Abstract
Given a finite, directed, connected graph \(\Gamma \) equipped with a weighting \(\mu \) on its edges, we provide a construction of a von Neumann algebra equipped with a faithful, normal, positive linear functional \((\mathcal {M}(\Gamma ,\mu ),\varphi )\). When the weighting \(\mu \) is instead on the vertices of \(\Gamma \), the first author showed the isomorphism class of \((\mathcal {M}(\Gamma ,\mu ),\varphi )\) depends only on the data \((\Gamma ,\mu )\) and is an interpolated free group factor equipped with a scaling of its unique trace (possibly direct sum copies of \(\mathbb {C}\)). Moreover, the free dimension of the interpolated free group factor is easily computed from \(\mu \). In this paper, we show for a weighting \(\mu \) on the edges of \(\Gamma \) that the isomorphism class of \((\mathcal {M}(\Gamma ,\mu ),\varphi )\) depends only on the data \((\Gamma ,\mu )\), and is either as in the vertex weighting case or is a free Araki–Woods factor equipped with a scaling of its free quasi-free state (possibly direct sum copies of \(\mathbb {C}\)). The latter occurs when the subgroup of \({\mathbb {R}}^+\) generated by \(\mu (e_1)\cdots \mu (e_n)\) for loops \(e_1\cdots e_n\) in \(\Gamma \) is non-trivial, and in this case the point spectrum of the free quasi-free state will be precisely this subgroup. As an application, we give the isomorphism type of some infinite index subfactors considered previously by Jones and Penneys.
Similar content being viewed by others
References
Connes, A.: Almost periodic states and factors of type \({\rm III}_{1}\). J. Funct. Anal. 16, 415–445 (1974)
Kenneth, J.: Dykema, Free products of hyperfinite von Neumann algebras and free dimension. Duke Math. J. 69(1), 97–119 (1993). https://doi.org/10.1215/S0012-7094-93-06905-0
Dykema, K.J.: Free products of finite-dimensional and other von Neumann algebras with respect to non-tracial states, Free probability theory (Waterloo, ON, 1995), Fields Inst. Commun., vol. 12, Amer. Math. Soc., Providence, RI, pp. 41–88 (1997)
Guionnet, A., Jones, V.F.R., Shlyakhtenko, D.: Random matrices, free probability, planar algebras and subfactors, Quanta of maths, Clay Math. Proc., vol. 11, Amer. Math. Soc., Providence, RI. arXiv:0712.2904v2, pp. 201–239 (2010)
Guionnet, A., Jones, V.F.R., Shlyakhtenko, D.: A semi-finite algebra associated to a subfactor planar algebra. J. Funct. Anal. 261(5), 1345–1360 (2011). https://doi.org/10.1016/j.jfa.2011.05.004. arXiv:0911.4728
Guionnet, A., Shlyakhtenko, D.: Free monotone transport. Invent. Math. 197(3), 613–661 (2014)
Hartglass, M.: Free product von Neumann algebras associated to graphs, and Guionnet, Jones, Shlyakhtenko subfactors in infinite depth. J. Funct. Anal. 265(12), 3305–3324 (2013). https://doi.org/10.1016/j.jfa.2013.09.011
Hartglass, M.: Free product \({\rm C}^*\)-algebras associated with graphs, free differentials, and laws of loops. Can. J. Math. 69(3), 548–578 (2017)
Hartglass, M., Nelson, B.: Free products of finite-dimensional and other von Neumann algebras in terms of free Araki-Woods factors. ArXiv e-prints arXiv:1810.01924 (2018)
Hartglass, M., Nelson, B.: Free transport for interpolated free group factors. J. Funct. Anal. 274(1), 222–251 (2018)
Houdayer, C.: On some free products of von Neumann algebras which are free Araki-Woods factors, Int. Math. Res. Not. IMRN, no. 23, Art. ID rnm098, 21 (2007)
Hartglass, M., Penneys, D.: \(C^*\)-algebras from planar algebras II: The Guionnet-Jones-Shlyakhtenko \(C^*\)-algebras. J. Funct. Anal. 267(10), 3859–3893 (2014)
Jones, C., Penneys, D.: Realizations of algebra objects and discrete subfactors. Adv. Math. 350, 588–661 (2019)
Nelson, B.: On finite free Fisher information for eigenvectors of a modular operator. J. Funct. Anal. 273(7), 2292–2352 (2017)
Popa, S.: An axiomatization of the lattice of higher relative commutants of a subfactor. Invent. Math. 120(3), 427–445 (1995)
Shlyakhtenko, D.: Free quasi-free states. Pac. J. Math. 177(2), 329–368 (1997)
Shlyakhtenko, D.: \(A\)-valued semicircular systems. J. Funct. Anal. 166(1), 1–47 (1999). https://doi.org/10.1006/jfan.1999.3424
Acknowledgements
We would like to thank Corey Jones and David Penneys for initially suggesting the idea of this paper. We would like to thank Dimitri Shlyakhtenko for many helpful conversations about free Araki–Woods factors. We would also like to thank the anonymous reviewers who provided many helpful comments and corrections, including a substantially shorter proof of Lemma 3.1. This work was initiated at the Mathematical Sciences Research Institute (MSRI) Summer School on Subfactors: planar algebras, quantum symmetries, and random matrices, and continued while Brent Nelson was visiting the Institute for Pure and Applied Mathematics (IPAM) during the Long Program on Quantitative Linear Algebra, both of which are supported by the National Science Foundation. Brent Nelson’s work was also supported by NSF Grant DMS-1502822.
