xref: /petsc/src/mat/impls/aij/seq/klu/klu.c (revision ebead697dbf761eb322f829370bbe90b3bd93fa3)
1 
2 /*
3    Provides an interface to the KLUv1.2 sparse solver
4 
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.
9 
10 */
11 #include <../src/mat/impls/aij/seq/aij.h>
12 
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
78 
79 EXTERN_C_BEGIN
80 #include <klu.h>
81 EXTERN_C_END
82 
83 static const char *KluOrderingTypes[] = {"AMD","COLAMD"};
84 static const char *scale[] ={"NONE","SUM","MAX"};
85 
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;
95 
96 static PetscErrorCode MatDestroy_KLU(Mat A)
97 {
98   Mat_KLU    *lu=(Mat_KLU*)A->data;
99 
100   PetscFunctionBegin;
101   if (lu->CleanUpKLU) {
102     klu_K_free_symbolic(&lu->Symbolic,&lu->Common);
103     klu_K_free_numeric(&lu->Numeric,&lu->Common);
104     PetscCall(PetscFree2(lu->perm_r,lu->perm_c));
105   }
106   PetscCall(PetscObjectComposeFunction((PetscObject)A,"MatFactorGetSolverType_C",NULL));
107   PetscCall(PetscFree(A->data));
108   PetscFunctionReturn(0);
109 }
110 
111 static PetscErrorCode MatSolveTranspose_KLU(Mat A,Vec b,Vec x)
112 {
113   Mat_KLU     *lu = (Mat_KLU*)A->data;
114   PetscScalar *xa;
115   PetscInt     status;
116 
117   PetscFunctionBegin;
118   /* KLU uses a column major format, solve Ax = b by klu_*_solve */
119   /* ----------------------------------*/
120   PetscCall(VecCopy(b,x)); /* klu_solve stores the solution in rhs */
121   PetscCall(VecGetArray(x,&xa));
122   status = klu_K_solve(lu->Symbolic,lu->Numeric,A->rmap->n,1,(PetscReal*)xa,&lu->Common);
123   PetscCheck(status == 1,PETSC_COMM_SELF,PETSC_ERR_LIB,"KLU Solve failed");
124   PetscCall(VecRestoreArray(x,&xa));
125   PetscFunctionReturn(0);
126 }
127 
128 static PetscErrorCode MatSolve_KLU(Mat A,Vec b,Vec x)
129 {
130   Mat_KLU     *lu = (Mat_KLU*)A->data;
131   PetscScalar *xa;
132   PetscInt     status;
133 
134   PetscFunctionBegin;
135   /* KLU uses a column major format, solve A^Tx = b by klu_*_tsolve */
136   /* ----------------------------------*/
137   PetscCall(VecCopy(b,x)); /* klu_solve stores the solution in rhs */
138   PetscCall(VecGetArray(x,&xa));
139 #if defined(PETSC_USE_COMPLEX)
140   PetscInt conj_solve=1;
141   status = klu_K_tsolve(lu->Symbolic,lu->Numeric,A->rmap->n,1,(PetscReal*)xa,conj_solve,&lu->Common); /* conjugate solve */
142 #else
143   status = klu_K_tsolve(lu->Symbolic,lu->Numeric,A->rmap->n,1,xa,&lu->Common);
144 #endif
145   PetscCheck(status == 1,PETSC_COMM_SELF,PETSC_ERR_LIB,"KLU Solve failed");
146   PetscCall(VecRestoreArray(x,&xa));
147   PetscFunctionReturn(0);
148 }
149 
150 static PetscErrorCode MatLUFactorNumeric_KLU(Mat F,Mat A,const MatFactorInfo *info)
151 {
152   Mat_KLU        *lu = (Mat_KLU*)(F)->data;
153   Mat_SeqAIJ     *a  = (Mat_SeqAIJ*)A->data;
154   PetscInt       *ai = a->i,*aj=a->j;
155   PetscScalar    *av = a->a;
156 
157   PetscFunctionBegin;
158   /* numeric factorization of A' */
159   /* ----------------------------*/
160 
161   if (lu->flg == SAME_NONZERO_PATTERN && lu->Numeric) {
162     klu_K_free_numeric(&lu->Numeric,&lu->Common);
163   }
164   lu->Numeric = klu_K_factor(ai,aj,(PetscReal*)av,lu->Symbolic,&lu->Common);
165   PetscCheck(lu->Numeric,PETSC_COMM_SELF,PETSC_ERR_LIB,"KLU Numeric factorization failed");
166 
167   lu->flg                = SAME_NONZERO_PATTERN;
168   lu->CleanUpKLU         = PETSC_TRUE;
169   F->ops->solve          = MatSolve_KLU;
170   F->ops->solvetranspose = MatSolveTranspose_KLU;
171   PetscFunctionReturn(0);
172 }
173 
