1solve_evp_real_2stage_doubleL(i3b)rary Functions Mansuoallve_evp_real_2stage_double(3)
2
6 solve_evp_real_2stage_double - solve the double-precision real
7 eigenvalue problem with the 2-stage ELPA solver (legacy interface)
8
11 FORTRAN INTERFACE
12 use elpa1 use elpa2
13 success = solve_evp_real_2stage_double (na, nev, a(lda,matrixCols),
14 ev(nev), q(ldq, matrixCols), ldq, nblk, matrixCols, mpi_comm_rows,
15 mpi_comm_cols, mpi_comm_all, THIS_REAL_ELPA_KERNEL, useQR, useGPU)
16 With the definitions of the input and output variables:
18 integer, intent(in) na: global dimension of
20 quadratic matrix a to solve
21 integer, intent(in) nev: number of eigenvalues to
22 be computed; the first nev eigenvalules are calculated
23 real*8, intent(inout) a: locally distributed part
24 of the matrix a. The local dimensions are lda x matrixCols
25 integer, intent(in) lda: leading dimension of
26 locally distributed matrix a
27 real*8, intent(inout) ev: on output the first nev
28 computed eigenvalues
29 real*8, intent(inout) q: on output the first nev
30 computed eigenvectors
31 integer, intent(in) ldq: leading dimension of
32 matrix q which stores the eigenvectors
33 integer, intent(in) nblk: blocksize of block cyclic
34 distributin, must be the same in both directions
35 integer, intent(in) matrixCols: number of columns of
36 locally distributed matrices a and q
37 integer, intent(in) mpi_comm_rows: communicator for
38 communication in rows. Constructed with elpa_get_communicators(3)
39 integer, intent(in) mpi_comm_cols: communicator for
40 communication in colums. Constructed with elpa_get_communicators(3)
41 integer, intent(in) mpi_comm_all: communicator for all
42 processes in the processor set involved in ELPA
43 logical, intent(in), optional: useQR: optional argument;
44 switches to QR-decomposition if set to .true.
45 logical, intent(in), optional: useGPU: decide whether GPUs
46 should be used or not
47 logical success: return value indicating
48 success or failure
49 C INTERFACE
51 #include "elpa_legacy.h"
52 success = solve_evp_real_2stage_double_precision (int na, int nev,
54 double *a, int lda, double *ev, double *q, int ldq, int nblk, int
55 matrixCols, int mpi_comm_rows, int mpi_comm_cols, int mpi_comm_all, int
56 THIS_ELPA_REAL_KERNEL, int useQR, int useGPU);
57 With the definitions of the input and output variables:
59 int na: global dimension of quadratic matrix a to solve
61 int nev: number of eigenvalues to be computed; the first
62 nev eigenvalules are calculated
63 double *a: pointer to locally distributed part of the
64 matrix a. The local dimensions are lda x matrixCols
65 int lda: leading dimension of locally distributed matrix
66 a
67 double *ev: pointer to memory containing on output the first
68 nev computed eigenvalues
69 double *q: pointer to memory containing on output the first
70 nev computed eigenvectors
71 int ldq: leading dimension of matrix q which stores the
72 eigenvectors
73 int nblk: blocksize of block cyclic distributin, must be
74 the same in both directions
75 int matrixCols: number of columns of locally distributed
76 matrices a and q
77 int mpi_comm_rows: communicator for communication in rows.
78 Constructed with elpa_get_communicators(3)
79 int mpi_comm_cols: communicator for communication in colums.
80 Constructed with elpa_get_communicators(3)
81 int mpi_comm_all: communicator for all processes in the processor
82 set involved in ELPA
83 int useQR: if set to 1 switch to QR-decomposition
84 int useGPU: decide whether GPUs should be used or nor
85 int success: return value indicating success (1) or failure
86 (0)
87
90 Solve the real eigenvalue problem with the 2-stage solver. The ELPA
91 communicators mpi_comm_rows and mpi_comm_cols are obtained with the
92 elpa_get_communicators(3) function. The distributed quadratic marix a
93 has global dimensions na x na, and a local size lda x matrixCols. The
94 solver will compute the first nev eigenvalues, which will be stored on
95 exit in ev. The eigenvectors corresponding to the eigenvalues will be
96 stored in q. All memory of the arguments must be allocated outside the
97 call to the solver.
98 This function is part of the legacy API of the ELPA library. Better use
99 the current API.
100
102 Old interface: elpa_get_communicators(3)
103 solve_evp_real_1stage_double(3) solve_evp_real_1stage_single(3)
104 solve_evp_complex_1stage_double(3) solve_evp_complex_1stage_single(3)
105 solve_evp_complex_2stage_double(3) solve_evp_complex_2stage_single(3)
106 solve_evp_real_2stage_single(3)
107 Current interface: elpa2_print_kernels(1)
108ELPA Wed May 17 2017solve_evp_real_2stage_double(3)