary states are included in the set \eqref{eq:bs_schom}.
760: They are obtained via a perturbation of the form \eqref{def_action},
761: with the particular choice $\Gamma\,=\, e^{- i\pi J_0^1}$\cite{Callan:1994ub}:
762: \begin{equation}
763: \label{eq:dir_bs}
764: \Vert D(0) \Rangle_{s.d.}\,=\,
765: e^{- i\pi J_0^1}\Vert N(0) \Rangle_{s.d.}
766: \end{equation}
767:
768: Boundary states in eq. \eqref{eq:bs_schom} satisfy the rotated gluing
769: condition:
770: \begin{equation}
771: \label{eq:rot_gc}
772: \left(
773: J^a_m + \Omega\circ\gamma_{\bar{J}}(\bar{J}^a_{-m})
774: \right)\Vert B(g) \Rangle \,=\,0
775: \end{equation}
776: with $\gamma_{\bar{J}}(\bar{J}^a)\,:=\,
777: e^{-i \sum_b g_b J^b_0}\bar{J}^a e^{i \sum_b g_b J^b_0}$
778:
779:
780: Cardy's boundary states \eqref{eq:cbs} can be easily retrieved in
781: the set \eqref{eq:bs_schom}.
782: As a matter of fact,
783: when the unperturbed boundary state is a Neumann one with the Wilson
784: line parametrized by $\tilde{t}_+$, the general gluing condition
785: \begin{equation}
786: \label{eq:part_gluing}
787: E^\pm(\zeta) \,=\, e^{\pm i \sqrt{2} \tilde{t}_+} \bar{E}^\pm(\bar{\zeta})
788: \end{equation}
789: is invariant under the shift $\tilde{t}_+ \rightarrow \tilde{t}_+ +
790: \pi\sqrt{2}$. The marginal perturbation implementing this
791: shift, $\Gamma\,=\,\mbox{exp}(i\frac{\pi}{\sqrt{2}}J^3_0)$ acts non
792: trivially on a Neumann boundary state $\Vert N(\tilde{t}_+) \Rangle$,
793: producing exactly a switch of the sign in front of the Ishibashi
794: state built on the $j=\frac{1}{2}$ $\hat{su}(2)_{1}$ module. Thus we
795: can conclude that, in the description of eq. \eqref{eq:bs_schom},
796: Cardy's boundary corresponds to $\Vert N(0) \Rangle_{s.d.}$ and $\Vert
797: N(\pi/\sqrt{2}) \Rangle_{s.d.}$:
798: \begin{subequations}
799: \begin{align}
800: \Vert 0 \Rangle &
801: \,=\,
802: 2^\frac{1}{4}
803: \left(
804: \vert 0 \Rangle + \vert \frac{1}{2} \Rangle
805: \right) \,\equiv\, \Vert N(0) \Rangle_{s.d.} \\
806: \Vert \frac{1}{2} \Rangle &
807: \,=\,
808: 2^\frac{1}{4}
809: \left(
810: \vert 0 \Rangle - \vert \frac{1}{2} \Rangle
811: \right) \,\equiv\, \Vert N(\pi/\sqrt{2}) \Rangle_{s.d.}
812: \end{align}
813: \end{subequations}
814:
815: This result holds for any gluing condition we decide to fix on the
816: boundary of the cylinder. The independent gluing conditions are then
817: parametrized by $SO(3)$, because the elements which yield trivial
818: gluing automorphism are those associated to the center of
819: $SU(2)$. Cardy's boundary states are then those associated to the
820: central elements of $SU(2)$.
821:
822: Authors of \cite{Gaberdiel:2001xm} introduced an alternative
823: description in which they showed that the boundary states generated by
824: the perturbation \eqref{def_action} can be directly parametrized by
825: means of the associated $SU(2)$ elements $g$ and described via the
826: following formula:
827: \begin{equation}
828: \label{eq:su2_bs_2}
829: \Vert g \Rangle
830: \,=\,
831: \sum_{j\,\in\,\mathcal{Y}}
832: \sum_{\substack{-j\,\leq\,m\,\leq\,j\\-j\,\leq\,n\,\leq\,j}}
833: D^j_{m,\,-n}(g) \vert j;\,m,\,n\Rangle
834: \end{equation}
835: where $\vert j;\,m,\,n\Rangle$ is the Ishibashi state associated to
836: the discrete $h\,=\,j^2$ module of the $su(2)$ algebra and
837: $D^j_{m,\,-n}(g)$ is the matrix elements of the $j$-representation of
838: $g\,\in\,SU(2)$, parametrized as $g\,=\,\left(
839: \begin{smallmatrix}
840: a & b \\
841: -b* & a*
842: \end{smallmatrix}
843: \right)$. In this description, the generating boundary state $\Vert
844: N(0) \Rangle_{s.d.}$ is associated to the $SU(2)$ identity, $g =
845: \mathbb{I}$ $N(0)\,=\, \left(
846: \begin{smallmatrix}
847: 1 & 0 \\
848: 0 & 1
849: \end{smallmatrix}
850: \right)$.
851:
852: We will not use this construction to define BIO, however we report it
853: because it allows to write the annulus amplitude with different
854: boundary conditions associate respectively to $\boum$ and $\boup$ in a
855: quite simple way. Let us consider the transition amplitude between two
856: boundary states $|\alpha\rangle$ and $|\beta\rangle$ which can be
857: defined via the action of $g \in SU(2)$ on $|\alpha\rangle$,
858: $|\beta\rangle\,=\, g |\alpha\rangle$. From the above definitions of
859: boundary states, it can be shown that $A_{\alpha, g \alpha}$ depends
860: only on the conjugacy classes of $g$ (for a detailed demonstration see
861: \cite{Recknagel:1998ih}, or \cite{Gaberdiel:2001xm}, section
862: 4). Therefore, we can choose to deform the boundary state with an
863: element in a given torus of $SU(2)$: $t \,=\, h^{-1} g h \,=\,
864: \left(
865: \begin{smallmatrix}
866: e^{4 \pi i \lambda} & 0 \\
867: 0 & e^{- 4 \pi i \lambda}
868: \end{smallmatrix}
869: \right)$ following the detailed analysis in \cite{Gaberdiel:2001xm}, we
870: can finally write the amplitude as
871: \begin{equation}
872: \label{eq:amppp}
873: \mathcal{A}(p) \,=\, \frac{1}{\sqrt{2}}
874: \sum_{j \in \frac{1}{2} \mathbb{Z_+}}
875: cos(8 \pi j \lambda) \vartheta_{2 \tilde{q}} (\tilde{q})
876: \end{equation}
877: where $\tilde{q} = e^{ - \frac{2 \pi i}{\tau}}$
878:
879: A boundary perturbation will affect the boundary operators spectrum
880: too. Generically, when the perturbing field is truly marginal, the
881: study of the deformation of a correlator containing both bulk and
882: boundary fields allows to define the image $\tilde{\psi}_j$ of a
883: boundary field $\psi_j$ under a rotation generated by the perturbing
884: field $\psi$ as\cite{Recknagel:1998ih} (see appendix \ref{ch:bcft_deform}):
885: \begin{equation}
886: \label{eq:bf_pert}
887: \tilde{\psi}_j\,=\,
888: \left[
889: e^{\frac{1}{2} g \psi} \psi_j
890: \right](u)\,:=\,
891: \sum_{n=0}^\infty \frac{g^n}{2^n n!}
892: \oint_{C_1} \frac{dx_1}{2\pi} \cdots \oint_{C_n} \frac{dx_n}{2\pi}
893: \psi_j(u)\psi(x_n)\cdots\psi(x_1)
894: \end{equation}
895: where $C_i$ are small circles surrounding the insertion points of the
896: operators $\psi(x_i)$ on the boundary.
