xref: /petsc/src/tao/leastsquares/tests/chwirut2.c (revision 503c0ea9b45bcfbcebbb1ea5341243bbc69f0bea)
1 /*
2    Include "petsctao.h" so that we can use TAO solvers.  Note that this
3    file automatically includes libraries such as:
4      petsc.h       - base PETSc routines   petscvec.h - vectors
5      petscsys.h    - system routines        petscmat.h - matrices
6      petscis.h     - index sets            petscksp.h - Krylov subspace methods
7      petscviewer.h - viewers               petscpc.h  - preconditioners
8 
9  This version tests correlated terms using both vector and listed forms
10 */
11 
12 #include <petsctao.h>
13 
14 /*
15 Description:   These data are the result of a NIST study involving
16                ultrasonic calibration.  The response variable is
17                ultrasonic response, and the predictor variable is
18                metal distance.
19 
20 Reference:     Chwirut, D., NIST (197?).
21                Ultrasonic Reference Block Study.
22 */
23 
24 static char help[]="Finds the nonlinear least-squares solution to the model \n\
25             y = exp[-b1*x]/(b2+b3*x)  +  e \n";
26 
27 /*T
28    Concepts: TAO^Solving a system of nonlinear equations, nonlinear least squares
29    Routines: TaoCreate();
30    Routines: TaoSetType();
31    Routines: TaoSetResidualRoutine();
32    Routines: TaoSetJacobianRoutine();
33    Routines: TaoSetSolution();
34    Routines: TaoSetFromOptions();
35    Routines: TaoSetConvergenceHistory(); TaoGetConvergenceHistory();
36    Routines: TaoSolve();
37    Routines: TaoView(); TaoDestroy();
38    Processors: 1
39 T*/
40 
41 #define NOBSERVATIONS 214
42 #define NPARAMETERS 3
43 
44 /* User-defined application context */
45 typedef struct {
46   /* Working space */
47   PetscReal t[NOBSERVATIONS];   /* array of independent variables of observation */
48   PetscReal y[NOBSERVATIONS];   /* array of dependent variables */
49   PetscReal j[NOBSERVATIONS][NPARAMETERS]; /* dense jacobian matrix array*/
50   PetscInt idm[NOBSERVATIONS];  /* Matrix indices for jacobian */
51   PetscInt idn[NPARAMETERS];
52 } AppCtx;
53 
54 /* User provided Routines */
55 PetscErrorCode InitializeData(AppCtx *user);
56 PetscErrorCode FormStartingPoint(Vec);
57 PetscErrorCode EvaluateFunction(Tao, Vec, Vec, void *);
58 PetscErrorCode EvaluateJacobian(Tao, Vec, Mat, Mat, void *);
59 
60 /*--------------------------------------------------------------------*/
61 int main(int argc,char **argv)
62 {
63   PetscInt       wtype=0;
64   Vec            x, f;               /* solution, function */
65   Vec            w;                  /* weights */
66   Mat            J;                  /* Jacobian matrix */
67   Tao            tao;                /* Tao solver context */
68   PetscInt       i;               /* iteration information */
69   PetscReal      hist[100],resid[100];
70   PetscInt       lits[100];
71   PetscInt       w_row[NOBSERVATIONS]; /* explicit weights */
72   PetscInt       w_col[NOBSERVATIONS];
73   PetscReal      w_vals[NOBSERVATIONS];
74   PetscBool      flg;
