Actual source code: glle.c
petsc-3.13.6 2020-09-29
2: #include <../src/ts/impls/implicit/glle/glle.h>
3: #include <petscdm.h>
4: #include <petscblaslapack.h>
6: static const char *TSGLLEErrorDirections[] = {"FORWARD","BACKWARD","TSGLLEErrorDirection","TSGLLEERROR_",0};
7: static PetscFunctionList TSGLLEList;
8: static PetscFunctionList TSGLLEAcceptList;
9: static PetscBool TSGLLEPackageInitialized;
10: static PetscBool TSGLLERegisterAllCalled;
12: /* This function is pure */
13: static PetscScalar Factorial(PetscInt n)
14: {
15: PetscInt i;
16: if (n < 12) { /* Can compute with 32-bit integers */
17: PetscInt f = 1;
18: for (i=2; i<=n; i++) f *= i;
19: return (PetscScalar)f;
20: } else {
21: PetscScalar f = 1.;
22: for (i=2; i<=n; i++) f *= (PetscScalar)i;
23: return f;
24: }
25: }
27: /* This function is pure */
28: static PetscScalar CPowF(PetscScalar c,PetscInt p)
29: {
30: return PetscPowRealInt(PetscRealPart(c),p)/Factorial(p);
31: }
33: static PetscErrorCode TSGLLEGetVecs(TS ts,DM dm,Vec *Z,Vec *Ydotstage)
34: {
35: TS_GLLE *gl = (TS_GLLE*)ts->data;
39: if (Z) {
40: if (dm && dm != ts->dm) {
41: DMGetNamedGlobalVector(dm,"TSGLLE_Z",Z);
42: } else *Z = gl->Z;
43: }
44: if (Ydotstage) {
45: if (dm && dm != ts->dm) {
46: DMGetNamedGlobalVector(dm,"TSGLLE_Ydot",Ydotstage);
47: } else *Ydotstage = gl->Ydot[gl->stage];
48: }
49: return(0);
50: }
53: static PetscErrorCode TSGLLERestoreVecs(TS ts,DM dm,Vec *Z,Vec *Ydotstage)
54: {
58: if (Z) {
59: if (dm && dm != ts->dm) {
60: DMRestoreNamedGlobalVector(dm,"TSGLLE_Z",Z);
61: }
62: }
63: if (Ydotstage) {
65: if (dm && dm != ts->dm) {
66: DMRestoreNamedGlobalVector(dm,"TSGLLE_Ydot",Ydotstage);
67: }
68: }
69: return(0);
70: }
72: static PetscErrorCode DMCoarsenHook_TSGLLE(DM fine,DM coarse,void *ctx)
73: {
75: return(0);
76: }
78: static PetscErrorCode DMRestrictHook_TSGLLE(DM fine,Mat restrct,Vec rscale,Mat inject,DM coarse,void *ctx)
79: {
80: TS ts = (TS)ctx;
82: Vec Ydot,Ydot_c;
85: TSGLLEGetVecs(ts,fine,NULL,&Ydot);
86: TSGLLEGetVecs(ts,coarse,NULL,&Ydot_c);
87: MatRestrict(restrct,Ydot,Ydot_c);
88: VecPointwiseMult(Ydot_c,rscale,Ydot_c);
89: TSGLLERestoreVecs(ts,fine,NULL,&Ydot);
90: TSGLLERestoreVecs(ts,coarse,NULL,&Ydot_c);
91: return(0);
92: }
94: static PetscErrorCode DMSubDomainHook_TSGLLE(DM dm,DM subdm,void *ctx)
95: {
97: return(0);
98: }
100: static PetscErrorCode DMSubDomainRestrictHook_TSGLLE(DM dm,VecScatter gscat, VecScatter lscat,DM subdm,void *ctx)
101: {
102: TS ts = (TS)ctx;
104: Vec Ydot,Ydot_s;
107: TSGLLEGetVecs(ts,dm,NULL,&Ydot);
108: TSGLLEGetVecs(ts,subdm,NULL,&Ydot_s);
110: VecScatterBegin(gscat,Ydot,Ydot_s,INSERT_VALUES,SCATTER_FORWARD);
111: VecScatterEnd(gscat,Ydot,Ydot_s,INSERT_VALUES,SCATTER_FORWARD);
113: TSGLLERestoreVecs(ts,dm,NULL,&Ydot);
114: TSGLLERestoreVecs(ts,subdm,NULL,&Ydot_s);
115: return(0);
116: }
118: static PetscErrorCode TSGLLESchemeCreate(PetscInt p,PetscInt q,PetscInt r,PetscInt s,const PetscScalar *c,
119: const PetscScalar *a,const PetscScalar *b,const PetscScalar *u,const PetscScalar *v,TSGLLEScheme *inscheme)
120: {
121: TSGLLEScheme scheme;
122: PetscInt j;
126: if (p < 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"Scheme order must be positive");
127: if (r < 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"At least one item must be carried between steps");
128: if (s < 1) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"At least one stage is required");
130: *inscheme = 0;
131: PetscNew(&scheme);
132: scheme->p = p;
133: scheme->q = q;
134: scheme->r = r;
135: scheme->s = s;
137: PetscMalloc5(s,&scheme->c,s*s,&scheme->a,r*s,&scheme->b,r*s,&scheme->u,r*r,&scheme->v);
138: PetscArraycpy(scheme->c,c,s);
139: for (j=0; j<s*s; j++) scheme->a[j] = (PetscAbsScalar(a[j]) < 1e-12) ? 0 : a[j];
140: for (j=0; j<r*s; j++) scheme->b[j] = (PetscAbsScalar(b[j]) < 1e-12) ? 0 : b[j];
141: for (j=0; j<s*r; j++) scheme->u[j] = (PetscAbsScalar(u[j]) < 1e-12) ? 0 : u[j];
142: for (j=0; j<r*r; j++) scheme->v[j] = (PetscAbsScalar(v[j]) < 1e-12) ? 0 : v[j];
144: PetscMalloc6(r,&scheme->alpha,r,&scheme->beta,r,&scheme->gamma,3*s,&scheme->phi,3*r,&scheme->psi,r,&scheme->stage_error);
145: {
146: PetscInt i,j,k,ss=s+2;
147: PetscBLASInt m,n,one=1,*ipiv,lwork=4*((s+3)*3+3),info,ldb;
148: PetscReal rcond,*sing,*workreal;
149: PetscScalar *ImV,*H,*bmat,*workscalar,*c=scheme->c,*a=scheme->a,*b=scheme->b,*u=scheme->u,*v=scheme->v;
150: PetscBLASInt rank;
151: PetscMalloc7(PetscSqr(r),&ImV,3*s,&H,3*ss,&bmat,lwork,&workscalar,5*(3+r),&workreal,r+s,&sing,r+s,&ipiv);
153: /* column-major input */
154: for (i=0; i<r-1; i++) {
155: for (j=0; j<r-1; j++) ImV[i+j*r] = 1.0*(i==j) - v[(i+1)*r+j+1];
156: }
157: /* Build right hand side for alpha (tp - glm.B(2:end,:)*(glm.c.^(p)./factorial(p))) */
158: for (i=1; i<r; i++) {
159: scheme->alpha[i] = 1./Factorial(p+1-i);
160: for (j=0; j<s; j++) scheme->alpha[i] -= b[i*s+j]*CPowF(c[j],p);
161: }
162: PetscBLASIntCast(r-1,&m);
163: PetscBLASIntCast(r,&n);
164: PetscStackCallBLAS("LAPACKgesv",LAPACKgesv_(&m,&one,ImV,&n,ipiv,scheme->alpha+1,&n,&info));
165: if (info < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_LIB,"Bad argument to GESV");
166: if (info > 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_MAT_LU_ZRPVT,"Bad LU factorization");
168: /* Build right hand side for beta (tp1 - glm.B(2:end,:)*(glm.c.^(p+1)./factorial(p+1)) - e.alpha) */
169: for (i=1; i<r; i++) {
170: scheme->beta[i] = 1./Factorial(p+2-i) - scheme->alpha[i];
171: for (j=0; j<s; j++) scheme->beta[i] -= b[i*s+j]*CPowF(c[j],p+1);
172: }
173: PetscStackCallBLAS("LAPACKgetrs",LAPACKgetrs_("No transpose",&m,&one,ImV,&n,ipiv,scheme->beta+1,&n,&info));
174: if (info < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_LIB,"Bad argument to GETRS");
175: if (info > 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_LIB,"Should not happen");
177: /* Build stage_error vector
178: xi = glm.c.^(p+1)/factorial(p+1) - glm.A*glm.c.^p/factorial(p) + glm.U(:,2:end)*e.alpha;
179: */
180: for (i=0; i<s; i++) {
181: scheme->stage_error[i] = CPowF(c[i],p+1);
182: for (j=0; j<s; j++) scheme->stage_error[i] -= a[i*s+j]*CPowF(c[j],p);
183: for (j=1; j<r; j++) scheme->stage_error[i] += u[i*r+j]*scheme->alpha[j];
184: }
186: /* alpha[0] (epsilon in B,J,W 2007)
187: epsilon = 1/factorial(p+1) - B(1,:)*c.^p/factorial(p) + V(1,2:end)*e.alpha;
188: */
189: scheme->alpha[0] = 1./Factorial(p+1);
190: for (j=0; j<s; j++) scheme->alpha[0] -= b[0*s+j]*CPowF(c[j],p);
191: for (j=1; j<r; j++) scheme->alpha[0] += v[0*r+j]*scheme->alpha[j];
193: /* right hand side for gamma (glm.B(2:end,:)*e.xi - e.epsilon*eye(s-1,1)) */
194: for (i=1; i<r; i++) {
195: scheme->gamma[i] = (i==1 ? -1. : 0)*scheme->alpha[0];
196: for (j=0; j<s; j++) scheme->gamma[i] += b[i*s+j]*scheme->stage_error[j];
197: }
198: PetscStackCallBLAS("LAPACKgetrs",LAPACKgetrs_("No transpose",&m,&one,ImV,&n,ipiv,scheme->gamma+1,&n,&info));
199: if (info < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_LIB,"Bad argument to GETRS");
200: if (info > 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_LIB,"Should not happen");
202: /* beta[0] (rho in B,J,W 2007)
203: e.rho = 1/factorial(p+2) - glm.B(1,:)*glm.c.^(p+1)/factorial(p+1) ...
