Dimension of the space of unitary equivariant translation invariant tensor valuations (Q2217512)

Suppose that \({\mathbb V}, {\mathbb W}\) are finite-dimensional real vector spaces and \(Z\) is a translation-invariant, continuous valuation on the space \({\mathcal K}({\mathbb V})\) of convex bodies in \({\mathbb V}\) (equipped with the Hausdorff metric) with values in \({\mathbb W}\). If \(G\subset\mathrm{GL}({\mathbb V})\) is a closed subgroup acting on \({\mathbb W}\), the valuation \(Z\) is called \(G\)-equivariant if \(Z(\varphi(K))=\varphi Z(K)\) for all \(\varphi\in G\) and \(K\in {\mathcal K}({\mathbb V})\). In the present paper, \({\mathbb V}={\mathbb R}^{2m}\) (where \(m\ge 2\)), which can be identified with \({\mathbb C}^m\), further \(G=\mathrm{U}(m)\) is the unitary group, and \({\mathbb W}\) is the space \({\mathbb S}^d({\mathbb R}^{2m})\) of symmetric tensors of rank \(d\) on \({\mathbb V}\). With these choices, the paper determines the (finite) dimensions of the spaces of \(k\)-homogeneous, translation-invariant, continuous, \(\mathrm{U}(m)\)-equivariant valuations. The dimensions are expressed explicitly in terms of \(m,d,k\). For \(d=0\), the result is due to \textit{S. Alesker} [J. Differ. Geom. 63, No. 1, 63--95 (2003; Zbl 1073.52004)], and for \(d=1\) to \textit{T. Wannerer} [J. Differ. Geom. 96, No. 1, 141--182 (2014; Zbl 1296.53149)]. The proofs make heavy use of representation theory, and also Wannerer's [loc. cit.] theory of unitarily invariant area measures proves useful.