This file is indexed.

/usr/include/trilinos/ml_aggregate.h is in libtrilinos-ml-dev 12.12.1-5.

This file is owned by root:root, with mode 0o644.

The actual contents of the file can be viewed below.

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
/* ************************************************************************* */
/* See the file COPYRIGHT for a complete copyright notice, contact person    */
/* and disclaimer.                                                           */
/* ************************************************************************* */

/* ************************************************************************* */
/* data structure to hold aggregation information                            */
/* ************************************************************************* */
/* Author        : Charles Tong (LLNL)                                       */
/* Date          : August, 1999                                              */
/* ************************************************************************* */

#ifndef __MLAGGH__
#define __MLAGGH__

#include <stdio.h>
/* #include <stdlib.h> */

#define ML_AGGR_UNCOUPLED    1
#define ML_AGGR_COUPLED      2
#define ML_AGGR_MIS          3
#define ML_AGGR_DD           4
#define ML_AGGR_HYBRIDUC     5
#define ML_AGGR_HYBRIDUM     6
/*MS*/
#define ML_AGGR_METIS        7
#define ML_AGGR_PARMETIS     8
#define ML_AGGR_ZOLTAN       9
#define ML_AGGR_VBMETIS     10  /*mgee*/
#define ML_AGGR_USER        11
/*ms*/

/*cms*/
#define ML_AGG_ANY_MATERIAL -1

#define ML_ZOLTAN_TYPE_RCB               1
#define ML_ZOLTAN_TYPE_HYPERGRAPH        2
#define ML_ZOLTAN_TYPE_FAST_HYPERGRAPH   3

/* ************************************************************************* */
/** definition of the aggregate structure                                     */
/* ------------------------------------------------------------------------- */

typedef struct ML_Aggregate_Struct
{
   int    ML_id;
   double print_flag;
   int    max_coarse_size;             /**<  maximum size of coarsest grid */
   int    ordering;                    /**<  natural, random, graph        */
   int    min_nodes_per_aggregate;     /**<  aggregate size control        */
   int    max_neigh_already_selected;  /**<  complexity control            */
   double threshold;                   /**<  for pruning matrix            */
   double curr_threshold;              /**<  adjusted for levels           */
   double drop_tol_for_smoothing;      /**<  self-explanatory              */
   int    attach_scheme;               /**<  aggregate shape control       */
   double smoothP_damping_factor;      /**<  for prolongator smoother      */
   int   * smoothP_damping_sweeps;     /**<  #prolongator smoother sweeps  */
   int    smoothP_type;                /**<  point, block                  */
   int    coarsen_scheme;              /**<  Uncoupled, Coupled, MIS       */
   int   * coarsen_scheme_level;
   int    num_PDE_eqns;                /**<  block size                    */
   int    nullspace_dim;               /**<  self-explanatory              */
   double *nullspace_vect;             /**<  for null space vectors        */
   int    nullspace_corrupted;         /**< indicates whether fine grid   */
                                       /**<  nullspace has been overwritten*/
   int    *aggr_count;                 /**<  no. aggregates at each level  */
  int    keep_agg_information;         /**<  1: keep aggr_info             */
                                       /**<  otherwise: free it            */
   int    **aggr_info;                 /**<  node to aggregate map         */
   int    max_levels;                  /**<  maximum number of levels      */
   int    begin_level;                 /**<  finest grid level             */
   int    cur_level;                   /**<  temporary variable            */
   double operator_complexity;         /**<  sum of nnz for all A's        */
   double fine_complexity;             /**<  nnz of the finest A           */
   int    nvblocks;                    /**<  for variable blocks (finest)  */
   int    *vblock_info;                /**<  for variable blocks (finest)  */
   int    keep_P_tentative;            /**<  keeping tentative prolongator */
   struct ML_Operator_Struct **P_tentative; /**<  so it can be reused later.*/
   int    smooth_existing_P_tentative; /**<  already have P tent, don't create it*/
  int use_transpose;                   /**<  Used to build restriction by doing */
  int Restriction_smoothagg_transpose; /**<  smoothed aggregation on A^T */
/*MS*/
  void *aggr_options;                  /**<  option about METIS and ParMETIS    */
  void *aggr_viz_and_stats;            /**<  information about the aggregates   */
                                       /**<  only if the user explicitely
                                          < requires them                      */
  void * field_of_values;
  int    block_scaled_SA;              /**<  = 1 indicates that the prolongator
                                          smoother should use block diagonal
                                          scaling (blocksize = num_PDE_eqns) */