204: + glm.V(1,2:end)*e.beta;% - e.epsilon;
205: % Note: The paper (B,J,W 2007) includes the last term in their definition
206: * */
207: scheme->beta[0] = 1./Factorial(p+2);
208: for (j=0; j<s; j++) scheme->beta[0] -= b[0*s+j]*CPowF(c[j],p+1);
209: for (j=1; j<r; j++) scheme->beta[0] += v[0*r+j]*scheme->beta[j];
211: /* gamma[0] (sigma in B,J,W 2007)
212: * e.sigma = glm.B(1,:)*e.xi + glm.V(1,2:end)*e.gamma;
213: * */
214: scheme->gamma[0] = 0.0;
215: for (j=0; j<s; j++) scheme->gamma[0] += b[0*s+j]*scheme->stage_error[j];
216: for (j=1; j<r; j++) scheme->gamma[0] += v[0*s+j]*scheme->gamma[j];
218: /* Assemble H
219: * % Determine the error estimators phi
220: H = [[cpow(glm.c,p) + C*e.alpha] [cpow(glm.c,p+1) + C*e.beta] ...
221: [e.xi - C*(e.gamma + 0*e.epsilon*eye(s-1,1))]]';
222: % Paper has formula above without the 0, but that term must be left
223: % out to satisfy the conditions they propose and to make the
224: % example schemes work
225: e.H = H;
226: e.phi = (H \ [1 0 0;1 1 0;0 0 -1])';
227: e.psi = -e.phi*C;
228: * */
229: for (j=0; j<s; j++) {
230: H[0+j*3] = CPowF(c[j],p);
231: H[1+j*3] = CPowF(c[j],p+1);
232: H[2+j*3] = scheme->stage_error[j];
233: for (k=1; k<r; k++) {
234: H[0+j*3] += CPowF(c[j],k-1)*scheme->alpha[k];
235: H[1+j*3] += CPowF(c[j],k-1)*scheme->beta[k];
236: H[2+j*3] -= CPowF(c[j],k-1)*scheme->gamma[k];
237: }
238: }
239: bmat[0+0*ss] = 1.; bmat[0+1*ss] = 0.; bmat[0+2*ss] = 0.;
240: bmat[1+0*ss] = 1.; bmat[1+1*ss] = 1.; bmat[1+2*ss] = 0.;
241: bmat[2+0*ss] = 0.; bmat[2+1*ss] = 0.; bmat[2+2*ss] = -1.;
242: m = 3;
243: PetscBLASIntCast(s,&n);
244: PetscBLASIntCast(ss,&ldb);
245: rcond = 1e-12;
246: #if defined(PETSC_USE_COMPLEX)
247: /* ZGELSS( M, N, NRHS, A, LDA, B, LDB, S, RCOND, RANK, WORK, LWORK, RWORK, INFO) */
248: PetscStackCallBLAS("LAPACKgelss",LAPACKgelss_(&m,&n,&m,H,&m,bmat,&ldb,sing,&rcond,&rank,workscalar,&lwork,workreal,&info));
249: #else
250: /* DGELSS( M, N, NRHS, A, LDA, B, LDB, S, RCOND, RANK, WORK, LWORK, INFO) */
251: PetscStackCallBLAS("LAPACKgelss",LAPACKgelss_(&m,&n,&m,H,&m,bmat,&ldb,sing,&rcond,&rank,workscalar,&lwork,&info));
252: #endif
253: if (info < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_LIB,"Bad argument to GELSS");
254: if (info > 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_LIB,"SVD failed to converge");
256: for (j=0; j<3; j++) {
257: for (k=0; k<s; k++) scheme->phi[k+j*s] = bmat[k+j*ss];
258: }
260: /* the other part of the error estimator, psi in B,J,W 2007 */
261: scheme->psi[0*r+0] = 0.;
262: scheme->psi[1*r+0] = 0.;
263: scheme->psi[2*r+0] = 0.;
264: for (j=1; j<r; j++) {
265: scheme->psi[0*r+j] = 0.;
266: scheme->psi[1*r+j] = 0.;
267: scheme->psi[2*r+j] = 0.;
268: for (k=0; k<s; k++) {
269: scheme->psi[0*r+j] -= CPowF(c[k],j-1)*scheme->phi[0*s+k];
270: scheme->psi[1*r+j] -= CPowF(c[k],j-1)*scheme->phi[1*s+k];
271: scheme->psi[2*r+j] -= CPowF(c[k],j-1)*scheme->phi[2*s+k];
272: }
273: }
274: PetscFree7(ImV,H,bmat,workscalar,workreal,sing,ipiv);
275: }
276: /* Check which properties are satisfied */
277: scheme->stiffly_accurate = PETSC_TRUE;
278: if (scheme->c[s-1] != 1.) scheme->stiffly_accurate = PETSC_FALSE;
279: for (j=0; j<s; j++) if (a[(s-1)*s+j] != b[j]) scheme->stiffly_accurate = PETSC_FALSE;
280: for (j=0; j<r; j++) if (u[(s-1)*r+j] != v[j]) scheme->stiffly_accurate = PETSC_FALSE;
281: scheme->fsal = scheme->stiffly_accurate; /* FSAL is stronger */
282: for (j=0; j<s-1; j++) if (r>1 && b[1*s+j] != 0.) scheme->fsal = PETSC_FALSE;
283: if (b[1*s+r-1] != 1.) scheme->fsal = PETSC_FALSE;
284: for (j=0; j<r; j++) if (r>1 && v[1*r+j] != 0.) scheme->fsal = PETSC_FALSE;
286: *inscheme = scheme;
287: return(0);
288: }
290: static PetscErrorCode TSGLLESchemeDestroy(TSGLLEScheme sc)
291: {
295: PetscFree5(sc->c,sc->a,sc->b,sc->u,sc->v);
296: PetscFree6(sc->alpha,sc->beta,sc->gamma,sc->phi,sc->psi,sc->stage_error);
297: PetscFree(sc);
298: return(0);
299: }
301: static PetscErrorCode TSGLLEDestroy_Default(TS_GLLE *gl)
302: {
304: PetscInt i;
307: for (i=0; i<gl->nschemes; i++) {
308: if (gl->schemes[i]) {TSGLLESchemeDestroy(gl->schemes[i]);}
309: }
310: PetscFree(gl->schemes);
311: gl->nschemes = 0;
312: PetscMemzero(gl->type_name,sizeof(gl->type_name));
313: return(0);
314: }
316: static PetscErrorCode TSGLLEViewTable_Private(PetscViewer viewer,PetscInt m,PetscInt n,const PetscScalar a[],const char name[])
317: {
319: PetscBool iascii;
320: PetscInt i,j;
323: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);
324: if (iascii) {
325: PetscViewerASCIIPrintf(viewer,"%30s = [",name);
326: for (i=0; i<m; i++) {
327: if (i) {PetscViewerASCIIPrintf(viewer,"%30s [","");}
328: PetscViewerASCIIUseTabs(viewer,PETSC_FALSE);
329: for (j=0; j<n; j++) {
330: PetscViewerASCIIPrintf(viewer," %12.8g",PetscRealPart(a[i*n+j]));
331: }
332: PetscViewerASCIIPrintf(viewer,"]\n");
333: PetscViewerASCIIUseTabs(viewer,PETSC_TRUE);
334: }
335: }
336: return(0);
337: }
340: static PetscErrorCode TSGLLESchemeView(TSGLLEScheme sc,PetscBool view_details,PetscViewer viewer)
341: {
343: PetscBool iascii;
346: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);
347: if (iascii) {
348: PetscViewerASCIIPrintf(viewer,"GL scheme p,q,r,s = %d,%d,%d,%d\n",sc->p,sc->q,sc->r,sc->s);
349: PetscViewerASCIIPushTab(viewer);
350: PetscViewerASCIIPrintf(viewer,"Stiffly accurate: %s, FSAL: %s\n",sc->stiffly_accurate ? "yes" : "no",sc->fsal ? "yes" : "no");
351: PetscViewerASCIIPrintf(viewer,"Leading error constants: %10.3e %10.3e %10.3e\n",
352: PetscRealPart(sc->alpha[0]),PetscRealPart(sc->beta[0]),PetscRealPart(sc->gamma[0]));
353: TSGLLEViewTable_Private(viewer,1,sc->s,sc->c,"Abscissas c");
354: if (view_details) {
355: TSGLLEViewTable_Private(viewer,sc->s,sc->s,sc->a,"A");
356: TSGLLEViewTable_Private(viewer,sc->r,sc->s,sc->b,"B");
357: TSGLLEViewTable_Private(viewer,sc->s,sc->r,sc->u,"U");
358: TSGLLEViewTable_Private(viewer,sc->r,sc->r,sc->v,"V");
360: TSGLLEViewTable_Private(viewer,3,sc->s,sc->phi,"Error estimate phi");
361: TSGLLEViewTable_Private(viewer,3,sc->r,sc->psi,"Error estimate psi");
362: TSGLLEViewTable_Private(viewer,1,sc->r,sc->alpha,"Modify alpha");
363: TSGLLEViewTable_Private(viewer,1,sc->r,sc->beta,"Modify beta");
364: TSGLLEViewTable_Private(viewer,1,sc->r,sc->gamma,"Modify gamma");
365: TSGLLEViewTable_Private(viewer,1,sc->s,sc->stage_error,"Stage error xi");
366: }
367: PetscViewerASCIIPopTab(viewer);
368: } else SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_SUP,"Viewer type %s not supported",((PetscObject)viewer)->type_name);
