1: \section{Bisch-Jones' Planar Algebras}
2:
3: In this section, we shall apply our results to
4: identify the planar algebras of Bisch and Jones.
5:
6: Let $S = \{ s_n; n \geq 0 \}$ be the fusion rule set of
7: (the Tannaka dual of) $SU(2)$:
8: \[
9: s_ms_n = \sum_{k=0}^{\min\{m,n\}} s_{|m-n| + 2k},
10: \qquad
11: s_n^* = s_n.
12: \]
13:
14: According to Kazhdan and Wenzl (\cite{KW}),
15: rigid semisimple tensor categories (over the field $\C$)
16: having the fusion rule set $S$ is classified in the following way:
17: Let $X$ be an object representing $s_1 \in S$.
18: Then, by the fusion rule, we have
19: \[
20: X\otimes X \cong I \oplus X'
21: \]
22: with $X'$ representing $s_2$. Let $e \in \End(X\otimes X)$ be the
23: idempotent to the $I$-component and set $e_1 = e\otimes 1_X$,
24: $e_2 = 1_X\otimes e$. Then we have
25: \[
26: e_1e_2e_1 = (q + q^{-1} + 2)^{-1} e_1,
27: \quad
28: e_2e_1e_2 = (q + q^{-1} + 2)^{-1} e_2,
29: \]
30: with $q \in \C$ satisfying $1 + q + q^2 + \dots + q^N \not= 0$ for
31: any integer $N \geq 1$.
32:
33: Let $g = q(1-e) - e \in \End(X\otimes X)$ and set
34: \[
35: g_i = 1_{X^{i-1}}\otimes g\otimes 1 = q - (1+q)e_i,
36: \quad
37: i = 1, 2, \dots.
38: \]
39: From the above relations for $e_1$ and $e_2$,
40: the family $\{ g_i \}_{i \geq i}$ is the generator of the Hecke algebra:
41: \begin{gather*}
42: g_ig_{i+1}g_i = g_{i+1}g_ig_{i+1}, \qquad i \geq 1,\\
43: g_ig_j = g_jg_i, \qquad |i-j| \geq 2,\\
44: (g_i - q)(g_i + 1) = 0, \qquad i \geq 1.
45: \end{gather*}
46:
47: Let $T = g_2g_1 \in \End(X\otimes X\otimes X)$. Then we have
48: \[
49: T(1_X\otimes \d) = r (\d\otimes 1_X)
50: \qquad
51: \text{for $\d \in \Hom(I,X\otimes X)$,}
52: \]
53: where $r \in \C$ satisfies $r^2 = q^3$
54: (it is claimed in \cite[Proposition~5.1]{KW} that $r^2 = 1$ with $N =
55: 2$ but this should be corrected to $r^N = q^{N(N+1)/2}$).
56: Note that, letting $t = r/q$, we have $q = t^2$, $r = t^3$ and,
57: if $q$ is replaced with $q^{-1}$, then $r$ and therefore $t$ are
58: changed into their inverses.
59:
60: The tensor category is now completely classified by the complex number
61: $t$ (or the pair $(q,r)$):
62: Given a complex number $t$ satisfying
63: $1 + t^2 + t^4 + \dots + t^{2n} \not= 0$ for $n = 1, 2, \dots$,
64: there exists a unique rigid semisimple tensor category $\cC(t)$ with
65: the fusion rule governed by $S$.
66: Two tensor categories $\cC(t)$ and $\cC(t')$ are isomorphic
67: as monoidal categories if and only if either $t = t'$ or
68: $t^{-1} = t'$.
69:
70: These are in fact the Tannaka dual of the quantum group $SL_t(2,\C)$:
71: The tensor category
72: $\cC(t)$ is isomorphic to the one generated by the fundamental
73: representation $V$ of the universal enveloping algebra $U_t$.
74: Recall the Hopf algebra $U_t$ is defined by the following relations on
75: generators $\{ K, E, F \}$ (see \cite{Kas} for example):
76: \begin{gather*}
77: KEK^{-1} = t^2 E, \quad
78: KFK^{-1} = t^{-2} F, \quad
79: [E,F] = \frac{K - K^{-1}}{t - t^{-1}},\\
80: \D(K) = K\otimes K, \quad
81: \D(E) = E\otimes K + 1\otimes E, \quad
82: \D(F) = F\otimes 1 + K^{-1}\otimes F.
