On the uniqueness of best approximation in non-Archimedean spaces (Q1179062)

\textit{A. F. Monna} [Indag. Math. 30, 484--496 (1968; Zbl 0172.39302)] has shown that 1. No subspace other than the trivial one of a non-Archimedean normed linear space (n.a. n.l.s.) \(X\) over a non-Archimedean non-trivially valued field \(K\), can be Chebyshev and 2. Strict convexity in non-Archimedean normed linear spaces can not be defined analogously to that available in normed linear spaces over the reals. The authors first show by an example that a subset as against a subspace can be Chebyshev. They take \(X= Q_2\) (2-adic field) over itself and the subset to be \(\{0,1/2,1/2^2,1/2^3,1/2^4,\ldots\}\). This subset is claimed to be compact which is obviously wrong as it is not even bounded in \(X\). The authors say that a n.a. n.l.s. \(X\) such that \(\| X\|=| K|\), is strictly convex if \(\| x+ y\|=\max(\| x\|,\| y\|)\) and \(\| x\|= \| y\|= 1\) imply \(x=y\). They have shown that if the characteristic of the residue class field of \(K\) is 2 then no n.a.n.l.s. can be strictly convex. Now let the characteristic be different from 2, and \(\| X\|=| K|\). Choose \(e\), \(x\in X\) such that \(\| e\|=1\), \(\| x\|<1\). Clearly \(e+ x\neq e- x\) while they meet the other requirements of the definition. This means that no n.a.n.l.s. can be strictly convex in the sense of the definition given by the authors.