Geodesic
An automatic scheḿe for updafing the block size in the block conjugate gradient method for solving linear systems
Zapiski Nauchnykh Seminarov POMI, Computational methods and algorithms. Part XIV, Tome 268 (2000), pp. 159-175 
A. A. Nikishin; A. Yu. Yeremin. An automatic scheḿe for updafing the block size in the block conjugate gradient method for solving linear systems. Zapiski Nauchnykh Seminarov POMI, Computational methods and algorithms. Part XIV, Tome 268 (2000), pp. 159-175. http://geodesic.mathdoc.fr/item/ZNSL_2000_268_a9/
@article{ZNSL_2000_268_a9,
     author = {A. A. Nikishin and A. Yu. Yeremin},
     title = {An automatic scheḿe for updafing the block size in the block conjugate gradient method for solving linear systems},
     journal = {Zapiski Nauchnykh Seminarov POMI},
     pages = {159--175},
     year = {2000},
     volume = {268},
     language = {ru},
     url = {http://geodesic.mathdoc.fr/item/ZNSL_2000_268_a9/}
}
TY  - JOUR
AU  - A. A. Nikishin
AU  - A. Yu. Yeremin
TI  - An automatic scheḿe for updafing the block size in the block conjugate gradient method for solving linear systems
JO  - Zapiski Nauchnykh Seminarov POMI
PY  - 2000
SP  - 159
EP  - 175
VL  - 268
UR  - http://geodesic.mathdoc.fr/item/ZNSL_2000_268_a9/
LA  - ru
ID  - ZNSL_2000_268_a9
ER  - 
%0 Journal Article
%A A. A. Nikishin
%A A. Yu. Yeremin
%T An automatic scheḿe for updafing the block size in the block conjugate gradient method for solving linear systems
%J Zapiski Nauchnykh Seminarov POMI
%D 2000
%P 159-175
%V 268
%U http://geodesic.mathdoc.fr/item/ZNSL_2000_268_a9/
%G ru
%F ZNSL_2000_268_a9

Voir la notice du chapitre de livre provenant de la source Math-Net.Ru

The paper considers the problem of constructing on efficient automatic procedure for reducing the block size in the block conjugate gradient method providing for the rate of convergence comparable with that of the block conjugate gradient method with constant block size. Results of numerical experiments show that, independently of the type of distribution of the smallest eigenvalues of the preconditioned matrix, the procedure suggested always leads to a decrease of arithmetic costs with respect to those of the block method with constant block size.