83: \end{gather*}
84:
85: Note here that the relation
86: \[
87: \t(K) = K^{-1}, \quad
88: \t(E) = -t^{-1}EK^{-1}, \quad
89: \t(F) = -tKF
90: \]
91: defines an antimultiplicative and anticomultiplicative involution of $U_t$.
92: % and the antipode is given by
93: %$\text{Ad}\,K^{1/2}\circ \t = \t\circ \text{Ad}\,K^{1/2}$.
94:
95: If we define the fundamental representation of $U_t$
96: by $V = \C v_1 + \C v_2$ with
97: \[
98: K =
99: \begin{pmatrix}
100: t & 0\\ 0 & t^{-1}
101: \end{pmatrix},
102: \quad
103: E =
104: \begin{pmatrix}
105: 0 & 1\\ 0 & 0
106: \end{pmatrix},
107: \quad
108: F =
109: \begin{pmatrix}
110: 0 & 0\\ 1 & 0
111: \end{pmatrix},
112: \]
113: then we can easily check that
114: the tensor category $\cR$ generated by $V$ has the fusion rule $S$
115: with the invariant $(q,r) = (t^2,t^3)$, whence it is isomorphic to $\cC(t)$.
116:
117: With this model in hand, it is immediate to see that the tensor
118: category $\cC(t)$ admits a Frobenius duality (\cite{FRTC}):
119: We first extend the Hopf algebra $U_t$ to $\widetilde U_t$
120: by adding a formal square root $K^{1/2}$ of $K$.
121: By choosing a square root $t^{1/2}$ of $t$,
122: we can extend $U_t$-modules in $\cR$ to $\widetilde U_t$-modules
123: without modifying the monoidal structure.
124:
125: Given a $\widetilde U_t$-module $V$, the dual vector space $V^*$
126: is again a $\widetilde U_t$-module by
127: \[
128: \langle xv^*, v \rangle = \langle v^*, \t(x)v \rangle
129: \]
130: and, given a $\widetilde U_t$-linear map $V \to W$,
131: the usual transposed map ${}^tf$ is also $\widetilde U_t$-linear.
132: Moreover, the obvious identification of the second dual $V^{**}$ with
133: $V$ is $\widetilde U_t$-linear as $\t$ being involutive.
134:
135: Now, together with the involution explained so far, it is easy to see
136: that
137: the $\widetilde U_t$-linear map $\e_V: V\otimes V^* \to \C$ defined by
138: \[
139: \e_V(v\otimes v^*) = \langle K^{1/2}v, v^* \rangle
140: \]
141: for a $\widetilde U_t$-module $V$ gives a Frobenius duality in
142: $\cR \cong \cC(t)$.
143:
144: The quantum dimension of the fundamental representation $V$ is
145: computed by
146: \[
147: \e_V{}^t\e_V = \trace_V(K) = t + t^{-1}.
148: \]
149:
150: \begin{Remark}
151: Given an integer $l \geq 3$, a similar analysis works without
152: much difficulties for tensor categories of the truncated fusion rule
153: of level $l$.
154: \end{Remark}
155:
156: Given a finite family $\{ \cC(t_j) \}_{1 \leq j \leq m}$ of
157: tensor categories of type $A$
158: with their spectrum sets $\{ S_j \}_{1 \leq j \leq m}$,
159: consider its free product
160: $\cC = \cC(t_1)* \dots *\cC(t_m)$. The spectrum $S$ of $\cC$ is
161: then the free product $S = S_1*\dots *S_m$.
162:
163: Letting $X_j$ be the fundamental generator of $\cC(t_j)$
164: with the associated class $x_j \in S_j$,
165: set $X = X_1\otimes \dots\otimes X_m$ in $\cC$. Then we have
166: the ascending sequence of semisimple algebras
167: \[
168: \End(X) \subset \End(X\otimes X^*) \subset
169: \End(X\otimes X^*\otimes X) \subset \dots.
170: \]
171:
172: We shall naturally identify the inductive limit algebra
173: with the Bisch-Jones' planar algebra
174: \[
175: FC(t_1 + t_1^{-1}, \dots, t_m + t_m^{-1})
176: = \bigcup_{n \geq 1}
177: FC_n(t_1 + t_1^{-1}, \dots, t_m + t_m^{-1}),
178: \]
179: where the coloring is specified by taking $\{ x_1, \dots, x_m \}$
180: as the color set.
