No label defined (Q4350136)

Let \(X\) be a real Banach space and \(J: X\to\mathbb{R}\). The functional \(J\) is said to be convex if for \(u\), \(v\) in \(X\) and for \(\lambda,\mu> 0\) with \(\lambda+\mu= 1\), \(J(\lambda u+\mu v)\leq \lambda J(u)+ \mu J(v)\). If for \(u\), \(v\) \((u\neq v)\) in \(X\) and \(\lambda,\mu> 0\) with \(\lambda+\mu= 1\), the strict inequality \(J(\lambda u+ \mu v)< \lambda J(u)+ \mu J(v)\) holds, then \(J\) is called strictly convex. \(J\) is said to be coercive if \(J(v)\to +\infty\) as \(\| v\|\to +\infty\). Let \(u\in X\). If there is a bounded linear functional \(P_u:X\to \mathbb{R}\) such that for every \(v\) in \(X\) and \(\alpha\) in \(\mathbb{R}\) we can express \[ J(u+\alpha v)- J(u)= \alpha[P_u(v)+ \phi(\alpha, v)], \] where \(\phi(\alpha, v)\to 0\) (uniformly in \(v\) on every bounded subset of \(X\)) as \(\alpha\to 0\), then we say that \(J\) is differentiable at \(u\) and \(P_u\) is the Fréchet derivative of \(J\) at \(u\) and write \(J'(u)\) for \(P_u\). In this lecture, I have considered the minimization of the functional \(J:X\to\mathbb{R}\) over a closed convex subset of \(X\) when it is coercive and weakly lower semicontinuous. For the uniqueness of the minimum point strict convexity of \(J\) is necessary. Then the results are applied to find the solution of some standard boundary value problems. For the entire collection see [Zbl 0867.00015].