75   AppCtx         user;               /* user-defined work context */
76 
77   PetscCall(PetscInitialize(&argc,&argv,(char *)0,help));
78   PetscCall(PetscOptionsGetInt(NULL,NULL,"-wtype",&wtype,&flg));
79   PetscCall(PetscPrintf(PETSC_COMM_WORLD,"wtype=%d\n",wtype));
80   /* Allocate vectors */
81   PetscCall(VecCreateSeq(MPI_COMM_SELF,NPARAMETERS,&x));
82   PetscCall(VecCreateSeq(MPI_COMM_SELF,NOBSERVATIONS,&f));
83 
84   PetscCall(VecDuplicate(f,&w));
85 
86   /* no correlation, but set in different ways */
87   PetscCall(VecSet(w,1.0));
88   for (i=0;i<NOBSERVATIONS;i++) {
89     w_row[i]=i; w_col[i]=i; w_vals[i]=1.0;
90   }
91 
92   /* Create the Jacobian matrix. */
93   PetscCall(MatCreateSeqDense(MPI_COMM_SELF,NOBSERVATIONS,NPARAMETERS,NULL,&J));
94 
95   for (i=0;i<NOBSERVATIONS;i++) user.idm[i] = i;
96 
97   for (i=0;i<NPARAMETERS;i++) user.idn[i] = i;
98 
99   /* Create TAO solver and set desired solution method */
100   PetscCall(TaoCreate(PETSC_COMM_SELF,&tao));
101   PetscCall(TaoSetType(tao,TAOPOUNDERS));
102 
103  /* Set the function and Jacobian routines. */
104   PetscCall(InitializeData(&user));
105   PetscCall(FormStartingPoint(x));
106   PetscCall(TaoSetSolution(tao,x));
107   PetscCall(TaoSetResidualRoutine(tao,f,EvaluateFunction,(void*)&user));
108   if (wtype == 1) {
109     PetscCall(TaoSetResidualWeights(tao,w,0,NULL,NULL,NULL));
110   } else if (wtype == 2) {
111     PetscCall(TaoSetResidualWeights(tao,NULL,NOBSERVATIONS,w_row,w_col,w_vals));
112   }
113   PetscCall(TaoSetJacobianResidualRoutine(tao, J, J, EvaluateJacobian, (void*)&user));
114   PetscCall(TaoSetTolerances(tao,1e-5,0.0,PETSC_DEFAULT));
115 
116   /* Check for any TAO command line arguments */
117   PetscCall(TaoSetFromOptions(tao));
118 
119   PetscCall(TaoSetConvergenceHistory(tao,hist,resid,0,lits,100,PETSC_TRUE));
120   /* Perform the Solve */
121   PetscCall(TaoSolve(tao));
122 
123   /* Free TAO data structures */
124   PetscCall(TaoDestroy(&tao));
125 
126    /* Free PETSc data structures */
127   PetscCall(VecDestroy(&x));
128   PetscCall(VecDestroy(&w));
129   PetscCall(VecDestroy(&f));
130   PetscCall(MatDestroy(&J));
131 
132   PetscCall(PetscFinalize());
133   return 0;
134 }
135 
136 /*--------------------------------------------------------------------*/
137 PetscErrorCode EvaluateFunction(Tao tao, Vec X, Vec F, void *ptr)
138 {
139   AppCtx          *user = (AppCtx *)ptr;
140   PetscInt        i;
141   PetscReal       *y=user->y,*f,*t=user->t;
142   const PetscReal *x;
143 
144   PetscFunctionBegin;
145   PetscCall(VecGetArrayRead(X,&x));
146   PetscCall(VecGetArray(F,&f));
147 
148   for (i=0;i<NOBSERVATIONS;i++) {
149     f[i] = y[i] - PetscExpScalar(-x[0]*t[i])/(x[1] + x[2]*t[i]);
150   }
151   PetscCall(VecRestoreArrayRead(X,&x));
152   PetscCall(VecRestoreArray(F,&f));
153   PetscLogFlops(6*NOBSERVATIONS);
154   PetscFunctionReturn(0);
155 }
156 
157 /*------------------------------------------------------------*/
158 /* J[i][j] = df[i]/dt[j] */
159 PetscErrorCode EvaluateJacobian(Tao tao, Vec X, Mat J, Mat Jpre, void *ptr)
