The cone of nonnegative \(c\)-numerical range and its preservers (Q5929758)

Let \(M_n\) be the vector space of all \(n\times n\) complex matrices and let \(c = (c_1,\dots,c_n)\in {\mathbb R}^n\). For each \(A\in M_n\), the \(c\)-numerical range of \(A\) is defined as \(W_c(A)=\big \{ \sum^n_{i=1}c_i\langle Ax_i,x_i\rangle\): \(\{x_1,\dots,x_n\}\) is an orthonormal set in \({\mathbb C}^n\big\}\). Let \({\mathcal P}_n\) be the cone of all matrices \(A\) with non-negative \(W_c(A)\), i.e., \({\mathcal P}_n=\{A\in M_n |W_c(A)\subset [0,\infty)\}\). The authors study extremal properties of \({\mathcal P}_n\) and give a characterization of linear mappings \(T:M_n\to M_n\) and \(T:H_n\to H_n\) for which \(T({\mathcal P}_n)= {\mathcal P}_n\) (here \(H_n\) denotes the vector space of all \(n\times n\) Hermitian matrices).