897:
898: As they stands, representations \eqref{eq:bs_schom} and
899: \eqref{eq:su2_bs_2} of boundary conditions do not allow to
900: successfully explain how the transition between pairwise adjacent
901: boundary conditions take place. Thus, we have introduced a new
902: representation for the infinite set of boundary conditions which can be
903: applied to the compactified boson at the self-dual radius. This
904: representation merges the infinite choice of boundary conditions with
905: the the necessity to have a BIO acting ``\emph{a l\'a Cardy}'', ie mediating
906: between the diffenten boundary conditions exploiting the fusion rules
907: of the associated chiral algebra.
908:
909: We can exploit construction \eqref{eq:bs_schom} to
910: parametrize the generic boundary condition defined over the (inner or
911: outer) boundary of the $k$-th cylindrical end, represented by the
912: boundary state $\Vert g(k) \Rangle$, with a couple of elements:
913: \begin{equation}
914: \label{eq:boundary_par}
915: \left(
916: \Vert J(k) \Rangle,\, \Gamma(k)
917: \right) \qquad \text{with}\quad
918: \begin{cases}
919: J(k) & \,\in\,\mathcal{A} \\
920: \Gamma(k) & \,\in\,\frac{SU(2)}{\mathbb{Z}_2}
921: \end{cases}
922: \end{equation}
923: being
924: $\Vert J(k) \Rangle$ a Cardy's boundary state (thus
925: corresponding to an element in the center of $SU(2)$) and $\Gamma$ an
926: $\frac{SU(2)}{\mathbb{Z}_2}$ group element, such that:
927: \begin{equation}
928: \label{eq:poly_bs}
929: \Vert g(k) \Rangle
930: \,=\,
931: \Gamma \, \Vert J(k) \Rangle.
932: \end{equation}
933:
934: In this connection, the model can be defined not as a truly marginal
935: deformation of a open string theory by means of the action of elements of the
936: affine chiral $SU(2)$ , but as a truly marginal deformation of the
937: $\hat{su}(2)_{k=1}$ by means of $SO(3)$ elements.
938:
939: Thus, BIO for the compactified boson with $\Omega(p) = \sqrt{2}$ are
940: actually given by the consequent deformation induced by the two
941: adjacent boundary conditions on $\hat{\mathfrak{su(2)}}_{k=1}$ WZW
942: model boundary insertion operators which we introduced in formula
943: \eqref{eq:bio}
944:
945: To understand how this deformation affects and define boundary
946: insertion operators, let us consider the $(p,q)$-edge of the ribbon
947: graph. Let the two adjacent boundary conditions be defined as:
948: \begin{subequations}
949: \label{bcpq}
950: \begin{align}
951: \Vert g_1 (p) \Rangle & \,=\, \Gamma_1(p) \Vert J_1 (p) \Rangle \\
952: \Vert g_2 (q) \Rangle & \,=\, \Gamma_2(q) \Vert J_2 (q) \Rangle
953: \end{align}
954: \end{subequations}
955:
956: According to the parametrization of boundary condition introduced
957: above, BIO must mediate both between Cardy's boundary states and
958: between the $\frac{SU(2)}{\mathbb{Z}_2}$ elements $\Gamma_1(p)$ and
959: $\Gamma_2(q)$. While the former action is achieved trough the fusion
960: prefactor $\mathcal{N}_{J(p)\,j(p,q)}^{J(q)}$, the latter can be
961: understood deforming BIO with the action of both the boundary
962: potentials which we are adding on the $(p,q)$-edge. As a matter of
963: fact, according to equation \eqref{bcpq}, the theory on the $(p)$-th
964: polytope is deformed by the action of the boundary term $S_{B(p)} =
965: \int dx(p,q) \mathbf{J}_1(\zeta(p))\vert_{|\zeta(p)|=\frac{2 \pi}{2
966: \pi - \varepsilon(p)}}$, while the theory on the $(q)$-th polytope
967: is deformed by the boundary term $S_{B(q)} = \int dx(q,p)
968: \mathbf{J}_2(\zeta(q))\vert_{|\zeta(q)|=\frac{2 \pi}{2 \pi -
969: \varepsilon(q)}}$. Recalling that $x(q,p) = - x(p,q)$, (the
970: functional part of) boundary insertion operators, we propose
971: $\psi_{j(p,q)}$ to be deformed by a suitable combination of the
972: $SO(3)$ operators which are associated to the above boundary terms. We
973: ask this combination to cancel, on the boundary, the global effect of
974: the boundary deformation, to let the two ends glue dynamically in such
975: a way that this dynamic is actually governed by the fusion rules of
976: the WZW model. This correspond to a perform over $\psi_{j(p,q)}$ a
977: rotation induced by the $\frac{SU(2)}{\mathbb{Z}_2}$ element
978: $\overline{\Gamma} \,=\, \Gamma_2\Gamma_1^{-1}$, with $\Gamma_i =
979: e^{i\mathbf{J}_i}$. In the same way, $\psi_{j(q,p)}$ will be deformed
980: by the action of by the action of $\overline{\Gamma}^{-1} \,=\,
981: \Gamma_1\Gamma_2^{-1}$.
