typedef enum { DS_MAT_A,
DS_MAT_B,
DS_MAT_C,
DS_MAT_T,
DS_MAT_D,
DS_MAT_Q,
DS_MAT_Z,
DS_MAT_X,
DS_MAT_Y,
DS_MAT_U,
DS_MAT_V,
DS_MAT_W,
DS_MAT_E0,
DS_MAT_E1,
DS_MAT_E2,
DS_MAT_E3,
DS_MAT_E4,
DS_MAT_E5,
DS_MAT_E6,
DS_MAT_E7,
DS_MAT_E8,
DS_MAT_E9,
DS_NUM_MAT } DSMatType;
| DS_MAT_A | - first matrix of eigenproblem/singular value problem | |
| DS_MAT_B | - second matrix of a generalized eigenproblem | |
| DS_MAT_C | - third matrix of a quadratic eigenproblem (deprecated) | |
| DS_MAT_T | - tridiagonal matrix | |
| DS_MAT_D | - diagonal matrix | |
| DS_MAT_Q | - orthogonal matrix of (right) Schur vectors | |
| DS_MAT_Z | - orthogonal matrix of left Schur vectors | |
| DS_MAT_X | - right eigenvectors | |
| DS_MAT_Y | - left eigenvectors | |
| DS_MAT_U | - left singular vectors | |
| DS_MAT_V | - right singular vectors | |
| DS_MAT_W | - workspace matrix | |
| DS_MAT_Ex | - extra matrices (x=0,..,9) |
All matrices can have space to hold ld x ld elements, except for DS_MAT_T that has space for 3 x ld elements (ld = leading dimension) and DS_MAT_D that has space for just ld elements.
In DSPEP problems, matrices A, B, W can have space for d*ld x d*ld, where d is the polynomial degree, and X can have ld x d*ld. Also DSNEP has exceptions. Check the manual page of each DS type for details.