Best approximation using a peak norm (Q1185534)

The authors define for each \(\alpha\) with \(0<\alpha\leq 1\) the peak norm or \(\alpha\)-norm \(\|\dots\|^{(\alpha)}\) on the space \(C[0,1]\) by \[ \| f\|^{(\alpha)}={1\over\alpha}\sup_{m(A)=\alpha}\int_ A| g|, \] where the supremum is taken over all subsets of \([0,1]\) with Lebesgue measure \(m(A)=\alpha\). Obviously this norm is equal to the \(L^ 1\) norm on \(C[0,1]\) if \(\alpha=1\). Now a \(L^ 1\)-type characterization theorem and a uniqueness theorem are proved.