An orthogonality in normed linear spaces based on angular distance inequality (Q284688)

Let \(X:=(X,\|\cdot\|)\) be a normed space. The angular distance between two nonzero vectors \(x,y\) in \(X\) is defined by \(\alpha[x,y]:=\|x'-y'\|\) where \(x':=x/\|x\|\) and \(y':=y/\|y\|\). In this paper, the authors present a new notion of orthogonality in a normed space based on an inequality concerning the angular distance. For two vectors \(x,y\in X\), say that \(x\) is orthogonal to \(y\) and denote it by \(x\perp_+ y\) if \(\|x\|\|y\|=0\) or the following two statements hold: (a) \(\{x,y\}\) is linearly independent; (b) \(\alpha[x+ty]+\alpha[x+ty,-y]\leq\alpha[x,y]+\alpha[x,-y]\) for all real numbers \(t\). It is proved that this notion is equivalent to the standard definition of orthogonality when the underlying space is an inner product space. Moreover, the authors also prove that orthogonality in the sense of Singer can be expressed in terms of the angular distance inequality. Finally, a characterization of inner product spaces in terms of the additivity of the orthogonalities in the sense of Singer [\textit{P.-K. Lin}, Math. Nachr. 160, 325--328 (1993; Zbl 0919.46011)] and of Alsina-Shikorska-Tomàs [\textit{C. Alsina} et al., Norm derivatives and characterizations of inner product spaces. Hackensack, NJ: World Scientific (2010; Zbl 1196.46001)] is given.