Actual source code: ex30.c
1: static char help[] = "Grid based Landau collision operator with PIC interface with OpenMP setup. (one species per grid)\n";
3: /*
4: Support 2.5V with axisymmetric coordinates
5: - r,z coordinates
6: - Domain and species data input by Landau operator
7: - "radius" for each grid, normalized with electron thermal velocity
8: - Domain: (0,radius) x (-radius,radius), thus first coordinate x[0] is perpendicular velocity and 2pi*x[0] term is added for axisymmetric
9: Supports full 3V
11: */
13: #include <petscdmplex.h>
14: #include <petscds.h>
15: #include <petscdmswarm.h>
16: #include <petscksp.h>
17: #include <petsc/private/petscimpl.h>
18: #if defined(PETSC_HAVE_OPENMP) && defined(PETSC_HAVE_THREADSAFETY)
19: #include <omp.h>
20: #endif
21: #include <petsclandau.h>
22: #include <petscdmcomposite.h>
24: typedef struct {
25: Mat MpTrans;
26: Mat Mp;
27: Vec ff;
28: Vec uu;
29: } MatShellCtx;
31: typedef struct {
32: PetscInt v_target;
33: PetscInt g_target;
34: PetscInt global_vertex_id_0;
35: DM *globSwarmArray;
36: LandauCtx *ctx;
37: DM *grid_dm;
38: Mat *g_Mass;
39: Mat *globMpArray;
40: Vec *globXArray;
41: PetscBool print;
42: PetscBool print_entropy;
43: } PrintCtx;
45: PetscErrorCode MatMultMtM_SeqAIJ(Mat MtM, Vec xx, Vec yy)
46: {
47: MatShellCtx *matshellctx;
49: PetscFunctionBeginUser;
50: PetscCall(MatShellGetContext(MtM, &matshellctx));
51: PetscCheck(matshellctx, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "No context");
52: PetscCall(MatMult(matshellctx->Mp, xx, matshellctx->ff));
53: PetscCall(MatMult(matshellctx->MpTrans, matshellctx->ff, yy));
54: PetscFunctionReturn(PETSC_SUCCESS);
55: }
57: PetscErrorCode MatMultAddMtM_SeqAIJ(Mat MtM, Vec xx, Vec yy, Vec zz)
58: {
59: MatShellCtx *matshellctx;
61: PetscFunctionBeginUser;
62: PetscCall(MatShellGetContext(MtM, &matshellctx));
63: PetscCheck(matshellctx, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "No context");
64: PetscCall(MatMult(matshellctx->Mp, xx, matshellctx->ff));
65: PetscCall(MatMultAdd(matshellctx->MpTrans, matshellctx->ff, yy, zz));
66: PetscFunctionReturn(PETSC_SUCCESS);
67: }
69: PetscErrorCode createSwarm(const DM dm, PetscInt dim, DM *sw)
70: {
71: PetscInt Nc = 1;
73: PetscFunctionBeginUser;
74: PetscCall(DMCreate(PETSC_COMM_SELF, sw));
75: PetscCall(DMSetType(*sw, DMSWARM));
76: PetscCall(DMSetDimension(*sw, dim));
77: PetscCall(DMSwarmSetType(*sw, DMSWARM_PIC));
78: PetscCall(DMSwarmSetCellDM(*sw, dm));
79: PetscCall(DMSwarmRegisterPetscDatatypeField(*sw, "w_q", Nc, PETSC_REAL));
80: PetscCall(DMSwarmFinalizeFieldRegister(*sw));
81: PetscCall(DMSetFromOptions(*sw));
82: PetscCall(PetscObjectSetName((PetscObject)*sw, "Particle Grid"));
83: PetscFunctionReturn(PETSC_SUCCESS);
84: }
86: static PetscErrorCode makeSwarm(DM sw, const PetscInt dim, const PetscInt Np, const PetscReal xx[], const PetscReal yy[], const PetscReal zz[])
87: {
88: PetscReal *coords;
89: PetscDataType dtype;
90: PetscInt bs, p, zero = 0;
92: PetscFunctionBeginUser;
93: PetscCall(DMSwarmSetLocalSizes(sw, Np, zero));
94: PetscCall(DMSwarmGetField(sw, "DMSwarmPIC_coor", &bs, &dtype, (void **)&coords));