  double phase3_agg_creation;          /**<  Steers how the MIS  and Uncoupled
                                          handle phase 3 of aggregation.
                                          Values near 0 create few additional
                                          aggregates.Large values create many
                                          additional aggregates. Convergence
                                          can be improve convergence by new
                                          aggregates but nonzero fill-in
                                          increases on coarse meshes.
                                          Default: .5                        */
  double **nodal_coord;                 /**< Coordinates of fine-grid nodes
					  and aggregates */
  int N_dimensions;
/*ms*/
/*mgee*/
  void  *vblock_data;                 /**< holds data structure aggr_vblock */
  int minimizing_energy, cheap_minimizing_energy;
  double minimizing_energy_droptol;
  double   *old_RowOmegas;
  int    coarsen_rate;               /* used only for semicoarsening */
  int    semicoarsen_levels;         /* number of levels where semicoarsening */
                                     /* will be applied                       */
  char   semicoarsen_coordinate;     /* coordinate direction to be coarsened  */


} ML_Aggregate;

/* ************************************************************************* */
/* other ML include files                                               */
/* ------------------------------------------------------------------------- */

#include "ml_common.h"
#include "ml_defs.h"
#include "ml_comm.h"
#include "ml_memory.h"
#include "ml_operator.h"


/* ************************************************************************* */
/* linked list structures for holding free node information             */
/* ------------------------------------------------------------------------- */

typedef struct ML_Node_Struct
{
   int    node_id;
   struct ML_Node_Struct *next;
} ML_Node;

/* ************************************************************************* */
/* definition of the structure for holding aggregate information        */
/* ------------------------------------------------------------------------- */

typedef struct ML_SuperNode_Struct
{
   int    length;
   int    maxlength;
   int    index;
   int    *list;
   struct ML_SuperNode_Struct *next;

} ML_SuperNode;

/* ************************************************************************* */
/* definition of the structure for holding communication information in */
/* aggregation procedure                                                */
/* ------------------------------------------------------------------------- */

typedef struct ML_Aggregate_Comm_Struct
{
   int     N_send_neighbors, N_recv_neighbors;
   int     local_nrows;
   int     *send_neighbors, *recv_neighbors;
   int     *send_leng, *recv_leng;
   int     *send_list;
   ML_Comm *comm;

} ML_Aggregate_Comm;

typedef struct ML_agg_indx_comm_struct {
  int N_neighbors;
  int *temp_leng, *send_leng, *recv_leng, *send_list, *recv_list, *tem2_index,
    *neighbors, *temp_index;
} ML_agg_indx_comm;

/* ************************************************************************* */
/* ************************************************************************* */
/* functions to manipulate the aggregate data structure                      */
/* ------------------------------------------------------------------------- */

#ifndef ML_CPP
#ifdef __cplusplus
extern "C"
{
#endif
#endif

/* ------------------------------------------------------------------------- */
/* constructor/destructor and level control                                  */
/* ------------------------------------------------------------------------- */

int ML_Aggregate_Create( ML_Aggregate ** );
int ML_Aggregate_Destroy( ML_Aggregate ** );

int ML_Aggregate_Set_OutputLevel( ML_Aggregate *, double level );
int ML_Aggregate_Set_Reuse(ML_Aggregate *ag);

int ML_Aggregate_Set_MaxLevels( ML_Aggregate *, int level );
int ML_Aggregate_Set_CurrentLevel( ML_Aggregate *, int level );
int ML_Aggregate_Set_StartLevel( ML_Aggregate *, int level );

/* ------------------------------------------------------------------------- */
/* aggregate size and shape control                                          */
/* ------------------------------------------------------------------------- */

int ML_Aggregate_Set_MinNodesPerAggregate(ML_Aggregate *,int n);
int ML_Aggregate_Set_AttachScheme_MaxLink( ML_Aggregate * );
int ML_Aggregate_Set_AttachScheme_MinRank( ML_Aggregate * );
int ML_Aggregate_Set_Phase3AggregateCreationAggressiveness(
				   ML_Aggregate *ag, double factor);

