Actual source code: chwirut2.c
petsc-3.9.0 2018-04-07
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>
16: #include <mpi.h>
19: /*
20: Description: These data are the result of a NIST study involving
21: ultrasonic calibration. The response variable is
22: ultrasonic response, and the predictor variable is
23: metal distance.
25: Reference: Chwirut, D., NIST (197?).
26: Ultrasonic Reference Block Study.
27: */
29: static char help[]="Finds the nonlinear least-squares solution to the model \n\
30: y = exp[-b1*x]/(b2+b3*x) + e \n";
32: /* T
33: Concepts: TAO^Solving a system of nonlinear equations, nonlinear least squares
34: Routines: TaoCreate();
35: Routines: TaoSetType();
36: Routines: TaoSetSeparableObjectiveRoutine();
37: Routines: TaoSetMonitor();
38: Routines: TaoSetInitialVector();
39: Routines: TaoSetFromOptions();
40: Routines: TaoSolve();
41: Routines: TaoDestroy();
42: Processors: n
43: T*/
45: #define NOBSERVATIONS 214
46: #define NPARAMETERS 3
48: #define DIE_TAG 2000
49: #define IDLE_TAG 1000
51: /* User-defined application context */
52: typedef struct {
53: /* Working space */
54: PetscReal t[NOBSERVATIONS]; /* array of independent variables of observation */
55: PetscReal y[NOBSERVATIONS]; /* array of dependent variables */
56: PetscMPIInt size,rank;
57: } AppCtx;
59: /* User provided Routines */
60: PetscErrorCode InitializeData(AppCtx *user);
61: PetscErrorCode FormStartingPoint(Vec);
62: PetscErrorCode EvaluateFunction(Tao, Vec, Vec, void *);
63: PetscErrorCode TaskWorker(AppCtx *user);
64: PetscErrorCode StopWorkers(AppCtx *user);
65: PetscErrorCode RunSimulation(PetscReal *x, PetscInt i, PetscReal*f, AppCtx *user);
67: /*--------------------------------------------------------------------*/
68: int main(int argc,char **argv)
69: {
71: Vec x, f; /* solution, function */
72: Tao tao; /* Tao solver context */
73: AppCtx user; /* user-defined work context */
75: /* Initialize TAO and PETSc */
76: PetscInitialize(&argc,&argv,(char *)0,help);
77: MPI_Comm_size(MPI_COMM_WORLD,&user.size);
78: MPI_Comm_rank(MPI_COMM_WORLD,&user.rank);
79: InitializeData(&user);
81: /* Run optimization on rank 0 */
82: if (user.rank == 0) {
83: /* Allocate vectors */
84: VecCreateSeq(PETSC_COMM_SELF,NPARAMETERS,&x);
85: VecCreateSeq(PETSC_COMM_SELF,NOBSERVATIONS,&f);
87: /* TAO code begins here */
89: /* Create TAO solver and set desired solution method */
90: TaoCreate(PETSC_COMM_SELF,&tao);
91: TaoSetType(tao,TAOPOUNDERS);
93: /* Set the function and Jacobian routines. */
94: FormStartingPoint(x);
95: TaoSetInitialVector(tao,x);
96: TaoSetSeparableObjectiveRoutine(tao,f,EvaluateFunction,(void*)&user);
98: /* Check for any TAO command line arguments */
99: TaoSetFromOptions(tao);
101: /* Perform the Solve */
102: TaoSolve(tao);
104: /* Free TAO data structures */
105: TaoDestroy(&tao);
107: /* Free PETSc data structures */
108: VecDestroy(&x);
