2017
DOI: 10.1007/s13675-017-0080-8
|View full text |Cite
|
Sign up to set email alerts
|

Nonsmooth spectral gradient methods for unconstrained optimization

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1

Citation Types

0
1
0
1

Year Published

2018
2018
2024
2024

Publication Types

Select...
4
1
1

Relationship

0
6

Authors

Journals

citations
Cited by 6 publications
(2 citation statements)
references
References 23 publications
0
1
0
1
Order By: Relevance
“…The idea of employing a quasi-Newton approach, inspired by the success of quasi-Newton methods in practice for nonsmooth optimization (see [LO13]), has also been explored further in [CQ15,CRZ17]. Another approach, motivated by the encouraging results obtained when employing spectral gradient methods to solve smooth [Fle05,BMR14] and nonsmooth [CLR07] optimization problems, has been to employ a Barzilai-Borwein (BB) strategy for computing initial stepsizes in a GS approach; see [LACR17] and the background in [BB88,Ray93,Ray97]. Using a BB strategy can be viewed as choosing H k = α k I for all k ∈ N where the scalar α k is set according to iterate and gradient displacements in the latest iteration.…”
Section: Second-order-type Variantsmentioning
confidence: 99%
“…The idea of employing a quasi-Newton approach, inspired by the success of quasi-Newton methods in practice for nonsmooth optimization (see [LO13]), has also been explored further in [CQ15,CRZ17]. Another approach, motivated by the encouraging results obtained when employing spectral gradient methods to solve smooth [Fle05,BMR14] and nonsmooth [CLR07] optimization problems, has been to employ a Barzilai-Borwein (BB) strategy for computing initial stepsizes in a GS approach; see [LACR17] and the background in [BB88,Ray93,Ray97]. Using a BB strategy can be viewed as choosing H k = α k I for all k ∈ N where the scalar α k is set according to iterate and gradient displacements in the latest iteration.…”
Section: Second-order-type Variantsmentioning
confidence: 99%
“…Embora os métodos de feixe possuam um papel extremamente relevante na minimização de funções não suaves, limitaremos nosso estudo aos métodos de otimização que fazem uso de técnicas amostrais, isto é, somente algoritmos que estejam relacionados às ideas iniciais do método GS. Durante estes mais de doze anos, diversas variantes deste algoritmo foram propostas [12,13,14,15,16,17]. Aqui, estudaremos algumas destas variantes aplicadas a funções largamente conhecidas na literatura de otimização não suave.…”
Section: Introductionunclassified