xref: /petsc/src/tao/complementarity/impls/asls/asils.c (revision 9f4d3c52fa2fe0bb72fec4f4e85d8e495867af35) !
1 #include <../src/tao/complementarity/impls/ssls/ssls.h>
2 /*
3    Context for ASXLS
4      -- active-set      - reduced matrices formed
5                           - inherit properties of original system
6      -- semismooth (S)  - function not differentiable
7                         - merit function continuously differentiable
8                         - Fischer-Burmeister reformulation of complementarity
9                           - Billups composition for two finite bounds
10      -- infeasible (I)  - iterates not guaranteed to remain within bounds
11      -- feasible (F)    - iterates guaranteed to remain within bounds
12      -- linesearch (LS) - Armijo rule on direction
13 
14    Many other reformulations are possible and combinations of
15    feasible/infeasible and linesearch/trust region are possible.
16 
17    Basic theory
18      Fischer-Burmeister reformulation is semismooth with a continuously
19      differentiable merit function and strongly semismooth if the F has
20      lipschitz continuous derivatives.
21 
22      Every accumulation point generated by the algorithm is a stationary
23      point for the merit function.  Stationary points of the merit function
24      are solutions of the complementarity problem if
25        a.  the stationary point has a BD-regular subdifferential, or
26        b.  the Schur complement F'/F'_ff is a P_0-matrix where ff is the
27            index set corresponding to the free variables.
28 
29      If one of the accumulation points has a BD-regular subdifferential then
30        a.  the entire sequence converges to this accumulation point at
31            a local q-superlinear rate
32        b.  if in addition the reformulation is strongly semismooth near
33            this accumulation point, then the algorithm converges at a
34            local q-quadratic rate.
35 
36    The theory for the feasible version follows from the feasible descent
37    algorithm framework.
38 
39    References:
40      Billups, "Algorithms for Complementarity Problems and Generalized
41        Equations," Ph.D thesis, University of Wisconsin  Madison, 1995.
42      De Luca, Facchinei, Kanzow, "A Semismooth Equation Approach to the
43        Solution of Nonlinear Complementarity Problems," Mathematical
44        Programming, 75, 1996.
45      Ferris, Kanzow, Munson, "Feasible Descent Algorithms for Mixed
46        Complementarity Problems," Mathematical Programming, 86,
47        1999.
48      Fischer, "A Special Newton type Optimization Method," Optimization,
49        24, 1992
50      Munson, Facchinei, Ferris, Fischer, Kanzow, "The Semismooth Algorithm
51        for Large Scale Complementarity Problems," Technical Report,
52        University of Wisconsin  Madison, 1999.
53 */
54 
55 
56 static PetscErrorCode TaoSetUp_ASILS(Tao tao)
57 {
58   TAO_SSLS       *asls = (TAO_SSLS *)tao->data;
59   PetscErrorCode ierr;
60 
61   PetscFunctionBegin;
62   ierr = VecDuplicate(tao->solution,&tao->gradient);CHKERRQ(ierr);
63   ierr = VecDuplicate(tao->solution,&tao->stepdirection);CHKERRQ(ierr);