982:
983: To show how the above rotation alter the functional part of boundary
984: insertion operators, let us consider the explicit expression of
985: components of $\hat{\mathfrak{su}}(2)_1$ BIO which we introduced in
986: \eqref{eq:bio}. Let us drop for a while the dependence from the fusion
987: rule factor $\mathcal{N}_{J_1(p)\,j(p,q)}^{J_2(q)}$. Such components
988: are labelled by two (semi-)integers $j = 0, \frac{1}{2}$ and $-j < m <
989: j$. For $j = 0$ the unique component is the identity operator
990: $\psi_{0,0}x(p,q) = \mathbb{I}$, while for $j = \frac{1}{2}$ the two
991: components are $\psi_{\frac{1}{2}\,\pm\frac{1}{2}} = \left.e^{\pm
992: \frac{i}{\sqrt{2}}X[\zeta(p)(z(p,q))]}\right\vert_{y(p,q)=0}$.
993:
994: Let us consider the action of the deformation on $\psi_{j(p,q)}$
995: generated by $\overline{\Gamma} =
996: e^{i\overline{J}}$
997: (since we are not moving to a definite representation, we can omit the
998: quantum number $m$).
999: According to \eqref{eq:bf_pert}, the rotated boundary operator will
1000: be:
1001: \begin{equation}
1002: \label{eq:part_def}
1003: \tilde{\psi}_{j(p,q)}(u(p,q))\,=\,
1004: \left[
1005: e^{\frac{1}{2} \overline{J}} \psi_{j(p,q)}
1006: \right](u(p,q))
1007: \end{equation}
1008:
1009: We can compute explicitly the expression of $\tilde{\psi}_{j(p,q)}$
1010: thanks to the self-locality of the boundary operators, and to the OPE
1011: between the truly marginal fields in the chiral algebra and a boundary
1012: operator:
1013: \begin{equation}
1014: \mathbf{J}(x)\psi_j(u) \sim
1015: \frac{\mathbf{X}_{\overline{J}}^j}{x - u} \psi_j(u)
1016: \end{equation}
1017: where $\mathbf{X}_{\overline{J}}^j$ is the natural action of the chiral
1018: algebra on a state of the $h=j^2$ $\hat{su}(2)_1$ module (see formula
1019: \eqref{eq:hor_ac}.
1020:
1021: An order by order computation in \eqref{eq:part_def} gives:
1022: \begin{equation}
1023: \label{boh}
1024: \tilde{\psi}_{j(p,q)}(u(p,q))\,=\,
1025: e^{\frac{i}{2} \mathbf{X}_{\overline{J}}^j} \psi_{j(p,q)}
1026: \end{equation}
1027: \ie{} the natural action of the chiral algebra on the vertex algebra
1028: fields translates into the natural action of an element of
1029: $SU(2)/\mathbb{Z}_2$ on the components of the primary field associated
1030: to a given $\hat{su}(2)_1$'s module. Moving to a specific
1031: representation, equation \eqref{boh} becomes:
1032: \begin{equation}
1033: \label{eq:rotbio_exp}
1034: \tilde{\psi}_{(j,\,m)(p,q)}(u(p,q))\,=\,
1035: D^j_{m n}(\Gamma) \, \psi_{(j,n)(p,q)}
1036: \end{equation}
1037: where $D^j_{mn}$ are the Wigner functions associated to
1038: $\Gamma\,=\,\left(
1039: \begin{smallmatrix}
1040: a & b \\
1041: -b* & a*
1042: \end{smallmatrix}
1043: \right)$:
1044: \begin{multline}
1045: \label{eq:su2bs_coeff}
1046: D^j_{m,\,n}(\Gamma)
1047: \,=\,
1048: \sum_{l\,=\,\text{max}(0,n-m)}^{\text{min}(j-m,j-n)}
1049: \frac{
1050: \left[
1051: (j\,+\,m)!\,
1052: (j\,-\,m)!\,
1053: (j\,+\,n)!\,
1054: (j\,-\,n)!\,
1055: \right]^{\frac{1}{2}}}
1056: {(j\,-\,m\,-\,l)!\,
1057: (j\,+\,n\,-\,l)!\,
1058: l!\,
1059: (m\,-\,n\,+\,l)!\,} \\
1060: \,\times\,
1061: a^{j\,+\,n\,-\,l}\,
1062: (a*)^{j\,-\,m\,-\,l}\,
1063: b^l\,
1064: (-b*)^{m\,-\,n\,+\,l}.
1065: \end{multline}
1066:
1067: Obviously, the spectrum of
1068: $\hat{su}(2)_1$ boundary primary fields must be invariant under the
1069: action of $\mathbb{Z}_2$. A first intuition about this comes once we
1070: remember the gluing condition \eqref{eq:part_gluing} being invariant
1071: under the shift generating the $\Vert1/2\Rangle$ Cardy's state: the
1072: spectrum of boundary operators is generated the action of a copy of
1073: the chiral algebra:
1074: \begin{equation}
1075: \mathbb{W}(\zeta)\,=\,
1076: \begin{cases}
1077: W(\zeta) & \Im{\zeta}\,\geq\,0\\
1078: \Omega\circ\gamma_{\bar{\Gamma}}(\bar{W}(\bar{\zeta})) & \Im{\zeta}\,<\,0
1079: \end{cases}.
1080: \end{equation}
1081: The gluing automorphism itself generates the boundary operators'
1082: spectrum, thus a boundary condensate which leaves invariant the gluing
1083: automorphism automatically leaves invariant the boundary operators'
1084: spectrum too. An explicit computation via formula
1085: \eqref{eq:rotbio_exp} of Cardy's boundary
1086: operators rotated by the action of the boundary condensate
1087: correspondent to the $\Vert N(\pi/\sqrt{2})\Rangle_{s.d.}$ boundary
1088: state confirms this statement: rotated boundary operators are obtained
1089: by multiplication by an inessential phase factor.
1090:
1091: Finally, restoring the fusion multiplicative coefficient, we have the
1092: following expression for boundary insertion operators for the
1093: compactified boson at the self dual radius:
1094: \begin{equation}
1095: \label{eq:final_bio}
1096: \psi^{[J_2,\,\Gamma_2](q)\,[J_1,\,\Gamma_1](p)}_{[j,\,m](p,q)}\,=\,
1097: \sum_{n=-j}^j D^{j(p,q)}_{m\,n(p,q)}(\Gamma_2{\Gamma_1}^{-1}) \,
1098: \psi_{[j\,n](p,q)}^{J_2(q)\,J_1(p)}.
1099: \end{equation}
1100: where $\psi_{j(p,q)}^{J(p)\,J(q)} (x(p,q)) \,=\,
1101: \mathcal{N}_{J(p)\,j(p,q)}^{J(q)}\,\psi_{j(p,q)}(x(p,q))$
1102:
1103:
1104: \section{The algebra of rotated Boundary Insertion Operators}
1105:
1106:
1107: \begin{comment}
1108: We have characterized each possible boundary condition on the $p$ edge of
1109: a ribbon graph by a couple of elements:
1110: \begin{equation}
1111: \label{eq:bc}
1112: \Vert g(p) \Rangle \,=\, \Gamma(p) \Vert J(p) \Rangle
1113: \end{equation}
1114: $\Vert J(p) \Rangle$ be
1115: ing a Cardy's boundary state and
1116: $\Gamma(p)\in\frac{SU(2)}{\mathbb{Z}_2}$.
