xref: /petsc/src/ts/impls/explicit/ssp/ssp.c (revision bd7c7dddb24a67b8eebe9f5ca9730c2e146ac916)
1 #define PETSCTS_DLL
2 
3 /*
4        Code for Timestepping with explicit SSP.
5 */
6 #include "private/tsimpl.h"                /*I   "petscts.h"   I*/
7 
8 PetscFList TSSSPList = 0;
9 #define TSSSPType char*
10 
11 #define TSSSPRKS2  "rks2"
12 #define TSSSPRKS3  "rks3"
13 #define TSSSPRK104 "rk104"
14 
15 typedef struct {
16   PetscErrorCode (*onestep)(TS,PetscReal,PetscReal,Vec);
17   PetscInt nstages;
18   Vec xdot;
19   Vec *work;
20   PetscInt nwork;
21   PetscTruth workout;
22 } TS_SSP;
23 
24 
25 #undef __FUNCT__
26 #define __FUNCT__ "SSPGetWorkVectors"
27 static PetscErrorCode SSPGetWorkVectors(TS ts,PetscInt n,Vec **work)
28 {
29   TS_SSP *ssp = (TS_SSP*)ts->data;
30   PetscErrorCode ierr;
31 
32   PetscFunctionBegin;
33   if (ssp->workout) SETERRQ(PETSC_ERR_PLIB,"Work vectors already gotten");
34   if (ssp->nwork < n) {
35     if (ssp->nwork > 0) {
36       ierr = VecDestroyVecs(ssp->work,ssp->nwork);CHKERRQ(ierr);
37     }
38     ierr = VecDuplicateVecs(ts->vec_sol,n,&ssp->work);CHKERRQ(ierr);
39     ssp->nwork = n;
40   }
41   *work = ssp->work;
42   ssp->workout = PETSC_TRUE;
43   PetscFunctionReturn(0);
44 }
45 
46 #undef __FUNCT__
47 #define __FUNCT__ "SSPRestoreWorkVectors"
48 static PetscErrorCode SSPRestoreWorkVectors(TS ts,PetscInt n,Vec **work)
49 {
50   TS_SSP *ssp = (TS_SSP*)ts->data;
51 
52   PetscFunctionBegin;
53   if (!ssp->workout) SETERRQ(PETSC_ERR_ORDER,"Work vectors have not been gotten");
54   if (*work != ssp->work) SETERRQ(PETSC_ERR_PLIB,"Wrong work vectors checked out");
55   ssp->workout = PETSC_FALSE;
56   *work = PETSC_NULL;
57   PetscFunctionReturn(0);
58 }
59 
60 
61 #undef __FUNCT__
62 #define __FUNCT__ "SSPStep_RK_2"
63 /* Optimal second order SSP Runge-Kutta, low-storage, c_eff=(s-1)/s */
64 /* Pseudocode 2 of Ketcheson 2008 */
65 static PetscErrorCode SSPStep_RK_2(TS ts,PetscReal t0,PetscReal dt,Vec sol)
66 {
67   TS_SSP *ssp = (TS_SSP*)ts->data;
68   Vec *work,F;
69   PetscInt i,s;
70   PetscErrorCode ierr;
71 
72   PetscFunctionBegin;
73   s = ssp->nstages;
74   ierr = SSPGetWorkVectors(ts,2,&work);CHKERRQ(ierr);
75   F = work[1];
76   ierr = VecCopy(sol,work[0]);CHKERRQ(ierr);
77   for (i=0; i<s-1; i++) {
78     ierr = TSComputeRHSFunction(ts,t0+dt*(i/(s-1.)),work[0],F);CHKERRQ(ierr);
79     ierr = VecAXPY(work[0],dt/(s-1.),F);CHKERRQ(ierr);
80   }
81   ierr = TSComputeRHSFunction(ts,t0+dt,work[0],F);CHKERRQ(ierr);
82   ierr = VecAXPBYPCZ(sol,(s-1.)/s,dt/s,1./s,work[0],F);CHKERRQ(ierr);
83   ierr = SSPRestoreWorkVectors(ts,2,&work);CHKERRQ(ierr);
84   PetscFunctionReturn(0);
85 }
86 
87 #undef __FUNCT__
88 #define __FUNCT__ "SSPStep_RK_3"
89 /* Optimal third order SSP Runge-Kutta, low-storage, c_eff=(sqrt(s)-1)/sqrt(s), where sqrt(s) is an integer */
90 /* Pseudocode 2 of Ketcheson 2008 */
91 static PetscErrorCode SSPStep_RK_3(TS ts,PetscReal t0,PetscReal dt,Vec sol)
92 {
93   TS_SSP *ssp = (TS_SSP*)ts->data;
94   Vec *work,F;
95   PetscInt i,s,n,r;
96   PetscReal c;
97   PetscErrorCode ierr;
98 
99   PetscFunctionBegin;
100   s = ssp->nstages;