181:
182: Given a periodic coloring
183: \[
184: x_1x_2\dots x_mx_m^* \dots x_1^*x_1\dots x_m \dots
185: \]
186: (although $x_j = x_j^*$, we dare to indicate $*$ depending on
187: the parity of times of occurrence),
188: we denote by $w_n$ its subword of the first $n$-colors for $n \geq 1$;
189: $w_1 = x_1$, $w_2 = x_1x_2$, $w_{m+1} = x_1\dots x_m x_m^*$ and so on.
190:
191: According to \cite{BJ1}, we introduce the algebra $A_n$ of planar
192: string diagrams with both of the top and the bottom vertices colored by
193: $w_n$ and the evaluation parameters given by
194: \[
195: \begin{minipage}{1.5cm}
196: \vspace{3mm}
197: \input lvalue.tpc
198:
199: \vspace{3mm}
200: \end{minipage}
201: \quad
202: =
203: \quad
204: t_j + t_j^{-1}
205: \quad
206: =
207: \quad
208: \begin{minipage}{1.5cm}
209: \vspace{3mm}
210: \input rvalue.tpc
211:
212: \vspace{3mm}
213: \end{minipage}
214: \]
215:
216: By definition, we have
217: $A_{mn} = FC_n(t_1 + t_1^{-1}, \dots, t_m + t_m^{-1})$.
218: In this sense, the algebras $A_n$ interpolate the Bisch-Jones planar
219: algebras $FC_n$. As seen in \cite[\S~3.2]{BJ1},
220: the dimension of $A_n$
221: (which is equal to the number of planar diagrams in $A_n$) is given by
222: \[
223: \frac{l+1}{k(m+1) + l + 1} \binom{k(m+1)+l+1}{k}
224: \]
225: for $n = km + l$ with $0 \leq l < m$.
226:
227: By counting the parity of the number of end-points of strings, we have
228:
229: \begin{Lemma}
230: In the planar algebra $A_n$, planar pairings among the source
231: vertices (or the target vertices) occur only by coupling $x_j$ with $x_j^*$
232: (pairings of $x_j$ with $x_j$ or $x_j^*$ with $x_j^*$ being
233: prohibited) while through strings connect $x_j$ with $x_j$ or
234: $x_j^*$ with $x_j^*$.
235: \end{Lemma}
236:
237: Let $W_n$ be the associated object of the word $w_n$ in
238: $\cC = \cC_1*\dots *\cC_m$:
239: $W_1 = X_1$, $W_2 = X_1\otimes X_2$,
240: $W_{m+1} = X_1\otimes X_m\otimes X_m^*$ and son on.
241:
242: By the obvious imbedding
243: $\End(W_n) = \End(W_n)\otimes 1 \subset \End(W_{n+1})$,
244: we have the ascending sequence of semisimple algebras
245: \[
246: \End(W_m) \subset \End(W_{m+1}) \subset
247: \End(W_{m+2}) \subset \dots.
248: \]
249:
250: By the above lemma and hook identities for Frobenius duality,
251: we can introduce an algebra homomorphism
252: \[
253: \phi:
254: \bigcup_{n \geq 1} A_n
255: \to
256: \bigcup_{n \geq 1} \End(W_n)
257: \]
258: %\lim_{\longrightarrow}
259: %\Bigl(
260: %\End(X) \to \End(XX^*) \to \End(XX^*X)
261: %\to \dots
262: %\Bigr)
263: %\]
264: so that
265:
266: \begin{figure}[htbp]
267: \vspace*{-0.2cm}
268: \hspace*{0.5cm}
269: \input{planar.tpc}
270: %\vspace*{0.5cm}
271: %\caption[]{}
272: %\label{squeezed}
273: \end{figure}
274:
275: \noindent(diagrams stream up to down).
276:
277: Given a simple object $s \in S = S_1*\dots *S_m$,
278: the vector space $\Hom(s,W_n)$ with the obvious action of the algebra
279: $\End(W_n)$ is a simple $\End(W_n)$-module whenever it is non-trivial.
280:
281: In what follows, we shall use the letter $y$ to indicate one of $x_j$
282: ($1 \leq j \leq m$) and define simple objects $\{ y_k \}_{k \geq 0}$
283: inductively so that $y_0 = 1$, $y_1 = y$ and
284: $y_ky = yy_k = y_{k-1} + y_{k+1}$.