95: for (p = 0; p < Np; p++) {
96: coords[p * dim + 0] = xx[p];
97: coords[p * dim + 1] = yy[p];
98: if (dim == 3) coords[p * dim + 2] = zz[p];
99: }
100: PetscCall(DMSwarmRestoreField(sw, "DMSwarmPIC_coor", &bs, &dtype, (void **)&coords));
101: PetscCall(DMSwarmVectorDefineField(sw, "w_q"));
102: PetscFunctionReturn(PETSC_SUCCESS);
103: }
105: static PetscErrorCode createMp(const DM dm, DM sw, Mat *Mp_out)
106: {
107: PetscBool removePoints = PETSC_TRUE;
108: Mat M_p;
110: PetscFunctionBeginUser;
111: // migrate after coords are set
112: PetscCall(DMSwarmMigrate(sw, removePoints));
113: //
114: PetscCall(PetscObjectSetName((PetscObject)sw, "Particle Grid"));
116: /* PetscInt N,*count,nmin=10000,nmax=0,ntot=0; */
117: /* // count */
118: /* PetscCall(DMSwarmCreatePointPerCellCount(sw, &N, &count)); */
119: /* for (int i=0, n; i< N ; i++) { */
120: /* if ((n=count[i]) > nmax) nmax = n; */
121: /* if (n < nmin) nmin = n; */
122: /* PetscCall(PetscInfo(dm, " %d) %d particles\n", i, n)); */
123: /* ntot += n; */
124: /* } */
125: /* PetscCall(PetscFree(count)); */
126: /* PetscCall(PetscInfo(dm, " %" PetscInt_FMT " max particle / cell, and %" PetscInt_FMT " min, ratio = %g, %" PetscInt_FMT " total\n", nmax, nmin, (double)nmax/(double)nmin,ntot)); */
128: /* This gives M f = \int_\Omega \phi f, which looks like a rhs for a PDE */
129: PetscCall(DMCreateMassMatrix(sw, dm, &M_p));
130: PetscCall(DMViewFromOptions(sw, NULL, "-ex30_sw_view"));
131: // output
132: *Mp_out = M_p;
133: PetscFunctionReturn(PETSC_SUCCESS);
134: }
136: static PetscErrorCode particlesToGrid(const DM dm, DM sw, const PetscInt a_tid, const PetscInt dim, const PetscReal a_wp[], Vec rho, Mat M_p)
137: {
138: PetscReal *wq;
139: PetscDataType dtype;
140: Vec ff;
141: PetscInt bs, p, Np;
143: PetscFunctionBeginUser;
144: PetscCall(DMSwarmGetField(sw, "w_q", &bs, &dtype, (void **)&wq));
145: PetscCall(DMSwarmGetLocalSize(sw, &Np));
146: for (p = 0; p < Np; p++) wq[p] = a_wp[p];
147: PetscCall(DMSwarmRestoreField(sw, "w_q", &bs, &dtype, (void **)&wq));
148: PetscCall(PetscObjectSetName((PetscObject)rho, "rho"));
149: PetscCall(DMSwarmCreateGlobalVectorFromField(sw, "w_q", &ff));
150: PetscCall(PetscObjectSetName((PetscObject)ff, "weights"));
151: PetscCall(MatMultTranspose(M_p, ff, rho));
152: PetscCall(DMSwarmDestroyGlobalVectorFromField(sw, "w_q", &ff));
153: PetscFunctionReturn(PETSC_SUCCESS);
154: }
156: //
157: // add grid to arg 'sw.w_q'
158: //
159: PetscErrorCode gridToParticles(const DM dm, DM sw, const Vec rhs, Vec work_ferhs, Mat M_p, Mat Mass)
160: {
161: PetscBool is_lsqr;
162: KSP ksp;
163: Mat PM_p = NULL, MtM, D = NULL;
164: Vec ff;
165: PetscInt N, M, nzl;
166: MatShellCtx *matshellctx = NULL;
167: PC pc;
169: PetscFunctionBeginUser;
170: // 1) apply M in, for Moore-Penrose with mass: Mp (Mp' Mp)^-1 M
171: PetscCall(MatMult(Mass, rhs, work_ferhs));
172: // 2) pseudo-inverse, first part: (Mp' Mp)^-1
173: PetscCall(KSPCreate(PETSC_COMM_SELF, &ksp));
174: PetscCall(KSPSetType(ksp, KSPCG));
175: PetscCall(KSPGetPC(ksp, &pc));
176: PetscCall(PCSetType(pc, PCJACOBI));
177: PetscCall(KSPSetOptionsPrefix(ksp, "ftop_"));