64   ierr = VecDuplicate(tao->solution,&asls->ff);CHKERRQ(ierr);
65   ierr = VecDuplicate(tao->solution,&asls->dpsi);CHKERRQ(ierr);
66   ierr = VecDuplicate(tao->solution,&asls->da);CHKERRQ(ierr);
67   ierr = VecDuplicate(tao->solution,&asls->db);CHKERRQ(ierr);
68   ierr = VecDuplicate(tao->solution,&asls->t1);CHKERRQ(ierr);
69   ierr = VecDuplicate(tao->solution,&asls->t2);CHKERRQ(ierr);
70   asls->fixed = NULL;
71   asls->free = NULL;
72   asls->J_sub = NULL;
73   asls->Jpre_sub = NULL;
74   asls->w = NULL;
75   asls->r1 = NULL;
76   asls->r2 = NULL;
77   asls->r3 = NULL;
78   asls->dxfree = NULL;
79   PetscFunctionReturn(0);
80 }
81 
82 static PetscErrorCode Tao_ASLS_FunctionGradient(TaoLineSearch ls, Vec X, PetscReal *fcn,  Vec G, void *ptr)
83 {
84   Tao            tao = (Tao)ptr;
85   TAO_SSLS       *asls = (TAO_SSLS *)tao->data;
86   PetscErrorCode ierr;
87 
88   PetscFunctionBegin;
89   ierr = TaoComputeConstraints(tao, X, tao->constraints);CHKERRQ(ierr);
90   ierr = VecFischer(X,tao->constraints,tao->XL,tao->XU,asls->ff);CHKERRQ(ierr);
91   ierr = VecNorm(asls->ff,NORM_2,&asls->merit);CHKERRQ(ierr);
92   *fcn = 0.5*asls->merit*asls->merit;
93 
94   ierr = TaoComputeJacobian(tao,tao->solution,tao->jacobian,tao->jacobian_pre);CHKERRQ(ierr);
95   ierr = MatDFischer(tao->jacobian, tao->solution, tao->constraints,tao->XL, tao->XU, asls->t1, asls->t2,asls->da, asls->db);CHKERRQ(ierr);
96   ierr = VecPointwiseMult(asls->t1, asls->ff, asls->db);CHKERRQ(ierr);
97   ierr = MatMultTranspose(tao->jacobian,asls->t1,G);CHKERRQ(ierr);
98   ierr = VecPointwiseMult(asls->t1, asls->ff, asls->da);CHKERRQ(ierr);
99   ierr = VecAXPY(G,1.0,asls->t1);CHKERRQ(ierr);
100   PetscFunctionReturn(0);
101 }
102 
103 static PetscErrorCode TaoDestroy_ASILS(Tao tao)
104 {
105   TAO_SSLS       *ssls = (TAO_SSLS *)tao->data;
106   PetscErrorCode ierr;
107 
108   PetscFunctionBegin;
109   ierr = VecDestroy(&ssls->ff);CHKERRQ(ierr);
110   ierr = VecDestroy(&ssls->dpsi);CHKERRQ(ierr);
111   ierr = VecDestroy(&ssls->da);CHKERRQ(ierr);
112   ierr = VecDestroy(&ssls->db);CHKERRQ(ierr);
113   ierr = VecDestroy(&ssls->w);CHKERRQ(ierr);
114   ierr = VecDestroy(&ssls->t1);CHKERRQ(ierr);
115   ierr = VecDestroy(&ssls->t2);CHKERRQ(ierr);
116   ierr = VecDestroy(&ssls->r1);CHKERRQ(ierr);
117   ierr = VecDestroy(&ssls->r2);CHKERRQ(ierr);
118   ierr = VecDestroy(&ssls->r3);CHKERRQ(ierr);
119   ierr = VecDestroy(&ssls->dxfree);CHKERRQ(ierr);
120   ierr = MatDestroy(&ssls->J_sub);CHKERRQ(ierr);
121   ierr = MatDestroy(&ssls->Jpre_sub);CHKERRQ(ierr);
122   ierr = ISDestroy(&ssls->fixed);CHKERRQ(ierr);
123   ierr = ISDestroy(&ssls->free);CHKERRQ(ierr);
124   ierr = PetscFree(tao->data);CHKERRQ(ierr);
125   PetscFunctionReturn(0);
126 }
127 
128 static PetscErrorCode TaoSolve_ASILS(Tao tao)
129 {
130   TAO_SSLS                     *asls = (TAO_SSLS *)tao->data;
131   PetscReal                    psi,ndpsi, normd, innerd, t=0;
132   PetscInt                     nf;
133   PetscErrorCode               ierr;
134   TaoConvergedReason           reason;
135   TaoLineSearchConvergedReason ls_reason;
136 
137   PetscFunctionBegin;
138   /* Assume that Setup has been called!