160 {
161   AppCtx          *user = (AppCtx *)ptr;
162   PetscInt        i;
163   PetscReal       *t=user->t;
164   const PetscReal *x;
165   PetscReal       base;
166 
167   PetscFunctionBegin;
168   PetscCall(VecGetArrayRead(X,&x));
169   for (i=0;i<NOBSERVATIONS;i++) {
170     base = PetscExpScalar(-x[0]*t[i])/(x[1] + x[2]*t[i]);
171 
172     user->j[i][0] = t[i]*base;
173     user->j[i][1] = base/(x[1] + x[2]*t[i]);
174     user->j[i][2] = base*t[i]/(x[1] + x[2]*t[i]);
175   }
176 
177   /* Assemble the matrix */
178   PetscCall(MatSetValues(J,NOBSERVATIONS,user->idm, NPARAMETERS, user->idn,(PetscReal *)user->j,INSERT_VALUES));
179   PetscCall(MatAssemblyBegin(J,MAT_FINAL_ASSEMBLY));
180   PetscCall(MatAssemblyEnd(J,MAT_FINAL_ASSEMBLY));
181 
182   PetscCall(VecRestoreArrayRead(X,&x));
183   PetscLogFlops(NOBSERVATIONS * 13);
184   PetscFunctionReturn(0);
185 }
186 
187 /* ------------------------------------------------------------ */
188 PetscErrorCode FormStartingPoint(Vec X)
189 {
190   PetscReal      *x;
191 
192   PetscFunctionBegin;
193   PetscCall(VecGetArray(X,&x));
194   x[0] = 1.19;
195   x[1] = -1.86;
196   x[2] = 1.08;
197   PetscCall(VecRestoreArray(X,&x));
198   PetscFunctionReturn(0);
199 }
200 
201 /* ---------------------------------------------------------------------- */
202 PetscErrorCode InitializeData(AppCtx *user)
203 {
204   PetscReal *t=user->t,*y=user->y;
205   PetscInt  i=0;
206 
207   PetscFunctionBegin;
208   y[i] =   92.9000;   t[i++] =  0.5000;
209   y[i] =    78.7000;  t[i++] =   0.6250;
210   y[i] =    64.2000;  t[i++] =   0.7500;
211   y[i] =    64.9000;  t[i++] =   0.8750;
212   y[i] =    57.1000;  t[i++] =   1.0000;
213   y[i] =    43.3000;  t[i++] =   1.2500;
214   y[i] =    31.1000;   t[i++] =  1.7500;
215   y[i] =    23.6000;   t[i++] =  2.2500;
216   y[i] =    31.0500;   t[i++] =  1.7500;
217   y[i] =    23.7750;   t[i++] =  2.2500;
218   y[i] =    17.7375;   t[i++] =  2.7500;
219   y[i] =    13.8000;   t[i++] =  3.2500;
220   y[i] =    11.5875;   t[i++] =  3.7500;
221   y[i] =     9.4125;   t[i++] =  4.2500;
222   y[i] =     7.7250;   t[i++] =  4.7500;
223   y[i] =     7.3500;   t[i++] =  5.2500;
224   y[i] =     8.0250;   t[i++] =  5.7500;
225   y[i] =    90.6000;   t[i++] =  0.5000;
226   y[i] =    76.9000;   t[i++] =  0.6250;
227   y[i] =    71.6000;   t[i++] = 0.7500;
228   y[i] =    63.6000;   t[i++] =  0.8750;
229   y[i] =    54.0000;   t[i++] =  1.0000;
230   y[i] =    39.2000;   t[i++] =  1.2500;
231   y[i] =    29.3000;   t[i++] = 1.7500;
232   y[i] =    21.4000;   t[i++] =  2.2500;
233   y[i] =    29.1750;   t[i++] =  1.7500;
234   y[i] =    22.1250;   t[i++] =  2.2500;
235   y[i] =    17.5125;   t[i++] =  2.7500;
236   y[i] =    14.2500;   t[i++] =  3.2500;
237   y[i] =     9.4500;   t[i++] =  3.7500;
238   y[i] =     9.1500;   t[i++] =  4.2500;
239   y[i] =     7.9125;   t[i++] =  4.7500;
240   y[i] =     8.4750;   t[i++] =  5.2500;
241   y[i] =     6.1125;   t[i++] =  5.7500;
242   y[i] =    80.0000;   t[i++] =  0.5000;
