• 検索結果がありません。

CENTRALIZING GROUP-LIKE OBJECTS IN TENSOR CATEGORIES AND THE INVARIANT $\chi$(Bimodules in Operator Algebras)

N/A
N/A
Protected

Academic year: 2021

シェア "CENTRALIZING GROUP-LIKE OBJECTS IN TENSOR CATEGORIES AND THE INVARIANT $\chi$(Bimodules in Operator Algebras)"

Copied!
4
0
0

読み込み中.... (全文を見る)

全文

(1)

CENTRALIZING GROUP-LIKE OBJECTS IN

TENSOR CATEGORIES AND THE INVARIANT $\chi$

YAMAGAMI SHIGERU

In this note, we shall present a tensor-categorical interpretation ofthe invariant

$\chi$for subfactors, which can be applied to compute theinvariant forgroup-subgroup

subfactors.

Automorphisms in Subfactors Given a subfactor $N\subset M$, set

$Aut(M, N)=\{\theta\in Aut(M);\theta(N)=N\}$,

Int$(M, N)=$

{

$Adu\in Aut(M,$ $N);u$ is a unitary in $N$

}.

Each $\theta\in AUt(M, N)$ is inductively extended to automorphisms of the Jones tower

$N\subset M\subset M_{1}\subset M_{2}\subset\cdots$ by

$\theta(e_{i})=e_{i}$, $i=1,2,3,$ $\cdots$

and hence induces

$Loi(\theta)=\mathrm{t}\mathrm{h}\mathrm{e}$ family of induced automorphisms on $N’\cap M\subset N’\cap M_{1}\subset\cdots$ .

Remark. We can use automorphisms on $M’\cap M_{1}\subset M’\cap M_{1}\subset\cdots$ as well, which

contains the equivalent information.

Theorem (Popa, Loi). Let $M,$ $N$ be $AFDII_{1}$

-factors

and $N\subset M$ be amenable.

Then

for

$\theta\in Aut(M, N)_{;}$

(i) $\theta$ is centrally trivial

iff

$\theta$ is inner at some

$M_{k;}i.e.,$ $\exists 0\neq u\in M_{k}$ such that $\theta(x)u=ux$

for

$x\in M$.

(ii) $Loi(\theta)=1$

iff

$\theta\in Int(M, N)$.

According to Y. Kawahigashi, we define the group

$\chi(M, N)=\frac{Cnt(M,N)\cap\overline{Int(M,N)}}{Int(M,N)}$

as the $\chi$-invariant for subfactors.

数理解析研究所講究録

(2)

Theorem (Kawahigashi). For

subfactors

of

index $<4_{\rangle}$

$\chi=\{$

$\mathbb{Z}_{2}$

for

$A_{2n+1}(n\geq 2)$ and $E_{6}$,

$\mathbb{Z}_{2}\oplus \mathbb{Z}_{2}$

for

$A_{3}$,

$\mathbb{Z}_{3}\oplus \mathbb{Z}_{3}$

for

$D_{4_{2}}$ $0$ otherwise.

Interpretations with bimodules

For a factor $N$, the correspondance

$\alpha\in Aut(N)\infty X_{\alpha}=NL^{2}(N)\alpha_{N}$

induces

$X_{\alpha}^{*}=x\alpha^{-}1$

$X_{\alpha}\otimes^{N}X\beta=x_{\alpha\beta}$.

Here the right $N$-action in $X_{\alpha}$ is modified by $\alpha$ compared to the standard bimodule

$L^{2}(N)$.

The bimodule $X_{\alpha}$ satisfies $X_{\alpha}\otimes X_{\alpha}^{*}\cong NL^{2}(N)_{N}$. The converse is not always

true: Let $R$be an AFD$\mathrm{I}\mathrm{I}_{1}$-factor,

$e$ bea non-trivialprojection in$R$and$v$ : $Rarrow eRe$

be an isomorphism. Then the bimodule $X=ReL^{2}(.R)_{R}$

. gives an example, where

the right action is induced from the isomorphism $\varphi$.

