1996
DOI: 10.1002/(sici)1099-1506(199607/08)3:4<329::aid-nla86>3.3.co;2-#
|Get access via publisher |Summarize |Cite
|
Sign up to set email alerts

DQGMRES: a Direct Quasi‐minimal Residual Algorithm Based on Incomplete Orthogonalization

Search citation statements

Order By: Relevance

Paper Sections

Select...
6
2
0
0

Citation Types

0
6
0
0

Year Published

1998
1998
2024
2024

Publication Types

Select...
7
1

Relationship

1
7

Authors

Journals

citations

Cited by 8 publications

(6 citation statements)
references

References 0 publications

0
6
0
0
Order By: Relevance
How this paper cites the one you are viewing
“…Regardless of finite precision, the mathematical equivalence of LCG and FOM can be easily established as follows. Actually, the similar review has been discussed by Saad et al (1996). He presented the direct implementation of truncated GMRES by combining quasiminimization and incomplete orthogonalization.…”
Section: Algorithm 22 (Full Orthogonalization Method)
mentioning
confidence: 91%
How this paper cites the one you are viewing
“…Regardless of finite precision, the mathematical equivalence of LCG and FOM can be easily established as follows. Actually, the similar review has been discussed by Saad et al (1996). He presented the direct implementation of truncated GMRES by combining quasiminimization and incomplete orthogonalization.…”
Section: Algorithm 22 (Full Orthogonalization Method)
mentioning
confidence: 91%
How this paper cites the one you are viewing
“…• The nonsymmetric Lanczos method [19] and its variations, e.g., QMR [10]; • Incomplete (truncated) methods [26], [27, §6.4.2 and §6.5.6], [29], and restarted methods [27], [31];…”
Section: Notation Descriptions and Examples
mentioning
confidence: 99%
“…We refer to the number k as the truncation parameter. In this case, the upper Hessenberg matrix T m+1,m in (2.1) is banded with k nonzero superdiagonals; see, e.g., [26], [27, §6.4.2 and §6.5.6], [29].…”
Section: Notation Descriptions and Examples
mentioning
confidence: 99%
How this paper cites the one you are viewing
“…It was argued in Reference [9] that if scaling results in a deterioration of the degree of normality of the matrix, measured by the number AA T −A T A/AA T , this will result in worse performances. Diagonal scaling consists of scaling all the rows of the matrix and then scaling all the columns (or the same operations done in reverse order).…”
Section: Other Techniques
mentioning
confidence: 99%