Actual source code: chwirut2.c
petsc-3.13.6 2020-09-29
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 - sysem routines petscmat.h - matrices
6: petscis.h - index sets petscksp.h - Krylov subspace methods
7: petscviewer.h - viewers petscpc.h - preconditioners
9: */
11: #include <petsctao.h>
13: /*
14: Description: These data are the result of a NIST study involving
15: ultrasonic calibration. The response variable is
16: ultrasonic response, and the predictor variable is
17: metal distance.
19: Reference: Chwirut, D., NIST (197?).
20: Ultrasonic Reference Block Study.
21: */
23: static char help[]="Finds the nonlinear least-squares solution to the model \n\
24: y = exp[-b1*x]/(b2+b3*x) + e \n";
26: /* T
27: Concepts: TAO^Solving a system of nonlinear equations, nonlinear least squares
28: Routines: TaoCreate();
29: Routines: TaoSetType();
30: Routines: TaoSetResidualRoutine();
31: Routines: TaoSetMonitor();
32: Routines: TaoSetInitialVector();
33: Routines: TaoSetFromOptions();
34: Routines: TaoSolve();
35: Routines: TaoDestroy();
36: Processors: n
37: T*/
39: #define NOBSERVATIONS 214
40: #define NPARAMETERS 3
42: #define DIE_TAG 2000
43: #define IDLE_TAG 1000
45: /* User-defined Section 1.5 Writing Application Codes with PETSc context */
46: typedef struct {
47: /* Working space */
48: PetscReal t[NOBSERVATIONS]; /* array of independent variables of observation */
49: PetscReal y[NOBSERVATIONS]; /* array of dependent variables */
50: PetscMPIInt size,rank;
51: } AppCtx;
53: /* User provided Routines */
54: PetscErrorCode InitializeData(AppCtx *user);
55: PetscErrorCode FormStartingPoint(Vec);
56: PetscErrorCode EvaluateFunction(Tao, Vec, Vec, void *);
57: PetscErrorCode TaskWorker(AppCtx *user);
58: PetscErrorCode StopWorkers(AppCtx *user);
59: PetscErrorCode RunSimulation(PetscReal *x, PetscInt i, PetscReal*f, AppCtx *user);
61: /*--------------------------------------------------------------------*/
62: int main(int argc,char **argv)
63: {
65: Vec x, f; /* solution, function */
66: Tao tao; /* Tao solver context */
67: AppCtx user; /* user-defined work context */
69: /* Initialize TAO and PETSc */
70: PetscInitialize(&argc,&argv,(char *)0,help);if (ierr) return ierr;
71: MPI_Comm_size(MPI_COMM_WORLD,&user.size);
72: MPI_Comm_rank(MPI_COMM_WORLD,&user.rank);
73: InitializeData(&user);
75: /* Run optimization on rank 0 */
76: if (user.rank == 0) {
77: /* Allocate vectors */
78: VecCreateSeq(PETSC_COMM_SELF,NPARAMETERS,&x);
79: VecCreateSeq(PETSC_COMM_SELF,NOBSERVATIONS,&f);
81: /* TAO code begins here */
83: /* Create TAO solver and set desired solution method */
84: TaoCreate(PETSC_COMM_SELF,&tao);
85: TaoSetType(tao,TAOPOUNDERS);
87: /* Set the function and Jacobian routines. */
88: FormStartingPoint(x);
89: TaoSetInitialVector(tao,x);
90: TaoSetResidualRoutine(tao,f,EvaluateFunction,(void*)&user);
92: /* Check for any TAO command line arguments */
93: TaoSetFromOptions(tao);
95: /* Perform the Solve */
96: TaoSolve(tao);