1117: These boundary conditions reflect the presence of a boundary
1118: condensate
1119: \begin{equation}
1120: S_B\,=\,\int dx g(\zeta(p))\vert_{|\zeta|=1}
1121: \end{equation}
1122: perturbing (in an open string configuration) a Lagrangian for a free
1123: field defined on a segment $y \in
1124: \left[0,\,\frac{L(p)}{\theta(p)}\right]$.
1125:
1126: With boundary conditions
1127: given as in equation \eqref{eq:bc}, the spectrum of associated
1128: boundary operators is defined via a rotation of the unperturbed
1129: spectrum, this latter being made up by Cardy's boundary operators.
1130: \begin{equation}
1131: \psi_{[j,m](p,q)}^{[J,\Gamma](p)} \,=\,
1132: D_{m\,n}^j(\Gamma) \,
1133: \psi_{[j,m](p,q)}^{J(p)}
1134: \end{equation}
1135:
1136: The definition of the rotated BIO which mediates the changing in
1137: boundary conditions between two adjacent polytopes is straightforward.
1138: Boundary condensates like in equation \eqref{def_action} corresponds
1139: to the simultaneous deformation of both the cylinder boundary
1140: conditions with the same operator. The obvious generalization would
1141: involve a cylindrical end with different boundary conditions at the
1142: two boundaries (see fig. \ref{fig:cyltouhp}). This corresponds to
1143: different boundary operators integrated on different portions of
1144: boundary:
1145: \begin{equation}
1146: S_B\,=\,\int dx\, g_1(\zeta(p))\vert_{|\zeta|=1} \,+\,
1147: \int dx\, g_2(\zeta(p))\vert_{|\zeta|=\q}
1148: \end{equation}
1149: \begin{figure}[!t]
1150: \centering
1151: \includegraphics[width=.7\textwidth]{immagini/cyltoplane}
1152: \caption{bla bla}
1153: \label{fig:cyltouhp}
1154: \end{figure}
1155: In this case, the boundary states formalism allows us to define the
1156: cylinder amplitude as a sum over twisted characters of the symmetry
1157: algebra, \ie{} characters of representation twisted by an inner
1158: automorphism $Ad_U$, with $U\,=\,e^{i(g_2 - g_1)}$.
1159:
1160: Likewise, when we deal with the amplitude define on the full open
1161: surface, we have to take into account transition in boundary
1162: conditions which take place on each edge of the ribbon graph. Let us
1163: consider the $(p,q)$-edge of the ribbon graph. Let the adjacent
1164: boundary conditions be defined as:
1165: \begin{subequations}
1166: \begin{align}
1167: \Vert g_1 (p) \Rangle & \,=\, \Gamma_1(p) \Vert J_1 (p) \Rangle \\
1168: \Vert g_2 (q) \Rangle & \,=\, \Gamma_2(q) \Vert J_2 (q) \Rangle
1169: \end{align}
1170: \end{subequations}
1171: The BIOs must act both on the Cardy's boundary state and on the
1172: $\frac{SU(2)}{\mathbb{Z}_2}$ group element. The former action is as
1173: usual obtained via the action of Cardy's boundary conditions changing
1174: operators $\psi_{j(p,q)}^{J_2(q)\,J_1(p)}\,\propto\,
1175: \mathcal{N}_{J_1(p)\,j(p,q)}^{J_2(q)}$, while the latter involves the
1176: action of an algebra inner automorphism represented by
1177: $D^{j(p,q)}_{m\,n(p,q)}(\Gamma_2{\Gamma_1}^{-1})$.
1178: \end{comment}
1179:
1180:
1181: According to last section remarks, boundary insertion operators for
1182: the compactified boson at enhanced symmetry values of the self-dual
1183: radius have the following expression:
1184: \begin{equation}
1185: \label{eq:final_bio.}
1186: \psi^{[J_2,\,\Gamma_2](q)\,[J_1,\,\Gamma_1](p)}_{[j,\,m](p,q)}\,=\,
1187: \sum_{n=-j}^j D^{j(p,q)}_{m\,n(p,q)}(\Gamma_2{\Gamma_1}^{-1}) \,
1188: \psi_{[j\,n](p,q)}^{J_2(q)\,J_1(p)}.
1189: \end{equation}
1190:
1191: The aim of this, quite technical, section, is to show that with this
1192: choice, effects of boundary perturbations do not affect the
1193: algebra of boundary operators, thus they do not change the dynamic of
1194: the model. This is a check of consistency of the above choice for
1195: boundary insertion operators, since actually all deformations we have
1196: considered are induced by truly marginal operators, thus they must not
1197: break the $su(2)$ chiral algebra.
1198:
1199: Let us deal with the simplest case: the amplitude computed on the open
1200: surface associated with the sphere with three punctures. The amplitude
1201: over each cylindrical end will involve a sum over characters twisted
1202: by the action of conjugacy classes of $\frac{SU(2)}{\mathbb{Z}_2}$.
1203: The full amplitude will involve a sum over the intermediate
1204: channels associated with the three edges of the ribbon graph, each
1205: being associated to an automorphism by the operator $U\,=\, e^{i
1206: \left[g_1(q)-g_2(p)\right]}, \forall (p,q)\,=\,(1,2),(1,3),(2,3)$. Such an
1207: automorphism acts at BIOs' level: the BIO associated to the
1208: $(p,q)$-th edge will be the Cardy's one rotated by the action of the
1209: $\frac{SU(2)}{\mathbb{Z}_2}$ element $\Gamma_2{\Gamma_1}^{-1}$
1210:
1211: The algebra of rotated BIOs follows from their definition. Let us
1212: notice that rotated BIOs are just a superposition of the different
1213: components of Cardy's $\hat{su}(2)_1$ primary operators (with respect
1214: to the affine chiral algebra).