178: PetscCall(KSPSetFromOptions(ksp));
179: PetscCall(PetscObjectTypeCompare((PetscObject)ksp, KSPLSQR, &is_lsqr));
180: if (!is_lsqr) {
181: PetscCall(MatGetLocalSize(M_p, &M, &N));
182: if (N > M) {
183: PetscCall(PetscInfo(ksp, " M (%" PetscInt_FMT ") < M (%" PetscInt_FMT ") more vertices than particles: revert to lsqr\n", M, N));
184: is_lsqr = PETSC_TRUE;
185: PetscCall(KSPSetType(ksp, KSPLSQR));
186: PetscCall(PCSetType(pc, PCNONE)); // should not happen, but could solve stable (Mp^T Mp), move projection Mp before solve
187: } else {
188: PetscCall(PetscNew(&matshellctx));
189: PetscCall(MatCreateVecs(M_p, &matshellctx->uu, &matshellctx->ff));
190: if (0) {
191: PetscCall(MatTransposeMatMult(M_p, M_p, MAT_INITIAL_MATRIX, 4, &MtM));
192: PetscCall(KSPSetOperators(ksp, MtM, MtM));
193: PetscCall(PetscInfo(M_p, "createMtM KSP with explicit Mp'Mp\n"));
194: PetscCall(MatViewFromOptions(MtM, NULL, "-ftop2_MtM_mat_view"));
195: } else {
196: PetscCall(MatCreateShell(PetscObjectComm((PetscObject)dm), N, N, PETSC_DECIDE, PETSC_DECIDE, matshellctx, &MtM));
197: PetscCall(MatTranspose(M_p, MAT_INITIAL_MATRIX, &matshellctx->MpTrans));
198: matshellctx->Mp = M_p;
199: PetscCall(MatShellSetOperation(MtM, MATOP_MULT, (PetscErrorCodeFn *)MatMultMtM_SeqAIJ));
200: PetscCall(MatShellSetOperation(MtM, MATOP_MULT_ADD, (PetscErrorCodeFn *)MatMultAddMtM_SeqAIJ));
201: PetscCall(MatCreateSeqAIJ(PETSC_COMM_SELF, N, N, 1, NULL, &D));
202: PetscCall(MatViewFromOptions(matshellctx->MpTrans, NULL, "-ftop2_MpT_mat_view"));
203: for (PetscInt i = 0; i < N; i++) {
204: const PetscScalar *vals;
205: const PetscInt *cols;
206: PetscScalar dot = 0;
207: PetscCall(MatGetRow(matshellctx->MpTrans, i, &nzl, &cols, &vals));
208: for (PetscInt ii = 0; ii < nzl; ii++) dot += PetscSqr(vals[ii]);
209: if (dot < PETSC_MACHINE_EPSILON) {
210: PetscCall(PetscInfo(ksp, "empty row in pseudo-inverse %d\n", (int)i));
211: is_lsqr = PETSC_TRUE; // empty rows
212: PetscCall(KSPSetType(ksp, KSPLSQR));
213: PetscCall(PCSetType(pc, PCNONE)); // should not happen, but could solve stable (Mp Mp^T), move projection Mp before solve
214: // clean up
215: PetscCall(MatDestroy(&matshellctx->MpTrans));
216: PetscCall(VecDestroy(&matshellctx->ff));
217: PetscCall(VecDestroy(&matshellctx->uu));
218: PetscCall(MatDestroy(&D));
219: PetscCall(MatDestroy(&MtM));
220: PetscCall(PetscFree(matshellctx));
221: D = NULL;
222: break;
223: }
224: PetscCall(MatSetValue(D, i, i, dot, INSERT_VALUES));
225: }
226: if (D) {
227: PetscCall(MatAssemblyBegin(D, MAT_FINAL_ASSEMBLY));
228: PetscCall(MatAssemblyEnd(D, MAT_FINAL_ASSEMBLY));
229: PetscCall(PetscInfo(M_p, "createMtMKSP Have %" PetscInt_FMT " eqs, nzl = %" PetscInt_FMT "\n", N, nzl));
230: PetscCall(KSPSetOperators(ksp, MtM, D));
231: PetscCall(MatViewFromOptions(D, NULL, "-ftop2_D_mat_view"));
232: PetscCall(MatViewFromOptions(M_p, NULL, "-ftop2_Mp_mat_view"));
233: PetscCall(MatViewFromOptions(matshellctx->MpTrans, NULL, "-ftop2_MpTranspose_mat_view"));
234: PetscCall(MatViewFromOptions(MtM, NULL, "-ftop2_MtM_mat_view"));
235: }
236: }
237: }
238: }
239: if (is_lsqr) {
