An optimal inequality and extremal classes of affine spheres in centroaffine geometry (Q2487753)

This paper is very similar to the authors foregoing paper [\textit{B.-Y. Chen}, Proc. Japan Acad., Ser. A 80, No. 7, 123--128 (2004; Zbl 1076.53010)]. There the author considers locally strongly convex graph hypersurfaces, now he considers locally strongly convex centroaffine hypersurfaces. The integrability conditions for a centroaffine hypersurface imply the associated theorema egregium (see for the general relative case, e.g., p. 79 [\textit{U. Simon, A. Schwenk-Schellschmidt, H. Viesel}, Introduction to the affine differential geometry of hypersurfaces. Lecture Notes Science University Tokyo (SUT) (1992; Zbl 0780.53002)]): \[ \text{trace}_h \text{Ric} = \|K\|^2 + n(n-1)\varepsilon - n^2 \|T\|^2 \] where the norm is taken w.r.t. the centroaffine metric \(h\). Ric denotes its Ricci tensor, \(K:= \nabla - \nabla(h)\) the difference tensor between induced connection \(\nabla\) and Levi-Civita connection \(\nabla(h)\) of \(h\), \(nT\) the trace of \(K\), the so called Chebyshev form, and \(\varepsilon = +1\) for hyperbolic or \(\varepsilon = -1\) for elliptic hypersurfaces. The author states the following inequality as theorem 4.1: A centroaffine hypersurface satisfies \[ \text{trace}_h \text{Ric} \geq n(n-1) \epsilon +\frac{n^2 (1-n)}{n+2} \|T\|^2. \] In theorem 5.1 (resp. theorem 6.1) he gives a classification of all elliptic (hyperbolic, resp.) hypersurfaces satisfying equality in theorem 4.1, using some explicit coordinate representations. Reviewer's remarks. 1. As already stated in the review of Chen (loc.cit.), one can simplify the proof of theorem 4.1, inserting the traceless part \(\widetilde {K}\) of \(K\) into the theorema egregium; see e.g. Lemma 2.1 in [\textit{A.-M. Li} et al., Geom. Dedicata 66, No. 2, 203--221 (1997; Zbl 0878.53010)]. Equality in theorem 4.1 holds if and only if \(\tilde {K}= 0\), that means if the hypersurface is a quadric (see, section 7.1 in U. Simon et al. (loc. cit.) and A.-M. Li et al. (loc. cit.)). 2. From the foregoing remark equality in Chen's inequality characterizes quadrics. That is not stated in the author's classification, and most hypersurfaces do not appear in a quadratic representation therein. Concerning this, the reviewer contacted the author; the author now could verify that all hypersurfaces in his classification are quadrics, and he could give coordinate representations in terms of quadratic polynomials.