Precima/analytics
Precima/analytics: data analytics problem
| Name |
analytics |
| Group |
Precima |
| Matrix ID |
2851 |
|
Num Rows
|
303,813 |
|
Num Cols
|
303,813 |
|
Nonzeros
|
2,006,126 |
|
Pattern Entries
|
2,006,126 |
|
Kind
|
Data Analytics Problem |
|
Symmetric
|
Yes |
|
Date
|
2018 |
|
Author
|
C. Ugaz |
|
Editor
|
T. Davis |
| Structural Rank |
303,813 |
| Structural Rank Full |
true |
|
Num Dmperm Blocks
|
41,422 |
|
Strongly Connect Components
|
41,422 |
|
Num Explicit Zeros
|
0 |
|
Pattern Symmetry
|
100% |
|
Numeric Symmetry
|
100% |
|
Cholesky Candidate
|
no |
|
Positive Definite
|
no |
|
Type
|
real |
| Download |
MATLAB
Rutherford Boeing
Matrix Market
|
| Notes |
Precima: data analytics problem
Submitted by Carl Ugaz, Precima, Inc. CUgaz@precima.com
The matrix is nonsingular, ill-conditioned,, and symmetric indefinite.
If factorized with KLU, best results are obtained if the permutation to
block triangular form (BTF) is not used, because it breaks the symmetry
of the pattern.
Ignoring the diagonal (which has some zero entries and some nonzero)
the undirected graph of A divides into 41,422 connected components.
Of those, 40,966 are single nodes, and 451 components have between
4 and 90 nodes. The remaining 5 components have 93738, 89686, 40319,
15884, and 4289 nodes, respectively.
|