174 static PetscErrorCode MatLUFactorSymbolic_KLU(Mat F,Mat A,IS r,IS c,const MatFactorInfo *info)
175 {
176   Mat_SeqAIJ     *a  = (Mat_SeqAIJ*)A->data;
177   Mat_KLU        *lu = (Mat_KLU*)(F->data);
178   PetscInt       i,*ai = a->i,*aj = a->j,m=A->rmap->n,n=A->cmap->n;
179   const PetscInt *ra,*ca;
180 
181   PetscFunctionBegin;
182   if (lu->PetscMatOrdering) {
183     PetscCall(ISGetIndices(r,&ra));
184     PetscCall(ISGetIndices(c,&ca));
185     PetscCall(PetscMalloc2(m,&lu->perm_r,n,&lu->perm_c));
186     /* we cannot simply memcpy on 64 bit archs */
187     for (i = 0; i < m; i++) lu->perm_r[i] = ra[i];
188     for (i = 0; i < n; i++) lu->perm_c[i] = ca[i];
189     PetscCall(ISRestoreIndices(r,&ra));
190     PetscCall(ISRestoreIndices(c,&ca));
191   }
192 
193   /* symbolic factorization of A' */
194   /* ---------------------------------------------------------------------- */
195   if (r) {
196     lu->PetscMatOrdering = PETSC_TRUE;
197     lu->Symbolic = klu_K_analyze_given(n,ai,aj,lu->perm_c,lu->perm_r,&lu->Common);
198   } else { /* use klu internal ordering */
199     lu->Symbolic = klu_K_analyze(n,ai,aj,&lu->Common);
200   }
201   PetscCheck(lu->Symbolic,PETSC_COMM_SELF,PETSC_ERR_LIB,"KLU Symbolic Factorization failed");
202 
203   lu->flg                   = DIFFERENT_NONZERO_PATTERN;
204   lu->CleanUpKLU            = PETSC_TRUE;
205   (F)->ops->lufactornumeric = MatLUFactorNumeric_KLU;
206   PetscFunctionReturn(0);
207 }
208 
209 static PetscErrorCode MatView_Info_KLU(Mat A,PetscViewer viewer)
210 {
211   Mat_KLU       *lu= (Mat_KLU*)A->data;
212   klu_K_numeric *Numeric=(klu_K_numeric*)lu->Numeric;
213 
214   PetscFunctionBegin;
215   PetscCall(PetscViewerASCIIPrintf(viewer,"KLU stats:\n"));
216   PetscCall(PetscViewerASCIIPrintf(viewer,"  Number of diagonal blocks: %" PetscInt_FMT "\n",(PetscInt)(Numeric->nblocks)));
217   PetscCall(PetscViewerASCIIPrintf(viewer,"  Total nonzeros=%" PetscInt_FMT "\n",(PetscInt)(Numeric->lnz+Numeric->unz)));
218   PetscCall(PetscViewerASCIIPrintf(viewer,"KLU runtime parameters:\n"));
219   /* Control parameters used by numeric factorization */
220   PetscCall(PetscViewerASCIIPrintf(viewer,"  Partial pivoting tolerance: %g\n",lu->Common.tol));
221   /* BTF preordering */
222   PetscCall(PetscViewerASCIIPrintf(viewer,"  BTF preordering enabled: %" PetscInt_FMT "\n",(PetscInt)(lu->Common.btf)));
223   /* mat ordering */
224   if (!lu->PetscMatOrdering) {
225     PetscCall(PetscViewerASCIIPrintf(viewer,"  Ordering: %s (not using the PETSc ordering)\n",KluOrderingTypes[(int)lu->Common.ordering]));
226   }
227   /* matrix row scaling */
228   PetscCall(PetscViewerASCIIPrintf(viewer, "  Matrix row scaling: %s\n",scale[(int)lu->Common.scale]));
229   PetscFunctionReturn(0);
230 }
231 
232 static PetscErrorCode MatView_KLU(Mat A,PetscViewer viewer)
233 {
234   PetscBool         iascii;
235   PetscViewerFormat format;
236 
237   PetscFunctionBegin;
238   PetscCall(PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii));
239   if (iascii) {
240     PetscCall(PetscViewerGetFormat(viewer,&format));
241     if (format == PETSC_VIEWER_ASCII_INFO) {
242       PetscCall(MatView_Info_KLU(A,viewer));
243     }
244   }
245   PetscFunctionReturn(0);
246 }
247 
248 PetscErrorCode MatFactorGetSolverType_seqaij_klu(Mat A,MatSolverType *type)
249 {
250   PetscFunctionBegin;
251   *type = MATSOLVERKLU;
252   PetscFunctionReturn(0);
253 }
254 
255 /*MC
256   MATSOLVERKLU = "klu" - A matrix type providing direct solvers (LU) for sequential matrices
257   via the external package KLU.