/* ------------------------------------------------------------------------- */
/* aggregation node traversal control                                        */
/* ------------------------------------------------------------------------- */

int ML_Aggregate_Set_NaturalOrdering( ML_Aggregate * );
int ML_Aggregate_Set_RandomOrdering( ML_Aggregate * );
int ML_Aggregate_Set_GraphOrdering( ML_Aggregate * );

/* ------------------------------------------------------------------------- */
/* control when to stop coarsening                                           */
/* ------------------------------------------------------------------------- */

int ML_Aggregate_Set_MaxCoarseSize( ML_Aggregate *, int size );

/* ------------------------------------------------------------------------- */
/* different parallel coarsening schemes                                     */
/* ------------------------------------------------------------------------- */

int ML_Aggregate_Set_CoarsenScheme_Uncoupled( ML_Aggregate * );
int ML_Aggregate_Set_CoarsenScheme_Coupled( ML_Aggregate * );
int ML_Aggregate_Set_CoarsenScheme_MIS( ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenScheme_DD( ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenScheme_UncoupledMIS( ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenScheme_UncoupledCoupled( ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenScheme_METIS( ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenScheme_VBMETIS( ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenScheme_ParMETIS( ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenScheme_Zoltan( ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenScheme_User( ML_Aggregate *ag  );
int ML_Aggregate_Phase2_3_Cleanup(ML_Aggregate *ml_ag, ML_Operator *Amatrix,
				  int *aggr_count, int nvertices,
				  int *aggr_index, int exp_Nrows,
				  ML_Comm *comm, char *input_bdry,char *label,
                                  ML_agg_indx_comm *);
int ML_Aggregate_Set_CoarsenSchemeLevel( int level, int, ML_Aggregate *ag,
					 int choice );
int ML_Aggregate_Set_CoarsenSchemeLevel_Coupled( int level, int, ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenSchemeLevel_Uncoupled( int level, int, ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenSchemeLevel_UncoupledMIS( int level, int, ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenSchemeLevel_MIS( int level, int, ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenSchemeLevel_METIS( int level, int, ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenSchemeLevel_ParMETIS( int level, int, ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenSchemeLevel_Zoltan( int level, int, ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenSchemeLevel_User( int level, int, ML_Aggregate *ag  );
int ML_Aggregate_Set_Vblocks_CoarsenScheme_VBMETIS( ML_Aggregate *ag,const int level,
                                                    const int N_levels,const int nblocks,
                                                    const int *blocks,const int *block_pde,
                                                    const int block_dim);

/* ------------------------------------------------------------------------- */
/* set threshold for pruning matrix graph                                    */
/* ------------------------------------------------------------------------- */

int ML_Aggregate_Set_Threshold( ML_Aggregate *, double epsilon );
int ML_Aggregate_Reset_Threshold( ML_Aggregate * );

/* ------------------------------------------------------------------------- */
/* whether to smooth existing tentative prolongator                          */
/* ------------------------------------------------------------------------- */

extern int ML_Aggregate_Set_Flag_SmoothExistingTentativeP( ML_Aggregate *, int);
extern int ML_Aggregate_Get_Flag_SmoothExistingTentativeP( ML_Aggregate *);

/* ------------------------------------------------------------------------- */
/* damping factor for prolongator smoother                                   */
/* ------------------------------------------------------------------------- */

int ML_Aggregate_Set_DampingFactor( ML_Aggregate *, double factor );
int ML_Aggregate_Set_PSmootherType( ML_Aggregate *, int stype );
int ML_Aggregate_Set_PointDiagScaling( ML_Aggregate *ag);
int ML_Aggregate_Set_BlockDiagScaling( ML_Aggregate *ag);
int ML_Aggregate_Set_DampingSweeps( ML_Aggregate *ag, int numSweeps, int level );
int ML_Aggregate_Get_DampingSweeps( ML_Aggregate *ag, int level);

/* ------------------------------------------------------------------------- */
/* accessing aggregation information                                         */
/* ------------------------------------------------------------------------- */

int ML_Aggregate_Get_AggrCount( ML_Aggregate *, int level );
int ML_Aggregate_Get_AggrMap( ML_Aggregate *, int level, int**);
extern int ML_Gen_Blocks_Aggregates(ML_Aggregate *ag, int level,
                                    int *nblocks, int **block_list);