Theorem. Let $X$ be an N-N $b_{i}module$ such that $X\otimes X^{*}\cong L^{2}(N)$ and consider

one

of

the following cases. (i) $N$ is properly

infinite.

(ii) $X$ is a descendent

of

an irreducible bimodule $Z$

of finite

index. Then $\exists\alpha\in Aut(N)$ such that $X\cong X_{\alpha}$.

The bimodule $X_{\alpha}=NL^{2}(N)\alpha_{N}$ is simply denoted by $\alpha$ in the following. For a bimodule $AX_{B}$ and $\alpha\in Aut(A),$ $\beta\in Aut(B)$, set

$\alpha X\beta=\alpha\otimes X\otimes\beta$

and

$o_{ut}(A)\cross xOut(B)=\{([\alpha], [\beta])\in Out(A)\cross Out(B);\alpha x\cong x\beta\}$,

a subgroup ofOut$(A)\cross o_{ut}(B)$.

Theorem (Kosaki, Choda-Kosaki). For $Z=NL^{2}(M)_{M}$ with $N\subset M$ irreducible,

we have

(i) Out$(N)\cross zOut(M)\cong Aut(M, N)/Int(M, N)$.

(ii) $\theta$ is inner at some

$M_{k}$

iff

$ML^{2}(M)\theta_{M}$ appears in $z*z\cdots z*z(=(Z^{*}Z)^{k})$.

.$\cdot$

) (ii) Use $M_{3}=End(zZ^{*}z_{M})$ and the Frobenius reciprocity

$Hom(\theta z*ZZ*, z*zZ*)\cong Hom(L2(M)\theta, (Z^{*}z)^{3})$

for example. $\square$

(3)

.T

heorem (Goto).

.$Loi(\theta)=1$ implies $[\theta]=([\alpha], [\beta])\in Aut(M, N)/Int(M, N)$ is

$m$ the center

of fusion

$algebra\rangle$ $i.e.$,

$\alpha X\cong X\alpha$, $\beta Y\cong Y\beta$, $\alpha Z’\cong z’\beta$

for

descendants

$NX_{N},$ $MY_{M}$ and $NZ_{M}’$

of

$Z$.

Conversely the $central_{i}ty$ in the

fusion

algebra

forces

the triviality

of

$Loi$ invariant

as long as the principal graph (or the dualprincipal graph) is $multipli_{C}i.ty-free$.

Theorem. The following are equivalent. (i) $Loi(\theta)=1$.

(ii) $.F$or each bimodule $X$ in the tensor category generated by $Z$, we can

find

an

zsomorphism $I_{X}$ : $Xarrow\theta X\theta^{-1}$ such that $I_{X^{*}}=\overline{I_{X}},$ $I_{XY}=I_{X}\otimes I_{Y}$, and

the diagram

$Xarrow I_{X}\theta X\theta^{-1}$

$\tau\downarrow$ $\downarrow\theta T\theta^{-1}$

$Yarrow I_{Y}\theta Y\theta^{-1}$ commutes

for

$T\in Hom(x, Y)$

.

Applications to $\chi(G, H)$

For a subgroup $H\subset G$ of a finite group $G$ with an outer action on an AFD

$\mathrm{I}\mathrm{I}_{1}$-factor, set

$\chi(G, H)=\chi(R\lambda c, R\lambda H)$.

According to [KY], irreducible bimodules generated by $R\rangle\triangleleft HL^{2}(R\lambda G)_{R\rangle\triangleleft G}$ are

parametrized by

$R\lambda G-R\lambda G$ : $\hat{G}$ $R\lambda H- R\rangle\triangleleft G$: $\hat{H}$ $R\lambda H- R\lambda H$ :

$\dot{a}\in H\backslash G/\prod_{H}H\overline{\cap aH}a-1$.