243   y[i] =    79.0000;   t[i++] =  0.6250;
244   y[i] =    63.8000;   t[i++] =  0.7500;
245   y[i] =    57.2000;   t[i++] =  0.8750;
246   y[i] =    53.2000;   t[i++] =  1.0000;
247   y[i] =   42.5000;   t[i++] =  1.2500;
248   y[i] =   26.8000;   t[i++] =  1.7500;
249   y[i] =    20.4000;   t[i++] =  2.2500;
250   y[i] =    26.8500;  t[i++] =   1.7500;
251   y[i] =    21.0000;  t[i++] =   2.2500;
252   y[i] =    16.4625;  t[i++] =   2.7500;
253   y[i] =    12.5250;  t[i++] =   3.2500;
254   y[i] =    10.5375;  t[i++] =   3.7500;
255   y[i] =     8.5875;  t[i++] =   4.2500;
256   y[i] =     7.1250;  t[i++] =   4.7500;
257   y[i] =     6.1125;  t[i++] =   5.2500;
258   y[i] =     5.9625;  t[i++] =   5.7500;
259   y[i] =    74.1000;  t[i++] =   0.5000;
260   y[i] =    67.3000;  t[i++] =   0.6250;
261   y[i] =    60.8000;  t[i++] =   0.7500;
262   y[i] =    55.5000;  t[i++] =   0.8750;
263   y[i] =    50.3000;  t[i++] =   1.0000;
264   y[i] =    41.0000;  t[i++] =   1.2500;
265   y[i] =    29.4000;  t[i++] =   1.7500;
266   y[i] =    20.4000;  t[i++] =   2.2500;
267   y[i] =    29.3625;  t[i++] =   1.7500;
268   y[i] =    21.1500;  t[i++] =   2.2500;
269   y[i] =    16.7625;  t[i++] =   2.7500;
270   y[i] =    13.2000;  t[i++] =   3.2500;
271   y[i] =    10.8750;  t[i++] =   3.7500;
272   y[i] =     8.1750;  t[i++] =   4.2500;
273   y[i] =     7.3500;  t[i++] =   4.7500;
274   y[i] =     5.9625;  t[i++] =  5.2500;
275   y[i] =     5.6250;  t[i++] =   5.7500;
276   y[i] =    81.5000;  t[i++] =    .5000;
277   y[i] =    62.4000;  t[i++] =    .7500;
278   y[i] =    32.5000;  t[i++] =   1.5000;
279   y[i] =    12.4100;  t[i++] =   3.0000;
280   y[i] =    13.1200;  t[i++] =   3.0000;
281   y[i] =    15.5600;  t[i++] =   3.0000;
282   y[i] =     5.6300;  t[i++] =   6.0000;
283   y[i] =    78.0000;   t[i++] =   .5000;
284   y[i] =    59.9000;  t[i++] =    .7500;
285   y[i] =    33.2000;  t[i++] =   1.5000;
286   y[i] =    13.8400;  t[i++] =   3.0000;
287   y[i] =    12.7500;  t[i++] =   3.0000;
288   y[i] =    14.6200;  t[i++] =   3.0000;
289   y[i] =     3.9400;  t[i++] =   6.0000;
290   y[i] =    76.8000;  t[i++] =    .5000;
291   y[i] =    61.0000;  t[i++] =    .7500;
292   y[i] =    32.9000;  t[i++] =   1.5000;
293   y[i] =   13.8700;   t[i++] = 3.0000;
294   y[i] =    11.8100;  t[i++] =   3.0000;
295   y[i] =    13.3100;  t[i++] =   3.0000;
296   y[i] =     5.4400;  t[i++] =   6.0000;
297   y[i] =    78.0000;  t[i++] =    .5000;
298   y[i] =    63.5000;  t[i++] =    .7500;
299   y[i] =    33.8000;  t[i++] =   1.5000;
300   y[i] =    12.5600;  t[i++] =   3.0000;
301   y[i] =     5.6300;  t[i++] =   6.0000;
302   y[i] =    12.7500;  t[i++] =   3.0000;
303   y[i] =    13.1200;  t[i++] =   3.0000;
304   y[i] =     5.4400;  t[i++] =   6.0000;
305   y[i] =    76.8000;  t[i++] =    .5000;
306   y[i] =    60.0000;  t[i++] =    .7500;
307   y[i] =    47.8000;  t[i++] =   1.0000;
308   y[i] =    32.0000;  t[i++] =   1.5000;
309   y[i] =    22.2000;  t[i++] =   2.0000;
310   y[i] =    22.5700;  t[i++] =   2.0000;