369: return(0);
370: }
372: static PetscErrorCode TSGLLEEstimateHigherMoments_Default(TSGLLEScheme sc,PetscReal h,Vec Ydot[],Vec Xold[],Vec hm[])
373: {
375: PetscInt i;
378: if (sc->r > 64 || sc->s > 64) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Ridiculous number of stages or items passed between stages");
379: /* build error vectors*/
380: for (i=0; i<3; i++) {
381: PetscScalar phih[64];
382: PetscInt j;
383: for (j=0; j<sc->s; j++) phih[j] = sc->phi[i*sc->s+j]*h;
384: VecZeroEntries(hm[i]);
385: VecMAXPY(hm[i],sc->s,phih,Ydot);
386: VecMAXPY(hm[i],sc->r,&sc->psi[i*sc->r],Xold);
387: }
388: return(0);
389: }
391: static PetscErrorCode TSGLLECompleteStep_Rescale(TSGLLEScheme sc,PetscReal h,TSGLLEScheme next_sc,PetscReal next_h,Vec Ydot[],Vec Xold[],Vec X[])
392: {
394: PetscScalar brow[32],vrow[32];
395: PetscInt i,j,r,s;
398: /* Build the new solution from (X,Ydot) */
399: r = sc->r;
400: s = sc->s;
401: for (i=0; i<r; i++) {
402: VecZeroEntries(X[i]);
403: for (j=0; j<s; j++) brow[j] = h*sc->b[i*s+j];
404: VecMAXPY(X[i],s,brow,Ydot);
405: for (j=0; j<r; j++) vrow[j] = sc->v[i*r+j];
406: VecMAXPY(X[i],r,vrow,Xold);
407: }
408: return(0);
409: }
411: static PetscErrorCode TSGLLECompleteStep_RescaleAndModify(TSGLLEScheme sc,PetscReal h,TSGLLEScheme next_sc,PetscReal next_h,Vec Ydot[],Vec Xold[],Vec X[])
412: {
414: PetscScalar brow[32],vrow[32];
415: PetscReal ratio;
416: PetscInt i,j,p,r,s;
419: /* Build the new solution from (X,Ydot) */
420: p = sc->p;
421: r = sc->r;
422: s = sc->s;
423: ratio = next_h/h;
424: for (i=0; i<r; i++) {
425: VecZeroEntries(X[i]);
426: for (j=0; j<s; j++) {
427: brow[j] = h*(PetscPowRealInt(ratio,i)*sc->b[i*s+j]
428: + (PetscPowRealInt(ratio,i) - PetscPowRealInt(ratio,p+1))*(+ sc->alpha[i]*sc->phi[0*s+j])
429: + (PetscPowRealInt(ratio,i) - PetscPowRealInt(ratio,p+2))*(+ sc->beta [i]*sc->phi[1*s+j]
430: + sc->gamma[i]*sc->phi[2*s+j]));
431: }
432: VecMAXPY(X[i],s,brow,Ydot);
433: for (j=0; j<r; j++) {
434: vrow[j] = (PetscPowRealInt(ratio,i)*sc->v[i*r+j]
435: + (PetscPowRealInt(ratio,i) - PetscPowRealInt(ratio,p+1))*(+ sc->alpha[i]*sc->psi[0*r+j])
436: + (PetscPowRealInt(ratio,i) - PetscPowRealInt(ratio,p+2))*(+ sc->beta [i]*sc->psi[1*r+j]
437: + sc->gamma[i]*sc->psi[2*r+j]));
438: }
439: VecMAXPY(X[i],r,vrow,Xold);
440: }
441: if (r < next_sc->r) {
442: if (r+1 != next_sc->r) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Cannot accommodate jump in r greater than 1");
443: VecZeroEntries(X[r]);
444: for (j=0; j<s; j++) brow[j] = h*PetscPowRealInt(ratio,p+1)*sc->phi[0*s+j];
445: VecMAXPY(X[r],s,brow,Ydot);
446: for (j=0; j<r; j++) vrow[j] = PetscPowRealInt(ratio,p+1)*sc->psi[0*r+j];
447: VecMAXPY(X[r],r,vrow,Xold);
448: }
449: return(0);
450: }
452: static PetscErrorCode TSGLLECreate_IRKS(TS ts)
453: {
454: TS_GLLE *gl = (TS_GLLE*)ts->data;
458: gl->Destroy = TSGLLEDestroy_Default;
459: gl->EstimateHigherMoments = TSGLLEEstimateHigherMoments_Default;
460: gl->CompleteStep = TSGLLECompleteStep_RescaleAndModify;
461: PetscMalloc1(10,&gl->schemes);
462: gl->nschemes = 0;
464: {
465: /* p=1,q=1, r=s=2, A- and L-stable with error estimates of order 2 and 3
466: * Listed in Butcher & Podhaisky 2006. On error estimation in general linear methods for stiff ODE.
467: * irks(0.3,0,[.3,1],[1],1)
468: * Note: can be made to have classical order (not stage order) 2 by replacing 0.3 with 1-sqrt(1/2)
469: * but doing so would sacrifice the error estimator.
470: */
471: const PetscScalar c[2] = {3./10., 1.};
472: const PetscScalar a[2][2] = {{3./10., 0}, {7./10., 3./10.}};
473: const PetscScalar b[2][2] = {{7./10., 3./10.}, {0,1}};
474: const PetscScalar u[2][2] = {{1,0},{1,0}};
475: const PetscScalar v[2][2] = {{1,0},{0,0}};
476: TSGLLESchemeCreate(1,1,2,2,c,*a,*b,*u,*v,&gl->schemes[gl->nschemes++]);
477: }
479: {
480: /* p=q=2, r=s=3: irks(4/9,0,[1:3]/3,[0.33852],1) */
481: /* http://www.math.auckland.ac.nz/~hpod/atlas/i2a.html */
482: const PetscScalar c[3] = {1./3., 2./3., 1}
483: ,a[3][3] = {{4./9. ,0 , 0},
484: {1.03750643704090e+00 , 4./9., 0},
485: {7.67024779410304e-01 , -3.81140216918943e-01, 4./9.}}
486: ,b[3][3] = {{0.767024779410304, -0.381140216918943, 4./9.},
487: {0.000000000000000, 0.000000000000000, 1.000000000000000},
488: {-2.075048385225385, 0.621728385225383, 1.277197204924873}}
489: ,u[3][3] = {{1.0000000000000000, -0.1111111111111109, -0.0925925925925922},
490: {1.0000000000000000, -0.8152842148186744, -0.4199095530877056},
491: {1.0000000000000000, 0.1696709930641948, 0.0539741070314165}}
492: ,v[3][3] = {{1.0000000000000000, 0.1696709930641948, 0.0539741070314165},
493: {0.000000000000000, 0.000000000000000, 0.000000000000000},
494: {0.000000000000000, 0.176122795075129, 0.000000000000000}};
495: TSGLLESchemeCreate(2,2,3,3,c,*a,*b,*u,*v,&gl->schemes[gl->nschemes++]);
496: }
497: {
498: /* p=q=3, r=s=4: irks(9/40,0,[1:4]/4,[0.3312 1.0050],[0.49541 1;1 0]) */
499: const PetscScalar c[4] = {0.25,0.5,0.75,1.0}
500: ,a[4][4] = {{9./40. , 0, 0, 0},
501: {2.11286958887701e-01 , 9./40. , 0, 0},
502: {9.46338294287584e-01 , -3.42942861246094e-01, 9./40. , 0},
503: {0.521490453970721 , -0.662474225622980, 0.490476425459734, 9./40. }}
504: ,b[4][4] = {{0.521490453970721 , -0.662474225622980, 0.490476425459734, 9./40. },
505: {0.000000000000000 , 0.000000000000000, 0.000000000000000, 1.000000000000000},
506: {-0.084677029310348 , 1.390757514776085, -1.568157386206001, 2.023192696767826},
507: {0.465383797936408 , 1.478273530625148, -1.930836081010182, 1.644872111193354}}
508: ,u[4][4] = {{1.00000000000000000 , 0.02500000000001035, -0.02499999999999053, -0.00442708333332865},
509: {1.00000000000000000 , 0.06371304111232945, -0.04032173972189845, -0.01389438413189452},
510: {1.00000000000000000 , -0.07839543304147778, 0.04738685705116663, 0.02032603595928376},
511: {1.00000000000000000 , 0.42550734619251651, 0.10800718022400080, -0.01726712647760034}}
512: ,v[4][4] = {{1.00000000000000000 , 0.42550734619251651, 0.10800718022400080, -0.01726712647760034},
513: {0.000000000000000 , 0.000000000000000, 0.000000000000000, 0.000000000000000},
514: {0.000000000000000 , -1.761115796027561, -0.521284157173780, 0.258249384305463},
515: {0.000000000000000 , -1.657693358744728, -1.052227765232394, 0.521284157173780}};
516: TSGLLESchemeCreate(3,3,4,4,c,*a,*b,*u,*v,&gl->schemes[gl->nschemes++]);
517: }
518: {
519: /* p=q=4, r=s=5:
520: irks(3/11,0,[1:5]/5, [0.1715 -0.1238 0.6617],...