101   n = (PetscInt)sqrt(s);
102   r = s-n;
103   if (n*n != s) SETERRQ1(PETSC_ERR_SUP,"No support for optimal third order schemes with %d stages, must be a square number at least 4",s);
104   ierr = SSPGetWorkVectors(ts,3,&work);CHKERRQ(ierr);
105   F = work[2];
106   ierr = VecCopy(sol,work[0]);CHKERRQ(ierr);
107   for (i=0; i<(n-1)*(n-2)/2; i++) {
108     c = (i<n*(n+1)/2) ? 1.*i/(s-n) : (1.*i-n)/(s-n);
109     ierr = TSComputeRHSFunction(ts,t0+c*dt,work[0],F);CHKERRQ(ierr);
110     ierr = VecAXPY(work[0],dt/r,F);CHKERRQ(ierr);
111   }
112   ierr = VecCopy(work[0],work[1]);CHKERRQ(ierr);
113   for ( ; i<n*(n+1)/2-1; i++) {
114     c = (i<n*(n+1)/2) ? 1.*i/(s-n) : (1.*i-n)/(s-n);
115     ierr = TSComputeRHSFunction(ts,t0+c*dt,work[0],F);CHKERRQ(ierr);
116     ierr = VecAXPY(work[0],dt/r,F);CHKERRQ(ierr);
117   }
118   {
119     c = (i<n*(n+1)/2) ? 1.*i/(s-n) : (1.*i-n)/(s-n);
120     ierr = TSComputeRHSFunction(ts,t0+c*dt,work[0],F);CHKERRQ(ierr);
121     ierr = VecAXPBYPCZ(work[0],1.*n/(2*n-1.),(n-1.)*dt/(r*(2*n-1)),(n-1.)/(2*n-1.),work[1],F);CHKERRQ(ierr);
122     i++;
123   }
124   for ( ; i<s; i++) {
125     c = (i<n*(n+1)/2) ? 1.*i/(s-n) : (1.*i-n)/(s-n);
126     ierr = TSComputeRHSFunction(ts,t0+c*dt,work[0],F);CHKERRQ(ierr);
127     ierr = VecAXPY(work[0],dt/r,F);CHKERRQ(ierr);
128   }
129   ierr = VecCopy(work[0],sol);CHKERRQ(ierr);
130   ierr = SSPRestoreWorkVectors(ts,3,&work);CHKERRQ(ierr);
131   PetscFunctionReturn(0);
132 }
133 
134 #undef __FUNCT__
135 #define __FUNCT__ "SSPStep_RK_10_4"
136 /* Optimal fourth order SSP Runge-Kutta, low-storage (2N), c_eff=0.6 */
137 /* SSPRK(10,4), Pseudocode 3 of Ketcheson 2008 */
138 static PetscErrorCode SSPStep_RK_10_4(TS ts,PetscReal t0,PetscReal dt,Vec sol)
139 {
140   TS_SSP *ssp = (TS_SSP*)ts->data;
141   const PetscReal c[10] = {0, 1./6, 2./6, 3./6, 4./6, 2./6, 3./6, 4./6, 5./6, 1};
142   Vec *work,F;
143   PetscInt i,s;
144   PetscErrorCode ierr;
145 
146   PetscFunctionBegin;
147   s = ssp->nstages;
148   ierr = SSPGetWorkVectors(ts,3,&work);CHKERRQ(ierr);
149   F = work[2];
150   ierr = VecCopy(sol,work[0]);CHKERRQ(ierr);
151   for (i=0; i<5; i++) {
152     ierr = TSComputeRHSFunction(ts,t0+c[i],work[0],F);CHKERRQ(ierr);
153     ierr = VecAXPY(work[0],dt/6,F);CHKERRQ(ierr);
154   }
155   ierr = VecAXPBYPCZ(work[1],1./25,9./25,0,sol,work[0]);CHKERRQ(ierr);
156   ierr = VecAXPBY(work[0],15,-5,work[1]);CHKERRQ(ierr);
157   for ( ; i<9; i++) {
158     ierr = TSComputeRHSFunction(ts,t0+c[i],work[0],F);CHKERRQ(ierr);
159     ierr = VecAXPY(work[0],dt/6,F);CHKERRQ(ierr);
160   }
161   ierr = TSComputeRHSFunction(ts,t0+dt,work[0],F);CHKERRQ(ierr);
162   ierr = VecAXPBYPCZ(work[1],3./5,dt/10,1,work[0],F);CHKERRQ(ierr);
163   ierr = VecCopy(work[1],sol);CHKERRQ(ierr);
164   ierr = SSPRestoreWorkVectors(ts,3,&work);CHKERRQ(ierr);
165   PetscFunctionReturn(0);
166 }
167 
168 
169 #undef __FUNCT__
170 #define __FUNCT__ "TSSetUp_SSP"
171 static PetscErrorCode TSSetUp_SSP(TS ts)
172 {
173   /* TS_SSP       *ssp = (TS_SSP*)ts->data; */
174   /* PetscErrorCode ierr; */
175 
176   PetscFunctionBegin;
177   PetscFunctionReturn(0);
178 }