/* ------------------------------------------------------------------------- */
/* set null space for the finest grid                                        */
/* ------------------------------------------------------------------------- */

int ML_Aggregate_Set_NullSpace(ML_Aggregate *, int, int, double *, int);
int ML_Aggregate_Scale_NullSpace(ML_Aggregate *ag, double *scale_vect,
				 int length);

int ML_Aggregate_Coarsen( ML_Aggregate *, ML_Operator *A,
                          ML_Operator **P, ML_Comm *comm );

/* ------------------------------------------------------------------------- */
/* functions for performing aggregation                                      */
/* ------------------------------------------------------------------------- */

int ML_Aggregate_CoarsenMIS(ML_Aggregate *ml_ag,ML_Operator *Amatrix,
                            ML_Operator **Pmatrix, ML_Comm *comm);

int ML_Aggregate_CoarsenDomainDecomp(ML_Aggregate *ml_ag,
           ML_Operator *Amatrix, ML_Operator **Pmatrix, ML_Comm *comm);

int ML_Aggregate_CoarsenUncoupled(ML_Aggregate *ml_ag,
           ML_Operator *Amatrix, ML_Operator **Pmatrix, ML_Comm *comm);

int ML_Aggregate_CoarsenCoupled(ML_Aggregate *ml_ag,
           ML_Operator *Amatrix, ML_Operator **Pmatrix, ML_Comm *comm);

int ML_Aggregate_CoarsenMETIS(ML_Aggregate *ml_ag,
           ML_Operator *Amatrix, ML_Operator **Pmatrix, ML_Comm *comm);

int ML_Aggregate_CoarsenParMETIS(ML_Aggregate *ml_ag,
           ML_Operator *Amatrix, ML_Operator **Pmatrix, ML_Comm *comm);

  /* ------------------------------------------------------------------------- */
/* miscellaneous local functions                                             */
/* ------------------------------------------------------------------------- */
int ML_Aggregate_KeepInfo(ML_Aggregate *ag, int value);
int ML_Aggregate_LabelVertices(int, int *, int, char *,char *,int,
              int *, int *, int, int **, int, int **, int *, int *,
              int, int **, int *, int *, int **, int, ML_Comm *, int *);

int ML_Aggregate_UpdateVertexStates(int N_remaining_vertices,
              char vertex_state[], int recv_cnt, int recv_proc[],
              int recv_leng[], int **recv_buf, int **recv_list,
              int proc_flag[], int *NremainingRcvProcs, int send_cnt,
              int send_proc[], int send_leng[], int **send_buf,
              int *send_flag, USR_REQ *Request, ML_Comm *comm, int);

int ML_Aggregate_ExchangeBdry(double *data, void *comm);

int ML_Aggregate_ExchangeData(char *rbuf, char *sbuf, int N_neigh,
                              int *neigh, int *rleng, int *sleng,
                              int msgid, int datatype, ML_Comm *comm);

int ML_aggr_index_communicate(int N_neighbors, int temp_leng[], int send_leng[],
			   int recv_leng[], int send_list[], int recv_list[],
			   int nvertices,
			   ML_Comm *comm, int aggr_index[], int msgtype,
			   int tem2_index[], int neighbors[], int temp_index[],int);




int ML_Aggregate_Print(ML_Aggregate *);
int ML_Aggregate_Print_Complexity(ML_Aggregate *);

void ML_CSR_MSR_ML_memorydata_Destroy(void *data);


