Involutive operator algebras (Q2299365)

Let \(\mathcal{H}\) be a complex Hilbert space and \(\mathcal{A}\) be an operator algebra on \(\mathcal{H}\), that is, a closed subalgebra of \(\mathcal{B}(\mathcal{H})\). When \(\mathcal{A}\) is equipped with operator space norms inherited from \(\mathcal{B}(\mathcal{H})\) and a completely isometric algebra involution \(\dagger\), it is called an operator \(*\)-algebra. There are four types of involutions (bijections of period two) on an operator algebra, including the above one as the first type. In the paper under review, the authors investigate the structure of operator \(*\)-algebras and remark that their results are applicable to algebras with other types of involution, after appropriate modifications in the arguments. The paper begins with establishing some characterisations of operator algebras and operator \(*\)-algebras \(\mathcal{A}\), in terms of their biggest \(C^*\)-cover \(C^*_{\max}(\mathcal{A})\) and smallest \(C^*\)-cover \(C^*_e(\mathcal{A})\) of \(\mathcal{A}\). In the third section of the paper, a wide range of examples of operator \(*\)-algebras are presented, including some uniform algebras, operator algebras generated by a single element, and operator algebras obtained from operator systems by Paulsen's off diagonal technique. Then several characterisations of operator \(*\)-algebras with contractive approximate identities and, in particular, those with countable contractive approximate identities are given. Moreover, analogues of the Arens-Kadison theorem and Cohen's factorisation theorem for operator \(*\)-algebras are proved. The last section of the paper is devoted to hereditary subalgebras of operator \(*\)-algebras and their relations with Akemann's noncommutative topology. To be more precise, suppose that \(\mathcal{A}\) with the involution \(\dagger\) turns into an operator \(*\)-algebra. An orthogonal projection \(p\in\mathcal{A}^{**}\) is open in \(\mathcal{A}^{**}\) if there is a net \((x_t)\) in \(\mathcal{A}\) such that \[ x_t=px_t=x_tp\overset{w^*}\longrightarrow p. \] If, moreover, \(p=p^\dagger\), then we say that \(p\) is \(\dagger\)-open. In this case, the closed \(\dagger\)-subalgebra \(\mathcal{D}=p\mathcal{A}^{**}p\cap \mathcal{A}\) is called a \(\dagger\)-hereditary subalgebra of \(\mathcal{A}\) and \(p\) is called the support projection of \(\mathcal{D}\). The authors characterise \(\dagger\)-hereditary subalgebras in terms of one-sided ideals, analogous to a well-known characterisation of hereditary \(C^*\)-subalgebras. Besides, for every \(\dagger\)-hereditary subalgebra \(\mathcal{B}\) of \(\mathcal{A}\) a set \(E\) of real positive elements of \(\mathcal{A}\) is identified which generates \(\mathcal{B}\), that is, \(\mathcal{B}=\overline{E\mathcal{A} E}\). In particular separable \(\dagger\)-hereditary subalgebras or \(\dagger\)-hereditary subalgebras with countable contractive approximate identities are of the form \(\overline{x\mathcal{A} x}\) for certain \(x\in\mathcal{A}\). A projection \(q\in\mathcal{A}^{**}\) is called \(\dagger\)-compact in \(\mathcal{A}^{**}\) if \(1-q\) is open in \(\mathcal{A}^{**}\) and there is an \(x\) in the unit ball of \(\mathcal{A}\) such that \(q=qx\). The paper is concluded with characterisations of compact projections and noncommutative analogues of Urysohn's lemma and Tietze's extension theorem.