Actual source code: klu.c
2: /*
3: Provides an interface to the KLUv1.2 sparse solver
5: When build with PETSC_USE_64BIT_INDICES this will use SuiteSparse_long as the
6: integer type in KLU, otherwise it will use int. This means
7: all integers in this file are simply declared as PetscInt. Also it means
8: that KLU SuiteSparse_long version MUST be built with 64 bit integers when used.
10: */
11: #include <../src/mat/impls/aij/seq/aij.h>
13: #if defined(PETSC_USE_64BIT_INDICES)
14: #define klu_K_defaults klu_l_defaults
15: #define klu_K_analyze(a,b,c,d) klu_l_analyze((SuiteSparse_long)a,(SuiteSparse_long*)b,(SuiteSparse_long*)c,d)
16: #define klu_K_analyze_given(a,b,c,d,e,f) klu_l_analyze_given((SuiteSparse_long)a,(SuiteSparse_long*)b,(SuiteSparse_long*)c,(SuiteSparse_long*)d,(SuiteSparse_long*)e,f)
17: #define klu_K_free_symbolic klu_l_free_symbolic
18: #define klu_K_free_numeric klu_l_free_numeric
19: #define klu_K_common klu_l_common
20: #define klu_K_symbolic klu_l_symbolic
21: #define klu_K_numeric klu_l_numeric
22: #if defined(PETSC_USE_COMPLEX)
23: #define klu_K_factor(a,b,c,d,e) klu_zl_factor((SuiteSparse_long*)a,(SuiteSparse_long*)b,c,d,e);
24: #define klu_K_solve klu_zl_solve
25: #define klu_K_tsolve klu_zl_tsolve
26: #define klu_K_refactor klu_zl_refactor
27: #define klu_K_sort klu_zl_sort
28: #define klu_K_flops klu_zl_flops
29: #define klu_K_rgrowth klu_zl_rgrowth
30: #define klu_K_condest klu_zl_condest
31: #define klu_K_rcond klu_zl_rcond
32: #define klu_K_scale klu_zl_scale
33: #else
34: #define klu_K_factor(a,b,c,d,e) klu_l_factor((SuiteSparse_long*)a,(SuiteSparse_long*)b,c,d,e);
35: #define klu_K_solve klu_l_solve
36: #define klu_K_tsolve klu_l_tsolve
37: #define klu_K_refactor klu_l_refactor
38: #define klu_K_sort klu_l_sort
39: #define klu_K_flops klu_l_flops
40: #define klu_K_rgrowth klu_l_rgrowth
41: #define klu_K_condest klu_l_condest
42: #define klu_K_rcond klu_l_rcond
43: #define klu_K_scale klu_l_scale
44: #endif
45: #else
46: #define klu_K_defaults klu_defaults
47: #define klu_K_analyze klu_analyze
48: #define klu_K_analyze_given klu_analyze_given
49: #define klu_K_free_symbolic klu_free_symbolic
50: #define klu_K_free_numeric klu_free_numeric
51: #define klu_K_common klu_common
52: #define klu_K_symbolic klu_symbolic
53: #define klu_K_numeric klu_numeric
54: #if defined(PETSC_USE_COMPLEX)
55: #define klu_K_factor klu_z_factor
56: #define klu_K_solve klu_z_solve
57: #define klu_K_tsolve klu_z_tsolve
58: #define klu_K_refactor klu_z_refactor
59: #define klu_K_sort klu_z_sort
60: #define klu_K_flops klu_z_flops
61: #define klu_K_rgrowth klu_z_rgrowth
62: #define klu_K_condest klu_z_condest
63: #define klu_K_rcond klu_z_rcond
64: #define klu_K_scale klu_z_scale
65: #else
66: #define klu_K_factor klu_factor
67: #define klu_K_solve klu_solve
68: #define klu_K_tsolve klu_tsolve
69: #define klu_K_refactor klu_refactor