521: [ -0.0812 0.4079 1.0000
522: 1.0000 0 0
523: 0.8270 1.0000 0])
524: */
525: const PetscScalar c[5] = {0.2,0.4,0.6,0.8,1.0}
526: ,a[5][5] = {{2.72727272727352e-01 , 0.00000000000000e+00, 0.00000000000000e+00 , 0.00000000000000e+00 , 0.00000000000000e+00},
527: {-1.03980153733431e-01, 2.72727272727405e-01, 0.00000000000000e+00, 0.00000000000000e+00 , 0.00000000000000e+00},
528: {-1.58615400341492e+00, 7.44168951881122e-01, 2.72727272727309e-01, 0.00000000000000e+00 , 0.00000000000000e+00},
529: {-8.73658042865628e-01, 5.37884671894595e-01, -1.63298538799523e-01, 2.72727272726996e-01 , 0.00000000000000e+00},
530: {2.95489397443992e-01 , -1.18481693910097e+00 , -6.68029812659953e-01 , 1.00716687860943e+00 , 2.72727272727288e-01}}
531: ,b[5][5] = {{2.95489397443992e-01 , -1.18481693910097e+00 , -6.68029812659953e-01 , 1.00716687860943e+00 , 2.72727272727288e-01},
532: {0.00000000000000e+00 , 1.11022302462516e-16 , -2.22044604925031e-16 , 0.00000000000000e+00 , 1.00000000000000e+00},
533: {-4.05882503986005e+00, -4.00924006567769e+00, -1.38930610972481e+00, 4.45223930308488e+00 , 6.32331093108427e-01},
534: {8.35690179937017e+00 , -2.26640927349732e+00 , 6.86647884973826e+00 , -5.22595158025740e+00 , 4.50893068837431e+00},
535: {1.27656267027479e+01 , 2.80882153840821e+00 , 8.91173096522890e+00 , -1.07936444078906e+01 , 4.82534148988854e+00}}
536: ,u[5][5] = {{1.00000000000000e+00 , -7.27272727273551e-02 , -3.45454545454419e-02 , -4.12121212119565e-03 ,-2.96969696964014e-04},
537: {1.00000000000000e+00 , 2.31252881006154e-01 , -8.29487834416481e-03 , -9.07191207681020e-03 ,-1.70378403743473e-03},
538: {1.00000000000000e+00 , 1.16925777880663e+00 , 3.59268562942635e-02 , -4.09013451730615e-02 ,-1.02411119670164e-02},
539: {1.00000000000000e+00 , 1.02634463704356e+00 , 1.59375044913405e-01 , 1.89673015035370e-03 ,-4.89987231897569e-03},
540: {1.00000000000000e+00 , 1.27746320298021e+00 , 2.37186008132728e-01 , -8.28694373940065e-02 ,-5.34396510196430e-02}}
541: ,v[5][5] = {{1.00000000000000e+00 , 1.27746320298021e+00 , 2.37186008132728e-01 , -8.28694373940065e-02 ,-5.34396510196430e-02},
542: {0.00000000000000e+00 , -1.77635683940025e-15 , -1.99840144432528e-15 , -9.99200722162641e-16 ,-3.33066907387547e-16},
543: {0.00000000000000e+00 , 4.37280081906924e+00 , 5.49221645016377e-02 , -8.88913177394943e-02 , 1.12879077989154e-01},
544: {0.00000000000000e+00 , -1.22399504837280e+01 , -5.21287338448645e+00 , -8.03952325565291e-01 , 4.60298678047147e-01},
545: {0.00000000000000e+00 , -1.85178762883829e+01 , -5.21411849862624e+00 , -1.04283436528809e+00 , 7.49030161063651e-01}};
546: TSGLLESchemeCreate(4,4,5,5,c,*a,*b,*u,*v,&gl->schemes[gl->nschemes++]);
547: }
548: {
549: /* p=q=5, r=s=6;
550: irks(1/3,0,[1:6]/6,...
551: [-0.0489 0.4228 -0.8814 0.9021],...
552: [-0.3474 -0.6617 0.6294 0.2129
553: 0.0044 -0.4256 -0.1427 -0.8936
554: -0.8267 0.4821 0.1371 -0.2557
555: -0.4426 -0.3855 -0.7514 0.3014])
556: */
557: const PetscScalar c[6] = {1./6, 2./6, 3./6, 4./6, 5./6, 1.}
558: ,a[6][6] = {{ 3.33333333333940e-01, 0 , 0 , 0 , 0 , 0 },
559: { -8.64423857333350e-02, 3.33333333332888e-01, 0 , 0 , 0 , 0 },
560: { -2.16850174258252e+00, -2.23619072028839e+00, 3.33333333335204e-01, 0 , 0 , 0 },
561: { -4.73160970138997e+00, -3.89265344629268e+00, -2.76318716520933e-01, 3.33333333335759e-01, 0 , 0 },
562: { -6.75187540297338e+00, -7.90756533769377e+00, 7.90245051802259e-01, -4.48352364517632e-01, 3.33333333328483e-01, 0 },
563: { -4.26488287921548e+00, -1.19320395589302e+01, 3.38924509887755e+00, -2.23969848002481e+00, 6.62807710124007e-01, 3.33333333335440e-01}}
564: ,b[6][6] = {{ -4.26488287921548e+00, -1.19320395589302e+01, 3.38924509887755e+00, -2.23969848002481e+00, 6.62807710124007e-01, 3.33333333335440e-01},
565: { -8.88178419700125e-16, 4.44089209850063e-16, -1.54737334057131e-15, -8.88178419700125e-16, 0.00000000000000e+00, 1.00000000000001e+00},
566: { -2.87780425770651e+01, -1.13520448264971e+01, 2.62002318943161e+01, 2.56943874812797e+01, -3.06702268304488e+01, 6.68067773510103e+00},
567: { 5.47971245256474e+01, 6.80366875868284e+01, -6.50952588861999e+01, -8.28643975339097e+01, 8.17416943896414e+01, -1.17819043489036e+01},
568: { -2.33332114788869e+02, 6.12942539462634e+01, -4.91850135865944e+01, 1.82716844135480e+02, -1.29788173979395e+02, 3.09968095651099e+01},
569: { -1.72049132343751e+02, 8.60194713593999e+00, 7.98154219170200e-01, 1.50371386053218e+02, -1.18515423962066e+02, 2.50898277784663e+01}}
570: ,u[6][6] = {{ 1.00000000000000e+00, -1.66666666666870e-01, -4.16666666664335e-02, -3.85802469124815e-03, -2.25051440302250e-04, -9.64506172339142e-06},
571: { 1.00000000000000e+00, 8.64423857327162e-02, -4.11484912671353e-02, -1.11450903217645e-02, -1.47651050487126e-03, -1.34395070766826e-04},
572: { 1.00000000000000e+00, 4.57135912953434e+00, 1.06514719719137e+00, 1.33517564218007e-01, 1.11365952968659e-02, 6.12382756769504e-04},
573: { 1.00000000000000e+00, 9.23391519753404e+00, 2.22431212392095e+00, 2.91823807741891e-01, 2.52058456411084e-02, 1.22800542949647e-03},
574: { 1.00000000000000e+00, 1.48175480533865e+01, 3.73439117461835e+00, 5.14648336541804e-01, 4.76430038853402e-02, 2.56798515502156e-03},
575: { 1.00000000000000e+00, 1.50512347758335e+01, 4.10099701165164e+00, 5.66039141003603e-01, 3.91213893800891e-02, -2.99136269067853e-03}}
576: ,v[6][6] = {{ 1.00000000000000e+00, 1.50512347758335e+01, 4.10099701165164e+00, 5.66039141003603e-01, 3.91213893800891e-02, -2.99136269067853e-03},
577: { 0.00000000000000e+00, -4.88498130835069e-15, -6.43929354282591e-15, -3.55271367880050e-15, -1.22124532708767e-15, -3.12250225675825e-16},
578: { 0.00000000000000e+00, 1.22250171233141e+01, -1.77150760606169e+00, 3.54516769879390e-01, 6.22298845883398e-01, 2.31647447450276e-01},
579: { 0.00000000000000e+00, -4.48339457331040e+01, -3.57363126641880e-01, 5.18750173123425e-01, 6.55727990241799e-02, 1.63175368287079e-01},
580: { 0.00000000000000e+00, 1.37297394708005e+02, -1.60145272991317e+00, -5.05319555199441e+00, 1.55328940390990e-01, 9.16629423682464e-01},
581: { 0.00000000000000e+00, 1.05703241119022e+02, -1.16610260983038e+00, -2.99767252773859e+00, -1.13472315553890e-01, 1.09742849254729e+00}};
582: TSGLLESchemeCreate(5,5,6,6,c,*a,*b,*u,*v,&gl->schemes[gl->nschemes++]);
583: }
584: return(0);
585: }
587: /*@C
588: TSGLLESetType - sets the class of general linear method to use for time-stepping
590: Collective on TS
592: Input Parameters:
593: + ts - the TS context
594: - type - a method
596: Options Database Key:
597: . -ts_gl_type <type> - sets the method, use -help for a list of available method (e.g. irks)
599: Notes:
600: See "petsc/include/petscts.h" for available methods (for instance)
601: . TSGLLE_IRKS - Diagonally implicit methods with inherent Runge-Kutta stability (for stiff problems)
603: Normally, it is best to use the TSSetFromOptions() command and
604: then set the TSGLLE type from the options database rather than by using
605: this routine. Using the options database provides the user with
606: maximum flexibility in evaluating the many different solvers.