311   y[i] =    18.8200;  t[i++] =   2.5000;
312   y[i] =    13.9500;  t[i++] =   3.0000;
313   y[i] =    11.2500;  t[i++] =   4.0000;
314   y[i] =     9.0000;  t[i++] =   5.0000;
315   y[i] =     6.6700;  t[i++] =   6.0000;
316   y[i] =    75.8000;  t[i++] =    .5000;
317   y[i] =    62.0000;  t[i++] =    .7500;
318   y[i] =    48.8000;  t[i++] =   1.0000;
319   y[i] =    35.2000;  t[i++] =   1.5000;
320   y[i] =    20.0000;  t[i++] =   2.0000;
321   y[i] =    20.3200;  t[i++] =   2.0000;
322   y[i] =    19.3100;  t[i++] =   2.5000;
323   y[i] =    12.7500;  t[i++] =   3.0000;
324   y[i] =    10.4200;  t[i++] =   4.0000;
325   y[i] =     7.3100;  t[i++] =   5.0000;
326   y[i] =     7.4200;  t[i++] =   6.0000;
327   y[i] =    70.5000;  t[i++] =    .5000;
328   y[i] =    59.5000;  t[i++] =    .7500;
329   y[i] =    48.5000;  t[i++] =   1.0000;
330   y[i] =    35.8000;  t[i++] =   1.5000;
331   y[i] =    21.0000;  t[i++] =   2.0000;
332   y[i] =    21.6700;  t[i++] =   2.0000;
333   y[i] =    21.0000;  t[i++] =   2.5000;
334   y[i] =    15.6400;  t[i++] =   3.0000;
335   y[i] =     8.1700;  t[i++] =   4.0000;
336   y[i] =     8.5500;  t[i++] =   5.0000;
337   y[i] =    10.1200;  t[i++] =   6.0000;
338   y[i] =    78.0000;  t[i++] =    .5000;
339   y[i] =    66.0000;  t[i++] =    .6250;
340   y[i] =    62.0000;  t[i++] =    .7500;
341   y[i] =    58.0000;  t[i++] =    .8750;
342   y[i] =    47.7000;  t[i++] =   1.0000;
343   y[i] =    37.8000;  t[i++] =   1.2500;
344   y[i] =    20.2000;  t[i++] =   2.2500;
345   y[i] =    21.0700;  t[i++] =   2.2500;
346   y[i] =    13.8700;  t[i++] =   2.7500;
347   y[i] =     9.6700;  t[i++] =   3.2500;
348   y[i] =     7.7600;  t[i++] =   3.7500;
349   y[i] =    5.4400;   t[i++] =  4.2500;
350   y[i] =    4.8700;   t[i++] =  4.7500;
351   y[i] =     4.0100;  t[i++] =   5.2500;
352   y[i] =     3.7500;  t[i++] =   5.7500;
353   y[i] =    24.1900;  t[i++] =   3.0000;
354   y[i] =    25.7600;  t[i++] =   3.0000;
355   y[i] =    18.0700;  t[i++] =   3.0000;
356   y[i] =    11.8100;  t[i++] =   3.0000;
357   y[i] =    12.0700;  t[i++] =   3.0000;
358   y[i] =    16.1200;  t[i++] =   3.0000;
359   y[i] =    70.8000;  t[i++] =    .5000;
360   y[i] =    54.7000;  t[i++] =    .7500;
361   y[i] =    48.0000;  t[i++] =   1.0000;
362   y[i] =    39.8000;  t[i++] =   1.5000;
363   y[i] =    29.8000;  t[i++] =   2.0000;
364   y[i] =    23.7000;  t[i++] =   2.5000;
365   y[i] =    29.6200;  t[i++] =   2.0000;
366   y[i] =    23.8100;  t[i++] =   2.5000;
367   y[i] =    17.7000;  t[i++] =   3.0000;
368   y[i] =    11.5500;  t[i++] =   4.0000;
369   y[i] =    12.0700;  t[i++] =   5.0000;
370   y[i] =     8.7400;  t[i++] =   6.0000;
371   y[i] =    80.7000;  t[i++] =    .5000;
372   y[i] =    61.3000;  t[i++] =    .7500;
373   y[i] =    47.5000;  t[i++] =   1.0000;
374    y[i] =   29.0000;  t[i++] =   1.5000;
375    y[i] =   24.0000;  t[i++] =   2.0000;
376   y[i] =    17.7000;  t[i++] =   2.5000;
377   y[i] =    24.5600;  t[i++] =   2.0000;
378   y[i] =    18.6700;  t[i++] =   2.5000;