285:
286: We now introduce a one-to-one correspondance $s \leftrightarrow \s$
287: between elements in $S$ and words of the letter $\{ x_1, \dots, x_m
288: \}$, which is defined inductively so that (i) $s = \s$ if $|s| = 1$
289: and (ii) $\s = \s' y^k$ if $s = s'y_k$ with $s \| y_k$ and
290: $s' \leftrightarrow \s'$,
291: where $|s|$ denotes the length of $s \in S$.
292:
293: Given a word $\s$ and an integer $n \geq 1$, let $V_{\s}^{(n)}$
294: be the $A_n$-module of the middle pattern $\s$ defined by
295: \cite[Definition~3.1.16]{BJ1}. Recall that $V_{\s}^{(n)}$ is the free
296: vector space over the set of colored planar diagrams from $\s$ to
297: $w_n$ with no self-couplings among vertices in $\s$ and the action of
298: $A_n$ is defined by the obvious composition of diagrams with the
299: convention that, if composed diagrams get out of $V_\s^{(n)}$,
300: then it is set to be zero.
301:
302: We now introduce a linear map
303: $\Phi: V_\s^{(n)} \to \Hom(s,W_n)$ in the following way:
304: By replacing self-coupling inside $w_n$ by ${}^t\e$'s, each planar
305: diagram in $V_\s^{(n)}$ associates a morphism in
306: $\Hom(\otimes \s, W_n)$, where $\otimes \s$ denotes the tensor product
307: of objects appearing in $\s$ with $*$ placed according to
308: the parity of times of occurrence (for example,
309: $\otimes \s = X_1\otimes X_1^*\otimes X_1\otimes X_3\otimes X_3^*$ for
310: $\s = x_1x_1x_1x_3x_3$).
311: Now, for each $k \geq 1$ and each $y$ in $\{ x_1, \dots, x_m \}$,
312: choose a non-trivial morphism
313: \[
314: y_k \to y^{\otimes k} = y\otimes y^*\otimes y\otimes \dots
315: \]
316: once for all (which is determined up to scalar multiplication) and
317: form the tensor product for factors appearing in $\otimes \s$,
318: resulting a morphism $s \to \otimes\s$. By taking the composition with
319: this, we obtain the linear map
320: \[
321: \Phi:
322: \begin{CD}
323: V_\s^{(n)} @>>> \Hom(\otimes \s, W_n) @>>> \Hom(s,W_n).
324: \end{CD}
325: \]
326:
327: \begin{Lemma}
328: For $a \in A_n$, we have the commutative diagram
329: \[
330: \begin{CD}
331: V_\s^{(n)} @>>> \Hom(s,W_n)\\
332: @V{\s}VV @VV{\phi(a)}V\\
333: V_\s^{(n)} @>>> \Hom(s,W_n)
334: \end{CD}\ .
335: \]
336: \end{Lemma}
337:
338: \begin{proof}
339: If $a$ and $v$ are planar diagrams in $A_n$ and $V_\s^{(n)}$
340: respectively with the composed diagram $av$ out of $V_\s^{(n)}$,
341: then the image of $av$ in $\Hom(\otimes \s, W_n)$ is factored into the
342: form $f\circ g$, where $g \in \Hom(\otimes \s, \otimes \s')$ and
343: $f \in \Hom(\otimes \s', W_n)$ with $\s'$ a strict subsequence of
344: $\s$. Then, by an easy induction,
345: we see that $\Hom(s,\otimes \s') = 0$
346: (the highest spin part $y_k$ in $y^{\otimes k}$ does not appear in
347: $y^{\otimes j}$ for $j < k$), whence $av$ is in the kernel of
348: the linear map $V_\s^{(n)} \to \Hom(s,W_n)$.
349: \end{proof}
350:
351: \begin{Lemma}
352: The linear map $\Phi: V_\s^{(n)} \to \Hom(s,W_n)$ is an isomorphism
353: for any $\s$ and any $n \geq 1$.
354:
355: The representation of $A_n$ on $\Hom(s,W_n)$ via $\phi$ is irreducible and
356: inequivalent for different $s$ (as long as $\Hom(s,W_n) \not = 0$).
357: \end{Lemma}
358:
359: \begin{proof}
360: We prove the assertion by an induction on $n$.
361: Assume that the statement is true for $n$.