240: PC pc2;
241: PetscBool is_bjac;
242: PetscCall(KSPGetPC(ksp, &pc2));
243: PetscCall(PetscObjectTypeCompare((PetscObject)pc2, PCBJACOBI, &is_bjac));
244: if (is_bjac) {
245: PetscCall(DMSwarmCreateMassMatrixSquare(sw, dm, &PM_p));
246: PetscCall(KSPSetOperators(ksp, M_p, PM_p));
247: } else {
248: PetscCall(KSPSetOperators(ksp, M_p, M_p));
249: }
250: PetscCall(MatViewFromOptions(M_p, NULL, "-ftop2_Mp_mat_view"));
251: }
252: PetscCall(DMSwarmCreateGlobalVectorFromField(sw, "w_q", &ff)); // this grabs access
253: if (!is_lsqr) {
254: PetscCall(KSPSolve(ksp, work_ferhs, matshellctx->uu));
255: // 3) with Moore-Penrose apply Mp: M_p (Mp' Mp)^-1 M
256: PetscCall(MatMult(M_p, matshellctx->uu, ff));
257: if (D) PetscCall(MatDestroy(&D));
258: PetscCall(MatDestroy(&MtM));
259: if (matshellctx->MpTrans) PetscCall(MatDestroy(&matshellctx->MpTrans));
260: PetscCall(VecDestroy(&matshellctx->ff));
261: PetscCall(VecDestroy(&matshellctx->uu));
262: PetscCall(PetscFree(matshellctx));
263: } else {
264: // finally with LSQR apply M_p^\dagger
265: PetscCall(KSPSolveTranspose(ksp, work_ferhs, ff));
266: }
267: PetscCall(KSPDestroy(&ksp));
268: PetscCall(MatDestroy(&PM_p));
269: PetscCall(DMSwarmDestroyGlobalVectorFromField(sw, "w_q", &ff));
270: PetscFunctionReturn(PETSC_SUCCESS);
271: }
273: #define EX30_MAX_NUM_THRDS 12
274: #define EX30_MAX_BATCH_SZ 1024
275: //
276: // add grid to arg 'globSwarmArray[].w_q'
277: //
278: PetscErrorCode gridToParticles_private(DM grid_dm[], DM globSwarmArray[], const PetscInt dim, const PetscInt v_target, const PetscInt numthreads, const PetscInt num_vertices, const PetscInt global_vertex_id, Mat globMpArray[], Mat g_Mass[], Vec t_fhat[][EX30_MAX_NUM_THRDS], PetscReal moments[], Vec globXArray[], LandauCtx *ctx)
279: {
280: PetscErrorCode ierr = (PetscErrorCode)0; // used for inside thread loops
282: PetscFunctionBeginUser;
283: // map back to particles
284: for (PetscInt v_id_0 = 0; v_id_0 < ctx->batch_sz; v_id_0 += numthreads) {
285: PetscCall(PetscInfo(grid_dm[0], "g2p: global batch %" PetscInt_FMT " of %" PetscInt_FMT ", Landau batch %" PetscInt_FMT " of %" PetscInt_FMT ": map back to particles\n", global_vertex_id + 1, num_vertices, v_id_0 + 1, ctx->batch_sz));
286: //PetscPragmaOMP(parallel for)
287: for (PetscInt tid = 0; tid < numthreads; tid++) {
288: const PetscInt v_id = v_id_0 + tid, glb_v_id = global_vertex_id + v_id;
289: if (glb_v_id < num_vertices) {
290: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) { // add same particels for all grids
291: PetscErrorCode ierr_t;
292: ierr_t = PetscInfo(grid_dm[0], "gridToParticles: global batch %" PetscInt_FMT ", local batch b=%" PetscInt_FMT ", grid g=%" PetscInt_FMT ", index(b,g) %" PetscInt_FMT "\n", global_vertex_id, v_id, grid, LAND_PACK_IDX(v_id, grid));
293: ierr_t = gridToParticles(grid_dm[grid], globSwarmArray[LAND_PACK_IDX(v_id, grid)], globXArray[LAND_PACK_IDX(v_id, grid)], t_fhat[grid][tid], globMpArray[LAND_PACK_IDX(v_id, grid)], g_Mass[grid]);
294: if (ierr_t) ierr = ierr_t;
295: }
296: }
297: }
298: PetscCheck(!ierr, PETSC_COMM_WORLD, PETSC_ERR_PLIB, "Error in OMP loop. ierr = %d", (int)ierr);