1215:
1216: Let us focus our attention on the two points function. When we
1217: consider two BIOs both mediating a boundary condition changing in the
1218: $p$-to-$q$ direction, we deal with the following expression:
1219: \begin{multline}
1220: \label{eq:rot2points}
1221: \left\langle
1222: \psi^{[J_2,\,\Gamma_2](q)\,[J_1,\,\Gamma_1](p)}_{[j,\,m](p,q)}
1223: (x_1(p,q)) \,
1224: \psi^{[J_4,\,\Gamma_4](q)\,[J_3,\,\Gamma_3](p)}_{[j',\,m'](p,q)}
1225: (x_2(p,q))
1226: \right\rangle\,=\,\\
1227: \sum_{n\,n'}
1228: D^{j(p,q)}_{m\,n(p,q)}(\Gamma_2{\Gamma_1}^{-1})
1229: D^{j'(p,q)}_{m'\,n'(p,q)}(\Gamma_4{\Gamma_3}^{-1})
1230: \left\langle
1231: \psi^{J_2(q)\,J_1(p)}_{[j,\,n](p,q)}(x_1(p,q)) \,
1232: \psi^{J_4(q)\,J_3(p)}_{[j',\,n'](p,q)}(x_2(p,q))
1233: \right\rangle
1234: \end{multline}
1235: First of all, we must notice that a coherent gluing impose the two
1236: operators to mediate between the same boundary conditions (see eq.
1237: \eqref{eq:2points}), thus the above expression reduces to:
1238: \begin{multline}
1239: \label{eq:rot2points_sim}
1240: \left\langle
1241: \psi^{[J_2,\,\Gamma_2](q)\,[J_1,\,\Gamma_1](p)}_{[j,\,m](p,q)}
1242: (x_1(p,q)) \,
1243: \psi^{[J_2,\,\Gamma_2](q)\,[J_1,\,\Gamma_1](p)}_{[j',\,m'](p,q)}
1244: (x_2(p,q))
1245: \right\rangle\,=\,\\
1246: \sum_{n\,n'}
1247: D^{j(p,q)}_{m\,n(p,q)}(\Gamma_2{\Gamma_1}^{-1})
1248: D^{j'(p,q)}_{m'\,n'(p,q)}(\Gamma_2{\Gamma_1}^{-1})
1249: \left\langle
1250: \psi^{J_2(q)\,J_1(p)}_{[j,\,n](p,q)}
1251: \psi^{J_2(q)\,J_1(p)}_{[j',\,n'](p,q)}
1252: \right\rangle
1253: \end{multline}
1254: Equation \eqref{eq:rot2points_sim} shows that the net effect of the
1255: rotation on the two points function vanish, because we are actually
1256: implementing the same SU(2) rotation on all boundary fields entering
1257: in the unperturbed correlator. Thus the local $SU(2)$
1258: invariance ensures:
1259: \begin{multline}
1260: \left\langle
1261: \psi^{[J_2,\,\Gamma_2](q)\,[J_1,\,\Gamma_1](p)}_{[j,\,m](p,q)}
1262: (x_1(p,q)) \,
1263: \psi^{[J_2,\,\Gamma_2](q)\,[J_1,\,\Gamma_1](p)}_{[j',\,m'](p,q)}
1264: (x_2(p,q))
1265: \right\rangle\,=\,\\
1266: \left\langle
1267: \psi^{J_2(q)\,J_1(p)}_{[j,\,m](p,q)}(x_1(p,q)) \,
1268: \psi^{J_2(q)\,J_1(p)}_{[j',\,m'](p,q)}(x_2(p,q))
1269: \right\rangle
1270: \end{multline}
1271:
1272: Dealing with the two points function between a
1273: $p$-to-$q$ and $q$-to-$p$ mediating operators, the situation is
1274: slightly different. We have to compute:
1275: \begin{multline}
1276: \label{eq:rot2points2}
1277: \left\langle
1278: \psi^{[J_2,\,\Gamma_2](q)\,[J_1,\,\Gamma_1](p)}_{[j,\,m](p,q)}
1279: (x_1(p,q)) \,
1280: \psi^{[J_1,\,\Gamma_1](p)\,[J_2,\,\Gamma_2](q)}_{[j',\,m'](q,p)}
1281: (x_2(q,p))
1282: \right\rangle\,=\,\\
1283: \sum_{n\,n'}
1284: D^{j(p,q)}_{m\,n(p,q)}(\Gamma_2{\Gamma_1}^{-1})
1285: D^{j'(q,p)}_{m'\,n'(q,p)}(\Gamma_1{\Gamma_2}^{-1}) \times \\
1286: \left\langle
1287: \psi^{J_2(q)\,J_1(p)}_{[j,\,n](p,q)}(x_1(p,q)) \,
1288: \psi^{J_1(p)\,J_2(q)}_{[j',\,n'](q,p)}(x_2(q,p))
1289: \right\rangle
1290: \end{multline}
1291:
1292: As a matter of fact, in the previous expression we are dealing with a
1293: representation of diagonal subgroup of the direct product
1294: $\frac{SU(2)}{\mathbb{Z}_2}(p,q)\times\frac{SU(2)}{\mathbb{Z}_2}(q,p)$,
1295: thus it holds (see eq. \eqref{eq:dir_prod}:
1296: \begin{equation}
1297: \label{eq:dirprod_rep}
1298: D^{j(p,q)}_{m\,n(p,q)}(\Gamma_2{\Gamma_1}^{-1})
1299: D^{j'(q,p)}_{m'\,n'(q,p)}(\Gamma_1{\Gamma_2}^{-1})
1300: \,=\,D^{j \times j'}_{m\,n;\,m'\,n'}(\mathbb{I})
1301: \end{equation}
1302:
1303: The trivial Clebsh-Gordan expansion (eq. \eqref{eq:CG_series}) gives
1304: (we omit the polytope indices writing the Clebsh-Gordan coefficients):
1305: \begin{multline}
1306: \left\langle
1307: \psi^{[J_2,\,\Gamma_2](q)\,[J_1,\,\Gamma_1](p)}_{[j,\,m](p,q)}
1308: (x_1(p,q)) \,
1309: \psi^{[J_1,\,\Gamma_1](p)\,[J_2,\,\Gamma_2](q)}_{[j',\,m'](q,p)}
1310: (x_2(q,p))
1311: \right\rangle\,=\,\\
1312: \sum_{n\,n'}\,\sum_{J\,N} C_{j_1\,m_1\,j_2\,m_2}^{J\,N}
1313: C_{j_1\,n_1\,j_2\,n_2}^{J\,N}
1314: \left\langle
1315: \psi^{J_2(q)\,J_1(p)}_{[j,\,n](p,q)}(x_1(p,q)) \,
1316: \psi^{J_1(p)\,J_2(q)}_{[j',\,n'](q,p)}(x_2(q,p))
1317: \right\rangle\,=\,\\
1318: \left\langle
1319: \psi^{J_2(q)\,J_1(p)}_{[j,\,m](p,q)}(x_1(p,q)) \,
1320: \psi^{J_1(p)\,J_2(q)}_{[j',\,m'](q,p)}(x_2(q,p))
1321: \right\rangle.