379    y[i] =   16.2400;  t[i++] =   3.0000;
380   y[i] =     8.7400;  t[i++] =   4.0000;
381   y[i] =     7.8700;  t[i++] =   5.0000;
382   y[i] =     8.5100;  t[i++] =   6.0000;
383   y[i] =    66.7000;  t[i++] =    .5000;
384   y[i] =    59.2000;  t[i++] =    .7500;
385   y[i] =    40.8000;  t[i++] =   1.0000;
386   y[i] =    30.7000;  t[i++] =   1.5000;
387   y[i] =    25.7000;  t[i++] =   2.0000;
388   y[i] =    16.3000;  t[i++] =   2.5000;
389   y[i] =    25.9900;  t[i++] =   2.0000;
390   y[i] =    16.9500;  t[i++] =   2.5000;
391   y[i] =    13.3500;  t[i++] =   3.0000;
392   y[i] =     8.6200;  t[i++] =   4.0000;
393   y[i] =     7.2000;  t[i++] =   5.0000;
394   y[i] =     6.6400;  t[i++] =   6.0000;
395   y[i] =    13.6900;  t[i++] =   3.0000;
396   y[i] =    81.0000;  t[i++] =    .5000;
397   y[i] =    64.5000;  t[i++] =    .7500;
398   y[i] =    35.5000;  t[i++] =   1.5000;
399    y[i] =   13.3100;  t[i++] =   3.0000;
400   y[i] =     4.8700;  t[i++] =   6.0000;
401   y[i] =    12.9400;  t[i++] =   3.0000;
402   y[i] =     5.0600;  t[i++] =   6.0000;
403   y[i] =    15.1900;  t[i++] =   3.0000;
404   y[i] =    14.6200;  t[i++] =   3.0000;
405   y[i] =    15.6400;  t[i++] =   3.0000;
406   y[i] =    25.5000;  t[i++] =   1.7500;
407   y[i] =    25.9500;  t[i++] =   1.7500;
408   y[i] =    81.7000;  t[i++] =    .5000;
409   y[i] =    61.6000;  t[i++] =    .7500;
410   y[i] =    29.8000;  t[i++] =   1.7500;
411   y[i] =    29.8100;  t[i++] =   1.7500;
412   y[i] =    17.1700;  t[i++] =   2.7500;
413   y[i] =    10.3900;  t[i++] =   3.7500;
414   y[i] =    28.4000;  t[i++] =   1.7500;
415   y[i] =    28.6900;  t[i++] =   1.7500;
416   y[i] =    81.3000;  t[i++] =    .5000;
417   y[i] =    60.9000;  t[i++] =    .7500;
418   y[i] =    16.6500;  t[i++] =   2.7500;
419   y[i] =    10.0500;  t[i++] =   3.7500;
420   y[i] =    28.9000;  t[i++] =   1.7500;
421   y[i] =    28.9500;  t[i++] =   1.7500;
422   PetscFunctionReturn(0);
423 }
424 
425 /*TEST
426 
427      build:
428        requires: !complex
429 
430      test:
431        args:  -tao_smonitor -tao_max_it 100 -tao_type pounders -tao_pounders_delta 0.05 -tao_gatol 1.e-5
432        requires: !single
433        TODO: produces different output for many different systems
434 
435      test:
436        suffix: 2
437        args: -tao_smonitor -tao_max_it 100 -wtype 1 -tao_type pounders -tao_pounders_delta 0.05 -tao_gatol 1.e-5
438        requires: !single
439        TODO: produces different output for many different systems
440 
441      test:
442        suffix: 3
443        args: -tao_smonitor -tao_max_it 100 -wtype 2 -tao_type pounders -tao_pounders_delta 0.05 -tao_gatol 1.e-5
444        requires: !single
445        TODO: produces different output for many different systems
446 
447      test:
448        suffix: 4
449        args: -tao_smonitor -tao_max_it 100 -tao_type pounders -tao_pounders_delta 0.05 -pounders_subsolver_tao_type blmvm -tao_gatol 1.e-5
450        requires: !single
451        TODO: produces different output for many different systems
452 
453  TEST*/
454