Original title:
New Limited-Memory Variable Metric Methods for Unconstrained Optimization
Authors:
Vlček, Jan ; Lukšan, Ladislav Document type: Papers Conference/Event: SANM 2003. Summer School /15./, Hejnice (CZ), 2003-09-08 / 2003-09-12
Year:
2004
Language:
eng Abstract:
A new family of numerically efficient variable metric or quasi-Newton methods for unconstrained optimizaion is adapted for large-scale optimization. Global convergence of the method can be established for convex sufficiently smooth functions. Some encouraging numerical experience is reported. We refer to our report V-876 for proof and details.
Keywords:
global convergence; limited-memory methods; numerical results; unconstrained minimization; variable metric methods Project no.: CEZ:AV0Z1030915 (CEP), IAA1030103 (CEP) Funding provider: GA AV ČR Host item entry: Software and Algorithms of Numerical Mathematics, ISBN 80-7043-312-4
Institution: Institute of Computer Science AS ČR
(web)
Document availability information: Fulltext is available at the institute of the Academy of Sciences. Original record: http://hdl.handle.net/11104/0010612