98: /* Free TAO data structures */
99: TaoDestroy(&tao);
101: /* Free PETSc data structures */
102: VecDestroy(&x);
103: VecDestroy(&f);
104: StopWorkers(&user);
105: } else {
106: TaskWorker(&user);
107: }
108: PetscFinalize();
109: return ierr;
110: }
112: /*--------------------------------------------------------------------*/
113: PetscErrorCode EvaluateFunction(Tao tao, Vec X, Vec F, void *ptr)
114: {
115: AppCtx *user = (AppCtx *)ptr;
116: PetscInt i;
117: PetscReal *x,*f;
121: VecGetArray(X,&x);
122: VecGetArray(F,&f);
123: if (user->size == 1) {
124: /* Single processor */
125: for (i=0;i<NOBSERVATIONS;i++) {
126: RunSimulation(x,i,&f[i],user);
127: }
128: } else {
129: /* Multiprocessor master */
130: PetscMPIInt tag;
131: PetscInt finishedtasks,next_task,checkedin;
132: PetscReal f_i=0.0;
133: MPI_Status status;
135: next_task=0;
136: finishedtasks=0;
137: checkedin=0;
139: while(finishedtasks < NOBSERVATIONS || checkedin < user->size-1) {
140: MPI_Recv(&f_i,1,MPIU_REAL,MPI_ANY_SOURCE,MPI_ANY_TAG,PETSC_COMM_WORLD,&status);
141: if (status.MPI_TAG == IDLE_TAG) {
142: checkedin++;
143: } else {
145: tag = status.MPI_TAG;
146: f[tag] = (PetscReal)f_i;
147: finishedtasks++;
148: }
150: if (next_task<NOBSERVATIONS) {
151: MPI_Send(x,NPARAMETERS,MPIU_REAL,status.MPI_SOURCE,next_task,PETSC_COMM_WORLD);
152: next_task++;
154: } else {
155: /* Send idle message */
156: MPI_Send(x,NPARAMETERS,MPIU_REAL,status.MPI_SOURCE,IDLE_TAG,PETSC_COMM_WORLD);
157: }
158: }
159: }
160: VecRestoreArray(X,&x);
161: VecRestoreArray(F,&f);
162: PetscLogFlops(6*NOBSERVATIONS);
163: return(0);
164: }
166: /* ------------------------------------------------------------ */
167: PetscErrorCode FormStartingPoint(Vec X)
168: {
169: PetscReal *x;
173: VecGetArray(X,&x);
174: x[0] = 0.15;
175: x[1] = 0.008;
176: x[2] = 0.010;
177: VecRestoreArray(X,&x);
178: return(0);
179: }
181: /* ---------------------------------------------------------------------- */
182: PetscErrorCode InitializeData(AppCtx *user)
183: {
184: PetscReal *t=user->t,*y=user->y;
185: PetscInt i=0;
188: y[i] = 92.9000; t[i++] = 0.5000;
189: y[i] = 78.7000; t[i++] = 0.6250;
190: y[i] = 64.2000; t[i++] = 0.7500;
191: y[i] = 64.9000; t[i++] = 0.8750;
192: y[i] = 57.1000; t[i++] = 1.0000;
193: y[i] = 43.3000; t[i++] = 1.2500;
194: y[i] = 31.1000; t[i++] = 1.7500;
195: y[i] = 23.6000; t[i++] = 2.2500;
196: y[i] = 31.0500; t[i++] = 1.7500;
197: y[i] = 23.7750; t[i++] = 2.2500;
198: y[i] = 17.7375; t[i++] = 2.7500;
199: y[i] = 13.8000; t[i++] = 3.2500;
200: y[i] = 11.5875; t[i++] = 3.7500;
201: y[i] = 9.4125; t[i++] = 4.2500;
202: y[i] = 7.7250; t[i++] = 4.7500;
203: y[i] = 7.3500; t[i++] = 5.2500;
204: y[i] = 8.0250; t[i++] = 5.7500;
205: y[i] = 90.6000; t[i++] = 0.5000;
206: y[i] = 76.9000; t[i++] = 0.6250;
207: y[i] = 71.6000; t[i++] = 0.7500;
208: y[i] = 63.6000; t[i++] = 0.8750;
209: y[i] = 54.0000; t[i++] = 1.0000;
210: y[i] = 39.2000; t[i++] = 1.2500;
211: y[i] = 29.3000; t[i++] = 1.7500;
