On the uniqueness of the injective \(\mathrm{III}_{1}\) factor. With a preamble by Alain Connes. (Q321901)

As it was common for many great mathematicians before the online era, Uffe Haagerup had several unpublished notes/preprints of outstanding quality. With his untimely passing in 2015, several mathematicians close to him -- and with permission of his son Søren -- have taken to the task of publishing several of these wonderful pieces. I heard from him many years ago that he felt that his greatest achievement -- and there were many -- was the proof of the uniqueness of the injective type III\(_1\)-factor [Acta Math. 158, 95--148 (1987; Zbl 0628.46061)]. The uniqueness of all injective factors except the type III\(_1\) was achieved by the monumental work of \textit{A. Connes} [Ann. Math. (2) 104, 73--115 (1976; Zbl 0343.46042)]. Connes did work on the III\(_1\) case and was able to find an equivalent condition, the bicentralizer problem. For a \(\sigma\)-finite von Neumann algebra \(M\) and a faithful normal state \(\varphi\), the bicentralizer of \(\varphi\) is the set \[ B_\varphi=\{x\in M:\;\lim_n\|xa_n-a_nx\|_\varphi=0\;\text{ for all } (a_n)_{n-1}^\infty\;\text{ such that } \lim_n\|\varphi a_n-a_n\varphi\|=0\}. \] It was Uffe who was able to show that the bicentralizer of an injective type III\(_1\)-factor is trivial. As seen from the current paper, Uffe's work went beyond the proof he published in 1987, and here we get to see a bit more of his penetrating work. I will now defer to Alain Connes, from the preamble to the paper: ``Uffe Haagerup solved the hardest problem of the classification of factors, namely the uniqueness problem for injective factors of type III\(_1\). The present paper, taken from his unpublished notes, presents a direct proof of this uniqueness by showing that any injective factor of type III\(_1\) is an infinite tensor product of type I factors so that the uniqueness follows from the Araki-Woods classification. The proof is typical of Uffe's genius, the attack is direct, and combines his amazing control of completely positive maps and his sheer analytical power, together with his solution to the bicentralizer problem. After his tragic death, Hiroshi Ando volunteered to type the manuscript. Some pages were missing from the notes, but eventually Cyril Houdayer and Reiji Tomatsu suggested a missing proof of Lemma 3.4 and Theorem 3.1. We heartily thank Hiroshi, Cyril and Reiji for making the manuscript available to the community. We also thank Søren Haagerup for giving permission to publish his father's paper.'' The manuscript credits the typing to Hiroshi Ando (Chiba University) in cooperation with Cyril Houdayer (Université Paris-Sud), Toshihiko Masuda (Kyushu University), Reiji Tomatsu (Hokkaido University), Yoshimichi Ueda (Kyushu University) and Wojciech Szymanski (University of Southern Denmark).