70: #define klu_K_sort klu_sort
71: #define klu_K_flops klu_flops
72: #define klu_K_rgrowth klu_rgrowth
73: #define klu_K_condest klu_condest
74: #define klu_K_rcond klu_rcond
75: #define klu_K_scale klu_scale
76: #endif
77: #endif
79: EXTERN_C_BEGIN
80: #include <klu.h>
81: EXTERN_C_END
83: static const char *KluOrderingTypes[] = {"AMD","COLAMD"};
84: static const char *scale[] ={"NONE","SUM","MAX"};
86: typedef struct {
87: klu_K_common Common;
88: klu_K_symbolic *Symbolic;
89: klu_K_numeric *Numeric;
90: PetscInt *perm_c,*perm_r;
91: MatStructure flg;
92: PetscBool PetscMatOrdering;
93: PetscBool CleanUpKLU;
94: } Mat_KLU;
96: static PetscErrorCode MatDestroy_KLU(Mat A)
97: {
98: Mat_KLU *lu=(Mat_KLU*)A->data;
100: if (lu->CleanUpKLU) {
101: klu_K_free_symbolic(&lu->Symbolic,&lu->Common);
102: klu_K_free_numeric(&lu->Numeric,&lu->Common);
103: PetscFree2(lu->perm_r,lu->perm_c);
104: }
105: PetscFree(A->data);
106: return 0;
107: }
109: static PetscErrorCode MatSolveTranspose_KLU(Mat A,Vec b,Vec x)
110: {
111: Mat_KLU *lu = (Mat_KLU*)A->data;
112: PetscScalar *xa;
113: PetscInt status;
115: /* KLU uses a column major format, solve Ax = b by klu_*_solve */
116: /* ----------------------------------*/
117: VecCopy(b,x); /* klu_solve stores the solution in rhs */
118: VecGetArray(x,&xa);
119: status = klu_K_solve(lu->Symbolic,lu->Numeric,A->rmap->n,1,(PetscReal*)xa,&lu->Common);
121: VecRestoreArray(x,&xa);
122: return 0;
123: }
125: static PetscErrorCode MatSolve_KLU(Mat A,Vec b,Vec x)
126: {
127: Mat_KLU *lu = (Mat_KLU*)A->data;
128: PetscScalar *xa;
129: PetscInt status;
131: /* KLU uses a column major format, solve A^Tx = b by klu_*_tsolve */
132: /* ----------------------------------*/
133: VecCopy(b,x); /* klu_solve stores the solution in rhs */
134: VecGetArray(x,&xa);
135: #if defined(PETSC_USE_COMPLEX)
136: PetscInt conj_solve=1;
137: status = klu_K_tsolve(lu->Symbolic,lu->Numeric,A->rmap->n,1,(PetscReal*)xa,conj_solve,&lu->Common); /* conjugate solve */
138: #else
139: status = klu_K_tsolve(lu->Symbolic,lu->Numeric,A->rmap->n,1,xa,&lu->Common);
140: #endif
142: VecRestoreArray(x,&xa);
143: return 0;
144: }
146: static PetscErrorCode MatLUFactorNumeric_KLU(Mat F,Mat A,const MatFactorInfo *info)
147: {
148: Mat_KLU *lu = (Mat_KLU*)(F)->data;
149: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
150: PetscInt *ai = a->i,*aj=a->j;
151: PetscScalar *av = a->a;
153: /* numeric factorization of A' */
154: /* ----------------------------*/
156: if (lu->flg == SAME_NONZERO_PATTERN && lu->Numeric) {
157: klu_K_free_numeric(&lu->Numeric,&lu->Common);
158: }
159: lu->Numeric = klu_K_factor(ai,aj,(PetscReal*)av,lu->Symbolic,&lu->Common);
162: lu->flg = SAME_NONZERO_PATTERN;
163: lu->CleanUpKLU = PETSC_TRUE;
164: F->ops->solve = MatSolve_KLU;
165: F->ops->solvetranspose = MatSolveTranspose_KLU;
166: return 0;
167: }