139      Set the structure for the Jacobian and create a linear solver. */
140 
141   ierr = TaoComputeVariableBounds(tao);CHKERRQ(ierr);
142   ierr = TaoLineSearchSetObjectiveAndGradientRoutine(tao->linesearch,Tao_ASLS_FunctionGradient,tao);CHKERRQ(ierr);
143   ierr = TaoLineSearchSetObjectiveRoutine(tao->linesearch,Tao_SSLS_Function,tao);CHKERRQ(ierr);
144 
145   /* Calculate the function value and fischer function value at the
146      current iterate */
147   ierr = TaoLineSearchComputeObjectiveAndGradient(tao->linesearch,tao->solution,&psi,asls->dpsi);CHKERRQ(ierr);
148   ierr = VecNorm(asls->dpsi,NORM_2,&ndpsi);CHKERRQ(ierr);
149 
150   while (1) {
151     /* Check the termination criteria */
152     ierr = PetscInfo3(tao,"iter %D, merit: %g, ||dpsi||: %g\n",tao->niter, (double)asls->merit,  (double)ndpsi);CHKERRQ(ierr);
153     ierr = TaoMonitor(tao, tao->niter, asls->merit, ndpsi, 0.0, t, &reason);CHKERRQ(ierr);
154     if (TAO_CONTINUE_ITERATING != reason) break;
155     tao->niter++;
156 
157     /* We are going to solve a linear system of equations.  We need to
158        set the tolerances for the solve so that we maintain an asymptotic
159        rate of convergence that is superlinear.
160        Note: these tolerances are for the reduced system.  We really need
161        to make sure that the full system satisfies the full-space conditions.
162 
163        This rule gives superlinear asymptotic convergence
164        asls->atol = min(0.5, asls->merit*sqrt(asls->merit));
165        asls->rtol = 0.0;
166 
167        This rule gives quadratic asymptotic convergence
168        asls->atol = min(0.5, asls->merit*asls->merit);
169        asls->rtol = 0.0;
170 
171        Calculate a free and fixed set of variables.  The fixed set of
172        variables are those for the d_b is approximately equal to zero.
173        The definition of approximately changes as we approach the solution
174        to the problem.
175 
176        No one rule is guaranteed to work in all cases.  The following
177        definition is based on the norm of the Jacobian matrix.  If the
178        norm is large, the tolerance becomes smaller. */
179     ierr = MatNorm(tao->jacobian,NORM_1,&asls->identifier);CHKERRQ(ierr);
180     asls->identifier = PetscMin(asls->merit, 1e-2) / (1 + asls->identifier);
181 
182     ierr = VecSet(asls->t1,-asls->identifier);CHKERRQ(ierr);
183     ierr = VecSet(asls->t2, asls->identifier);CHKERRQ(ierr);
184 
185     ierr = ISDestroy(&asls->fixed);CHKERRQ(ierr);
186     ierr = ISDestroy(&asls->free);CHKERRQ(ierr);
187     ierr = VecWhichBetweenOrEqual(asls->t1, asls->db, asls->t2, &asls->fixed);CHKERRQ(ierr);
188     ierr = ISComplementVec(asls->fixed,asls->t1, &asls->free);CHKERRQ(ierr);
189 
190     ierr = ISGetSize(asls->fixed,&nf);CHKERRQ(ierr);
191     ierr = PetscInfo1(tao,"Number of fixed variables: %D\n", nf);CHKERRQ(ierr);
192 
193     /* We now have our partition.  Now calculate the direction in the
194        fixed variable space. */
195     ierr = TaoVecGetSubVec(asls->ff, asls->fixed, tao->subset_type, 0.0, &asls->r1);CHKERRQ(ierr);
196     ierr = TaoVecGetSubVec(asls->da, asls->fixed, tao->subset_type, 1.0, &asls->r2);CHKERRQ(ierr);
197     ierr = VecPointwiseDivide(asls->r1,asls->r1,asls->r2);CHKERRQ(ierr);
198     ierr = VecSet(tao->stepdirection,0.0);CHKERRQ(ierr);