362: Consider the $\End(W_{n+1})$-module $\Hom(s_{n+1},W_{n+1})$ with
363: $s_{n+1} \in S$ and let us begin with looking at how it is decomposed into
364: simple components when restricted to $\End(W_n)$.
365: Let $w_{n+1} = w_ny$ with $y \in \{ x_1, \dots, x_m \}$ and write
366: $W_{n+1} = W_n\otimes y$. By the obvious isomorphism
367: \[
368: \Hom(s_{n+1},W_{n+1}) \cong
369: \bigoplus_{s \in S} \Hom(s,W_n)\otimes \Hom(s_{n+1},s\otimes y),
370: \]
371: together with the Frobenius reciprocity
372: \[
373: \Hom(s_{n+1},s\otimes y) \cong
374: \Hom(s_{n+1}\otimes y^*,s),
375: \]
376: we need to decompose $s_{n+1}y$: If $s_{n+1} \| y$, $s_{n+1}y \in S$
377: and $s$ is forced to be $s_{n+1}y$. Then we have the isomorphism
378: \[
379: \Hom(s_{n+1}y,W_n) \ni w \mapsto (w\otimes 1_y)(1\otimes {}^t\e_{y^*})
380: \in \Hom(s_{n+1},W_{n+1})
381: \]
382: and the $A_n$-module $\Hom(s_{n+1},W_{n+1})$ is equivalent to
383: the $A_n$-module $\Hom(s_{n+1}y,W_n)$, which is irreducible by
384: the induction hypothesis. Thus the $A_{n+1}$-module
385: $\Hom(s_{n+1},W_{n+1})$ is irreducible as well.
386:
387: Next we consider the case that $s_{n+1}|y$ with $s_{n+1} = sy_k$
388: ($k \geq 1$). From $s_{n+1}y = sy_{k-1} + sy_{k+1}$, we have
389: two simple components $\Hom(sy_{k-1},W_n)$ and $\Hom(sy_{k+1},W_n)$,
390: which are mapped into $\Hom(s_{n+1},W_{n+1})$ by
391: \[
392: \Hom(s_n,W_n) \ni w \mapsto
393: (w\otimes 1_y)(1_s\otimes f) \in
394: \Hom(s_{n+1},W_{n+1})
395: \]
396: with $s_n = sy_{k-1}$ or $s_n = sy_{k+1}$ and $0 \not= f$ in
397: $\Hom(y_k,y_{k-1}\otimes y)$ or $\Hom(y_k,y_{k+1}\otimes y)$
398: respectively.
399:
400: The $A_n$-module $\Hom(s_{n+1},W_{n+1})$ is then decomposed into
401: a direct sum of two inequivalent $A_n$-modules
402: $\Hom(sy_{k-1},W_n)$ and $\Hom(sy_{k+1},W_n)$ by the induction
403: hypothesis.
404:
405: To see the irreducibility of the $A_{n+1}$-module
406: $\Hom(s_{n+1},W_{n+1})$,
407: we choose a planar diagram $v$ in $V_{\s y^{k+1}}$ and let
408: $w$ be its image in $\Hom(sy_{k+1},W_n)$, which corresponds to
409: \[
410: \widetilde w = (w\otimes 1_y)(1_s\otimes f)
411: \quad\text{with $f: y_k \to y_{k+1}\otimes y$}
412: \]
413: in $\Hom(s_{n+1},W_{n+1})$.
414: Let $\widetilde v \in V_{\s y^k}^{(n+1)}$ be the diagram
415: obtained by taking the contraction of
416: $v\otimes 1_y \in V_{\s y^{k+2}}^{(n+1)}$ for two through strings
417: colored by $y$ (and $y^*$) at the right end and set
418: $a = \widetilde v {\widetilde v}^* \in A_{n+1}$
419: (Fig.~\ref{contract}).
420: Then we have
421: $\phi(a)\widetilde w = \left( \prod \text{\rm O} \right)
422: \Phi(\widetilde v)\circ h$
423: with $h = (1_{y_k}\otimes \e_{y^*})(g\otimes 1_y)f$ and
424: $g: y_{k+1} \to y_k\otimes y^*$ (Fig.~\ref{lower}).