607: The TSGLLESetType() routine is provided for those situations where it
608: is necessary to set the timestepping solver independently of the
609: command line or options database. This might be the case, for example,
610: when the choice of solver changes during the execution of the
611: program, and the user's Section 1.5 Writing Application Codes with PETSc is taking responsibility for
612: choosing the appropriate method.
614: Level: intermediate
616: @*/
617: PetscErrorCode TSGLLESetType(TS ts,TSGLLEType type)
618: {
624: PetscTryMethod(ts,"TSGLLESetType_C",(TS,TSGLLEType),(ts,type));
625: return(0);
626: }
628: /*@C
629: TSGLLESetAcceptType - sets the acceptance test
631: Time integrators that need to control error must have the option to reject a time step based on local error
632: estimates. This function allows different schemes to be set.
634: Logically Collective on TS
636: Input Parameters:
637: + ts - the TS context
638: - type - the type
640: Options Database Key:
641: . -ts_gl_accept_type <type> - sets the method used to determine whether to accept or reject a step
643: Level: intermediate
645: .seealso: TS, TSGLLE, TSGLLEAcceptRegister(), TSGLLEAdapt, set type
646: @*/
647: PetscErrorCode TSGLLESetAcceptType(TS ts,TSGLLEAcceptType type)
648: {
654: PetscTryMethod(ts,"TSGLLESetAcceptType_C",(TS,TSGLLEAcceptType),(ts,type));
655: return(0);
656: }
658: /*@C
659: TSGLLEGetAdapt - gets the TSGLLEAdapt object from the TS
661: Not Collective
663: Input Parameter:
664: . ts - the TS context
666: Output Parameter:
667: . adapt - the TSGLLEAdapt context
669: Notes:
670: This allows the user set options on the TSGLLEAdapt object. Usually it is better to do this using the options
671: database, so this function is rarely needed.
673: Level: advanced
675: .seealso: TSGLLEAdapt, TSGLLEAdaptRegister()
676: @*/
677: PetscErrorCode TSGLLEGetAdapt(TS ts,TSGLLEAdapt *adapt)
678: {
684: PetscUseMethod(ts,"TSGLLEGetAdapt_C",(TS,TSGLLEAdapt*),(ts,adapt));
685: return(0);
686: }
688: static PetscErrorCode TSGLLEAccept_Always(TS ts,PetscReal tleft,PetscReal h,const PetscReal enorms[],PetscBool *accept)
689: {
691: *accept = PETSC_TRUE;
692: return(0);
693: }
695: static PetscErrorCode TSGLLEUpdateWRMS(TS ts)
696: {
697: TS_GLLE *gl = (TS_GLLE*)ts->data;
699: PetscScalar *x,*w;
700: PetscInt n,i;
703: VecGetArray(gl->X[0],&x);
704: VecGetArray(gl->W,&w);
705: VecGetLocalSize(gl->W,&n);
706: for (i=0; i<n; i++) w[i] = 1./(gl->wrms_atol + gl->wrms_rtol*PetscAbsScalar(x[i]));
707: VecRestoreArray(gl->X[0],&x);
708: VecRestoreArray(gl->W,&w);
709: return(0);
710: }
712: static PetscErrorCode TSGLLEVecNormWRMS(TS ts,Vec X,PetscReal *nrm)
713: {
714: TS_GLLE *gl = (TS_GLLE*)ts->data;
716: PetscScalar *x,*w;
717: PetscReal sum = 0.0,gsum;
718: PetscInt n,N,i;
721: VecGetArray(X,&x);
722: VecGetArray(gl->W,&w);
723: VecGetLocalSize(gl->W,&n);
724: for (i=0; i<n; i++) sum += PetscAbsScalar(PetscSqr(x[i]*w[i]));
725: VecRestoreArray(X,&x);
726: VecRestoreArray(gl->W,&w);
727: MPIU_Allreduce(&sum,&gsum,1,MPIU_REAL,MPIU_SUM,PetscObjectComm((PetscObject)ts));
728: VecGetSize(gl->W,&N);
729: *nrm = PetscSqrtReal(gsum/(1.*N));
730: return(0);
731: }
733: static PetscErrorCode TSGLLESetType_GLLE(TS ts,TSGLLEType type)
734: {
735: PetscErrorCode ierr,(*r)(TS);
736: PetscBool same;
737: TS_GLLE *gl = (TS_GLLE*)ts->data;
740: if (gl->type_name[0]) {
741: PetscStrcmp(gl->type_name,type,&same);
742: if (same) return(0);
743: (*gl->Destroy)(gl);
744: }
746: PetscFunctionListFind(TSGLLEList,type,&r);
747: if (!r) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_UNKNOWN_TYPE,"Unknown TSGLLE type \"%s\" given",type);
748: (*r)(ts);
749: PetscStrcpy(gl->type_name,type);
750: return(0);
751: }
753: static PetscErrorCode TSGLLESetAcceptType_GLLE(TS ts,TSGLLEAcceptType type)
754: {
755: PetscErrorCode ierr;
756: TSGLLEAcceptFunction r;
757: TS_GLLE *gl = (TS_GLLE*)ts->data;
760: PetscFunctionListFind(TSGLLEAcceptList,type,&r);
761: if (!r) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_UNKNOWN_TYPE,"Unknown TSGLLEAccept type \"%s\" given",type);
762: gl->Accept = r;
763: PetscStrncpy(gl->accept_name,type,sizeof(gl->accept_name));
764: return(0);
765: }
767: static PetscErrorCode TSGLLEGetAdapt_GLLE(TS ts,TSGLLEAdapt *adapt)
768: {
770: TS_GLLE *gl = (TS_GLLE*)ts->data;
773: if (!gl->adapt) {
774: TSGLLEAdaptCreate(PetscObjectComm((PetscObject)ts),&gl->adapt);
775: PetscObjectIncrementTabLevel((PetscObject)gl->adapt,(PetscObject)ts,1);
776: PetscLogObjectParent((PetscObject)ts,(PetscObject)gl->adapt);
777: }
778: *adapt = gl->adapt;
779: return(0);
780: }
782: static PetscErrorCode TSGLLEChooseNextScheme(TS ts,PetscReal h,const PetscReal hmnorm[],PetscInt *next_scheme,PetscReal *next_h,PetscBool *finish)
783: {
785: TS_GLLE *gl = (TS_GLLE*)ts->data;
786: PetscInt i,n,cur_p,cur,next_sc,candidates[64],orders[64];
787: PetscReal errors[64],costs[64],tleft;
790: cur = -1;
791: cur_p = gl->schemes[gl->current_scheme]->p;
792: tleft = ts->max_time - (ts->ptime + ts->time_step);
793: for (i=0,n=0; i<gl->nschemes; i++) {
794: TSGLLEScheme sc = gl->schemes[i];
795: if (sc->p < gl->min_order || gl->max_order < sc->p) continue;
796: if (sc->p == cur_p - 1) errors[n] = PetscAbsScalar(sc->alpha[0])*hmnorm[0];
797: else if (sc->p == cur_p) errors[n] = PetscAbsScalar(sc->alpha[0])*hmnorm[1];
798: else if (sc->p == cur_p+1) errors[n] = PetscAbsScalar(sc->alpha[0])*(hmnorm[2]+hmnorm[3]);
799: else continue;
800: candidates[n] = i;
801: orders[n] = PetscMin(sc->p,sc->q); /* order of global truncation error */
802: costs[n] = sc->s; /* estimate the cost as the number of stages */
803: if (i == gl->current_scheme) cur = n;
804: n++;
805: }
806: if (cur < 0 || gl->nschemes <= cur) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Current scheme not found in scheme list");
807: TSGLLEAdaptChoose(gl->adapt,n,orders,errors,costs,cur,h,tleft,&next_sc,next_h,finish);
808: *next_scheme = candidates[next_sc];
809: PetscInfo7(ts,"Adapt chose scheme %d (%d,%d,%d,%d) with step size %6.2e, finish=%d\n",*next_scheme,gl->schemes[*next_scheme]->p,gl->schemes[*next_scheme]->q,gl->schemes[*next_scheme]->r,gl->schemes[*next_scheme]->s,*next_h,*finish);
810: return(0);
811: }
813: static PetscErrorCode TSGLLEGetMaxSizes(TS ts,PetscInt *max_r,PetscInt *max_s)
814: {
815: TS_GLLE *gl = (TS_GLLE*)ts->data;
818: *max_r = gl->schemes[gl->nschemes-1]->r;
819: *max_s = gl->schemes[gl->nschemes-1]->s;
820: return(0);
821: }
823: static PetscErrorCode TSSolve_GLLE(TS ts)
824: {
825: TS_GLLE *gl = (TS_GLLE*)ts->data;
826: PetscInt i,k,its,lits,max_r,max_s;
827: PetscBool final_step,finish;
828: SNESConvergedReason snesreason;
829: PetscErrorCode ierr;
832: TSMonitor(ts,ts->steps,ts->ptime,ts->vec_sol);
834: TSGLLEGetMaxSizes(ts,&max_r,&max_s);
835: VecCopy(ts->vec_sol,gl->X[0]);
836: for (i=1; i<max_r; i++) {
837: VecZeroEntries(gl->X[i]);
838: }
839: TSGLLEUpdateWRMS(ts);
841: if (0) {
842: /* Find consistent initial data for DAE */
843: gl->stage_time = ts->ptime + ts->time_step;
844: gl->scoeff = 1.;
845: gl->stage = 0;
847: VecCopy(ts->vec_sol,gl->Z);
848: VecCopy(ts->vec_sol,gl->Y);
849: SNESSolve(ts->snes,NULL,gl->Y);
850: SNESGetIterationNumber(ts->snes,&its);
851: SNESGetLinearSolveIterations(ts->snes,&lits);
852: SNESGetConvergedReason(ts->snes,&snesreason);
854: ts->snes_its += its; ts->ksp_its += lits;
855: if (snesreason < 0 && ts->max_snes_failures > 0 && ++ts->num_snes_failures >= ts->max_snes_failures) {
856: ts->reason = TS_DIVERGED_NONLINEAR_SOLVE;
857: PetscInfo2(ts,"Step=%D, nonlinear solve solve failures %D greater than current TS allowed, stopping solve\n",ts->steps,ts->num_snes_failures);
858: return(0);
859: }
860: }
862: if (gl->current_scheme < 0) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_ORDER,"A starting scheme has not been provided");
864: for (k=0,final_step=PETSC_FALSE,finish=PETSC_FALSE; k<ts->max_steps && !finish; k++) {
865: PetscInt j,r,s,next_scheme = 0,rejections;
866: PetscReal h,hmnorm[4],enorm[3],next_h;
867: PetscBool accept;
868: const PetscScalar *c,*a,*u;
869: Vec *X,*Ydot,Y;
870: TSGLLEScheme scheme = gl->schemes[gl->current_scheme];
872: r = scheme->r; s = scheme->s;
873: c = scheme->c;
874: a = scheme->a; u = scheme->u;
875: h = ts->time_step;
876: X = gl->X; Ydot = gl->Ydot; Y = gl->Y;
878: if (ts->ptime > ts->max_time) break;
880: /*
881: We only call PreStep at the start of each STEP, not each STAGE. This is because it is
882: possible to fail (have to restart a step) after multiple stages.