1322: \end{multline}
1323: In the last equation we have used the unitarity of Clebsh-Gordan
1324: coefficients (see equation \eqref{eq:CG_unit2}).
1325:
1326: To calculate the OPE of rotated BIOs, let us notice that the rotation
1327: generated by the boundary condensate does not change them coordinate
1328: dependence. Let us consider the situation depicted in figure
1329: \ref{fig:BIO_ope-deformed}.
1330: \begin{figure}[!t]
1331: \centering
1332: \includegraphics[width=.5\textwidth]{immagini/BIO_ope-deformed}
1333: \caption{bla bla bla}
1334: \label{fig:BIO_ope-deformed}
1335: \end{figure}
1336:
1337: OPE between
1338: $\psi^{[J_1,\,\Gamma_1](p)\,[J_3,\,\Gamma_3](r)}_{[j_1,\,m_1](r,p)}$
1339: and $\psi^{[J_3,\,\Gamma_3](r)\,[J_2,\,\Gamma_2](q)}_{[j',\,m'](q,r)}$
1340: will mediate a change in boundary conditions from
1341: $[J_2,\,\Gamma_2](q)$ to $[J_1,\,\Gamma_1](p)$. In particular,
1342: \begin{multline}
1343: \notag
1344: \psi^{[J_1,\,\Gamma_1](p)\,[J_3,\,\Gamma_3](r)}_{[j_1,\,m_1](r,p)}
1345: (\omega_r) \,
1346: \psi^{[J_3,\,\Gamma_3](r)\,[J_2,\,\Gamma_2](q)}_{[j',\,m'](q,r)}
1347: (\omega_q)\,=\, \\
1348: \sum_{n_1(r,p)\,n_2(q,r)}\,
1349: D^{j_1(r,p)}_{m_1\,n_1(p,q)}(\Gamma_1{\Gamma_3}^{-1})\,
1350: D^{j_2(q,r)}_{m_2\,n_2(q,r)}(\Gamma_3{\Gamma_2}^{-1})\,
1351: \psi^{J_1(p)\,J_3(r)}_{[j_1,\,n_1](r,p)}(\omega_r) \,
1352: \psi^{J_3(r)\,J_2(q)}_{[j_2,\,m_2](q,r)}(\omega_q)
1353: \end{multline}
1354:
1355: We are dealing again with a representation of the diagonal subgroup
1356: of the direct product
1357: $\frac{SU(2)}{\mathbb{Z}_2}(r,p)\times\frac{SU(2)}{\mathbb{Z}_2}(q,r)$,
1358: thus applying \eqref{eq:dir_prod} and the Clebsh-Gordan series
1359: expansion \eqref{eq:CG_series} we are left with:
1360: \begin{multline}
1361: \label{interm1}
1362: \sum_{\substack{n_1(r,p)\\n_2(q,r)}}\,
1363: \sum_{\substack{j=|j_1-j_2|\\|m|\leq{}j\\|n|\leq{}j}}^{j_1+j_2}
1364: C^{j\,m}_{j_1(r,p)\,m_1(r,p)\,j_2(q,r)\,m_2(q,r)}\,
1365: D^j_{m\,n}(\Gamma_1\Gamma_2^{-1})\, \\ \times
1366: C^{j\,n}_{j_1(r,p)\,n_1(r,p)\,j_2(q,r)\,n_2(q,r)}\,
1367: \psi^{J_1(p)\,J_3(r)}_{[j_1,\,n_1](r,p)}(\omega_r) \,
1368: \psi^{J_3(r)\,J_2(q)}_{[j_2,\,n_2](q,r)}(\omega_q)
1369: \end{multline}
1370:
1371: The OPE between Cardy's boundary operators reads:
1372: \begin{multline}
1373: \psi^{J_1(p)\,J_3(r)}_{[j_1,\,n_1](r,p)}(\omega_r) \,
1374: \psi^{J_3(r)\,J_2(q)}_{[j_2,\,n_2](q,r)}(\omega_q)\,=\,
1375: \sum_{j_3\,n_3}\left\vert
1376: \omega_r\,-\,\omega_q
1377: \right\vert^{H(q,p)\,-\,H(r,p)\,-\,H(q,r)}\\
1378: C^{j_3\,n_3}_{j_1\,n_1\,j_2\,n_2}\,
1379: \mathcal{C}_{j_1\,j_2\,j_3}^{J_1(p)\,J_3(r)\,J_2(q)}\,
1380: \psi^{J_1(p)\,J_2(q)}_{[j_3,\,n_3](q,p)}(\omega_q).
1381: \end{multline}
1382: The Clebsh-Gordan coefficients
1383: $C^{j_3\,n_3}_{j_1\,n_1\,j_2\,n_2}$ compensate the fact that the LHS
1384: and RHS terms have different transformation behavior under the
1385: action of the horizontal $su(2)$ algebra, while the coefficients
1386: $\mathcal{C}_{j_1\,j_2\,j_3}^{J_1(p)\,J_3(r)\,J_2(q)}$ reflect the non
1387: trivial dynamic on each trivalent vertex of the ribbon graph.
1388:
1389: The inclusion of this last OPE into \eqref{interm1} and the Clebsh-Gordan
1390: coefficients' unitarity (equation \eqref{eq:CG_unit2}) leave us with:
1391: \begin{multline}
1392: \label{eq:rotated_ope}
1393: \psi^{[J_1,\,\Gamma_1](p)\,[J_3,\,\Gamma_3](r)}_{[j_1,\,m_1](r,p)}
1394: (\omega_r) \,
1395: \psi^{[J_3,\,\Gamma_3](r)\,[J_2,\,\Gamma_2](q)}_{[j',\,m'](q,r)}
1396: (\omega_q)\,=\, \\
1397: \sum_{j_3\,m}
1398: C^{j_3\,m}_{j_1\,m_1\,j_2\,m_2}\,
1399: \mathcal{C}_{j_1\,j_2\,j_3}^{J_1(p)\,J_3(r)\,J_2(q)}\,
1400: \psi^{[J_1,\,\Gamma_1](p)\,[J_2,\,\Gamma_2](q)}_{[j_1,\,m_1](q,p)}
1401: (\omega_p) \,
1402: \end{multline}
1403: We demonstrate that OPE between rotated BIOs is formally equal to OPE between
1404: unrotated BIOs. Thus, on the ribbon graph the non trivial dynamic is
1405: given by the fusion among the three representations entering in each
1406: trivalent vertex.
1407:
1408:
1409: \section{The action of BIOs at the self-dual radius}
1410:
1411: With the above remarks, we can investigate the properties of the four
1412: points functions of BIOs exploiting their crossing properties.