425: By Frobenius reciprocity, $h \in \End(y_k)$ has the expression
426: $(1_{y_k}\otimes \e_{y^*})(g\widetilde f\otimes 1_y)
427: (1_{y_k}\otimes {}^t\e_{y^*})$ and its quantum trace is calculated by
428: \[
429: \e_{y_k}(h\otimes 1) {}^t\e_{y_k}
430: = \e_{y_ky^*}(g\widetilde f\otimes 1) {}^t\e_{y_ky^*}
431: = \e_{y_{k+1}}(\widetilde f g\otimes 1) {}^t\e_{y_{k+1}}
432: = c \frac{t_j^{k+2} - t_j^{-k-2}}{t_j - t_j^{-1}}
433: \]
434: if $\widetilde f g = c 1_{y_{k+1}}$ and $y = x_j$, which does not vanish.
435: Thus $\phi(a)\widetilde w$ is a non-zero constant multiple of
436: $\Phi(\widetilde v)$, which belongs to the image of
437: $\Hom(sy_{k-1},W_n)$ in $\Hom(s_{n+1},W_{n+1})$;
438: $\Hom(sy_{k+1},W_n)$ and $\Hom(sy_{k-1},W_n)$ are mixed up by the action
439: $\phi(a)$ on $\Hom(s_{n+1},W_{n+1})$. This, together with the fact
440: $\phi(A_n)' \cong \C \oplus \C$, shows that $\phi(A_{n+1})' = \C$,
441: proving the irreducibility of the $A_{n+1}$-module
442: $\Hom(s_{n+1},W_{n+1})$.
443:
444: Since the image of $V_{\s y^k}^{(n+1)}$ in $\Hom(sy_k,W_{n+1})$ is clearly
445: non-trivial, the irreducibility particularly shows the surjectivity of $\Phi$.
446:
447: On the other hand, we know the exact sequence of $A_n$-modules
448: \[
449: \begin{CD}
450: 0 @>>> V_{\s y^{k-1}}^{(n)} @>>> V_{\s y^k}^{(n+1)}
451: @>>> V_{\s y^{k+1}}^{(n)} @>>> 0
452: \end{CD}
453: \]
454: (see the proof of \cite[Theorem~3.2.1]{BJ1}) and the induction hypothesis
455: implies
456: \begin{align*}
457: \dim V_{\s y^k}^{(n+1)}
458: &= \dim V_{\s y^{k-1}}^{(n)} + \dim V_{\s y^{k+1}}^{(n)}\\
459: &= \dim \Hom(sy_{k-1},W_n) + \dim \Hom(sy_{k+1},W_n)\\
460: &= \dim \Hom(sy_k,W_{n+1}),
461: \end{align*}
462: proving the bijectivity of the map
463: $\Phi: V_{\s y^k}^{(n+1)} \to \Hom(sy_k,W_{n+1})$.
464:
465: The inequivalence of $A_{n+1}$-modules
466: $\Hom(s_{n+1},W_{n+1})$ for different $s_{n+1}$
467: is now obvious because $s_{n+1} \in S$ can be recovered by
468: checking the irreducible decomposition of the restriction to
469: $A_n$.
470: \end{proof}
471:
472: \begin{figure}[htbp]
473: \vspace*{-0.2cm}
474: \hspace*{0.5cm}
475: \input{contract.tpc}
476: \vspace*{0.5cm}
477: \caption[]{}
478: \label{contract}
479: \end{figure}
480:
481:
482: \begin{figure}[htbp]
483: \vspace*{-0.2cm}
484: \hspace*{0.5cm}
485: \input{broom.tpc}
486: \vspace*{0.5cm}
487: \caption[]{}
488: \label{broom}
489: \end{figure}
490:
491:
492: \begin{figure}[htbp]
493: \vspace*{-0.2cm}
494: \hspace*{0.5cm}
495: \input{broom2.tpc}
496: \vspace*{0.5cm}
497: \caption[]{}
498: \label{broom2}
499: \end{figure}
500:
501:
502: \begin{figure}[htbp]
503: \vspace*{-0.2cm}
504: \hspace*{0.5cm}
505: \input{lower.tpc}
506: \vspace*{0.5cm}
507: \caption[]{}
508: \label{lower}
509: \end{figure}
510:
511: The following should be compared with \cite[Theorem~6.1.2]{BJ1}.
512:
513: \begin{Proposition}
514: The map $\phi$ is bijective: The Bisch-Jones' planar algebra
515: is identified with the inductive limit of the ascending sequence
516: of algebras associated to the simple object
517: $x = x_1x_2\dots x_n$ in the free product tensor category $\cC$.