883: */
884: TSPreStep(ts);
886: rejections = 0;
887: while (1) {
888: for (i=0; i<s; i++) {
889: PetscScalar shift;
890: gl->scoeff = 1./PetscRealPart(a[i*s+i]);
891: shift = gl->scoeff/ts->time_step;
892: gl->stage = i;
893: gl->stage_time = ts->ptime + PetscRealPart(c[i])*h;
895: /*
896: * Stage equation: Y = h A Y' + U X
897: * We assume that A is lower-triangular so that we can solve the stages (Y,Y') sequentially
898: * Build the affine vector z_i = -[1/(h a_ii)](h sum_j a_ij y'_j + sum_j u_ij x_j)
899: * Then y'_i = z + 1/(h a_ii) y_i
900: */
901: VecZeroEntries(gl->Z);
902: for (j=0; j<r; j++) {
903: VecAXPY(gl->Z,-shift*u[i*r+j],X[j]);
904: }
905: for (j=0; j<i; j++) {
906: VecAXPY(gl->Z,-shift*h*a[i*s+j],Ydot[j]);
907: }
908: /* Note: Z is used within function evaluation, Ydot = Z + shift*Y */
910: /* Compute an estimate of Y to start Newton iteration */
911: if (gl->extrapolate) {
912: if (i==0) {
913: /* Linear extrapolation on the first stage */
914: VecWAXPY(Y,c[i]*h,X[1],X[0]);
915: } else {
916: /* Linear extrapolation from the last stage */
917: VecAXPY(Y,(c[i]-c[i-1])*h,Ydot[i-1]);
918: }
919: } else if (i==0) { /* Directly use solution from the last step, otherwise reuse the last stage (do nothing) */
920: VecCopy(X[0],Y);
921: }
923: /* Solve this stage (Ydot[i] is computed during function evaluation) */
924: SNESSolve(ts->snes,NULL,Y);
925: SNESGetIterationNumber(ts->snes,&its);
926: SNESGetLinearSolveIterations(ts->snes,&lits);
927: SNESGetConvergedReason(ts->snes,&snesreason);
928: ts->snes_its += its; ts->ksp_its += lits;
929: if (snesreason < 0 && ts->max_snes_failures > 0 && ++ts->num_snes_failures >= ts->max_snes_failures) {
930: ts->reason = TS_DIVERGED_NONLINEAR_SOLVE;
931: PetscInfo2(ts,"Step=%D, nonlinear solve solve failures %D greater than current TS allowed, stopping solve\n",ts->steps,ts->num_snes_failures);
932: return(0);
933: }
934: }
936: gl->stage_time = ts->ptime + ts->time_step;
938: (*gl->EstimateHigherMoments)(scheme,h,Ydot,gl->X,gl->himom);
939: /* hmnorm[i] = h^{p+i}x^{(p+i)} with i=0,1,2; hmnorm[3] = h^{p+2}(dx'/dx) x^{(p+1)} */
940: for (i=0; i<3; i++) {
941: TSGLLEVecNormWRMS(ts,gl->himom[i],&hmnorm[i+1]);
942: }
943: enorm[0] = PetscRealPart(scheme->alpha[0])*hmnorm[1];
944: enorm[1] = PetscRealPart(scheme->beta[0]) *hmnorm[2];
945: enorm[2] = PetscRealPart(scheme->gamma[0])*hmnorm[3];
946: (*gl->Accept)(ts,ts->max_time-gl->stage_time,h,enorm,&accept);
947: if (accept) goto accepted;
948: rejections++;
949: PetscInfo3(ts,"Step %D (t=%g) not accepted, rejections=%D\n",k,gl->stage_time,rejections);
950: if (rejections > gl->max_step_rejections) break;
951: /*
952: There are lots of reasons why a step might be rejected, including solvers not converging and other factors that
953: TSGLLEChooseNextScheme does not support. Additionally, the error estimates may be very screwed up, so I'm not
954: convinced that it's safe to just compute a new error estimate using the same interface as the current adaptor
955: (the adaptor interface probably has to change). Here we make an arbitrary and naive choice. This assumes that
956: steps were written in Nordsieck form. The "correct" method would be to re-complete the previous time step with
957: the correct "next" step size. It is unclear to me whether the present ad-hoc method of rescaling X is stable.
958: */
959: h *= 0.5;
960: for (i=1; i<scheme->r; i++) {
961: VecScale(X[i],PetscPowRealInt(0.5,i));
962: }
963: }
964: SETERRQ3(PETSC_COMM_SELF,PETSC_ERR_CONV_FAILED,"Time step %D (t=%g) not accepted after %D failures\n",k,gl->stage_time,rejections);
966: accepted:
967: /* This term is not error, but it *would* be the leading term for a lower order method */
968: TSGLLEVecNormWRMS(ts,gl->X[scheme->r-1],&hmnorm[0]);
969: /* Correct scaling so that these are equivalent to norms of the Nordsieck vectors */
971: PetscInfo4(ts,"Last moment norm %10.2e, estimated error norms %10.2e %10.2e %10.2e\n",hmnorm[0],enorm[0],enorm[1],enorm[2]);
972: if (!final_step) {
973: TSGLLEChooseNextScheme(ts,h,hmnorm,&next_scheme,&next_h,&final_step);
974: } else {
975: /* Dummy values to complete the current step in a consistent manner */
976: next_scheme = gl->current_scheme;
977: next_h = h;
978: finish = PETSC_TRUE;
979: }
981: X = gl->Xold;
982: gl->Xold = gl->X;
983: gl->X = X;
984: (*gl->CompleteStep)(scheme,h,gl->schemes[next_scheme],next_h,Ydot,gl->Xold,gl->X);
986: TSGLLEUpdateWRMS(ts);
988: /* Post the solution for the user, we could avoid this copy with a small bit of cleverness */
989: VecCopy(gl->X[0],ts->vec_sol);
990: ts->ptime += h;
991: ts->steps++;
993: TSPostEvaluate(ts);
994: TSPostStep(ts);
995: TSMonitor(ts,ts->steps,ts->ptime,ts->vec_sol);
997: gl->current_scheme = next_scheme;
998: ts->time_step = next_h;
999: }
1000: return(0);
1001: }
1003: /*------------------------------------------------------------*/
1005: static PetscErrorCode TSReset_GLLE(TS ts)
1006: {
1007: TS_GLLE *gl = (TS_GLLE*)ts->data;
1008: PetscInt max_r,max_s;
1012: if (gl->setupcalled) {
1013: TSGLLEGetMaxSizes(ts,&max_r,&max_s);
1014: VecDestroyVecs(max_r,&gl->Xold);
1015: VecDestroyVecs(max_r,&gl->X);
1016: VecDestroyVecs(max_s,&gl->Ydot);
1017: VecDestroyVecs(3,&gl->himom);
1018: VecDestroy(&gl->W);
1019: VecDestroy(&gl->Y);
1020: VecDestroy(&gl->Z);
1021: }
1022: gl->setupcalled = PETSC_FALSE;
1023: return(0);
1024: }
1026: static PetscErrorCode TSDestroy_GLLE(TS ts)
1027: {
1028: TS_GLLE *gl = (TS_GLLE*)ts->data;
1032: TSReset_GLLE(ts);
1033: if (ts->dm) {
1034: DMCoarsenHookRemove(ts->dm,DMCoarsenHook_TSGLLE,DMRestrictHook_TSGLLE,ts);
1035: DMSubDomainHookRemove(ts->dm,DMSubDomainHook_TSGLLE,DMSubDomainRestrictHook_TSGLLE,ts);
1036: }
1037: if (gl->adapt) {TSGLLEAdaptDestroy(&gl->adapt);}
1038: if (gl->Destroy) {(*gl->Destroy)(gl);}
1039: PetscFree(ts->data);
1040: PetscObjectComposeFunction((PetscObject)ts,"TSGLLESetType_C",NULL);
1041: PetscObjectComposeFunction((PetscObject)ts,"TSGLLESetAcceptType_C",NULL);
1042: PetscObjectComposeFunction((PetscObject)ts,"TSGLLEGetAdapt_C",NULL);
1043: return(0);
1044: }
1046: /*
1047: This defines the nonlinear equation that is to be solved with SNES
1048: g(x) = f(t,x,z+shift*x) = 0
1049: */
1050: static PetscErrorCode SNESTSFormFunction_GLLE(SNES snes,Vec x,Vec f,TS ts)
1051: {
1052: TS_GLLE *gl = (TS_GLLE*)ts->data;
1054: Vec Z,Ydot;
1055: DM dm,dmsave;
1058: SNESGetDM(snes,&dm);
1059: TSGLLEGetVecs(ts,dm,&Z,&Ydot);
1060: VecWAXPY(Ydot,gl->scoeff/ts->time_step,x,Z);
1061: dmsave = ts->dm;
1062: ts->dm = dm;
1063: TSComputeIFunction(ts,gl->stage_time,x,Ydot,f,PETSC_FALSE);
1064: ts->dm = dmsave;
1065: TSGLLERestoreVecs(ts,dm,&Z,&Ydot);
1066: return(0);
1067: }
1069: static PetscErrorCode SNESTSFormJacobian_GLLE(SNES snes,Vec x,Mat A,Mat B,TS ts)
1070: {
1071: TS_GLLE *gl = (TS_GLLE*)ts->data;
1073: Vec Z,Ydot;
1074: DM dm,dmsave;
1077: SNESGetDM(snes,&dm);
1078: TSGLLEGetVecs(ts,dm,&Z,&Ydot);
1079: dmsave = ts->dm;
1080: ts->dm = dm;
1081: /* gl->Xdot will have already been computed in SNESTSFormFunction_GLLE */
1082: TSComputeIJacobian(ts,gl->stage_time,x,gl->Ydot[gl->stage],gl->scoeff/ts->time_step,A,B,PETSC_FALSE);
1083: ts->dm = dmsave;
1084: TSGLLERestoreVecs(ts,dm,&Z,&Ydot);
1085: return(0);
1086: }
1089: static PetscErrorCode TSSetUp_GLLE(TS ts)
1090: {
1091: TS_GLLE *gl = (TS_GLLE*)ts->data;
1092: PetscInt max_r,max_s;
1094: DM dm;
1097: gl->setupcalled = PETSC_TRUE;
1098: TSGLLEGetMaxSizes(ts,&max_r,&max_s);
1099: VecDuplicateVecs(ts->vec_sol,max_r,&gl->X);
1100: VecDuplicateVecs(ts->vec_sol,max_r,&gl->Xold);
1101: VecDuplicateVecs(ts->vec_sol,max_s,&gl->Ydot);
1102: VecDuplicateVecs(ts->vec_sol,3,&gl->himom);
1103: VecDuplicate(ts->vec_sol,&gl->W);
1104: VecDuplicate(ts->vec_sol,&gl->Y);
1105: VecDuplicate(ts->vec_sol,&gl->Z);
1107: /* Default acceptance tests and adaptivity */
1108: if (!gl->Accept) {TSGLLESetAcceptType(ts,TSGLLEACCEPT_ALWAYS);}
1109: if (!gl->adapt) {TSGLLEGetAdapt(ts,&gl->adapt);}
1111: if (gl->current_scheme < 0) {
1112: PetscInt i;
1113: for (i=0;; i++) {
1114: if (gl->schemes[i]->p == gl->start_order) break;
1115: if (i+1 == gl->nschemes) SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_OUTOFRANGE,"No schemes available with requested start order %d",i);
1116: }
1117: gl->current_scheme = i;
1118: }
1119: TSGetDM(ts,&dm);
1120: DMCoarsenHookAdd(dm,DMCoarsenHook_TSGLLE,DMRestrictHook_TSGLLE,ts);
1121: DMSubDomainHookAdd(dm,DMSubDomainHook_TSGLLE,DMSubDomainRestrictHook_TSGLLE,ts);
1122: return(0);
1123: }
1124: /*------------------------------------------------------------*/
1126: static PetscErrorCode TSSetFromOptions_GLLE(PetscOptionItems *PetscOptionsObject,TS ts)
1127: {
1128: TS_GLLE *gl = (TS_GLLE*)ts->data;
1129: char tname[256] = TSGLLE_IRKS,completef[256] = "rescale-and-modify";
1133: PetscOptionsHead(PetscOptionsObject,"General Linear ODE solver options");
1134: {
1135: PetscBool flg;
1136: PetscOptionsFList("-ts_gl_type","Type of GL method","TSGLLESetType",TSGLLEList,gl->type_name[0] ? gl->type_name : tname,tname,sizeof(tname),&flg);
1137: if (flg || !gl->type_name[0]) {
1138: TSGLLESetType(ts,tname);
1139: }
1140: PetscOptionsInt("-ts_gl_max_step_rejections","Maximum number of times to attempt a step","None",gl->max_step_rejections,&gl->max_step_rejections,NULL);
1141: PetscOptionsInt("-ts_gl_max_order","Maximum order to try","TSGLLESetMaxOrder",gl->max_order,&gl->max_order,NULL);
1142: PetscOptionsInt("-ts_gl_min_order","Minimum order to try","TSGLLESetMinOrder",gl->min_order,&gl->min_order,NULL);
1143: PetscOptionsInt("-ts_gl_start_order","Initial order to try","TSGLLESetMinOrder",gl->start_order,&gl->start_order,NULL);
1144: PetscOptionsEnum("-ts_gl_error_direction","Which direction to look when estimating error","TSGLLESetErrorDirection",TSGLLEErrorDirections,(PetscEnum)gl->error_direction,(PetscEnum*)&gl->error_direction,NULL);
1145: PetscOptionsBool("-ts_gl_extrapolate","Extrapolate stage solution from previous solution (sometimes unstable)","TSGLLESetExtrapolate",gl->extrapolate,&gl->extrapolate,NULL);
1146: PetscOptionsReal("-ts_gl_atol","Absolute tolerance","TSGLLESetTolerances",gl->wrms_atol,&gl->wrms_atol,NULL);
1147: PetscOptionsReal("-ts_gl_rtol","Relative tolerance","TSGLLESetTolerances",gl->wrms_rtol,&gl->wrms_rtol,NULL);
1148: PetscOptionsString("-ts_gl_complete","Method to use for completing the step","none",completef,completef,sizeof(completef),&flg);
1149: if (flg) {
1150: PetscBool match1,match2;
1151: PetscStrcmp(completef,"rescale",&match1);
1152: PetscStrcmp(completef,"rescale-and-modify",&match2);
1153: if (match1) gl->CompleteStep = TSGLLECompleteStep_Rescale;
1154: else if (match2) gl->CompleteStep = TSGLLECompleteStep_RescaleAndModify;
1155: else SETERRQ1(PETSC_COMM_SELF,PETSC_ERR_ARG_UNKNOWN_TYPE,"%s",completef);
1156: }
1157: {
1158: char type[256] = TSGLLEACCEPT_ALWAYS;
1159: PetscOptionsFList("-ts_gl_accept_type","Method to use for determining whether to accept a step","TSGLLESetAcceptType",TSGLLEAcceptList,gl->accept_name[0] ? gl->accept_name : type,type,sizeof(type),&flg);
1160: if (flg || !gl->accept_name[0]) {
1161: TSGLLESetAcceptType(ts,type);
1162: }
1163: }
1164: {
1165: TSGLLEAdapt adapt;
1166: TSGLLEGetAdapt(ts,&adapt);
1167: TSGLLEAdaptSetFromOptions(PetscOptionsObject,adapt);
1168: }
1169: }
1170: PetscOptionsTail();
1171: return(0);
1172: }
1174: static PetscErrorCode TSView_GLLE(TS ts,PetscViewer viewer)
1175: {
1176: TS_GLLE *gl = (TS_GLLE*)ts->data;
1177: PetscInt i;
1178: PetscBool iascii,details;
1182: PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);
1183: if (iascii) {
1184: PetscViewerASCIIPrintf(viewer," min order %D, max order %D, current order %D\n",gl->min_order,gl->max_order,gl->schemes[gl->current_scheme]->p);
1185: PetscViewerASCIIPrintf(viewer," Error estimation: %s\n",TSGLLEErrorDirections[gl->error_direction]);
1186: PetscViewerASCIIPrintf(viewer," Extrapolation: %s\n",gl->extrapolate ? "yes" : "no");
1187: PetscViewerASCIIPrintf(viewer," Acceptance test: %s\n",gl->accept_name[0] ? gl->accept_name : "(not yet set)");
1188: PetscViewerASCIIPushTab(viewer);
1189: TSGLLEAdaptView(gl->adapt,viewer);
1190: PetscViewerASCIIPopTab(viewer);
1191: PetscViewerASCIIPrintf(viewer," type: %s\n",gl->type_name[0] ? gl->type_name : "(not yet set)");
1192: PetscViewerASCIIPrintf(viewer,"Schemes within family (%d):\n",gl->nschemes);
1193: details = PETSC_FALSE;
1194: PetscOptionsGetBool(((PetscObject)ts)->options,((PetscObject)ts)->prefix,"-ts_gl_view_detailed",&details,NULL);
1195: PetscViewerASCIIPushTab(viewer);
1196: for (i=0; i<gl->nschemes; i++) {
1197: TSGLLESchemeView(gl->schemes[i],details,viewer);
1198: }
1199: if (gl->View) {
1200: (*gl->View)(gl,viewer);
1201: }
1202: PetscViewerASCIIPopTab(viewer);
1203: }
1204: return(0);
1205: }
1207: /*@C
1208: TSGLLERegister - adds a TSGLLE implementation
1210: Not Collective
1212: Input Parameters:
1213: + name_scheme - name of user-defined general linear scheme
1214: - routine_create - routine to create method context
1216: Notes:
1217: TSGLLERegister() may be called multiple times to add several user-defined families.