258 
259   ./configure --download-suitesparse to install PETSc to use KLU
260 
261   Use -pc_type lu -pc_factor_mat_solver_type klu to use this direct solver
262 
263   Consult KLU documentation for more information on the options database keys below.
264 
265   Options Database Keys:
266 + -mat_klu_pivot_tol <0.001>                  - Partial pivoting tolerance
267 . -mat_klu_use_btf <1>                        - Use BTF preordering
268 . -mat_klu_ordering <AMD>                     - KLU reordering scheme to reduce fill-in (choose one of) AMD COLAMD PETSC
269 - -mat_klu_row_scale <NONE>                   - Matrix row scaling (choose one of) NONE SUM MAX
270 
271    Note: KLU is part of SuiteSparse http://faculty.cse.tamu.edu/davis/suitesparse.html
272 
273    Level: beginner
274 
275 .seealso: `PCLU`, `MATSOLVERUMFPACK`, `MATSOLVERCHOLMOD`, `PCFactorSetMatSolverType()`, `MatSolverType`
276 M*/
277 
278 PETSC_INTERN PetscErrorCode MatGetFactor_seqaij_klu(Mat A,MatFactorType ftype,Mat *F)
279 {
280   Mat            B;
281   Mat_KLU       *lu;
282   PetscInt       m=A->rmap->n,n=A->cmap->n,idx = 0,status;
283   PetscBool      flg;
284 
285   PetscFunctionBegin;
286   /* Create the factorization matrix F */
287   PetscCall(MatCreate(PetscObjectComm((PetscObject)A),&B));
288   PetscCall(MatSetSizes(B,PETSC_DECIDE,PETSC_DECIDE,m,n));
289   PetscCall(PetscStrallocpy("klu",&((PetscObject)B)->type_name));
290   PetscCall(MatSetUp(B));
291 
292   PetscCall(PetscNewLog(B,&lu));
293 
294   B->data                  = lu;
295   B->ops->getinfo          = MatGetInfo_External;
296   B->ops->lufactorsymbolic = MatLUFactorSymbolic_KLU;
297   B->ops->destroy          = MatDestroy_KLU;
298   B->ops->view             = MatView_KLU;
299 
300   PetscCall(PetscObjectComposeFunction((PetscObject)B,"MatFactorGetSolverType_C",MatFactorGetSolverType_seqaij_klu));
301 
302   B->factortype   = MAT_FACTOR_LU;
303   B->assembled    = PETSC_TRUE;           /* required by -ksp_view */
304   B->preallocated = PETSC_TRUE;
305 
306   PetscCall(PetscFree(B->solvertype));
307   PetscCall(PetscStrallocpy(MATSOLVERKLU,&B->solvertype));
308   B->canuseordering = PETSC_TRUE;
309   PetscCall(PetscStrallocpy(MATORDERINGEXTERNAL,(char**)&B->preferredordering[MAT_FACTOR_LU]));
310 
311   /* initializations */
312   /* ------------------------------------------------*/
313   /* get the default control parameters */
314   status = klu_K_defaults(&lu->Common);
315   PetscCheck(status > 0,PETSC_COMM_SELF,PETSC_ERR_LIB,"KLU Initialization failed");
316 
317   lu->Common.scale = 0; /* No row scaling */
318 
319   PetscOptionsBegin(PetscObjectComm((PetscObject)B),((PetscObject)B)->prefix,"KLU Options","Mat");
320   /* Partial pivoting tolerance */
321   PetscCall(PetscOptionsReal("-mat_klu_pivot_tol","Partial pivoting tolerance","None",lu->Common.tol,&lu->Common.tol,NULL));
322   /* BTF pre-ordering */
323   PetscCall(PetscOptionsInt("-mat_klu_use_btf","Enable BTF preordering","None",(PetscInt)lu->Common.btf,(PetscInt*)&lu->Common.btf,NULL));
324   /* Matrix reordering */
325   PetscCall(PetscOptionsEList("-mat_klu_ordering","Internal ordering method","None",KluOrderingTypes,PETSC_STATIC_ARRAY_LENGTH(KluOrderingTypes),KluOrderingTypes[0],&idx,&flg));
326   lu->Common.ordering = (int)idx;
327   /* Matrix row scaling */
328   PetscCall(PetscOptionsEList("-mat_klu_row_scale","Matrix row scaling","None",scale,3,scale[0],&idx,&flg));
329   PetscOptionsEnd();
330   *F = B;
331   PetscFunctionReturn(0);
332 }
333