1413:
1414: \begin{figure}[!t]
1415: \centering
1416: \includegraphics[width=\textwidth]{immagini/BIO_4points}
1417: \caption{Four points function crossing symmetry}
1418: \label{fig:BIO_4points}
1419: \end{figure}
1420:
1421: First of all let us consider the natural picture in which the
1422: computation of a four points function arises. Let us consider two near
1423: trivalent vertexes. Due to the variable connectivity of the
1424: triangulation, the two configuration shown in figure
1425: \ref{fig:BIO_4points} are both admissible. The transition from the
1426: situation depicted in the lhs and the one in the rhs of the pictorial
1427: identity of figure \ref{fig:BIO_4points}, corresponds exactly to the
1428: transition between the $s$-channel and the $t$-channel of the four
1429: point blocks of a single copy of the bulk theory, thus the two
1430: factorization of the four points function
1431: $\langle
1432: \psi_{j_1(s,p)}^{J_1(p)\,J_4(s)}\,
1433: \psi_{j_2(r,s)}^{J_4(s)\,J_3(r)}\,
1434: \psi_{j_3(q,r)}^{J_3(r)\,J_2(q)}\,
1435: \psi_{j_4(p,q)}^{J_2(q)\,J_1(p)}
1436: \rangle$,
1437: pictorially represented in
1438: \ref{fig:BIO_4points}, are related by the bulk crossing matrices:
1439: \begin{equation}
1440: \label{fusion_matrices}
1441: F_{j_6(s,q)\,j_5(r,p)}
1442: \begin{bmatrix}
1443: j_4(p,s) & j_1(q,p) \\
1444: j_3(s,r) & j_2(r,q)
1445: \end{bmatrix}
1446: \end{equation}
1447: The explicit computation of the two factorization leads to the
1448: relation:
1449: \begin{multline}
1450: \label{eq:fact}
1451: \mathcal{C}^{J_4(s)\,J_3(r)\,J_2(q)}_{j_2(r,s)\,j_3(q,r)\,j_5(q,s)}\,
1452: \mathcal{C}^{J_1(p)\,J_4(s)\,J_2(q)}_{j_1(s,p)\,j_5(q,s)\,j_1(s,p)}\,
1453: \mathcal{C}^{J_1(p)\,J_2(q)\,J_1(p)}_{j_1(s,p)\,j_1(s,p)\,0}
1454: \,=\,\\
1455: \sum_{j_5(r,p)}
1456: F_{j_6(s,q)\,j_5(r,p)}
1457: \begin{bmatrix}
1458: j_4(p,s) & j_1(q,p) \\
1459: j_3(s,r) & j_2(r,q)
1460: \end{bmatrix}\,\times\\
1461: \mathcal{C}^{J_1(p)\,J_4(s)\,J_3(r)}_{j_1(s,p)\,j_2(r,s)\,j_6(r,p)}\,
1462: \mathcal{C}^{J_3(r)\,J_2(q)\,J_1(p)}_{j_3(q,r)\,j_4(p,q)\,j_6(p,r)}\,
1463: \mathcal{C}^{J_1(p)\,J_3(r)\,J_1(p)}_{j_6(r,p)\,j_6(p,r)\,0},
1464: \end{multline}
1465: \ie{} the usual BCFT sewing relation among boundary operators' OPEs.
1466:
1467: This complete our analysis of the conformal properties of the full
1468: theory arising by glueing together the BCFTs defined over each
1469: cylindrical ends: with the above construction, BIOs play exactly the
1470: role usual boundary operators play in BCFT.
1471:
1472: This analogy allows us to apply to BIOs all boundary operators
1473: properties. In particular, we can identify their OPE coefficients with
1474: the fusion matrices \eqref{fusion_matrices} with
1475: the following entries assignation:
1476: \begin{equation}
1477: \label{eq:OPE_fusion}
1478: \mathcal{C}^{J_1(p)\,J_2(q)\,J_3(s)}_{j_1(s,p)\,j_2(r,s)\,j_3(q,r)}
1479: \,=\,
1480: F_{J_2(q)\,j_3(q,r)}
1481: \begin{bmatrix}
1482: J_1(p) & J_3(s) \\
1483: j_1(s,p) & j_2(r,s)
1484: \end{bmatrix}
1485: \end{equation}
1486:
1487: Relation \eqref{eq:OPE_fusion}, obtained first in \cite{Runkel:1998pm}
1488: for the $A$-series minimal models exploiting the fact that both the
1489: primary and boundary conditions labels fall in the same set, has been
1490: extended to all minimal models and extended rational conformal field
1491: theories in \cite{Behrend:1999bn} and \cite{Felder:1999ka} noticing the
1492: full analogy between the equation \eqref{eq:fact} and the pentagon
1493: identity for the fusing matrices.
1494:
1495: According to \cite{Alvarez-Gaume:1988vr}, WZW-models fusion matrices
1496: coincide with the $6j$-symbols of the corresponding quantum group with
1497: deformation parameter given by the $(k\,+\,h^\vee)$-th root of the
1498: identity, where $k$ and $h^\vee$ are respectively the level and the
1499: dual Coxeter number of the extended algebra. Thus, with $k=1$ and
1500: $h^\vee=2$, the
1501: OPEs coefficients are the $SU(2)_{Q\,=\,e^{\frac{2}{3}\pi i}}$
1502: $6j$-symbols:
1503: \begin{equation}
1504: \label{eq:OPE_6j}
1505: \mathcal{C}^{J_1(p)\,J_2(q)\,J_3(s)}_{j_1(s,p)\,j_2(r,s)\,j_3(q,r)}
1506: \,=\,
1507: \begin{Bmatrix}
1508: j_1(s,p) & J_1(p) & J_2(q) \\
1509: J_3(s) & j_2(r,s) & j_3(q,r)
1510: \end{Bmatrix}_{Q\,=\,e^{\frac{2}{3}\pi i}}
1511: \end{equation}
1512:
1513: \section{Open string amplitude on a RRT}
1514:
1515:
1516: With the computation of OPE's coefficients we have all the building
1517: blocks to construct an open string amplitude on the domain defined by
1518: the open Riemann surface $M_\partial$.
1519:
1520: As first step, let us extend results of the previous section to higher
1521: dimensional target spaces. To this end, let us consider D scalar
1522: fields $X^\alpha, \alpha=1,\,\ldots,\,D$ which, as stated in
1523: \eqref{winding}, wind $\nu^\alpha$ times around the homology cycles of
1524: the compact target space manifold. Let us indicate with
1525: $E_{\alpha\beta} = G_{\alpha\beta} + B_{\alpha\beta}$ the background
1526: matrix on the compact target space manifold, where $G$ is the metric
1527: and $B$ the Kalb-Ramond field. The central charge of the model is
1528: $c=D$.