169: static PetscErrorCode MatLUFactorSymbolic_KLU(Mat F,Mat A,IS r,IS c,const MatFactorInfo *info)
170: {
171: Mat_SeqAIJ *a = (Mat_SeqAIJ*)A->data;
172: Mat_KLU *lu = (Mat_KLU*)(F->data);
173: PetscInt i,*ai = a->i,*aj = a->j,m=A->rmap->n,n=A->cmap->n;
174: const PetscInt *ra,*ca;
176: if (lu->PetscMatOrdering) {
177: ISGetIndices(r,&ra);
178: ISGetIndices(c,&ca);
179: PetscMalloc2(m,&lu->perm_r,n,&lu->perm_c);
180: /* we cannot simply memcpy on 64 bit archs */
181: for (i = 0; i < m; i++) lu->perm_r[i] = ra[i];
182: for (i = 0; i < n; i++) lu->perm_c[i] = ca[i];
183: ISRestoreIndices(r,&ra);
184: ISRestoreIndices(c,&ca);
185: }
187: /* symbolic factorization of A' */
188: /* ---------------------------------------------------------------------- */
189: if (r) {
190: lu->PetscMatOrdering = PETSC_TRUE;
191: lu->Symbolic = klu_K_analyze_given(n,ai,aj,lu->perm_c,lu->perm_r,&lu->Common);
192: } else { /* use klu internal ordering */
193: lu->Symbolic = klu_K_analyze(n,ai,aj,&lu->Common);
194: }
197: lu->flg = DIFFERENT_NONZERO_PATTERN;
198: lu->CleanUpKLU = PETSC_TRUE;
199: (F)->ops->lufactornumeric = MatLUFactorNumeric_KLU;
200: return 0;
201: }
203: static PetscErrorCode MatView_Info_KLU(Mat A,PetscViewer viewer)
204: {
205: Mat_KLU *lu= (Mat_KLU*)A->data;
206: klu_K_numeric *Numeric=(klu_K_numeric*)lu->Numeric;
208: PetscViewerASCIIPrintf(viewer,"KLU stats:\n");
209: PetscViewerASCIIPrintf(viewer," Number of diagonal blocks: %" PetscInt_FMT "\n",(PetscInt)(Numeric->nblocks));
210: PetscViewerASCIIPrintf(viewer," Total nonzeros=%" PetscInt_FMT "\n",(PetscInt)(Numeric->lnz+Numeric->unz));
211: PetscViewerASCIIPrintf(viewer,"KLU runtime parameters:\n");
212: /* Control parameters used by numeric factorization */
213: PetscViewerASCIIPrintf(viewer," Partial pivoting tolerance: %g\n",lu->Common.tol);
214: /* BTF preordering */
215: PetscViewerASCIIPrintf(viewer," BTF preordering enabled: %" PetscInt_FMT "\n",(PetscInt)(lu->Common.btf));
216: /* mat ordering */
217: if (!lu->PetscMatOrdering) {
218: PetscViewerASCIIPrintf(viewer," Ordering: %s (not using the PETSc ordering)\n",KluOrderingTypes[(int)lu->Common.ordering]);
219: }
220: /* matrix row scaling */
221: PetscViewerASCIIPrintf(viewer, " Matrix row scaling: %s\n",scale[(int)lu->Common.scale]);
222: return 0;
223: }
225: static PetscErrorCode MatView_KLU(Mat A,PetscViewer viewer)
226: {
227: PetscBool iascii;
228: PetscViewerFormat format;
230: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);
231: if (iascii) {
232: PetscViewerGetFormat(viewer,&format);
233: if (format == PETSC_VIEWER_ASCII_INFO) {
234: MatView_Info_KLU(A,viewer);
235: }
236: }
237: return 0;
238: }
240: PetscErrorCode MatFactorGetSolverType_seqaij_klu(Mat A,MatSolverType *type)
241: {
242: *type = MATSOLVERKLU;
243: return 0;
244: }
246: /*MC
247: MATSOLVERKLU = "klu" - A matrix type providing direct solvers (LU) for sequential matrices
248: via the external package KLU.