212: y[i] = 21.4000; t[i++] = 2.2500;
213: y[i] = 29.1750; t[i++] = 1.7500;
214: y[i] = 22.1250; t[i++] = 2.2500;
215: y[i] = 17.5125; t[i++] = 2.7500;
216: y[i] = 14.2500; t[i++] = 3.2500;
217: y[i] = 9.4500; t[i++] = 3.7500;
218: y[i] = 9.1500; t[i++] = 4.2500;
219: y[i] = 7.9125; t[i++] = 4.7500;
220: y[i] = 8.4750; t[i++] = 5.2500;
221: y[i] = 6.1125; t[i++] = 5.7500;
222: y[i] = 80.0000; t[i++] = 0.5000;
223: y[i] = 79.0000; t[i++] = 0.6250;
224: y[i] = 63.8000; t[i++] = 0.7500;
225: y[i] = 57.2000; t[i++] = 0.8750;
226: y[i] = 53.2000; t[i++] = 1.0000;
227: y[i] = 42.5000; t[i++] = 1.2500;
228: y[i] = 26.8000; t[i++] = 1.7500;
229: y[i] = 20.4000; t[i++] = 2.2500;
230: y[i] = 26.8500; t[i++] = 1.7500;
231: y[i] = 21.0000; t[i++] = 2.2500;
232: y[i] = 16.4625; t[i++] = 2.7500;
233: y[i] = 12.5250; t[i++] = 3.2500;
234: y[i] = 10.5375; t[i++] = 3.7500;
235: y[i] = 8.5875; t[i++] = 4.2500;
236: y[i] = 7.1250; t[i++] = 4.7500;
237: y[i] = 6.1125; t[i++] = 5.2500;
238: y[i] = 5.9625; t[i++] = 5.7500;
239: y[i] = 74.1000; t[i++] = 0.5000;
240: y[i] = 67.3000; t[i++] = 0.6250;
241: y[i] = 60.8000; t[i++] = 0.7500;
242: y[i] = 55.5000; t[i++] = 0.8750;
243: y[i] = 50.3000; t[i++] = 1.0000;
244: y[i] = 41.0000; t[i++] = 1.2500;
245: y[i] = 29.4000; t[i++] = 1.7500;
246: y[i] = 20.4000; t[i++] = 2.2500;
247: y[i] = 29.3625; t[i++] = 1.7500;
248: y[i] = 21.1500; t[i++] = 2.2500;
249: y[i] = 16.7625; t[i++] = 2.7500;
250: y[i] = 13.2000; t[i++] = 3.2500;
251: y[i] = 10.8750; t[i++] = 3.7500;
252: y[i] = 8.1750; t[i++] = 4.2500;
253: y[i] = 7.3500; t[i++] = 4.7500;
254: y[i] = 5.9625; t[i++] = 5.2500;
255: y[i] = 5.6250; t[i++] = 5.7500;
256: y[i] = 81.5000; t[i++] = .5000;
257: y[i] = 62.4000; t[i++] = .7500;
258: y[i] = 32.5000; t[i++] = 1.5000;
259: y[i] = 12.4100; t[i++] = 3.0000;
260: y[i] = 13.1200; t[i++] = 3.0000;
261: y[i] = 15.5600; t[i++] = 3.0000;
262: y[i] = 5.6300; t[i++] = 6.0000;
263: y[i] = 78.0000; t[i++] = .5000;
264: y[i] = 59.9000; t[i++] = .7500;
265: y[i] = 33.2000; t[i++] = 1.5000;
266: y[i] = 13.8400; t[i++] = 3.0000;
267: y[i] = 12.7500; t[i++] = 3.0000;
268: y[i] = 14.6200; t[i++] = 3.0000;
269: y[i] = 3.9400; t[i++] = 6.0000;
270: y[i] = 76.8000; t[i++] = .5000;
271: y[i] = 61.0000; t[i++] = .7500;
272: y[i] = 32.9000; t[i++] = 1.5000;
273: y[i] = 13.8700; t[i++] = 3.0000;
274: y[i] = 11.8100; t[i++] = 3.0000;
275: y[i] = 13.3100; t[i++] = 3.0000;
276: y[i] = 5.4400; t[i++] = 6.0000;
277: y[i] = 78.0000; t[i++] = .5000;
278: y[i] = 63.5000; t[i++] = .7500;
279: y[i] = 33.8000; t[i++] = 1.5000;
280: y[i] = 12.5600; t[i++] = 3.0000;
281: y[i] = 5.6300; t[i++] = 6.0000;
282: y[i] = 12.7500; t[i++] = 3.0000;
283: y[i] = 13.1200; t[i++] = 3.0000;
284: y[i] = 5.4400; t[i++] = 6.0000;
285: y[i] = 76.8000; t[i++] = .5000;
286: y[i] = 60.0000; t[i++] = .7500;
287: y[i] = 47.8000; t[i++] = 1.0000;
288: y[i] = 32.0000; t[i++] = 1.5000;
289: y[i] = 22.2000; t[i++] = 2.0000;
290: y[i] = 22.5700; t[i++] = 2.0000;
