×

A finite method to find a point in a set defined by a convex differentiable functional. (English. Russian original) Zbl 0669.49003

J. Sov. Math. 45, No. 4, 1273-1277 (1989); translation from Issled. Prikl. Mat. 13, 3-10 (1985).
It is required to find a point in the set \(Q=\{x\in H:\) \(\phi\) (x)\(\leq 0\}\), where \(\phi\) is a convex continuously Frechet-differentiable functional defined on the real Hilbert space H. We apply the following method to solve this problem: \[ (1)\quad x^{(k+1)}=x^{(k)}-\gamma_ k\frac{\phi (x^{(k)})}{\| g(x^{(k)})\|^ 2}g(x^{(k)}), \] where \(x^{(0)}\) is a point in H, \(g(x^{(k)})\) is the Frechet- derivative of the functional \(\phi\) at the point \(x^{(k)}\), \(\| \cdot \|\) is the norm in the space H induced by the scalar product. We assume that the process stops as soon as \(x^{(k)}\) reaches Q. We show that with \(\gamma_ k\) from the interval \([I+\epsilon,2]\) (\(\epsilon\) is any number from (0;1]), the iterative process (1) is finite and its last point belongs to Q if \(\phi\) satisfies the Slater condition.

MSC:

49J27 Existence theories for problems in abstract spaces
49J50 Fréchet and Gateaux differentiability in optimization
PDFBibTeX XMLCite
Full Text: DOI

References:

[1] I. I. Eremin, ?A relaxation method for the solution of systems of inequalities with convex functions in the left-hand sides,? Dokl. AN SSSR,160, No. 5, 994?996 (1965).
[2] E. V. Raik, ?Fejer-type methods in Hilbert space,? Izv. AN ESSR, Ser. Fiz. Mat.,16, No. 3, 286?293 (1967).
[3] V. R. Fazylov, ?A general method to find a point in a convex set,? Izv. Vyssh. Uchebn. Zaved., Mat., No. 6, 43?51 (1983).
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.