199     ierr = VecISAXPY(tao->stepdirection, asls->fixed,1.0,asls->r1);CHKERRQ(ierr);
200 
201     /* Our direction in the Fixed Variable Set is fixed.  Calculate the
202        information needed for the step in the Free Variable Set.  To
203        do this, we need to know the diagonal perturbation and the
204        right hand side. */
205 
206     ierr = TaoVecGetSubVec(asls->da, asls->free, tao->subset_type, 0.0, &asls->r1);CHKERRQ(ierr);
207     ierr = TaoVecGetSubVec(asls->ff, asls->free, tao->subset_type, 0.0, &asls->r2);CHKERRQ(ierr);
208     ierr = TaoVecGetSubVec(asls->db, asls->free, tao->subset_type, 1.0, &asls->r3);CHKERRQ(ierr);
209     ierr = VecPointwiseDivide(asls->r1,asls->r1, asls->r3);CHKERRQ(ierr);
210     ierr = VecPointwiseDivide(asls->r2,asls->r2, asls->r3);CHKERRQ(ierr);
211 
212     /* r1 is the diagonal perturbation
213        r2 is the right hand side
214        r3 is no longer needed
215 
216        Now need to modify r2 for our direction choice in the fixed
217        variable set:  calculate t1 = J*d, take the reduced vector
218        of t1 and modify r2. */
219 
220     ierr = MatMult(tao->jacobian, tao->stepdirection, asls->t1);CHKERRQ(ierr);
221     ierr = TaoVecGetSubVec(asls->t1,asls->free,tao->subset_type,0.0,&asls->r3);CHKERRQ(ierr);
222     ierr = VecAXPY(asls->r2, -1.0, asls->r3);CHKERRQ(ierr);
223 
224     /* Calculate the reduced problem matrix and the direction */
225     if (!asls->w && (tao->subset_type == TAO_SUBSET_MASK || tao->subset_type == TAO_SUBSET_MATRIXFREE)) {
226       ierr = VecDuplicate(tao->solution, &asls->w);CHKERRQ(ierr);
227     }
228     ierr = TaoMatGetSubMat(tao->jacobian, asls->free, asls->w, tao->subset_type,&asls->J_sub);CHKERRQ(ierr);
229     if (tao->jacobian != tao->jacobian_pre) {
230       ierr = TaoMatGetSubMat(tao->jacobian_pre, asls->free, asls->w, tao->subset_type, &asls->Jpre_sub);CHKERRQ(ierr);
231     } else {
232       ierr = MatDestroy(&asls->Jpre_sub);CHKERRQ(ierr);
233       asls->Jpre_sub = asls->J_sub;
234       ierr = PetscObjectReference((PetscObject)(asls->Jpre_sub));CHKERRQ(ierr);
235     }
236     ierr = MatDiagonalSet(asls->J_sub, asls->r1,ADD_VALUES);CHKERRQ(ierr);
237     ierr = TaoVecGetSubVec(tao->stepdirection, asls->free, tao->subset_type, 0.0, &asls->dxfree);CHKERRQ(ierr);
238     ierr = VecSet(asls->dxfree, 0.0);CHKERRQ(ierr);
239 
240     /* Calculate the reduced direction.  (Really negative of Newton
241        direction.  Therefore, rest of the code uses -d.) */
242     ierr = KSPReset(tao->ksp);CHKERRQ(ierr);
243     ierr = KSPSetOperators(tao->ksp, asls->J_sub, asls->Jpre_sub);CHKERRQ(ierr);
244     ierr = KSPSolve(tao->ksp, asls->r2, asls->dxfree);CHKERRQ(ierr);
245     ierr = KSPGetIterationNumber(tao->ksp,&tao->ksp_its);CHKERRQ(ierr);
246     tao->ksp_tot_its+=tao->ksp_its;
247 
248     /* Add the direction in the free variables back into the real direction. */
249     ierr = VecISAXPY(tao->stepdirection, asls->free, 1.0,asls->dxfree);CHKERRQ(ierr);
250 
251     /* Check the real direction for descent and if not, use the negative
252        gradient direction. */
253     ierr = VecNorm(tao->stepdirection, NORM_2, &normd);CHKERRQ(ierr);
254     ierr = VecDot(tao->stepdirection, asls->dpsi, &innerd);CHKERRQ(ierr);
255 