291: y[i] = 18.8200; t[i++] = 2.5000;
292: y[i] = 13.9500; t[i++] = 3.0000;
293: y[i] = 11.2500; t[i++] = 4.0000;
294: y[i] = 9.0000; t[i++] = 5.0000;
295: y[i] = 6.6700; t[i++] = 6.0000;
296: y[i] = 75.8000; t[i++] = .5000;
297: y[i] = 62.0000; t[i++] = .7500;
298: y[i] = 48.8000; t[i++] = 1.0000;
299: y[i] = 35.2000; t[i++] = 1.5000;
300: y[i] = 20.0000; t[i++] = 2.0000;
301: y[i] = 20.3200; t[i++] = 2.0000;
302: y[i] = 19.3100; t[i++] = 2.5000;
303: y[i] = 12.7500; t[i++] = 3.0000;
304: y[i] = 10.4200; t[i++] = 4.0000;
305: y[i] = 7.3100; t[i++] = 5.0000;
306: y[i] = 7.4200; t[i++] = 6.0000;
307: y[i] = 70.5000; t[i++] = .5000;
308: y[i] = 59.5000; t[i++] = .7500;
309: y[i] = 48.5000; t[i++] = 1.0000;
310: y[i] = 35.8000; t[i++] = 1.5000;
311: y[i] = 21.0000; t[i++] = 2.0000;
312: y[i] = 21.6700; t[i++] = 2.0000;
313: y[i] = 21.0000; t[i++] = 2.5000;
314: y[i] = 15.6400; t[i++] = 3.0000;
315: y[i] = 8.1700; t[i++] = 4.0000;
316: y[i] = 8.5500; t[i++] = 5.0000;
317: y[i] = 10.1200; t[i++] = 6.0000;
318: y[i] = 78.0000; t[i++] = .5000;
319: y[i] = 66.0000; t[i++] = .6250;
320: y[i] = 62.0000; t[i++] = .7500;
321: y[i] = 58.0000; t[i++] = .8750;
322: y[i] = 47.7000; t[i++] = 1.0000;
323: y[i] = 37.8000; t[i++] = 1.2500;
324: y[i] = 20.2000; t[i++] = 2.2500;
325: y[i] = 21.0700; t[i++] = 2.2500;
326: y[i] = 13.8700; t[i++] = 2.7500;
327: y[i] = 9.6700; t[i++] = 3.2500;
328: y[i] = 7.7600; t[i++] = 3.7500;
329: y[i] = 5.4400; t[i++] = 4.2500;
330: y[i] = 4.8700; t[i++] = 4.7500;
331: y[i] = 4.0100; t[i++] = 5.2500;
332: y[i] = 3.7500; t[i++] = 5.7500;
333: y[i] = 24.1900; t[i++] = 3.0000;
334: y[i] = 25.7600; t[i++] = 3.0000;
335: y[i] = 18.0700; t[i++] = 3.0000;
336: y[i] = 11.8100; t[i++] = 3.0000;
337: y[i] = 12.0700; t[i++] = 3.0000;
338: y[i] = 16.1200; t[i++] = 3.0000;
339: y[i] = 70.8000; t[i++] = .5000;
340: y[i] = 54.7000; t[i++] = .7500;
341: y[i] = 48.0000; t[i++] = 1.0000;
342: y[i] = 39.8000; t[i++] = 1.5000;
343: y[i] = 29.8000; t[i++] = 2.0000;
344: y[i] = 23.7000; t[i++] = 2.5000;
345: y[i] = 29.6200; t[i++] = 2.0000;
346: y[i] = 23.8100; t[i++] = 2.5000;
347: y[i] = 17.7000; t[i++] = 3.0000;
348: y[i] = 11.5500; t[i++] = 4.0000;
349: y[i] = 12.0700; t[i++] = 5.0000;
350: y[i] = 8.7400; t[i++] = 6.0000;
351: y[i] = 80.7000; t[i++] = .5000;
352: y[i] = 61.3000; t[i++] = .7500;
353: y[i] = 47.5000; t[i++] = 1.0000;
354: y[i] = 29.0000; t[i++] = 1.5000;
355: y[i] = 24.0000; t[i++] = 2.0000;
356: y[i] = 17.7000; t[i++] = 2.5000;
357: y[i] = 24.5600; t[i++] = 2.0000;
358: y[i] = 18.6700; t[i++] = 2.5000;
359: y[i] = 16.2400; t[i++] = 3.0000;
360: y[i] = 8.7400; t[i++] = 4.0000;
361: y[i] = 7.8700; t[i++] = 5.0000;
362: y[i] = 8.5100; t[i++] = 6.0000;
363: y[i] = 66.7000; t[i++] = .5000;
364: y[i] = 59.2000; t[i++] = .7500;
365: y[i] = 40.8000; t[i++] = 1.0000;
366: y[i] = 30.7000; t[i++] = 1.5000;
367: y[i] = 25.7000; t[i++] = 2.0000;
368: y[i] = 16.3000; t[i++] = 2.5000;
369: y[i] = 25.9900; t[i++] = 2.0000;