109: VecDestroy(&f);
110: StopWorkers(&user);
111: } else {
112: TaskWorker(&user);
113: }
114: PetscFinalize();
115: return ierr;
116: }
118: /*--------------------------------------------------------------------*/
119: PetscErrorCode EvaluateFunction(Tao tao, Vec X, Vec F, void *ptr)
120: {
121: AppCtx *user = (AppCtx *)ptr;
122: PetscInt i;
123: PetscReal *x,*f;
127: VecGetArray(X,&x);
128: VecGetArray(F,&f);
129: if (user->size == 1) {
130: /* Single processor */
131: for (i=0;i<NOBSERVATIONS;i++) {
132: RunSimulation(x,i,&f[i],user);
133: }
134: } else {
135: /* Multiprocessor master */
136: PetscMPIInt tag;
137: PetscInt finishedtasks,next_task,checkedin;
138: PetscReal f_i=0.0;
139: MPI_Status status;
141: next_task=0;
142: finishedtasks=0;
143: checkedin=0;
145: while(finishedtasks < NOBSERVATIONS || checkedin < user->size-1) {
146: MPI_Recv(&f_i,1,MPIU_REAL,MPI_ANY_SOURCE,MPI_ANY_TAG,PETSC_COMM_WORLD,&status);
147: if (status.MPI_TAG == IDLE_TAG) {
148: checkedin++;
149: } else {
151: tag = status.MPI_TAG;
152: f[tag] = (PetscReal)f_i;
153: finishedtasks++;
154: }
156: if (next_task<NOBSERVATIONS) {
157: MPI_Send(x,NPARAMETERS,MPIU_REAL,status.MPI_SOURCE,next_task,PETSC_COMM_WORLD);
158: next_task++;
160: } else {
161: /* Send idle message */
162: MPI_Send(x,NPARAMETERS,MPIU_REAL,status.MPI_SOURCE,IDLE_TAG,PETSC_COMM_WORLD);
163: }
164: }
165: }
166: VecRestoreArray(X,&x);
167: VecRestoreArray(F,&f);
168: PetscLogFlops(6*NOBSERVATIONS);
169: return(0);
170: }
172: /* ------------------------------------------------------------ */
173: PetscErrorCode FormStartingPoint(Vec X)
174: {
175: PetscReal *x;
179: VecGetArray(X,&x);
180: x[0] = 0.15;
181: x[1] = 0.008;
182: x[2] = 0.010;
183: VecRestoreArray(X,&x);
184: return(0);
185: }
187: /* ---------------------------------------------------------------------- */
188: PetscErrorCode InitializeData(AppCtx *user)
189: {
190: PetscReal *t=user->t,*y=user->y;
191: PetscInt i=0;
194: y[i] = 92.9000; t[i++] = 0.5000;
195: y[i] = 78.7000; t[i++] = 0.6250;
196: y[i] = 64.2000; t[i++] = 0.7500;
197: y[i] = 64.9000; t[i++] = 0.8750;
198: y[i] = 57.1000; t[i++] = 1.0000;
199: y[i] = 43.3000; t[i++] = 1.2500;
200: y[i] = 31.1000; t[i++] = 1.7500;
201: y[i] = 23.6000; t[i++] = 2.2500;
202: y[i] = 31.0500; t[i++] = 1.7500;
203: y[i] = 23.7750; t[i++] = 2.2500;
204: y[i] = 17.7375; t[i++] = 2.7500;
205: y[i] = 13.8000; t[i++] = 3.2500;
206: y[i] = 11.5875; t[i++] = 3.7500;
207: y[i] = 9.4125; t[i++] = 4.2500;
208: y[i] = 7.7250; t[i++] = 4.7500;
209: y[i] = 7.3500; t[i++] = 5.2500;
210: y[i] = 8.0250; t[i++] = 5.7500;
211: y[i] = 90.6000; t[i++] = 0.5000;
212: y[i] = 76.9000; t[i++] = 0.6250;
213: y[i] = 71.6000; t[i++] = 0.7500;
214: y[i] = 63.6000; t[i++] = 0.8750;
215: y[i] = 54.0000; t[i++] = 1.0000;
216: y[i] = 39.2000; t[i++] = 1.2500;
217: y[i] = 29.3000; t[i++] = 1.7500;
218: y[i] = 21.4000; t[i++] = 2.2500;
219: y[i] = 29.1750; t[i++] = 1.7500;
220: y[i] = 22.1250; t[i++] = 2.2500;
221: y[i] = 17.5125; t[i++] = 2.7500;
222: y[i] = 14.2500; t[i++] = 3.2500;
223: y[i] = 9.4500; t[i++] = 3.7500;
224: y[i] = 9.1500; t[i++] = 4.2500;
225: y[i] = 7.9125; t[i++] = 4.7500;
226: y[i] = 8.4750; t[i++] = 5.2500;
227: y[i] = 6.1125; t[i++] = 5.7500;
228: y[i] = 80.0000; t[i++] = 0.5000;
229: y[i] = 79.0000; t[i++] = 0.6250;
230: y[i] = 63.8000; t[i++] = 0.7500;
231: y[i] = 57.2000; t[i++] = 0.8750;
232: y[i] = 53.2000; t[i++] = 1.0000;
233: y[i] = 42.5000; t[i++] = 1.2500;
234: y[i] = 26.8000; t[i++] = 1.7500;
235: y[i] = 20.4000; t[i++] = 2.2500;
236: y[i] = 26.8500; t[i++] = 1.7500;
237: y[i] = 21.0000; t[i++] = 2.2500;
238: y[i] = 16.4625; t[i++] = 2.7500;
239: y[i] = 12.5250; t[i++] = 3.2500;
240: y[i] = 10.5375; t[i++] = 3.7500;
241: y[i] = 8.5875; t[i++] = 4.2500;
242: y[i] = 7.1250; t[i++] = 4.7500;
243: y[i] = 6.1125; t[i++] = 5.2500;
244: y[i] = 5.9625; t[i++] = 5.7500;
245: y[i] = 74.1000; t[i++] = 0.5000;
246: y[i] = 67.3000; t[i++] = 0.6250;
247: y[i] = 60.8000; t[i++] = 0.7500;
248: y[i] = 55.5000; t[i++] = 0.8750;
249: y[i] = 50.3000; t[i++] = 1.0000;
250: y[i] = 41.0000; t[i++] = 1.2500;
251: y[i] = 29.4000; t[i++] = 1.7500;
252: y[i] = 20.4000; t[i++] = 2.2500;
253: y[i] = 29.3625; t[i++] = 1.7500;
254: y[i] = 21.1500; t[i++] = 2.2500;
255: y[i] = 16.7625; t[i++] = 2.7500;
256: y[i] = 13.2000; t[i++] = 3.2500;
257: y[i] = 10.8750; t[i++] = 3.7500;
258: y[i] = 8.1750; t[i++] = 4.2500;
259: y[i] = 7.3500; t[i++] = 4.7500;
260: y[i] = 5.9625; t[i++] = 5.2500;
261: y[i] = 5.6250; t[i++] = 5.7500;
262: y[i] = 81.5000; t[i++] = .5000;
263: y[i] = 62.4000; t[i++] = .7500;
264: y[i] = 32.5000; t[i++] = 1.5000;
265: y[i] = 12.4100; t[i++] = 3.0000;
266: y[i] = 13.1200; t[i++] = 3.0000;
267: y[i] = 15.5600; t[i++] = 3.0000;
268: y[i] = 5.6300; t[i++] = 6.0000;
269: y[i] = 78.0000; t[i++] = .5000;
270: y[i] = 59.9000; t[i++] = .7500;
271: y[i] = 33.2000; t[i++] = 1.5000;
272: y[i] = 13.8400; t[i++] = 3.0000;
273: y[i] = 12.7500; t[i++] = 3.0000;
274: y[i] = 14.6200; t[i++] = 3.0000;
275: y[i] = 3.9400; t[i++] = 6.0000;
276: y[i] = 76.8000; t[i++] = .5000;
277: y[i] = 61.0000; t[i++] = .7500;
278: y[i] = 32.9000; t[i++] = 1.5000;
279: y[i] = 13.8700; t[i++] = 3.0000;
280: y[i] = 11.8100; t[i++] = 3.0000;
281: y[i] = 13.3100; t[i++] = 3.0000;
282: y[i] = 5.4400; t[i++] = 6.0000;
283: y[i] = 78.0000; t[i++] = .5000;
284: y[i] = 63.5000; t[i++] = .7500;
285: y[i] = 33.8000; t[i++] = 1.5000;
286: y[i] = 12.5600; t[i++] = 3.0000;
287: y[i] = 5.6300; t[i++] = 6.0000;
288: y[i] = 12.7500; t[i++] = 3.0000;
289: y[i] = 13.1200; t[i++] = 3.0000;
290: y[i] = 5.4400; t[i++] = 6.0000;
291: y[i] = 76.8000; t[i++] = .5000;
292: y[i] = 60.0000; t[i++] = .7500;
293: y[i] = 47.8000; t[i++] = 1.0000;
294: y[i] = 32.0000; t[i++] = 1.5000;
295: y[i] = 22.2000; t[i++] = 2.0000;
296: y[i] = 22.5700; t[i++] = 2.0000;
297: y[i] = 18.8200; t[i++] = 2.5000;
298: y[i] = 13.9500; t[i++] = 3.0000;
299: y[i] = 11.2500; t[i++] = 4.0000;
300: y[i] = 9.0000; t[i++] = 5.0000;
301: y[i] = 6.6700; t[i++] = 6.0000;
302: y[i] = 75.8000; t[i++] = .5000;
303: y[i] = 62.0000; t[i++] = .7500;
304: y[i] = 48.8000; t[i++] = 1.0000;
305: y[i] = 35.2000; t[i++] = 1.5000;
306: y[i] = 20.0000; t[i++] = 2.0000;
307: y[i] = 20.3200; t[i++] = 2.0000;
308: y[i] = 19.3100; t[i++] = 2.5000;
309: y[i] = 12.7500; t[i++] = 3.0000;
310: y[i] = 10.4200; t[i++] = 4.0000;
311: y[i] = 7.3100; t[i++] = 5.0000;
312: y[i] = 7.4200; t[i++] = 6.0000;
313: y[i] = 70.5000; t[i++] = .5000;
314: y[i] = 59.5000; t[i++] = .7500;
315: y[i] = 48.5000; t[i++] = 1.0000;
316: y[i] = 35.8000; t[i++] = 1.5000;
317: y[i] = 21.0000; t[i++] = 2.0000;
318: y[i] = 21.6700; t[i++] = 2.0000;
319: y[i] = 21.0000; t[i++] = 2.5000;
320: y[i] = 15.6400; t[i++] = 3.0000;
321: y[i] = 8.1700; t[i++] = 4.0000;
322: y[i] = 8.5500; t[i++] = 5.0000;
323: y[i] = 10.1200; t[i++] = 6.0000;
324: y[i] = 78.0000; t[i++] = .5000;
325: y[i] = 66.0000; t[i++] = .6250;
326: y[i] = 62.0000; t[i++] = .7500;
327: y[i] = 58.0000; t[i++] = .8750;
328: y[i] = 47.7000; t[i++] = 1.0000;
329: y[i] = 37.8000; t[i++] = 1.2500;
330: y[i] = 20.2000; t[i++] = 2.2500;
331: y[i] = 21.0700; t[i++] = 2.2500;
332: y[i] = 13.8700; t[i++] = 2.7500;
333: y[i] = 9.6700; t[i++] = 3.2500;
334: y[i] = 7.7600; t[i++] = 3.7500;
335: y[i] = 5.4400; t[i++] = 4.2500;
336: y[i] = 4.8700; t[i++] = 4.7500;
337: y[i] = 4.0100; t[i++] = 5.2500;
338: y[i] = 3.7500; t[i++] = 5.7500;
339: y[i] = 24.1900; t[i++] = 3.0000;
340: y[i] = 25.7600; t[i++] = 3.0000;
341: y[i] = 18.0700; t[i++] = 3.0000;
342: y[i] = 11.8100; t[i++] = 3.0000;
343: y[i] = 12.0700; t[i++] = 3.0000;
344: y[i] = 16.1200; t[i++] = 3.0000;
345: y[i] = 70.8000; t[i++] = .5000;
346: y[i] = 54.7000; t[i++] = .7500;
347: y[i] = 48.0000; t[i++] = 1.0000;
348: y[i] = 39.8000; t[i++] = 1.5000;
349: y[i] = 29.8000; t[i++] = 2.0000;
350: y[i] = 23.7000; t[i++] = 2.5000;
351: y[i] = 29.6200; t[i++] = 2.0000;
352: y[i] = 23.8100; t[i++] = 2.5000;
353: y[i] = 17.7000; t[i++] = 3.0000;
354: y[i] = 11.5500; t[i++] = 4.0000;
355: y[i] = 12.0700; t[i++] = 5.0000;
356: y[i] = 8.7400; t[i++] = 6.0000;
357: y[i] = 80.7000; t[i++] = .5000;
358: y[i] = 61.3000; t[i++] = .7500;
359: y[i] = 47.5000; t[i++] = 1.0000;
360: y[i] = 29.0000; t[i++] = 1.5000;
361: y[i] = 24.0000; t[i++] = 2.0000;
362: y[i] = 17.7000; t[i++] = 2.5000;
363: y[i] = 24.5600; t[i++] = 2.0000;
364: y[i] = 18.6700; t[i++] = 2.5000;
365: y[i] = 16.2400; t[i++] = 3.0000;
366: y[i] = 8.7400; t[i++] = 4.0000;
367: y[i] = 7.8700; t[i++] = 5.0000;
368: y[i] = 8.5100; t[i++] = 6.0000;
369: y[i] = 66.7000; t[i++] = .5000;
370: y[i] = 59.2000; t[i++] = .7500;
371: y[i] = 40.8000; t[i++] = 1.0000;
372: y[i] = 30.7000; t[i++] = 1.5000;
373: y[i] = 25.7000; t[i++] = 2.0000;
374: y[i] = 16.3000; t[i++] = 2.5000;
375: y[i] = 25.9900; t[i++] = 2.0000;
376: y[i] = 16.9500; t[i++] = 2.5000;
377: y[i] = 13.3500; t[i++] = 3.0000;
378: y[i] = 8.6200; t[i++] = 4.0000;
379: y[i] = 7.2000; t[i++] = 5.0000;
380: y[i] = 6.6400; t[i++] = 6.0000;
381: y[i] = 13.6900; t[i++] = 3.0000;
382: y[i] = 81.0000; t[i++] = .5000;
383: y[i] = 64.5000; t[i++] = .7500;
384: y[i] = 35.5000; t[i++] = 1.5000;
385: y[i] = 13.3100; t[i++] = 3.0000;
386: y[i] = 4.8700; t[i++] = 6.0000;
387: y[i] = 12.9400; t[i++] = 3.0000;
388: y[i] = 5.0600; t[i++] = 6.0000;
389: y[i] = 15.1900; t[i++] = 3.0000;
390: y[i] = 14.6200; t[i++] = 3.0000;
391: y[i] = 15.6400; t[i++] = 3.0000;
392: y[i] = 25.5000; t[i++] = 1.7500;
393: y[i] = 25.9500; t[i++] = 1.7500;
394: y[i] = 81.7000; t[i++] = .5000;
395: y[i] = 61.6000; t[i++] = .7500;
396: y[i] = 29.8000; t[i++] = 1.7500;
397: y[i] = 29.8100; t[i++] = 1.7500;
398: y[i] = 17.1700; t[i++] = 2.7500;
399: y[i] = 10.3900; t[i++] = 3.7500;
400: y[i] = 28.4000; t[i++] = 1.7500;
401: y[i] = 28.6900; t[i++] = 1.7500;
402: y[i] = 81.3000; t[i++] = .5000;
403: y[i] = 60.9000; t[i++] = .7500;
404: y[i] = 16.6500; t[i++] = 2.7500;
405: y[i] = 10.0500; t[i++] = 3.7500;
406: y[i] = 28.9000; t[i++] = 1.7500;
407: y[i] = 28.9500; t[i++] = 1.7500;
408: return(0);
409: }
411: PetscErrorCode TaskWorker(AppCtx *user)
412: {
413: PetscReal x[NPARAMETERS],f = 0.0;
414: PetscMPIInt tag=IDLE_TAG;
415: PetscInt index;
416: MPI_Status status;
420: /* Send check-in message to master */
422: MPI_Send(&f,1,MPIU_REAL,0,IDLE_TAG,PETSC_COMM_WORLD);
423: while (tag != DIE_TAG) {
424: MPI_Recv(x,NPARAMETERS,MPIU_REAL,0,MPI_ANY_TAG,PETSC_COMM_WORLD,&status);
425: tag = status.MPI_TAG;
426: if (tag == IDLE_TAG) {
427: MPI_Send(&f,1,MPIU_REAL,0,IDLE_TAG,PETSC_COMM_WORLD);
428: } else if (tag != DIE_TAG) {
429: index = (PetscInt)tag;
430: ierr=RunSimulation(x,index,&f,user);
431: ierr=MPI_Send(&f,1,MPIU_REAL,0,tag,PETSC_COMM_WORLD);
432: }
433: }
434: return(0);
435: }
437: PetscErrorCode RunSimulation(PetscReal *x, PetscInt i, PetscReal*f, AppCtx *user)
438: {
439: PetscReal *t = user->t;
440: PetscReal *y = user->y;
441: *f = y[i] - PetscExpScalar(-x[0]*t[i])/(x[1] + x[2]*t[i]);
442: return(0);
443: }
445: PetscErrorCode StopWorkers(AppCtx *user)
446: {
447: PetscInt checkedin;
448: MPI_Status status;
449: PetscReal f,x[NPARAMETERS];
453: checkedin=0;
454: while(checkedin < user->size-1) {
455: MPI_Recv(&f,1,MPIU_REAL,MPI_ANY_SOURCE,MPI_ANY_TAG,PETSC_COMM_WORLD,&status);
456: checkedin++;
457: PetscMemzero(x,NPARAMETERS*sizeof(PetscReal));
458: MPI_Send(x,NPARAMETERS,MPIU_REAL,status.MPI_SOURCE,DIE_TAG,PETSC_COMM_WORLD);
459: }
460: return(0);
461: }
464: /*TEST
466: build:
467: requires: !complex
469: test:
470: nsize: 3
471: args: -tao_smonitor -tao_max_it 100 -tao_type pounders
472: TODO: too many inconsistent results across machines
474: TEST*/