250: ./configure --download-suitesparse to install PETSc to use KLU
252: Use -pc_type lu -pc_factor_mat_solver_type klu to use this direct solver
254: Consult KLU documentation for more information on the options database keys below.
256: Options Database Keys:
257: + -mat_klu_pivot_tol <0.001> - Partial pivoting tolerance
258: . -mat_klu_use_btf <1> - Use BTF preordering
259: . -mat_klu_ordering <AMD> - KLU reordering scheme to reduce fill-in (choose one of) AMD COLAMD PETSC
260: - -mat_klu_row_scale <NONE> - Matrix row scaling (choose one of) NONE SUM MAX
262: Note: KLU is part of SuiteSparse http://faculty.cse.tamu.edu/davis/suitesparse.html
264: Level: beginner
266: .seealso: PCLU, MATSOLVERUMFPACK, MATSOLVERCHOLMOD, PCFactorSetMatSolverType(), MatSolverType
267: M*/
269: PETSC_INTERN PetscErrorCode MatGetFactor_seqaij_klu(Mat A,MatFactorType ftype,Mat *F)
270: {
271: Mat B;
272: Mat_KLU *lu;
274: PetscInt m=A->rmap->n,n=A->cmap->n,idx = 0,status;
275: PetscBool flg;
277: /* Create the factorization matrix F */
278: MatCreate(PetscObjectComm((PetscObject)A),&B);
279: MatSetSizes(B,PETSC_DECIDE,PETSC_DECIDE,m,n);
280: PetscStrallocpy("klu",&((PetscObject)B)->type_name);
281: MatSetUp(B);
283: PetscNewLog(B,&lu);
285: B->data = lu;
286: B->ops->getinfo = MatGetInfo_External;
287: B->ops->lufactorsymbolic = MatLUFactorSymbolic_KLU;
288: B->ops->destroy = MatDestroy_KLU;
289: B->ops->view = MatView_KLU;
291: PetscObjectComposeFunction((PetscObject)B,"MatFactorGetSolverType_C",MatFactorGetSolverType_seqaij_klu);
293: B->factortype = MAT_FACTOR_LU;
294: B->assembled = PETSC_TRUE; /* required by -ksp_view */
295: B->preallocated = PETSC_TRUE;
297: PetscFree(B->solvertype);
298: PetscStrallocpy(MATSOLVERKLU,&B->solvertype);
299: B->canuseordering = PETSC_TRUE;
300: PetscStrallocpy(MATORDERINGEXTERNAL,(char**)&B->preferredordering[MAT_FACTOR_LU]);
302: /* initializations */
303: /* ------------------------------------------------*/
304: /* get the default control parameters */
305: status = klu_K_defaults(&lu->Common);
308: lu->Common.scale = 0; /* No row scaling */
310: PetscOptionsBegin(PetscObjectComm((PetscObject)A),((PetscObject)A)->prefix,"KLU Options","Mat");
311: /* Partial pivoting tolerance */
312: PetscOptionsReal("-mat_klu_pivot_tol","Partial pivoting tolerance","None",lu->Common.tol,&lu->Common.tol,NULL);
313: /* BTF pre-ordering */
314: PetscOptionsInt("-mat_klu_use_btf","Enable BTF preordering","None",(PetscInt)lu->Common.btf,(PetscInt*)&lu->Common.btf,NULL);
315: /* Matrix reordering */
316: PetscOptionsEList("-mat_klu_ordering","Internal ordering method","None",KluOrderingTypes,sizeof(KluOrderingTypes)/sizeof(KluOrderingTypes[0]),KluOrderingTypes[0],&idx,&flg);
317: lu->Common.ordering = (int)idx;
318: /* Matrix row scaling */
319: PetscOptionsEList("-mat_klu_row_scale","Matrix row scaling","None",scale,3,scale[0],&idx,&flg);
320: PetscOptionsEnd();
321: *F = B;
322: return 0;
323: }