370: y[i] = 16.9500; t[i++] = 2.5000;
371: y[i] = 13.3500; t[i++] = 3.0000;
372: y[i] = 8.6200; t[i++] = 4.0000;
373: y[i] = 7.2000; t[i++] = 5.0000;
374: y[i] = 6.6400; t[i++] = 6.0000;
375: y[i] = 13.6900; t[i++] = 3.0000;
376: y[i] = 81.0000; t[i++] = .5000;
377: y[i] = 64.5000; t[i++] = .7500;
378: y[i] = 35.5000; t[i++] = 1.5000;
379: y[i] = 13.3100; t[i++] = 3.0000;
380: y[i] = 4.8700; t[i++] = 6.0000;
381: y[i] = 12.9400; t[i++] = 3.0000;
382: y[i] = 5.0600; t[i++] = 6.0000;
383: y[i] = 15.1900; t[i++] = 3.0000;
384: y[i] = 14.6200; t[i++] = 3.0000;
385: y[i] = 15.6400; t[i++] = 3.0000;
386: y[i] = 25.5000; t[i++] = 1.7500;
387: y[i] = 25.9500; t[i++] = 1.7500;
388: y[i] = 81.7000; t[i++] = .5000;
389: y[i] = 61.6000; t[i++] = .7500;
390: y[i] = 29.8000; t[i++] = 1.7500;
391: y[i] = 29.8100; t[i++] = 1.7500;
392: y[i] = 17.1700; t[i++] = 2.7500;
393: y[i] = 10.3900; t[i++] = 3.7500;
394: y[i] = 28.4000; t[i++] = 1.7500;
395: y[i] = 28.6900; t[i++] = 1.7500;
396: y[i] = 81.3000; t[i++] = .5000;
397: y[i] = 60.9000; t[i++] = .7500;
398: y[i] = 16.6500; t[i++] = 2.7500;
399: y[i] = 10.0500; t[i++] = 3.7500;
400: y[i] = 28.9000; t[i++] = 1.7500;
401: y[i] = 28.9500; t[i++] = 1.7500;
402: return(0);
403: }
405: PetscErrorCode TaskWorker(AppCtx *user)
406: {
407: PetscReal x[NPARAMETERS],f = 0.0;
408: PetscMPIInt tag=IDLE_TAG;
409: PetscInt index;
410: MPI_Status status;
414: /* Send check-in message to master */
416: MPI_Send(&f,1,MPIU_REAL,0,IDLE_TAG,PETSC_COMM_WORLD);
417: while (tag != DIE_TAG) {
418: MPI_Recv(x,NPARAMETERS,MPIU_REAL,0,MPI_ANY_TAG,PETSC_COMM_WORLD,&status);
419: tag = status.MPI_TAG;
420: if (tag == IDLE_TAG) {
421: MPI_Send(&f,1,MPIU_REAL,0,IDLE_TAG,PETSC_COMM_WORLD);
422: } else if (tag != DIE_TAG) {
423: index = (PetscInt)tag;
424: ierr=RunSimulation(x,index,&f,user);
425: ierr=MPI_Send(&f,1,MPIU_REAL,0,tag,PETSC_COMM_WORLD);
426: }
427: }
428: return(0);
429: }
431: PetscErrorCode RunSimulation(PetscReal *x, PetscInt i, PetscReal*f, AppCtx *user)
432: {
433: PetscReal *t = user->t;
434: PetscReal *y = user->y;
435: #if defined(PETSC_USE_REAL_SINGLE)
436: *f = y[i] - exp(-x[0]*t[i])/(x[1] + x[2]*t[i]); /* expf() for single-precision breaks this example on freebsd, valgrind errors on linux */
437: #else
438: *f = y[i] - PetscExpScalar(-x[0]*t[i])/(x[1] + x[2]*t[i]);
439: #endif
440: return(0);
441: }
443: PetscErrorCode StopWorkers(AppCtx *user)
444: {
445: PetscInt checkedin;
446: MPI_Status status;
447: PetscReal f,x[NPARAMETERS];
451: checkedin=0;
452: while(checkedin < user->size-1) {
453: MPI_Recv(&f,1,MPIU_REAL,MPI_ANY_SOURCE,MPI_ANY_TAG,PETSC_COMM_WORLD,&status);
454: checkedin++;
455: PetscArrayzero(x,NPARAMETERS);
456: MPI_Send(x,NPARAMETERS,MPIU_REAL,status.MPI_SOURCE,DIE_TAG,PETSC_COMM_WORLD);
457: }
458: return(0);
459: }
461: /*TEST
463: build:
464: requires: !complex
466: test:
467: nsize: 3
468: requires: !single
469: args: -tao_smonitor -tao_max_it 100 -tao_type pounders -tao_gatol 1.e-5
471: TEST*/