179 
180 #undef __FUNCT__
181 #define __FUNCT__ "TSStep_SSP"
182 static PetscErrorCode TSStep_SSP(TS ts,PetscInt *steps,PetscReal *ptime)
183 {
184   TS_SSP        *ssp = (TS_SSP*)ts->data;
185   Vec            sol = ts->vec_sol;
186   PetscErrorCode ierr;
187   PetscInt       i,max_steps = ts->max_steps;
188 
189   PetscFunctionBegin;
190   *steps = -ts->steps;
191   ierr = TSMonitor(ts,ts->steps,ts->ptime,sol);CHKERRQ(ierr);
192 
193   for (i=0; i<max_steps; i++) {
194     PetscReal dt = ts->time_step;
195 
196     ts->ptime += dt;
197     ierr = (*ssp->onestep)(ts,ts->ptime-dt,dt,sol);CHKERRQ(ierr);
198     ts->steps++;
199     ierr = TSMonitor(ts,ts->steps,ts->ptime,sol);CHKERRQ(ierr);
200     if (ts->ptime > ts->max_time) break;
201   }
202 
203   *steps += ts->steps;
204   *ptime  = ts->ptime;
205   PetscFunctionReturn(0);
206 }
207 /*------------------------------------------------------------*/
208 #undef __FUNCT__
209 #define __FUNCT__ "TSDestroy_SSP"
210 static PetscErrorCode TSDestroy_SSP(TS ts)
211 {
212   TS_SSP       *ssp = (TS_SSP*)ts->data;
213   PetscErrorCode ierr;
214 
215   PetscFunctionBegin;
216   if (ssp->work) {ierr = VecDestroyVecs(ssp->work,ssp->nwork);CHKERRQ(ierr);}
217   ierr = PetscFree(ssp);CHKERRQ(ierr);
218   PetscFunctionReturn(0);
219 }
220 /*------------------------------------------------------------*/
221 
222 #undef __FUNCT__
223 #define __FUNCT__ "TSSSPSetType"
224 static PetscErrorCode TSSSPSetType(TS ts,const TSSSPType type)
225 {
226   PetscErrorCode ierr,(*r)(TS,PetscReal,PetscReal,Vec);
227   TS_SSP *ssp = (TS_SSP*)ts->data;
228 
229   PetscFunctionBegin;
230   ierr = PetscFListFind(TSSSPList,((PetscObject)ts)->comm,type,(void(**)(void))&r);CHKERRQ(ierr);
231   if (!r) SETERRQ1(PETSC_ERR_ARG_UNKNOWN_TYPE,"Unknown TS_SSP type %s given",type);
232   ssp->onestep = r;
233   PetscFunctionReturn(0);
234 }
235 
236 #undef __FUNCT__
237 #define __FUNCT__ "TSSetFromOptions_SSP"
238 static PetscErrorCode TSSetFromOptions_SSP(TS ts)
239 {
240   char tname[256] = TSSSPRKS2;
241   TS_SSP *ssp = (TS_SSP*)ts->data;
242   PetscErrorCode ierr;
243   PetscTruth flg;
244 
245   PetscFunctionBegin;
246   ierr = PetscOptionsHead("SSP ODE solver options");CHKERRQ(ierr);
247   {
248     ierr = PetscOptionsList("-ts_ssp_type","Type of SSP method","TSSSPSetType",TSSSPList,tname,tname,sizeof(tname),&flg);CHKERRQ(ierr);
249     if (flg) {
250       ierr = TSSSPSetType(ts,tname);CHKERRQ(ierr);
251     }
252     ierr = PetscOptionsInt("-ts_ssp_nstages","Number of stages","TSSSPSetNumStages",ssp->nstages,&ssp->nstages,PETSC_NULL);CHKERRQ(ierr);
253   }
254   ierr = PetscOptionsTail();CHKERRQ(ierr);
255   PetscFunctionReturn(0);
256 }
257 
258 #undef __FUNCT__
259 #define __FUNCT__ "TSView_SSP"
260 static PetscErrorCode TSView_SSP(TS ts,PetscViewer viewer)
261 {
262   PetscFunctionBegin;
263   PetscFunctionReturn(0);
264 }
265 
266 /* ------------------------------------------------------------ */
267 
268 /*MC
269       TSSSP - Explicit strong stability preserving ODE solver
270 
271   Most hyperbolic conservation laws have exact solutions that are total variation diminishing (TVD) or total variation