299: /* Get moments */
300: PetscCall(PetscInfo(grid_dm[0], "Cleanup batches %" PetscInt_FMT " to %" PetscInt_FMT "\n", v_id_0, v_id_0 + numthreads));
301: for (PetscInt tid = 0; tid < numthreads; tid++) {
302: const PetscInt v_id = v_id_0 + tid, glb_v_id = global_vertex_id + v_id;
303: if (glb_v_id == v_target) {
304: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) {
305: PetscDataType dtype;
306: PetscReal *wp, *coords;
307: DM sw = globSwarmArray[LAND_PACK_IDX(v_id, grid)];
308: PetscInt npoints, bs = 1;
309: PetscCall(DMSwarmGetField(sw, "w_q", &bs, &dtype, (void **)&wp)); // take data out here
310: PetscCall(DMSwarmGetField(sw, "DMSwarmPIC_coor", &bs, &dtype, (void **)&coords));
311: PetscCall(DMSwarmGetLocalSize(sw, &npoints));
312: for (PetscInt p = 0; p < npoints; p++) {
313: PetscReal v2 = 0, fact = (dim == 2) ? 2.0 * PETSC_PI * coords[p * dim + 0] : 1, w = fact * wp[p] * ctx->n_0 * ctx->masses[ctx->species_offset[grid]];
314: for (PetscInt i = 0; i < dim; ++i) v2 += PetscSqr(coords[p * dim + i]);
315: moments[0] += w;
316: moments[1] += w * ctx->v_0 * coords[p * dim + 1]; // z-momentum
317: moments[2] += w * 0.5 * ctx->v_0 * ctx->v_0 * v2;
318: }
319: PetscCall(DMSwarmRestoreField(sw, "w_q", &bs, &dtype, (void **)&wp));
320: PetscCall(DMSwarmRestoreField(sw, "DMSwarmPIC_coor", &bs, &dtype, (void **)&coords));
321: }
322: const PetscReal N_inv = 1 / moments[0];
323: PetscCall(PetscInfo(grid_dm[0], "gridToParticles_private [%" PetscInt_FMT "], n = %g\n", v_id, (double)moments[0]));
324: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) {
325: PetscDataType dtype;
326: PetscReal *wp, *coords;
327: DM sw = globSwarmArray[LAND_PACK_IDX(v_id, grid)];
328: PetscInt npoints, bs = 1;
329: PetscCall(DMSwarmGetField(sw, "w_q", &bs, &dtype, (void **)&wp)); // take data out here
330: PetscCall(DMSwarmGetField(sw, "DMSwarmPIC_coor", &bs, &dtype, (void **)&coords));
331: PetscCall(DMSwarmGetLocalSize(sw, &npoints));
332: for (PetscInt p = 0; p < npoints; p++) {
333: const PetscReal fact = dim == 2 ? 2.0 * PETSC_PI * coords[p * dim + 0] : 1, w = fact * wp[p] * ctx->n_0 * ctx->masses[ctx->species_offset[grid]], ww = w * N_inv;
334: if (w > PETSC_REAL_MIN) {
335: moments[3] -= ww * PetscLogReal(ww);
336: PetscCheck(ww < 1 - PETSC_MACHINE_EPSILON, PETSC_COMM_WORLD, PETSC_ERR_PLIB, "ww (%g) > 1", (double)ww);
337: } else moments[4] -= w; // keep track of density that is lost
338: }
339: PetscCall(DMSwarmRestoreField(sw, "w_q", &bs, &dtype, (void **)&wp));
340: PetscCall(DMSwarmRestoreField(sw, "DMSwarmPIC_coor", &bs, &dtype, (void **)&coords));
341: }
342: }
343: } // thread batch
344: } // batch
345: PetscFunctionReturn(PETSC_SUCCESS);
346: }
348: static void maxwellian(PetscInt dim, const PetscReal x[], PetscReal kt_m, PetscReal n, PetscReal shift, PetscScalar *u)
349: {
350: PetscInt i;
351: PetscReal v2 = 0, theta = 2.0 * kt_m; /* theta = 2kT/mc^2 */
353: if (shift != 0.) {
354: v2 = 0;
355: for (i = 0; i < dim - 1; ++i) v2 += x[i] * x[i];
356: v2 += (x[dim - 1] - shift) * (x[dim - 1] - shift);
357: /* evaluate the shifted Maxwellian */