1219: Sample usage:
1220: .vb
1221: TSGLLERegister("my_scheme",MySchemeCreate);
1222: .ve
1224: Then, your scheme can be chosen with the procedural interface via
1225: $ TSGLLESetType(ts,"my_scheme")
1226: or at runtime via the option
1227: $ -ts_gl_type my_scheme
1229: Level: advanced
1231: .seealso: TSGLLERegisterAll()
1232: @*/
1233: PetscErrorCode TSGLLERegister(const char sname[],PetscErrorCode (*function)(TS))
1234: {
1238: TSGLLEInitializePackage();
1239: PetscFunctionListAdd(&TSGLLEList,sname,function);
1240: return(0);
1241: }
1243: /*@C
1244: TSGLLEAcceptRegister - adds a TSGLLE acceptance scheme
1246: Not Collective
1248: Input Parameters:
1249: + name_scheme - name of user-defined acceptance scheme
1250: - routine_create - routine to create method context
1252: Notes:
1253: TSGLLEAcceptRegister() may be called multiple times to add several user-defined families.
1255: Sample usage:
1256: .vb
1257: TSGLLEAcceptRegister("my_scheme",MySchemeCreate);
1258: .ve
1260: Then, your scheme can be chosen with the procedural interface via
1261: $ TSGLLESetAcceptType(ts,"my_scheme")
1262: or at runtime via the option
1263: $ -ts_gl_accept_type my_scheme
1265: Level: advanced
1267: .seealso: TSGLLERegisterAll()
1268: @*/
1269: PetscErrorCode TSGLLEAcceptRegister(const char sname[],TSGLLEAcceptFunction function)
1270: {
1274: PetscFunctionListAdd(&TSGLLEAcceptList,sname,function);
1275: return(0);
1276: }
1278: /*@C
1279: TSGLLERegisterAll - Registers all of the general linear methods in TSGLLE
1281: Not Collective
1283: Level: advanced
1285: .seealso: TSGLLERegisterDestroy()
1286: @*/
1287: PetscErrorCode TSGLLERegisterAll(void)
1288: {
1292: if (TSGLLERegisterAllCalled) return(0);
1293: TSGLLERegisterAllCalled = PETSC_TRUE;
1295: TSGLLERegister(TSGLLE_IRKS, TSGLLECreate_IRKS);
1296: TSGLLEAcceptRegister(TSGLLEACCEPT_ALWAYS,TSGLLEAccept_Always);
1297: return(0);
1298: }
1300: /*@C
1301: TSGLLEInitializePackage - This function initializes everything in the TSGLLE package. It is called
1302: from TSInitializePackage().
1304: Level: developer
1306: .seealso: PetscInitialize()
1307: @*/
1308: PetscErrorCode TSGLLEInitializePackage(void)
1309: {
1313: if (TSGLLEPackageInitialized) return(0);
1314: TSGLLEPackageInitialized = PETSC_TRUE;
1315: TSGLLERegisterAll();
1316: PetscRegisterFinalize(TSGLLEFinalizePackage);
1317: return(0);
1318: }
1320: /*@C
1321: TSGLLEFinalizePackage - This function destroys everything in the TSGLLE package. It is
1322: called from PetscFinalize().
1324: Level: developer
1326: .seealso: PetscFinalize()
1327: @*/
1328: PetscErrorCode TSGLLEFinalizePackage(void)
1329: {
1333: PetscFunctionListDestroy(&TSGLLEList);
1334: PetscFunctionListDestroy(&TSGLLEAcceptList);
1335: TSGLLEPackageInitialized = PETSC_FALSE;
1336: TSGLLERegisterAllCalled = PETSC_FALSE;
1337: return(0);
1338: }
1340: /* ------------------------------------------------------------ */
1341: /*MC
1342: TSGLLE - DAE solver using implicit General Linear methods
1344: These methods contain Runge-Kutta and multistep schemes as special cases. These special cases have some fundamental
1345: limitations. For example, diagonally implicit Runge-Kutta cannot have stage order greater than 1 which limits their
1346: applicability to very stiff systems. Meanwhile, multistep methods cannot be A-stable for order greater than 2 and BDF
1347: are not 0-stable for order greater than 6. GL methods can be A- and L-stable with arbitrarily high stage order and
1348: reliable error estimates for both 1 and 2 orders higher to facilitate adaptive step sizes and adaptive order schemes.
1349: All this is possible while preserving a singly diagonally implicit structure.
1351: Options database keys:
1352: + -ts_gl_type <type> - the class of general linear method (irks)
1353: . -ts_gl_rtol <tol> - relative error
1354: . -ts_gl_atol <tol> - absolute error
1355: . -ts_gl_min_order <p> - minimum order method to consider (default=1)
1356: . -ts_gl_max_order <p> - maximum order method to consider (default=3)
1357: . -ts_gl_start_order <p> - order of starting method (default=1)
1358: . -ts_gl_complete <method> - method to use for completing the step (rescale-and-modify or rescale)
1359: - -ts_adapt_type <method> - adaptive controller to use (none step both)
1361: Notes:
1362: This integrator can be applied to DAE.
1364: Diagonally implicit general linear (DIGL) methods are a generalization of diagonally implicit Runge-Kutta (DIRK).
1365: They are represented by the tableau
1367: .vb
1368: A | U
1369: -------
1370: B | V
1371: .ve
1373: combined with a vector c of abscissa. "Diagonally implicit" means that A is lower triangular.
1374: A step of the general method reads
1376: .vb
1377: [ Y ] = [A U] [ Y' ]
1378: [X^k] = [B V] [X^{k-1}]
1379: .ve
1381: where Y is the multivector of stage values, Y' is the multivector of stage derivatives, X^k is the Nordsieck vector of
1382: the solution at step k. The Nordsieck vector consists of the first r moments of the solution, given by
1384: .vb
1385: X = [x_0,x_1,...,x_{r-1}] = [x, h x', h^2 x'', ..., h^{r-1} x^{(r-1)} ]
1386: .ve
1388: If A is lower triangular, we can solve the stages (Y,Y') sequentially
1390: .vb
1391: y_i = h sum_{j=0}^{s-1} (a_ij y'_j) + sum_{j=0}^{r-1} u_ij x_j, i=0,...,{s-1}
1392: .ve
1394: and then construct the pieces to carry to the next step
1396: .vb
1397: xx_i = h sum_{j=0}^{s-1} b_ij y'_j + sum_{j=0}^{r-1} v_ij x_j, i=0,...,{r-1}
1398: .ve
1400: Note that when the equations are cast in implicit form, we are using the stage equation to define y'_i
1401: in terms of y_i and known stuff (y_j for j<i and x_j for all j).
1404: Error estimation
1406: At present, the most attractive GL methods for stiff problems are singly diagonally implicit schemes which posses
1407: Inherent Runge-Kutta Stability (IRKS). These methods have r=s, the number of items passed between steps is equal to
1408: the number of stages. The order and stage-order are one less than the number of stages. We use the error estimates
1409: in the 2007 paper which provide the following estimates
1411: .vb
1412: h^{p+1} X^{(p+1)} = phi_0^T Y' + [0 psi_0^T] Xold
1413: h^{p+2} X^{(p+2)} = phi_1^T Y' + [0 psi_1^T] Xold
1414: h^{p+2} (dx'/dx) X^{(p+1)} = phi_2^T Y' + [0 psi_2^T] Xold
1415: .ve
1417: These estimates are accurate to O(h^{p+3}).
1419: Changing the step size
1421: We use the generalized "rescale and modify" scheme, see equation (4.5) of the 2007 paper.
1423: Level: beginner
1425: References:
1426: + 1. - John Butcher and Z. Jackieweicz and W. Wright, On error propagation in general linear methods for
1427: ordinary differential equations, Journal of Complexity, Vol 23, 2007.
1428: - 2. - John Butcher, Numerical methods for ordinary differential equations, second edition, Wiley, 2009.
1430: .seealso: TSCreate(), TS, TSSetType()
1432: M*/
1433: PETSC_EXTERN PetscErrorCode TSCreate_GLLE(TS ts)
1434: {
1435: TS_GLLE *gl;
1439: TSGLLEInitializePackage();
1441: PetscNewLog(ts,&gl);
1442: ts->data = (void*)gl;
1444: ts->ops->reset = TSReset_GLLE;
1445: ts->ops->destroy = TSDestroy_GLLE;
1446: ts->ops->view = TSView_GLLE;
1447: ts->ops->setup = TSSetUp_GLLE;
1448: ts->ops->solve = TSSolve_GLLE;
1449: ts->ops->setfromoptions = TSSetFromOptions_GLLE;
1450: ts->ops->snesfunction = SNESTSFormFunction_GLLE;
1451: ts->ops->snesjacobian = SNESTSFormJacobian_GLLE;
1453: ts->usessnes = PETSC_TRUE;
1456: gl->max_step_rejections = 1;
1457: gl->min_order = 1;
1458: gl->max_order = 3;
1459: gl->start_order = 1;
1460: gl->current_scheme = -1;
1461: gl->extrapolate = PETSC_FALSE;
1463: gl->wrms_atol = 1e-8;
1464: gl->wrms_rtol = 1e-5;
1466: PetscObjectComposeFunction((PetscObject)ts,"TSGLLESetType_C", &TSGLLESetType_GLLE);
1467: PetscObjectComposeFunction((PetscObject)ts,"TSGLLESetAcceptType_C",&TSGLLESetAcceptType_GLLE);
1468: PetscObjectComposeFunction((PetscObject)ts,"TSGLLEGetAdapt_C", &TSGLLEGetAdapt_GLLE);
1469: return(0);
1470: }