272   bounded (TVB) although these solutions often contain discontinuities.  Spatial discretizations such as Godunov's
273   scheme and high-resolution finite volume methods (TVD limiters, ENO/WENO) are designed to preserve these properties,
274   but they are usually formulated using a forward Euler time discretization or by coupling the space and time
275   discretization as in the classical Lax-Wendroff scheme.  When the space and time discretization is coupled, it is very
276   difficult to produce schemes with high temporal accuracy while preserving TVD properties.  An alternative is the
277   semidiscrete formulation where we choose a spatial discretization that is TVD with forward Euler and then choose a
278   time discretization that preserves the TVD property.  Such integrators are called strong stability preserving (SSP).
279 
280   Let c_eff be the minimum number of function evaluations required to step as far as one step of forward Euler while
281   still being SSP.  Some theoretical bounds
282 
283   1. There are no explicit methods with c_eff > 1.
284 
285   2. There are no explicit methods beyond order 4 (for nonlinear problems) and c_eff > 0.
286 
287   3. There are no implicit methods with order greater than 1 and c_eff > 2.
288 
289   This integrator provides Runge-Kutta methods of order 2, 3, and 4 with maximal values of c_eff.  More stages allows
290   for larger values of c_eff which improves efficiency.  These implementations are low-memory and only use 2 or 3 work
291   vectors regardless of the total number of stages, so e.g. 25-stage 3rd order methods may be an excellent choice.
292 
293   Methods can be chosen with -ts_ssp_type {rks2,rks3,rk104}
294 
295   rks2: Second order methods with any number s>1 of stages.  c_eff = (s-1)/s
296 
297   rks3: Third order methods with s=n^2 stages, n>1.  c_eff = (s-n)/s
298 
299   rk104: A 10-stage fourth order method.  c_eff = 0.6
300 
301   Level: beginner
302 
303 .seealso:  TSCreate(), TS, TSSetType()
304 
305 M*/
306 EXTERN_C_BEGIN
307 #undef __FUNCT__
308 #define __FUNCT__ "TSCreate_SSP"
309 PetscErrorCode PETSCTS_DLLEXPORT TSCreate_SSP(TS ts)
310 {
311   TS_SSP       *ssp;
312   PetscErrorCode ierr;
313 
314   PetscFunctionBegin;
315   if (!TSSSPList) {
316     ierr = PetscFListAdd(&TSSSPList,TSSSPRKS2,  "SSPStep_RK_2",   (void(*)(void))SSPStep_RK_2);CHKERRQ(ierr);
317     ierr = PetscFListAdd(&TSSSPList,TSSSPRKS3,  "SSPStep_RK_3",   (void(*)(void))SSPStep_RK_3);CHKERRQ(ierr);
318     ierr = PetscFListAdd(&TSSSPList,TSSSPRK104, "SSPStep_RK_10_4",(void(*)(void))SSPStep_RK_10_4);CHKERRQ(ierr);
319   }
320 
321   ts->ops->setup           = TSSetUp_SSP;
322   ts->ops->step            = TSStep_SSP;
323   ts->ops->destroy         = TSDestroy_SSP;
324   ts->ops->setfromoptions  = TSSetFromOptions_SSP;
325   ts->ops->view            = TSView_SSP;
326 
327   ierr = PetscNewLog(ts,TS_SSP,&ssp);CHKERRQ(ierr);
328   ts->data = (void*)ssp;
329 
330   ierr = TSSSPSetType(ts,TSSSPRKS2);CHKERRQ(ierr);
331   ssp->nstages = 5;
332   PetscFunctionReturn(0);
333 }
334 EXTERN_C_END
335 
336 
337 
338 
339