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by Y. Kawahigashi.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix
Appendix
1.1 Some infinite index subfactors
Let \(\mathcal {T}\mathcal {L}_{\bullet }\) denote the Temperley Lieb planar algebra with parameter \(\delta \in [2, \infty )\). In [GJS10], one considers \({{\,\mathrm{Gr}\,}}(\mathcal {T}\mathcal {L}) = \oplus _{n=1}^{\infty } \mathcal {T}\mathcal {L}_{n}\). Recall from [GJS10] that \({{\,\mathrm{Gr}\,}}(\mathcal {T}\mathcal {L})\) has the structure of a graded algebra with graded multiplication \(\wedge \) and normalized Voiculescu trace \(\tau \):
Moreover, this graded algebra acts on itself by left and right multiplication, which is bounded with respect to the \(\Vert \cdot \Vert _2\)-norm induced by this trace. Let \(\mathcal {N}= ({{\,\mathrm{Gr}\,}}(\mathcal {T}\mathcal {L}), \tau )''\). It was shown in [Har13] that \(\mathcal {N}\cong L({\mathbb {F}}_{\infty })\).
Jones and Penneys in [JP19, Section 6.4] studied a directed graph \(\Gamma =(V, E)\) with edge-weighting \(\mu \) as in Section 2 with the additional requirement that \((\Gamma ,\mu )\) is balanced:
Fixing \(v \in V\), they studied the loop algebra \(\mathbf{A }_{v}\) which is spanned by formal linear combinations of loops based at v. It comes equipped with the following \(*\)-algebra structure and state \(\phi \):
-
\((e_{1}\cdots e_{n}) \cdot (f_{1}\cdots f_{m}) = e_{1}\cdots e_{n}f_{1}\cdots f_{m}\)
-
\((e_{1}\cdots e_{n})^{*} = \sqrt{\mu (e_{1})\cdots \mu (e_{n})}e_{n}\cdots e_{1}\)
-
With \(NC_{2}([n])\) the set of non-crossing pair partitions on \(\{1, \ldots , n\}\),
The following proposition is a standard combinatorial argument involving the elements \(Y_{e} \in (\mathcal {M}(\Gamma , \mu ), \varphi )\).
Proposition
The map \(p_{v}{\mathbb {C}}\langle Y_{e} : e \in E\rangle p_{v} \rightarrow \mathbf{A }_{v}\) given by \(Y_{e_{1}}\cdots Y_{e_{n}} \mapsto e_{1}\cdots e_{n}\) for every loop based at v is a \(*\)-algebra isomorphism statisfying \(\phi (e_{1}\cdots e_{n}) = \varphi (Y_{e_{1}}\cdots Y_{e_{n}})\).
This proposition allows us to conclude that \((\mathcal {M}, \phi ) := (\mathbf{A }_{v}, \phi )'' \cong (T_{H}, \varphi _{H})\) for \(H = H(\Gamma , \mu )\). (Note that \(\delta \ge 2\), and so requiring that \(\mu \) is balanced ensures \(\mathcal {M}(\Gamma ,\mu )\) is a factor.) Section 6.4 of [JP19] studied an inclusion of \(i : \mathcal {N}\hookrightarrow \mathcal {M}\) as follows. Identifying each diagram \(x\in {{\,\mathrm{Gr}\,}}(\mathcal {T}\mathcal {L})\) with a partition \(\pi \in NC_{2}([2n])\) in the natural way, one has
This inclusion preserves the tracial state \(\tau \) on \(\mathcal {N}\), and so \(\mathcal {N}\hookrightarrow \mathcal {M}^{\phi }\).
Jones and Penneys showed that the subfactor \(\mathcal {N}\subset \mathcal {M}\) is of infinite index, irreducible (\(\mathcal {N}' \cap \mathcal {M}= {\mathbb {C}}\)), discrete (\(L^{2}(\mathcal {M})\) decomposes as a direct sum of irreducible, finite index \(\mathcal {N}-\mathcal {N}\) bimodules), and that the subfactor \(\mathcal {N}\subset \mathcal {M}\) is determined by \((\Gamma , \mu )\). Our work shows that the inclusion \(\mathcal {N}\subset \mathcal {M}\) can be realized as an inclusion \(L({\mathbb {F}}_{\infty }) \subset (T_{H}, \varphi _{H}),\) where \(L({\mathbb {F}}_{\infty })\) lies inside \(T_{H}^{\varphi _{H}}\).
Rights and permissions
About this article
Cite this article
Hartglass, M., Nelson, B. Non-tracial Free Graph von Neumann Algebras. Commun. Math. Phys. 379, 1–40 (2020). https://doi.org/10.1007/s00220-020-03841-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00220-020-03841-x