256     if (innerd <= asls->delta*PetscPowReal(normd, asls->rho)) {
257       ierr = PetscInfo1(tao,"Gradient direction: %5.4e.\n", (double)innerd);CHKERRQ(ierr);
258       ierr = PetscInfo1(tao, "Iteration %D: newton direction not descent\n", tao->niter);CHKERRQ(ierr);
259       ierr = VecCopy(asls->dpsi, tao->stepdirection);CHKERRQ(ierr);
260       ierr = VecDot(asls->dpsi, tao->stepdirection, &innerd);CHKERRQ(ierr);
261     }
262 
263     ierr = VecScale(tao->stepdirection, -1.0);CHKERRQ(ierr);
264     innerd = -innerd;
265 
266     /* We now have a correct descent direction.  Apply a linesearch to
267        find the new iterate. */
268     ierr = TaoLineSearchSetInitialStepLength(tao->linesearch, 1.0);CHKERRQ(ierr);
269     ierr = TaoLineSearchApply(tao->linesearch, tao->solution, &psi,asls->dpsi, tao->stepdirection, &t, &ls_reason);CHKERRQ(ierr);
270     ierr = VecNorm(asls->dpsi, NORM_2, &ndpsi);CHKERRQ(ierr);
271   }
272   PetscFunctionReturn(0);
273 }
274 
275 /* ---------------------------------------------------------- */
276 /*MC
277    TAOASILS - Active-set infeasible linesearch algorithm for solving
278        complementarity constraints
279 
280    Options Database Keys:
281 + -tao_ssls_delta - descent test fraction
282 - -tao_ssls_rho - descent test power
283 
284   Level: beginner
285 M*/
286 PETSC_EXTERN PetscErrorCode TaoCreate_ASILS(Tao tao)
287 {
288   TAO_SSLS       *asls;
289   PetscErrorCode ierr;
290   const char     *armijo_type = TAOLINESEARCHARMIJO;
291 
292   PetscFunctionBegin;
293   ierr = PetscNewLog(tao,&asls);CHKERRQ(ierr);
294   tao->data = (void*)asls;
295   tao->ops->solve = TaoSolve_ASILS;
296   tao->ops->setup = TaoSetUp_ASILS;
297   tao->ops->view = TaoView_SSLS;
298   tao->ops->setfromoptions = TaoSetFromOptions_SSLS;
299   tao->ops->destroy = TaoDestroy_ASILS;
300   tao->subset_type = TAO_SUBSET_SUBVEC;
301   asls->delta = 1e-10;
302   asls->rho = 2.1;
303   asls->fixed = NULL;
304   asls->free = NULL;
305   asls->J_sub = NULL;
306   asls->Jpre_sub = NULL;
307   asls->w = NULL;
308   asls->r1 = NULL;
309   asls->r2 = NULL;
310   asls->r3 = NULL;
311   asls->t1 = NULL;
312   asls->t2 = NULL;
313   asls->dxfree = NULL;
314 
315   asls->identifier = 1e-5;
316 
317   ierr = TaoLineSearchCreate(((PetscObject)tao)->comm, &tao->linesearch);CHKERRQ(ierr);
318   ierr = TaoLineSearchSetType(tao->linesearch, armijo_type);CHKERRQ(ierr);
319   ierr = TaoLineSearchSetOptionsPrefix(tao->linesearch,tao->hdr.prefix);CHKERRQ(ierr);
320   ierr = TaoLineSearchSetFromOptions(tao->linesearch);CHKERRQ(ierr);
321 
322   ierr = KSPCreate(((PetscObject)tao)->comm, &tao->ksp);CHKERRQ(ierr);
323   ierr = KSPSetOptionsPrefix(tao->ksp,tao->hdr.prefix);CHKERRQ(ierr);
324   ierr = KSPSetFromOptions(tao->ksp);CHKERRQ(ierr);
325 
326   /* Override default settings (unless already changed) */
327   if (!tao->max_it_changed) tao->max_it = 2000;
328   if (!tao->max_funcs_changed) tao->max_funcs = 4000;
329   if (!tao->gttol_changed) tao->gttol = 0;
330   if (!tao->grtol_changed) tao->grtol = 0;
331 #if defined(PETSC_USE_REAL_SINGLE)
332   if (!tao->gatol_changed) tao->gatol = 1.0e-6;
333   if (!tao->fmin_changed)  tao->fmin = 1.0e-4;
334 #else
335   if (!tao->gatol_changed) tao->gatol = 1.0e-16;
336   if (!tao->fmin_changed) tao->fmin = 1.0e-8;
337 #endif
338   PetscFunctionReturn(0);
339 }
340 
341 
342