Note that the tensor category of $R\lambda$ H-R $xH$ bimodules contains the

Tannaka dual of$H$ as a subcategory. With this description,

we can deduce

$Cnt(M, N)/Int(M, N)\cong\Xi\cross(N_{G}(H)/H)$,

where

$\cup--=\{(\chi, \eta)\in H^{*}\cross G^{*} ; \chi=\eta|_{H}\}$

and $H^{*}$ and $G^{*}$ refer to the group of

1-dimensional representations.

Taking the

restricti.on

of centralizing morphisms to the Tannaka dual of $H$, we

can deduce the $\mathrm{f}\mathrm{o}\mathrm{l}1_{\mathrm{o}\mathrm{W}}1\mathrm{n}\mathrm{g}$.

(4)

Theorem. We have

$\cup--\mathrm{x}Z(G)H/H\subset\chi(G, H)\subset\cup--\mathrm{x}$

{

$\dot{c}\in C_{G}(H)H/H;\dot{c}$ acts trivially on $H\backslash G/H$

},

where $C_{G}(H)$ denotes the centralizer

of

$H$ in $G$ and $\dot{a}\in N_{G}(H)/H$ acts on$H\backslash G/H$

$by$

$HgH-+Haga^{-}H1$.

Corollary.

(i) $\chi(G, \{e\})\cong c*\cross Z(G)$.

(ii) $\chi(A_{\lambda}H, H)\cong\{(\chi, \eta)\in A^{*}\cross(A\rangle\triangleleft H)^{*} ; \chi=\eta|_{H}\}\cross A^{H}f$ where $A$ is an abelian group and $A^{H}=$

{

$a\in A;hah^{-1}=a$,

for

all $h\in H$

}.

(iii) $\chi(S_{n’ k}S)\cong \mathbb{Z}_{2}$. (iv) $\chi(A_{n}, A_{k})=\{e\}$.

REFERENCES

[CK] M. Choda and H. Kosaki, Strongly outer actions for an inclusion offactors, J. Funct.

Anal. 122 (1994), 315-332.

[Go] S. Goto, Commutatimty of automorphisms ofsubfactors modulo inner automorphisms, Proc. Amer. Math. Soc., to appear.

[Ka] Y. Kawahigashi, Centrally trivial automorphisms and an analogue ofConnes’$\chi(M)$ for

subfactors, Duke Math. J. 71 (1993), 93-118.

[Ko] H. Kosaki, Automorphisms in the irreducible decompositions of sectors, Quantum and

non-commutative analysis (H. Araki et al., eds.), Kluwer Academic, 1993, pp. 305-316.

[GSTC] S. Yamagami, Group symmetry in tensor categories, preprint (1995).

参照

関連したドキュメント

For a countable family {T n } ∞ n1 of strictly pseudo-contractions, a strong convergence of viscosity iteration is shown in order to find a common fixed point of { T n } ∞ n1 in

Definition An embeddable tiled surface is a tiled surface which is actually achieved as the graph of singular leaves of some embedded orientable surface with closed braid

Tensor distributions, algebras of generalized functions, gene- ralized tensor fields, Schwartz impossibility result, diffeomorphism invariant Colombeau algebras, calculus in

In recent work [23], authors proved local-in-time existence and uniqueness of strong solutions in H s for real s &gt; n/2 + 1 for the ideal Boussinesq equations in R n , n = 2, 3

Lemma 1.11 Let G be a finitely generated group with finitely generated sub- groups H and K , a non-trivial H –almost invariant subset X and a non-trivial K –almost invariant subset

If a non-saturated subset in the set of weights of the kth fundamental representation of SL(n) is found, then the analogous non-saturated subset exists in the set of weights of the

We relate group-theoretic constructions (´ etale-like objects) and Frobenioid-theoretic constructions (Frobenius-like objects) by transforming them into mono-theta environments (and

By using either Proposition 3.2 or Theorem 3.1, one can see easily that (g, %, r) is a Riemann-Poisson Lie algebra and hence (G, h , i, π) is a Riemann-Poisson Lie group where π is