Description: denormals in A+alpha*I as alpha > 0 varies, J. Castrillon, Teledyne
|(undirected graph drawing)|
|number of rows||89,400|
|number of columns||89,400|
|structural full rank?||yes|
|# of blocks from dmperm||1|
|# strongly connected comp.||1|
|explicit zero entries||0|
|nonzero pattern symmetry||symmetric|
|numeric value symmetry||symmetric|
The problem is x=(A+alpha*I)\b as alpha varies from 0 to 100. However, as this occurs, performance suffers because of the many denormals that appear below the diagonal in L.
|nnz(chol(P*(A+A'+s*I)*P')) with AMD||8,231,641|
|Cholesky flop count||2.5e+09|
|nnz(L+U), no partial pivoting, with AMD||16,373,882|
|nnz(V) for QR, upper bound nnz(L) for LU, with COLAMD||16,686,436|
|nnz(R) for QR, upper bound nnz(U) for LU, with COLAMD||30,789,903|
For a description of the statistics displayed above, click here.
Maintained by Tim Davis, last updated 12-Mar-2014.
Matrix pictures by cspy, a MATLAB function in the CSparse package.
Matrix graphs by Yifan Hu, AT&T Labs Visualization Group.