518: \end{Proposition}
519:
520: \begin{proof}
521: By the previous lemma, we see that $\phi: A_n \to \End(W_n)$ is surjective
522: because $\End(W_n)$ is semisimple and inequivalent irreducible representations of
523: $\End(W_n)$ gives rise to inequivalent irreducible representations of $A_n$.
524: (Use the double commutant theorem for the semisimple $A_n$-module
525: $\bigoplus_{s \in S} \Hom(s,W_n)$.)
526:
527: From the analysis in \cite{BJ1}, we know that
528: \begin{align*}
529: \dim A_n &= \sharp\{ \text{planar diagrams in $A_n$} \}\\
530: &= \text{the dimension of $\End(W_n)$ calculated by the fusion rule}\\
531: &= \dim \End(W_n),
532: \end{align*}
533: which is checked for generic evaluation parameters but the formula itself
534: clearly holds without restrictions.
535:
536: Since $\phi_n$ is surjective, the equality of dimensions shows that it is
537: in fact bijective.
538: \end{proof}
539:
540: \begin{Corollary}
541: The Bisch-Jones' planar algebra $FC_n(a_1,\dots,a_m)$ is semisimple
542: for all $n \geq 1$ if and only if none of $a_1, \dots, a_m$ belongs to the set
543: \[
544: \{ 2\cos(\pi r); r \in \Q \setminus \Z \}.
545: \]
546: \end{Corollary}
547:
548: \begin{proof}
549: With the choice $a = t + t^{-1}$,
550: the condition $1 + t^2 + t^4 + \dots + t^{2n} \not= 0$
551: for any $n \geq 1$ is equivalent to require that $a$ does not
552: belong to the set specified above.
553:
554: Conversely assume that the Fuss-Catalan algebra $FC_n$ is
555: semisimple for any $n \geq 1$ and look at the coloring $x_j$
556: ($1 \leq j \leq m$).
557: In the planar algebra $A_{mn} = FC_n$ with $n \geq 1$ an odd integer,
558: taking all the possible pairings of the form
559: $\e_{x_{j+1}\dots x_m}$ or $\e_{x_{j-1}\dots x_1}$
560: (multiplied by $a_{j+1}^{-1}\dots a_m^{-1}$ or
561: $a_1^{-1}\dots a_{j-1}^{-1}$ respectively), we obtain
562: an idempotent $e \in A_{mn}$ of middle pattern
563: $x_1\dots x_{j-1}x_j^nx_{j+1}\dots x_m$.
564:
565: By the choice, the reduced algebra $eA_{mn}e$ is isomorphic to
566: the Temperley-Lieb algebra, which is semisimple by our assumption.
567: It is well-known (\cite{W, GHJ}) that the Temperley-Lieb algebra
568: of $n$ strings (\cite{Kau})
569: is semisimple for all odd $n \geq 1$ (if and) only if the parameter
570: $a_j$ is out of the range of numbers in question.
571: \end{proof}
572:
573: The following should be compared with \cite[Theorem~3.3.4]{BJ1}
574: (the explicit formula for minimal projections is not needed here).
575:
576: \begin{Corollary}
577: If the quantum (non-normalized) trace is defined on $FC_n$ by
578: multiplying evaluation parameters of loops obtained after closing
579: diagrams, then the value $d_\s$ of the trace on a minimal projection
580: corresponding to the irreducible representation $V_\s^{(mn)}$
581: ($\s$ being a word of letters $x_1, \dots, x_m$) is inductively
582: calculated by the rule $d_\s = t_j + t_j^{-1}$ for $\s = x_j$ and
583: \[
584: d_\s = d_{\s'} \frac{t_j^{k+1} - t_j^{-k-1}}{t_j - t_j^{-1}}
585: \]
586: if $\s = \s'x_j^k$ with $x_j$ different from the last letter of
587: $\s'$.
588:
589: Moreover, we have the identity
590: \[
591: (t_1+t_1^{-1})^n \dots (t_m+t_m^{-1})^n
592: = \sum_\s d_\s \dim V_\s^{(mn)}.
593: \]
594: Note that $\dim V_\s^{(mn)}$ is equal to the number of possible
595: planar diagrams inside.
596: \end{Corollary}
597:
598:
599:
600: %%% Local Variables:
601: %%% mode: japanese-latex
602: %%% TeX-master: "ftc"
603: %%% End:
604: