Actual source code: chwirut2.c
petsc-3.9.4 2018-09-11
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: /*T
12: requires: !single
13: T*/
15: #include <petsctao.h>
18: /*
19: Description: These data are the result of a NIST study involving
20: ultrasonic calibration. The response variable is
21: ultrasonic response, and the predictor variable is
22: metal distance.
24: Reference: Chwirut, D., NIST (197?).
25: Ultrasonic Reference Block Study.
26: */
28: static char help[]="Finds the nonlinear least-squares solution to the model \n\
29: y = exp[-b1*x]/(b2+b3*x) + e \n";
31: /* T
32: Concepts: TAO^Solving a system of nonlinear equations, nonlinear least squares
33: Routines: TaoCreate();
34: Routines: TaoSetType();
35: Routines: TaoSetSeparableObjectiveRoutine();
36: Routines: TaoSetMonitor();
37: Routines: TaoSetInitialVector();
38: Routines: TaoSetFromOptions();
39: Routines: TaoSolve();
40: Routines: TaoDestroy();
41: Processors: n
42: T*/
44: #define NOBSERVATIONS 214
45: #define NPARAMETERS 3
47: #define DIE_TAG 2000
48: #define IDLE_TAG 1000
50: /* User-defined application context */
51: typedef struct {
52: /* Working space */
53: PetscReal t[NOBSERVATIONS]; /* array of independent variables of observation */
54: PetscReal y[NOBSERVATIONS]; /* array of dependent variables */
55: PetscMPIInt size,rank;
56: } AppCtx;
58: /* User provided Routines */
59: PetscErrorCode InitializeData(AppCtx *user);
60: PetscErrorCode FormStartingPoint(Vec);
61: PetscErrorCode EvaluateFunction(Tao, Vec, Vec, void *);
62: PetscErrorCode TaskWorker(AppCtx *user);
63: PetscErrorCode StopWorkers(AppCtx *user);
64: PetscErrorCode RunSimulation(PetscReal *x, PetscInt i, PetscReal*f, AppCtx *user);
66: /*--------------------------------------------------------------------*/
67: int main(int argc,char **argv)
68: {
70: Vec x, f; /* solution, function */
71: Tao tao; /* Tao solver context */
72: AppCtx user; /* user-defined work context */
74: /* Initialize TAO and PETSc */
75: PetscInitialize(&argc,&argv,(char *)0,help);
76: MPI_Comm_size(MPI_COMM_WORLD,&user.size);
77: MPI_Comm_rank(MPI_COMM_WORLD,&user.rank);
78: InitializeData(&user);
80: /* Run optimization on rank 0 */
81: if (user.rank == 0) {
82: /* Allocate vectors */
83: VecCreateSeq(PETSC_COMM_SELF,NPARAMETERS,&x);
84: VecCreateSeq(PETSC_COMM_SELF,NOBSERVATIONS,&f);
86: /* TAO code begins here */
88: /* Create TAO solver and set desired solution method */
89: TaoCreate(PETSC_COMM_SELF,&tao);
90: TaoSetType(tao,TAOPOUNDERS);
92: /* Set the function and Jacobian routines. */
93: FormStartingPoint(x);
94: TaoSetInitialVector(tao,x);
95: TaoSetSeparableObjectiveRoutine(tao,f,EvaluateFunction,(void*)&user);
97: /* Check for any TAO command line arguments */
98: TaoSetFromOptions(tao);
100: /* Perform the Solve */
101: TaoSolve(tao);
103: /* Free TAO data structures */
104: TaoDestroy(&tao);
106: /* Free PETSc data structures */
107: VecDestroy(&x);
108: VecDestroy(&f);
109: StopWorkers(&user);
110: } else {
111: TaskWorker(&user);
112: }
113: PetscFinalize();
114: return ierr;
115: }
117: /*--------------------------------------------------------------------*/
118: PetscErrorCode EvaluateFunction(Tao tao, Vec X, Vec F, void *ptr)
119: {
120: AppCtx *user = (AppCtx *)ptr;
121: PetscInt i;
122: PetscReal *x,*f;
126: VecGetArray(X,&x);
127: VecGetArray(F,&f);
128: if (user->size == 1) {
129: /* Single processor */
130: for (i=0;i<NOBSERVATIONS;i++) {
131: RunSimulation(x,i,&f[i],user);
132: }
133: } else {
134: /* Multiprocessor master */
135: PetscMPIInt tag;
136: PetscInt finishedtasks,next_task,checkedin;
137: PetscReal f_i=0.0;
138: MPI_Status status;
140: next_task=0;
141: finishedtasks=0;
142: checkedin=0;
144: while(finishedtasks < NOBSERVATIONS || checkedin < user->size-1) {
145: MPI_Recv(&f_i,1,MPIU_REAL,MPI_ANY_SOURCE,MPI_ANY_TAG,PETSC_COMM_WORLD,&status);
146: if (status.MPI_TAG == IDLE_TAG) {
147: checkedin++;
148: } else {
150: tag = status.MPI_TAG;
151: f[tag] = (PetscReal)f_i;
152: finishedtasks++;
153: }
155: if (next_task<NOBSERVATIONS) {
156: MPI_Send(x,NPARAMETERS,MPIU_REAL,status.MPI_SOURCE,next_task,PETSC_COMM_WORLD);
157: next_task++;
159: } else {
160: /* Send idle message */
161: MPI_Send(x,NPARAMETERS,MPIU_REAL,status.MPI_SOURCE,IDLE_TAG,PETSC_COMM_WORLD);
162: }
163: }
164: }
165: VecRestoreArray(X,&x);
166: VecRestoreArray(F,&f);
167: PetscLogFlops(6*NOBSERVATIONS);
168: return(0);
169: }
171: /* ------------------------------------------------------------ */
172: PetscErrorCode FormStartingPoint(Vec X)
173: {
174: PetscReal *x;
178: VecGetArray(X,&x);
179: x[0] = 0.15;
180: x[1] = 0.008;
181: x[2] = 0.010;
182: VecRestoreArray(X,&x);
183: return(0);
184: }
186: /* ---------------------------------------------------------------------- */
187: PetscErrorCode InitializeData(AppCtx *user)
188: {
189: PetscReal *t=user->t,*y=user->y;
190: PetscInt i=0;
193: y[i] = 92.9000; t[i++] = 0.5000;
194: y[i] = 78.7000; t[i++] = 0.6250;
195: y[i] = 64.2000; t[i++] = 0.7500;
196: y[i] = 64.9000; t[i++] = 0.8750;
197: y[i] = 57.1000; t[i++] = 1.0000;
198: y[i] = 43.3000; t[i++] = 1.2500;
199: y[i] = 31.1000; t[i++] = 1.7500;
200: y[i] = 23.6000; t[i++] = 2.2500;
201: y[i] = 31.0500; t[i++] = 1.7500;
202: y[i] = 23.7750; t[i++] = 2.2500;
203: y[i] = 17.7375; t[i++] = 2.7500;
204: y[i] = 13.8000; t[i++] = 3.2500;
205: y[i] = 11.5875; t[i++] = 3.7500;
206: y[i] = 9.4125; t[i++] = 4.2500;
207: y[i] = 7.7250; t[i++] = 4.7500;
208: y[i] = 7.3500; t[i++] = 5.2500;
209: y[i] = 8.0250; t[i++] = 5.7500;
210: y[i] = 90.6000; t[i++] = 0.5000;
211: y[i] = 76.9000; t[i++] = 0.6250;
212: y[i] = 71.6000; t[i++] = 0.7500;
213: y[i] = 63.6000; t[i++] = 0.8750;
214: y[i] = 54.0000; t[i++] = 1.0000;
215: y[i] = 39.2000; t[i++] = 1.2500;
216: y[i] = 29.3000; t[i++] = 1.7500;
217: y[i] = 21.4000; t[i++] = 2.2500;
218: y[i] = 29.1750; t[i++] = 1.7500;
219: y[i] = 22.1250; t[i++] = 2.2500;
220: y[i] = 17.5125; t[i++] = 2.7500;
221: y[i] = 14.2500; t[i++] = 3.2500;
222: y[i] = 9.4500; t[i++] = 3.7500;
223: y[i] = 9.1500; t[i++] = 4.2500;
224: y[i] = 7.9125; t[i++] = 4.7500;
225: y[i] = 8.4750; t[i++] = 5.2500;
226: y[i] = 6.1125; t[i++] = 5.7500;
227: y[i] = 80.0000; t[i++] = 0.5000;
228: y[i] = 79.0000; t[i++] = 0.6250;
229: y[i] = 63.8000; t[i++] = 0.7500;
230: y[i] = 57.2000; t[i++] = 0.8750;
231: y[i] = 53.2000; t[i++] = 1.0000;
232: y[i] = 42.5000; t[i++] = 1.2500;
233: y[i] = 26.8000; t[i++] = 1.7500;
234: y[i] = 20.4000; t[i++] = 2.2500;
235: y[i] = 26.8500; t[i++] = 1.7500;
236: y[i] = 21.0000; t[i++] = 2.2500;
237: y[i] = 16.4625; t[i++] = 2.7500;
238: y[i] = 12.5250; t[i++] = 3.2500;
239: y[i] = 10.5375; t[i++] = 3.7500;
240: y[i] = 8.5875; t[i++] = 4.2500;
241: y[i] = 7.1250; t[i++] = 4.7500;
242: y[i] = 6.1125; t[i++] = 5.2500;
243: y[i] = 5.9625; t[i++] = 5.7500;
244: y[i] = 74.1000; t[i++] = 0.5000;
245: y[i] = 67.3000; t[i++] = 0.6250;
246: y[i] = 60.8000; t[i++] = 0.7500;
247: y[i] = 55.5000; t[i++] = 0.8750;
248: y[i] = 50.3000; t[i++] = 1.0000;
249: y[i] = 41.0000; t[i++] = 1.2500;
250: y[i] = 29.4000; t[i++] = 1.7500;
251: y[i] = 20.4000; t[i++] = 2.2500;
252: y[i] = 29.3625; t[i++] = 1.7500;
253: y[i] = 21.1500; t[i++] = 2.2500;
254: y[i] = 16.7625; t[i++] = 2.7500;
255: y[i] = 13.2000; t[i++] = 3.2500;
256: y[i] = 10.8750; t[i++] = 3.7500;
257: y[i] = 8.1750; t[i++] = 4.2500;
258: y[i] = 7.3500; t[i++] = 4.7500;
259: y[i] = 5.9625; t[i++] = 5.2500;
260: y[i] = 5.6250; t[i++] = 5.7500;
261: y[i] = 81.5000; t[i++] = .5000;
262: y[i] = 62.4000; t[i++] = .7500;
263: y[i] = 32.5000; t[i++] = 1.5000;
264: y[i] = 12.4100; t[i++] = 3.0000;
265: y[i] = 13.1200; t[i++] = 3.0000;
266: y[i] = 15.5600; t[i++] = 3.0000;
267: y[i] = 5.6300; t[i++] = 6.0000;
268: y[i] = 78.0000; t[i++] = .5000;
269: y[i] = 59.9000; t[i++] = .7500;
270: y[i] = 33.2000; t[i++] = 1.5000;
271: y[i] = 13.8400; t[i++] = 3.0000;
272: y[i] = 12.7500; t[i++] = 3.0000;
273: y[i] = 14.6200; t[i++] = 3.0000;
274: y[i] = 3.9400; t[i++] = 6.0000;
275: y[i] = 76.8000; t[i++] = .5000;
276: y[i] = 61.0000; t[i++] = .7500;
277: y[i] = 32.9000; t[i++] = 1.5000;
278: y[i] = 13.8700; t[i++] = 3.0000;
279: y[i] = 11.8100; t[i++] = 3.0000;
280: y[i] = 13.3100; t[i++] = 3.0000;
281: y[i] = 5.4400; t[i++] = 6.0000;
282: y[i] = 78.0000; t[i++] = .5000;
283: y[i] = 63.5000; t[i++] = .7500;
284: y[i] = 33.8000; t[i++] = 1.5000;
285: y[i] = 12.5600; t[i++] = 3.0000;
286: y[i] = 5.6300; t[i++] = 6.0000;
287: y[i] = 12.7500; t[i++] = 3.0000;
288: y[i] = 13.1200; t[i++] = 3.0000;
289: y[i] = 5.4400; t[i++] = 6.0000;
290: y[i] = 76.8000; t[i++] = .5000;
291: y[i] = 60.0000; t[i++] = .7500;
292: y[i] = 47.8000; t[i++] = 1.0000;
293: y[i] = 32.0000; t[i++] = 1.5000;
294: y[i] = 22.2000; t[i++] = 2.0000;
295: y[i] = 22.5700; t[i++] = 2.0000;
296: y[i] = 18.8200; t[i++] = 2.5000;
297: y[i] = 13.9500; t[i++] = 3.0000;
298: y[i] = 11.2500; t[i++] = 4.0000;
299: y[i] = 9.0000; t[i++] = 5.0000;
300: y[i] = 6.6700; t[i++] = 6.0000;
301: y[i] = 75.8000; t[i++] = .5000;
302: y[i] = 62.0000; t[i++] = .7500;
303: y[i] = 48.8000; t[i++] = 1.0000;
304: y[i] = 35.2000; t[i++] = 1.5000;
305: y[i] = 20.0000; t[i++] = 2.0000;
306: y[i] = 20.3200; t[i++] = 2.0000;
307: y[i] = 19.3100; t[i++] = 2.5000;
308: y[i] = 12.7500; t[i++] = 3.0000;
309: y[i] = 10.4200; t[i++] = 4.0000;
310: y[i] = 7.3100; t[i++] = 5.0000;
311: y[i] = 7.4200; t[i++] = 6.0000;
312: y[i] = 70.5000; t[i++] = .5000;
313: y[i] = 59.5000; t[i++] = .7500;
314: y[i] = 48.5000; t[i++] = 1.0000;
315: y[i] = 35.8000; t[i++] = 1.5000;
316: y[i] = 21.0000; t[i++] = 2.0000;
317: y[i] = 21.6700; t[i++] = 2.0000;
318: y[i] = 21.0000; t[i++] = 2.5000;
319: y[i] = 15.6400; t[i++] = 3.0000;
320: y[i] = 8.1700; t[i++] = 4.0000;
321: y[i] = 8.5500; t[i++] = 5.0000;
322: y[i] = 10.1200; t[i++] = 6.0000;
323: y[i] = 78.0000; t[i++] = .5000;
324: y[i] = 66.0000; t[i++] = .6250;
325: y[i] = 62.0000; t[i++] = .7500;
326: y[i] = 58.0000; t[i++] = .8750;
327: y[i] = 47.7000; t[i++] = 1.0000;
328: y[i] = 37.8000; t[i++] = 1.2500;
329: y[i] = 20.2000; t[i++] = 2.2500;
330: y[i] = 21.0700; t[i++] = 2.2500;
331: y[i] = 13.8700; t[i++] = 2.7500;
332: y[i] = 9.6700; t[i++] = 3.2500;
333: y[i] = 7.7600; t[i++] = 3.7500;
334: y[i] = 5.4400; t[i++] = 4.2500;
335: y[i] = 4.8700; t[i++] = 4.7500;
336: y[i] = 4.0100; t[i++] = 5.2500;
337: y[i] = 3.7500; t[i++] = 5.7500;
338: y[i] = 24.1900; t[i++] = 3.0000;
339: y[i] = 25.7600; t[i++] = 3.0000;
340: y[i] = 18.0700; t[i++] = 3.0000;
341: y[i] = 11.8100; t[i++] = 3.0000;
342: y[i] = 12.0700; t[i++] = 3.0000;
343: y[i] = 16.1200; t[i++] = 3.0000;
344: y[i] = 70.8000; t[i++] = .5000;
345: y[i] = 54.7000; t[i++] = .7500;
346: y[i] = 48.0000; t[i++] = 1.0000;
347: y[i] = 39.8000; t[i++] = 1.5000;
348: y[i] = 29.8000; t[i++] = 2.0000;
349: y[i] = 23.7000; t[i++] = 2.5000;
350: y[i] = 29.6200; t[i++] = 2.0000;
351: y[i] = 23.8100; t[i++] = 2.5000;
352: y[i] = 17.7000; t[i++] = 3.0000;
353: y[i] = 11.5500; t[i++] = 4.0000;
354: y[i] = 12.0700; t[i++] = 5.0000;
355: y[i] = 8.7400; t[i++] = 6.0000;
356: y[i] = 80.7000; t[i++] = .5000;
357: y[i] = 61.3000; t[i++] = .7500;
358: y[i] = 47.5000; t[i++] = 1.0000;
359: y[i] = 29.0000; t[i++] = 1.5000;
360: y[i] = 24.0000; t[i++] = 2.0000;
361: y[i] = 17.7000; t[i++] = 2.5000;
362: y[i] = 24.5600; t[i++] = 2.0000;
363: y[i] = 18.6700; t[i++] = 2.5000;
364: y[i] = 16.2400; t[i++] = 3.0000;
365: y[i] = 8.7400; t[i++] = 4.0000;
366: y[i] = 7.8700; t[i++] = 5.0000;
367: y[i] = 8.5100; t[i++] = 6.0000;
368: y[i] = 66.7000; t[i++] = .5000;
369: y[i] = 59.2000; t[i++] = .7500;
370: y[i] = 40.8000; t[i++] = 1.0000;
371: y[i] = 30.7000; t[i++] = 1.5000;
372: y[i] = 25.7000; t[i++] = 2.0000;
373: y[i] = 16.3000; t[i++] = 2.5000;
374: y[i] = 25.9900; t[i++] = 2.0000;
375: y[i] = 16.9500; t[i++] = 2.5000;
376: y[i] = 13.3500; t[i++] = 3.0000;
377: y[i] = 8.6200; t[i++] = 4.0000;
378: y[i] = 7.2000; t[i++] = 5.0000;
379: y[i] = 6.6400; t[i++] = 6.0000;
380: y[i] = 13.6900; t[i++] = 3.0000;
381: y[i] = 81.0000; t[i++] = .5000;
382: y[i] = 64.5000; t[i++] = .7500;
383: y[i] = 35.5000; t[i++] = 1.5000;
384: y[i] = 13.3100; t[i++] = 3.0000;
385: y[i] = 4.8700; t[i++] = 6.0000;
386: y[i] = 12.9400; t[i++] = 3.0000;
387: y[i] = 5.0600; t[i++] = 6.0000;
388: y[i] = 15.1900; t[i++] = 3.0000;
389: y[i] = 14.6200; t[i++] = 3.0000;
390: y[i] = 15.6400; t[i++] = 3.0000;
391: y[i] = 25.5000; t[i++] = 1.7500;
392: y[i] = 25.9500; t[i++] = 1.7500;
393: y[i] = 81.7000; t[i++] = .5000;
394: y[i] = 61.6000; t[i++] = .7500;
395: y[i] = 29.8000; t[i++] = 1.7500;
396: y[i] = 29.8100; t[i++] = 1.7500;
397: y[i] = 17.1700; t[i++] = 2.7500;
398: y[i] = 10.3900; t[i++] = 3.7500;
399: y[i] = 28.4000; t[i++] = 1.7500;
400: y[i] = 28.6900; t[i++] = 1.7500;
401: y[i] = 81.3000; t[i++] = .5000;
402: y[i] = 60.9000; t[i++] = .7500;
403: y[i] = 16.6500; t[i++] = 2.7500;
404: y[i] = 10.0500; t[i++] = 3.7500;
405: y[i] = 28.9000; t[i++] = 1.7500;
406: y[i] = 28.9500; t[i++] = 1.7500;
407: return(0);
408: }
410: PetscErrorCode TaskWorker(AppCtx *user)
411: {
412: PetscReal x[NPARAMETERS],f = 0.0;
413: PetscMPIInt tag=IDLE_TAG;
414: PetscInt index;
415: MPI_Status status;
419: /* Send check-in message to master */
421: MPI_Send(&f,1,MPIU_REAL,0,IDLE_TAG,PETSC_COMM_WORLD);
422: while (tag != DIE_TAG) {
423: MPI_Recv(x,NPARAMETERS,MPIU_REAL,0,MPI_ANY_TAG,PETSC_COMM_WORLD,&status);
424: tag = status.MPI_TAG;
425: if (tag == IDLE_TAG) {
426: MPI_Send(&f,1,MPIU_REAL,0,IDLE_TAG,PETSC_COMM_WORLD);
427: } else if (tag != DIE_TAG) {
428: index = (PetscInt)tag;
429: ierr=RunSimulation(x,index,&f,user);
430: ierr=MPI_Send(&f,1,MPIU_REAL,0,tag,PETSC_COMM_WORLD);
431: }
432: }
433: return(0);
434: }
436: PetscErrorCode RunSimulation(PetscReal *x, PetscInt i, PetscReal*f, AppCtx *user)
437: {
438: PetscReal *t = user->t;
439: PetscReal *y = user->y;
440: *f = y[i] - PetscExpScalar(-x[0]*t[i])/(x[1] + x[2]*t[i]);
441: return(0);
442: }
444: PetscErrorCode StopWorkers(AppCtx *user)
445: {
446: PetscInt checkedin;
447: MPI_Status status;
448: PetscReal f,x[NPARAMETERS];
452: checkedin=0;
453: while(checkedin < user->size-1) {
454: MPI_Recv(&f,1,MPIU_REAL,MPI_ANY_SOURCE,MPI_ANY_TAG,PETSC_COMM_WORLD,&status);
455: checkedin++;
456: PetscMemzero(x,NPARAMETERS*sizeof(PetscReal));
457: MPI_Send(x,NPARAMETERS,MPIU_REAL,status.MPI_SOURCE,DIE_TAG,PETSC_COMM_WORLD);
458: }
459: return(0);
460: }
463: /*TEST
465: build:
466: requires: !complex
468: test:
469: nsize: 3
470: args: -tao_smonitor -tao_max_it 100 -tao_type pounders
471: TODO: too many inconsistent results across machines
473: TEST*/