1529:
1530: Let us find out which is the moduli space of inequivalent
1531: compactifications. This can be achieved by considering the torus
1532: partition function of the model: it factorizes into the product of
1533: contributions of each compact direction:
1534: \begin{equation}
1535: Z_{T_d} \,=\,
1536: \frac{1}{|\eta(\tau)|^2}
1537: \sum_{\Gamma_{D,D}}
1538: q^\frac{p_L^2}{4}
1539: \overline{q}^\frac{p_R^2}{4}
1540: \end{equation}
1541: Asking for modular invariance let the total momentum
1542: $\hat{p}=
1543: \left(
1544: \begin{smallmatrix}
1545: p_L \\ p_R
1546: \end{smallmatrix}
1547: \right)$
1548: take values
1549: into a self-dual, even-integer, Lorentzian lattice
1550: $\Gamma_{D,D}$\cite{Johnson:Dbranes}.
1551: The space of such inequivalent lattices is locally isomorphic to:
1552: \begin{equation}
1553: \label{lattice_modspace}
1554: \mathcal{M}
1555: \,=\,
1556: O(d,d,\mathbb{Z}) \backslash O(d,d) /[O(d) \times O(d)],
1557: \end{equation}
1558: which thus is the moduli space of inequivalent toroidal
1559: compactifications in a $D$ dimensional targer
1560: space\cite{Giveon:1994fu,Johnson:Dbranes}. The different orbits in
1561: this moduli space give rise to different theories in which the
1562: fundamental $U(1)_L \times U(1)_R$ current symmetry can be enhanced to
1563: different symmetry groups of rank at least $D$. This group plays the
1564: role of gauge group in the target space. Our choice is to compactify
1565: each direction at the self dual radius, because this allow to exploit
1566: the previous construction and define a coherent gluing of the
1567: conformal theory along the ribbon graph. This means to choice a
1568: specific orbit into \eqref{lattice_modspace}, \ie{} to fix
1569: definitively the target space modular structure. Moreover, the
1570: background gauge group coming from the closed string sector defined on
1571: the cylindrical ends is $[SU(2)_L \times SU(2)_R]^D$. On the
1572: contrary, we could choose not to fix the compactification radius as
1573: the self-dual one. This would give us more freedom in choosing the
1574: target space modular and metrical structure, and consequently the
1575: enhanced symmetry group. However, in doing so we can loose information
1576: about the dynamic of the theory on the ribbon graph, because, with the
1577: exception of some particular cases which we will introduce in the
1578: following chapter, combinatorial factor of BIOs would be defined only
1579: as a formal map.
1580:
1581: Thus, let us consider each direction compactified at the self dual
1582: radius.
1583: The amplitude on each \cyl{p} will
1584: receive a contribution from every direction:
1585: \begin{equation}
1586: \label{multi_ampl}
1587: \mathcal{A}_\text{\cyl{p}}
1588: \,=\,\frac{1}{
1589: 2^\frac{D}{2}
1590: \left[
1591: \eta\left(
1592: e^{-5\frac{4\pi}{\theta(p)}}
1593: \right)
1594: \right]^D}
1595: \prod_{\alpha=1}^D
1596: \sum_{j(p)=0,\frac{1}{2}}
1597: \cos{(8\pi j(p) \lambda^\alpha(p))}
1598: e^{-\frac{4\pi}{\theta(p)}{j(p)}^{2}}
1599: \end{equation}
1600:
1601: Moreover, Boundary Insertion Operators are primaries of the conformal theory,
1602: thus they also factorize into the contribution of each direction. This
1603: means that the full boundary theory factorizes into the contribution
1604: of each compact direction. In this connection, we can exploit a
1605: construction introduced in \cite{Carfora2}, which, exploiting a edge
1606: vertex factorization of the most general correlator we can write on
1607: the ribbon graph, allows to write the contribution to the amplitude
1608: from each compact direction as:
1609: \begin{multline}
1610: Z(|P_{T_{l}}|) \,=\,\\
1611: \left( \frac{1}{\sqrt{2}}\right) ^{N_{0}(T)}\sum_{\{j_{p}\in \frac{1}{2}
1612: \mathbb{Z}_{+}\}}\sum_{\{j_{(r,p)}\}}\prod_{\{\rho
1613: ^{0}(p,q,r)\}}^{N_{2}(T)}\left\{
1614: \begin{array}{ccc}
1615: j_{(r,p)} & j_{p} & j_{r} \\
1616: j_{q} & j_{(q,r)} & j_{(p,q)}
1617: \end{array}
1618: \right\} _{Q=e^{\frac{\pi }{3}i}}\times \\
1619: \times \prod_{\{\rho ^{1}(p,r)\}}^{N_{1}(T)}\left(
1620: b_{j_{(r,p)}}^{j_{p}j_{r}}\right) ^{2}L(p,r)^{-2H_{j_{(r,p)}}}
1621: \end{multline}
1622:
1623: Collecting the contribution of each direction and applying this
1624: results on the $N_(0)$ channels defined by \eqref{multi_ampl} we
1625: finally have:
1626: \begin{multline}
1627: Z(|P_{T_{l}}|, D) \,=\, \\
1628: \frac{1}{2^\frac{D N_0(T)}{2}}
1629: \prod_{\alpha=1}^D
1630: \left[\sum_{\{j_{p}\in \frac{1}{2}
1631: \mathbb{Z}_{+}\}}\sum_{\{j_{(r,p)}\}}\prod_{\{\rho
1632: ^{0}(p,q,r)\}}^{N_{2}(T)}\left\{
1633: \begin{array}{ccc}
1634: j_{(r,p)} & j_{p} & j_{r} \\
1635: j_{q} & j_{(q,r)} & j_{(p,q)}
1636: \end{array}
1637: \right\} _{Q=e^{\frac{\pi }{3}i}}\times \right.\\\left.
1638: \times \prod_{\{\rho ^{1}(p,r)\}}^{N_{1}(T)}\left(
1639: b_{j_{(r,p)}}^{j_{p}j_{r}}\right) ^{2}L(p,r)^{-2H_{j_{(r,p)}}}\cos
1640: (8\pi j_{p}\lambda (i))\frac{e^{-\frac{4\pi }{\theta
1641: (i)}j_{p}^{2}}}{\eta (e^{- 5\frac{4\pi }{\theta (i)}})}
1642: \right]_{(\alpha)}
1643: \end{multline}
1644: where the subscript $(\alpha)$ indicates the contribution of the
1645: $\alpha$-th direction.
1646:
1647: