Actual source code: tron.c
petsc-3.12.5 2020-03-29
1: #include <../src/tao/bound/impls/tron/tron.h>
2: #include <../src/tao/matrix/submatfree.h>
5: /* TRON Routines */
6: static PetscErrorCode TronGradientProjections(Tao,TAO_TRON*);
7: /*------------------------------------------------------------*/
8: static PetscErrorCode TaoDestroy_TRON(Tao tao)
9: {
10: TAO_TRON *tron = (TAO_TRON *)tao->data;
14: VecDestroy(&tron->X_New);
15: VecDestroy(&tron->G_New);
16: VecDestroy(&tron->Work);
17: VecDestroy(&tron->DXFree);
18: VecDestroy(&tron->R);
19: VecDestroy(&tron->diag);
20: VecScatterDestroy(&tron->scatter);
21: ISDestroy(&tron->Free_Local);
22: MatDestroy(&tron->H_sub);
23: MatDestroy(&tron->Hpre_sub);
24: PetscFree(tao->data);
25: return(0);
26: }
28: /*------------------------------------------------------------*/
29: static PetscErrorCode TaoSetFromOptions_TRON(PetscOptionItems *PetscOptionsObject,Tao tao)
30: {
31: TAO_TRON *tron = (TAO_TRON *)tao->data;
33: PetscBool flg;
36: PetscOptionsHead(PetscOptionsObject,"Newton Trust Region Method for bound constrained optimization");
37: PetscOptionsInt("-tao_tron_maxgpits","maximum number of gradient projections per TRON iterate","TaoSetMaxGPIts",tron->maxgpits,&tron->maxgpits,&flg);
38: PetscOptionsTail();
39: TaoLineSearchSetFromOptions(tao->linesearch);
40: KSPSetFromOptions(tao->ksp);
41: return(0);
42: }
44: /*------------------------------------------------------------*/
45: static PetscErrorCode TaoView_TRON(Tao tao, PetscViewer viewer)
46: {
47: TAO_TRON *tron = (TAO_TRON *)tao->data;
48: PetscBool isascii;
49: PetscErrorCode ierr;
52: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&isascii);
53: if (isascii) {
54: PetscViewerASCIIPrintf(viewer,"Total PG its: %D,",tron->total_gp_its);
55: PetscViewerASCIIPrintf(viewer,"PG tolerance: %g \n",(double)tron->pg_ftol);
56: }
57: return(0);
58: }
60: /* ---------------------------------------------------------- */
61: static PetscErrorCode TaoSetup_TRON(Tao tao)
62: {
64: TAO_TRON *tron = (TAO_TRON *)tao->data;
68: /* Allocate some arrays */
69: VecDuplicate(tao->solution, &tron->diag);
70: VecDuplicate(tao->solution, &tron->X_New);
71: VecDuplicate(tao->solution, &tron->G_New);
72: VecDuplicate(tao->solution, &tron->Work);
73: VecDuplicate(tao->solution, &tao->gradient);
74: VecDuplicate(tao->solution, &tao->stepdirection);
75: if (!tao->XL) {
76: VecDuplicate(tao->solution, &tao->XL);
77: VecSet(tao->XL, PETSC_NINFINITY);
78: }
79: if (!tao->XU) {
80: VecDuplicate(tao->solution, &tao->XU);
81: VecSet(tao->XU, PETSC_INFINITY);
82: }
83: return(0);
84: }
86: static PetscErrorCode TaoSolve_TRON(Tao tao)
87: {
88: TAO_TRON *tron = (TAO_TRON *)tao->data;
89: PetscErrorCode ierr;
90: PetscInt its;
91: TaoLineSearchConvergedReason ls_reason = TAOLINESEARCH_CONTINUE_ITERATING;
92: PetscReal prered,actred,delta,f,f_new,rhok,gdx,xdiff,stepsize;
95: tron->pgstepsize = 1.0;
96: tao->trust = tao->trust0;
97: /* Project the current point onto the feasible set */
98: TaoComputeVariableBounds(tao);
99: TaoLineSearchSetVariableBounds(tao->linesearch,tao->XL,tao->XU);
101: /* Project the initial point onto the feasible region */
102: VecMedian(tao->XL,tao->solution,tao->XU,tao->solution);
104: /* Compute the objective function and gradient */
105: TaoComputeObjectiveAndGradient(tao,tao->solution,&tron->f,tao->gradient);
106: VecNorm(tao->gradient,NORM_2,&tron->gnorm);
107: if (PetscIsInfOrNanReal(tron->f) || PetscIsInfOrNanReal(tron->gnorm)) SETERRQ(PETSC_COMM_SELF,1, "User provided compute function generated Inf or NaN");
109: /* Project the gradient and calculate the norm */
110: VecBoundGradientProjection(tao->gradient,tao->solution,tao->XL,tao->XU,tao->gradient);
111: VecNorm(tao->gradient,NORM_2,&tron->gnorm);
113: /* Initialize trust region radius */
114: tao->trust=tao->trust0;
115: if (tao->trust <= 0) {
116: tao->trust=PetscMax(tron->gnorm*tron->gnorm,1.0);
117: }
119: /* Initialize step sizes for the line searches */
120: tron->pgstepsize=1.0;
121: tron->stepsize=tao->trust;
123: tao->reason = TAO_CONTINUE_ITERATING;
124: TaoLogConvergenceHistory(tao,tron->f,tron->gnorm,0.0,tao->ksp_its);
125: TaoMonitor(tao,tao->niter,tron->f,tron->gnorm,0.0,tron->stepsize);
126: (*tao->ops->convergencetest)(tao,tao->cnvP);
127: while (tao->reason==TAO_CONTINUE_ITERATING){
128: /* Call general purpose update function */
129: if (tao->ops->update) {
130: (*tao->ops->update)(tao, tao->niter, tao->user_update);
131: }
132:
133: /* Perform projected gradient iterations */
134: TronGradientProjections(tao,tron);
136: VecBoundGradientProjection(tao->gradient,tao->solution,tao->XL,tao->XU,tao->gradient);
137: VecNorm(tao->gradient,NORM_2,&tron->gnorm);
139: tao->ksp_its=0;
140: f=tron->f; delta=tao->trust;
141: tron->n_free_last = tron->n_free;
142: TaoComputeHessian(tao,tao->solution,tao->hessian,tao->hessian_pre);
144: /* Generate index set (IS) of which bound constraints are active */
145: ISDestroy(&tron->Free_Local);
146: VecWhichInactive(tao->XL,tao->solution,tao->gradient,tao->XU,PETSC_TRUE,&tron->Free_Local);
147: ISGetSize(tron->Free_Local, &tron->n_free);
149: /* If no free variables */
150: if (tron->n_free == 0) {
151: VecNorm(tao->gradient,NORM_2,&tron->gnorm);
152: TaoLogConvergenceHistory(tao,tron->f,tron->gnorm,0.0,tao->ksp_its);
153: TaoMonitor(tao,tao->niter,tron->f,tron->gnorm,0.0,delta);
154: (*tao->ops->convergencetest)(tao,tao->cnvP);
155: if (!tao->reason) {
156: tao->reason = TAO_CONVERGED_STEPTOL;
157: }
158: break;
159: }
160: /* use free_local to mask/submat gradient, hessian, stepdirection */
161: TaoVecGetSubVec(tao->gradient,tron->Free_Local,tao->subset_type,0.0,&tron->R);
162: TaoVecGetSubVec(tao->gradient,tron->Free_Local,tao->subset_type,0.0,&tron->DXFree);
163: VecSet(tron->DXFree,0.0);
164: VecScale(tron->R, -1.0);
165: TaoMatGetSubMat(tao->hessian, tron->Free_Local, tron->diag, tao->subset_type, &tron->H_sub);
166: if (tao->hessian == tao->hessian_pre) {
167: MatDestroy(&tron->Hpre_sub);
168: PetscObjectReference((PetscObject)(tron->H_sub));
169: tron->Hpre_sub = tron->H_sub;
170: } else {
171: TaoMatGetSubMat(tao->hessian_pre, tron->Free_Local, tron->diag, tao->subset_type,&tron->Hpre_sub);
172: }
173: KSPReset(tao->ksp);
174: KSPSetOperators(tao->ksp, tron->H_sub, tron->Hpre_sub);
175: while (1) {
177: /* Approximately solve the reduced linear system */
178: KSPCGSetRadius(tao->ksp,delta);
180: KSPSolve(tao->ksp, tron->R, tron->DXFree);
181: KSPGetIterationNumber(tao->ksp,&its);
182: tao->ksp_its+=its;
183: tao->ksp_tot_its+=its;
184: VecSet(tao->stepdirection,0.0);
186: /* Add dxfree matrix to compute step direction vector */
187: VecISAXPY(tao->stepdirection,tron->Free_Local,1.0,tron->DXFree);
189: VecDot(tao->gradient, tao->stepdirection, &gdx);
190: VecCopy(tao->solution, tron->X_New);
191: VecCopy(tao->gradient, tron->G_New);
193: stepsize=1.0;f_new=f;
195: TaoLineSearchSetInitialStepLength(tao->linesearch,1.0);
196: TaoLineSearchApply(tao->linesearch, tron->X_New, &f_new, tron->G_New, tao->stepdirection,&stepsize,&ls_reason);
197: TaoAddLineSearchCounts(tao);
199: MatMult(tao->hessian, tao->stepdirection, tron->Work);
200: VecAYPX(tron->Work, 0.5, tao->gradient);
201: VecDot(tao->stepdirection, tron->Work, &prered);
202: actred = f_new - f;
203: if ((PetscAbsScalar(actred) <= 1e-6) && (PetscAbsScalar(prered) <= 1e-6)) {
204: rhok = 1.0;
205: } else if (actred<0) {
206: rhok=PetscAbs(-actred/prered);
207: } else {
208: rhok=0.0;
209: }
211: /* Compare actual improvement to the quadratic model */
212: if (rhok > tron->eta1) { /* Accept the point */
213: /* d = x_new - x */
214: VecCopy(tron->X_New, tao->stepdirection);
215: VecAXPY(tao->stepdirection, -1.0, tao->solution);
217: VecNorm(tao->stepdirection, NORM_2, &xdiff);
218: xdiff *= stepsize;
220: /* Adjust trust region size */
221: if (rhok < tron->eta2 ){
222: delta = PetscMin(xdiff,delta)*tron->sigma1;
223: } else if (rhok > tron->eta4 ){
224: delta= PetscMin(xdiff,delta)*tron->sigma3;
225: } else if (rhok > tron->eta3 ){
226: delta=PetscMin(xdiff,delta)*tron->sigma2;
227: }
228: VecBoundGradientProjection(tron->G_New,tron->X_New, tao->XL, tao->XU, tao->gradient);
229: ISDestroy(&tron->Free_Local);
230: VecWhichInactive(tao->XL,tron->X_New,tao->gradient,tao->XU,PETSC_TRUE,&tron->Free_Local);
231: f=f_new;
232: VecNorm(tao->gradient,NORM_2,&tron->gnorm);
233: VecCopy(tron->X_New, tao->solution);
234: VecCopy(tron->G_New, tao->gradient);
235: break;
236: }
237: else if (delta <= 1e-30) {
238: break;
239: }
240: else {
241: delta /= 4.0;
242: }
243: } /* end linear solve loop */
245: tron->f=f; tron->actred=actred; tao->trust=delta;
246: tao->niter++;
247: TaoLogConvergenceHistory(tao,tron->f,tron->gnorm,0.0,tao->ksp_its);
248: TaoMonitor(tao,tao->niter,tron->f,tron->gnorm,0.0,stepsize);
249: (*tao->ops->convergencetest)(tao,tao->cnvP);
250: } /* END MAIN LOOP */
251: return(0);
252: }
254: static PetscErrorCode TronGradientProjections(Tao tao,TAO_TRON *tron)
255: {
256: PetscErrorCode ierr;
257: PetscInt i;
258: TaoLineSearchConvergedReason ls_reason;
259: PetscReal actred=-1.0,actred_max=0.0;
260: PetscReal f_new;
261: /*
262: The gradient and function value passed into and out of this
263: routine should be current and correct.
265: The free, active, and binding variables should be already identified
266: */
269: for (i=0;i<tron->maxgpits;++i){
271: if (-actred <= (tron->pg_ftol)*actred_max) break;
273: ++tron->gp_iterates;
274: ++tron->total_gp_its;
275: f_new=tron->f;
277: VecCopy(tao->gradient,tao->stepdirection);
278: VecScale(tao->stepdirection,-1.0);
279: TaoLineSearchSetInitialStepLength(tao->linesearch,tron->pgstepsize);
280: TaoLineSearchApply(tao->linesearch, tao->solution, &f_new, tao->gradient, tao->stepdirection,
281: &tron->pgstepsize, &ls_reason);
282: TaoAddLineSearchCounts(tao);
284: VecBoundGradientProjection(tao->gradient,tao->solution,tao->XL,tao->XU,tao->gradient);
285: VecNorm(tao->gradient,NORM_2,&tron->gnorm);
287: /* Update the iterate */
288: actred = f_new - tron->f;
289: actred_max = PetscMax(actred_max,-(f_new - tron->f));
290: tron->f = f_new;
291: }
292: return(0);
293: }
295: static PetscErrorCode TaoComputeDual_TRON(Tao tao, Vec DXL, Vec DXU) {
297: TAO_TRON *tron = (TAO_TRON *)tao->data;
304: if (!tron->Work || !tao->gradient) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"Dual variables don't exist yet or no longer exist.\n");
306: VecBoundGradientProjection(tao->gradient,tao->solution,tao->XL,tao->XU,tron->Work);
307: VecCopy(tron->Work,DXL);
308: VecAXPY(DXL,-1.0,tao->gradient);
309: VecSet(DXU,0.0);
310: VecPointwiseMax(DXL,DXL,DXU);
312: VecCopy(tao->gradient,DXU);
313: VecAXPY(DXU,-1.0,tron->Work);
314: VecSet(tron->Work,0.0);
315: VecPointwiseMin(DXU,tron->Work,DXU);
316: return(0);
317: }
319: /*------------------------------------------------------------*/
320: /*MC
321: TAOTRON - The TRON algorithm is an active-set Newton trust region method
322: for bound-constrained minimization.
324: Options Database Keys:
325: + -tao_tron_maxgpits - maximum number of gradient projections per TRON iterate
326: - -tao_subset_type - "subvec","mask","matrix-free", strategies for handling active-sets
328: Level: beginner
329: M*/
330: PETSC_EXTERN PetscErrorCode TaoCreate_TRON(Tao tao)
331: {
332: TAO_TRON *tron;
334: const char *morethuente_type = TAOLINESEARCHMT;
337: tao->ops->setup = TaoSetup_TRON;
338: tao->ops->solve = TaoSolve_TRON;
339: tao->ops->view = TaoView_TRON;
340: tao->ops->setfromoptions = TaoSetFromOptions_TRON;
341: tao->ops->destroy = TaoDestroy_TRON;
342: tao->ops->computedual = TaoComputeDual_TRON;
344: PetscNewLog(tao,&tron);
345: tao->data = (void*)tron;
347: /* Override default settings (unless already changed) */
348: if (!tao->max_it_changed) tao->max_it = 50;
349: if (!tao->trust0_changed) tao->trust0 = 1.0;
350: if (!tao->steptol_changed) tao->steptol = 0.0;
352: /* Initialize pointers and variables */
353: tron->n = 0;
354: tron->maxgpits = 3;
355: tron->pg_ftol = 0.001;
357: tron->eta1 = 1.0e-4;
358: tron->eta2 = 0.25;
359: tron->eta3 = 0.50;
360: tron->eta4 = 0.90;
362: tron->sigma1 = 0.5;
363: tron->sigma2 = 2.0;
364: tron->sigma3 = 4.0;
366: tron->gp_iterates = 0; /* Cumulative number */
367: tron->total_gp_its = 0;
368: tron->n_free = 0;
370: tron->DXFree=NULL;
371: tron->R=NULL;
372: tron->X_New=NULL;
373: tron->G_New=NULL;
374: tron->Work=NULL;
375: tron->Free_Local=NULL;
376: tron->H_sub=NULL;
377: tron->Hpre_sub=NULL;
378: tao->subset_type = TAO_SUBSET_SUBVEC;
380: TaoLineSearchCreate(((PetscObject)tao)->comm, &tao->linesearch);
381: PetscObjectIncrementTabLevel((PetscObject)tao->linesearch, (PetscObject)tao, 1);
382: TaoLineSearchSetType(tao->linesearch,morethuente_type);
383: TaoLineSearchUseTaoRoutines(tao->linesearch,tao);
384: TaoLineSearchSetOptionsPrefix(tao->linesearch,tao->hdr.prefix);
386: KSPCreate(((PetscObject)tao)->comm, &tao->ksp);
387: PetscObjectIncrementTabLevel((PetscObject)tao->ksp, (PetscObject)tao, 1);
388: KSPSetOptionsPrefix(tao->ksp, tao->hdr.prefix);
389: KSPSetType(tao->ksp,KSPSTCG);
390: return(0);
391: }