Complement of forms (Q2377407)

Given two sesquilinear forms \(\mathfrak{t}\) and \(\mathfrak{w}\) on a complex linear space \(\mathfrak{D}\), the form \(\mathfrak{w_{t}}\) (if it exists) is called the \textit{complement} of \(\mathfrak{w}\) with respect to \(\mathfrak{t}\) if \(\mathfrak{w_{t}}[\varphi]=\sup_{g\in \mathfrak{D}}[\mathfrak{t}\, (\varphi, g)+\mathfrak{t}(g ,\varphi ) - \mathfrak{w}[g]] \text{ for all } \varphi \in \mathfrak{D}\). The authors consider the main properties of complements, in particular some properties which extend known operator versions (see \textit{K. Nishio} [Acta Sci. Math. 42, 143--154 (1980; Zbl 0461.47010)] and \textit{E. L. Pekarev} and \textit{Ju. L. Smul'jan} [Math. USSR, Izv. 10(1976), 351--370 (1977); translation from Izv. Akad. Nauk SSSR, Ser. Mat. 40, 366--378 (1976; Zbl 0351.47031)]). The notion of quasi-unit for non-negative sesquilinear forms is introduced and used for a description of the extreme points of the convex set \(\{\mathfrak{z}:0\leq \mathfrak{z}\leq \mathfrak{t}\}\) (cf.\ \textit{S. L. Eriksson} and \textit{H. Leutwiler} [Math. Ann. 274, 301--317 (1986; Zbl 0571.31005)]). The notions of parallel sum, parallel difference and complement of two non-negative sesquilinear forms that were introduced and studied by \textit{S. Hassi} et al. [Operator Theory: Advances and Applications 198, 211--227 (2010; Zbl 1200.47004); J. Funct. Anal. 257, No. 12, 3858--3894 (2009; Zbl 1185.47002)] are used systematically. Reviewer's remark. An error in Lemma 2(g) of this paper has been corrected in an erratum [ibid., No. 3, 941--943 (2013; Zbl 1273.47007)].