/* ************************************************************************* */
/* internal function defined later on in this file                           */
/* ------------------------------------------------------------------------- */

extern int ML_modified_matvec(void *Amat_in, int, double *,int , double *,int);
extern int ML_random_global_subset(ML_Operator *Amat, double reduction,
                                   int **list, int *length, int num_PDE_eqns);
extern int ML_repartition_matrix(ML_Operator *mat, ML_Operator **new_mat,
                                 ML_Operator **permutation,
                                 ML_Operator **permt, int num_PDE_eqns,int,
                                 double *, double *, double *, int,
                                 ML_Partitioner);
extern ML_Operator** ML_repartition_Acoarse(ML *ml, int fine, int coarse,
                 ML_Aggregate *ag, int R_is_Ptranspose, int ReturnPerm);



int ML_Aggregate_Compress_Matrix(ML_GetrowFunc *getrow_obj, int *mat_indx,
           int num_PDEs, ML_Comm *comm, int **new_mat_indx,
           int *N_neighbors, int **neighbors, int **recv_leng,
           int **send_leng, int **send_list, int **recv_list, int *bc_array);
int ML_Aggregate_CoarsenCoupledCore(ML_Aggregate *, ML_Comm *comm,
           int *amal_mat_indx, int *aggr_count, int **aggr_index2,
           int N_neighbors, int *neighbors, int *recv_leng, int *send_leng,
           int *send_list,int *,int **, int *bc_array);
int  ML_Aggregate_ExchangeStatus(char *recvbuf, char *sendbuf,
           int N_neighbors, int *neighbors, int *recv_leng,
           int *send_leng, int *recv_list, int Nrows, int msgid,
           int datatype, ML_Comm *comm);
int ML_Aggregate_ComposeExpandedCommInfo(ML_GetrowFunc *getrow_obj,
           int num_PDEs, ML_Comm *comm,
           int *new_N_neighbors, int **new_neighbors, int **new_recv_leng,
           int **new_send_leng, int **new_send_list, int **new_recv_list);
int ML_Aggregate_ComposeRecvFromSend(int nprocs, int mypid, int new_N_send,
           int *new_send_leng, int *new_send_neighbors, int *N_rcv,
           int **recv_leng, int **recv_neighbors, ML_Comm *comm);
int ML_Aggregate_Form_Aggregates(int phaseID, int phaseAFlag, int Nrows,
           int *mat_indx, int *aggr_index, int *aggr_stat,
           int *node_type, int *node_type2, int *order_array,
           int *aggr_count_in, int *aggr_cnt_leng_in,
           int **aggr_cnt_array_in, int max_row_nnz, int min_agg_size,
           int max_neigh_selected, int N_neighbors, int *neighbors,
           int *send_leng, int *send_list, int *recv_leng, int *recv_list,
           int *sendlist_proc, ML_Comm *comm, double printflag);
int ML_Aggregate_PutInto_Aggregates(char phaseID, int attach_scheme,
           int *mat_indx, int *aggr_index, int *aggr_stat,
           int *aggr_count_in, int **aggr_cnt_array_in,
           int N_neighbors, int *neighbors, int *send_leng, int *send_list,
           int *recv_leng, int *recv_list, ML_Comm *comm, double printflag);
int ML_Graph_CreateFromMatrix(ML_Aggregate *ml_ag, ML_Operator *Amatrix,
           int **mat_indx_out, ML_Comm *comm, double epsilon, int *nrows,
           int **bc_array);

int  ML_Aggregate_CoarsenUncoupledCore(ML_Aggregate *, ML_Comm *,
                ML_Operator *, int *mat_indx, int *bdry_array,
                int *aggr_count_in, int **aggr_index_in, char *true_bdry);
int ML_Aggregate_Set_CoarsenScheme_METIS( ML_Aggregate *ag  );
int ML_Aggregate_Set_CoarsenScheme_ParMETIS( ML_Aggregate *ag  );

int ML_Aggregate_Set_SmoothRestrictionWithA( ML_Aggregate *ag );
int ML_Aggregate_Set_SmoothRestrictionWithAT( ML_Aggregate *ag );

extern int ML_Aggregate_Set_NodalCoordinates(ML* ml, ML_Aggregate *ag, double *ptr);
extern int ML_Aggregate_Set_Dimensions(ML_Aggregate *ag, int N_dimensions);
extern int ML_Aggregate_Get_CoarsenScheme( ML_Aggregate *ag  );

/* ------------------------------------------------------------------------- */
/* functions for visualization                                               */
/* ------------------------------------------------------------------------- */

extern int ML_Aggregate_VizAndStats_Setup( ML *ml);
extern int ML_Aggregate_VizAndStats_Clean( ML *ml);
#ifndef ML_CPP
#ifdef __cplusplus
}
#endif
#endif

#endif