358: u[0] += n * PetscPowReal(PETSC_PI * theta, -1.5) * (PetscExpReal(-v2 / theta));
359: } else {
360: /* compute the exponents, v^2 */
361: for (i = 0; i < dim; ++i) v2 += x[i] * x[i];
362: /* evaluate the Maxwellian */
363: u[0] += n * PetscPowReal(PETSC_PI * theta, -1.5) * (PetscExpReal(-v2 / theta));
364: }
365: }
367: static PetscErrorCode PostStep(TS ts)
368: {
369: PetscInt n, dim, nDMs, v_id;
370: PetscReal t;
371: LandauCtx *ctx;
372: Vec X;
373: PrintCtx *printCtx;
374: DM pack;
375: PetscReal moments[5], e_grid[LANDAU_MAX_GRIDS];
377: PetscFunctionBeginUser;
378: PetscCall(TSGetApplicationContext(ts, &printCtx));
379: if (!printCtx->print && !printCtx->print_entropy) PetscFunctionReturn(PETSC_SUCCESS);
380: ctx = printCtx->ctx;
381: if (printCtx->v_target < printCtx->global_vertex_id_0 || printCtx->v_target >= printCtx->global_vertex_id_0 + ctx->batch_sz) PetscFunctionReturn(PETSC_SUCCESS);
382: for (PetscInt i = 0; i < 5; i++) moments[i] = 0;
383: for (PetscInt i = 0; i < LANDAU_MAX_GRIDS; i++) e_grid[i] = 0;
384: v_id = printCtx->v_target % ctx->batch_sz;
385: PetscCall(TSGetDM(ts, &pack));
386: PetscCall(DMGetDimension(pack, &dim));
387: PetscCall(DMCompositeGetNumberDM(pack, &nDMs)); // number of vertices * number of grids
388: PetscCall(TSGetSolution(ts, &X));
389: PetscCall(TSGetStepNumber(ts, &n));
390: PetscCall(TSGetTime(ts, &t));
391: PetscCall(DMCompositeGetAccessArray(pack, X, nDMs, NULL, printCtx->globXArray));
392: if (printCtx->print_entropy && printCtx->v_target >= 0 && 0) {
393: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) {
394: PetscDataType dtype;
395: PetscReal *wp, *coords;
396: DM sw = printCtx->globSwarmArray[LAND_PACK_IDX(v_id, grid)];
397: Vec work, subX = printCtx->globXArray[LAND_PACK_IDX(v_id, grid)];
398: PetscInt bs, NN;
399: // C-G moments
400: PetscCall(VecDuplicate(subX, &work));
401: PetscCall(gridToParticles(printCtx->grid_dm[grid], sw, subX, work, printCtx->globMpArray[LAND_PACK_IDX(v_id, grid)], printCtx->g_Mass[grid]));
402: PetscCall(VecDestroy(&work));
403: // moments
404: PetscCall(DMSwarmGetField(sw, "DMSwarmPIC_coor", &bs, &dtype, (void **)&coords));
405: PetscCall(DMSwarmGetLocalSize(sw, &NN));
406: PetscCall(DMSwarmGetField(sw, "w_q", &bs, &dtype, (void **)&wp));
407: for (PetscInt pp = 0; pp < NN; pp++) {
408: PetscReal v2 = 0, fact = (dim == 2) ? 2.0 * PETSC_PI * coords[pp * dim + 0] : 1, w = fact * wp[pp] * ctx->n_0 * ctx->masses[ctx->species_offset[grid]];
409: for (PetscInt i = 0; i < dim; ++i) v2 += PetscSqr(coords[pp * dim + i]);
410: moments[0] += w;
411: moments[1] += w * ctx->v_0 * coords[pp * dim + 1]; // z-momentum
412: moments[2] += w * 0.5 * ctx->v_0 * ctx->v_0 * v2;
413: e_grid[grid] += w * 0.5 * ctx->v_0 * ctx->v_0 * v2;
414: }
415: PetscCall(DMSwarmRestoreField(sw, "DMSwarmPIC_coor", &bs, &dtype, (void **)&coords));
416: PetscCall(DMSwarmRestoreField(sw, "w_q", &bs, &dtype, (void **)&wp));
417: }
418: // entropy
419: const PetscReal N_inv = 1 / moments[0];
420: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) {
421: PetscDataType dtype;
422: PetscReal *wp, *coords;
423: DM sw = printCtx->globSwarmArray[LAND_PACK_IDX(v_id, grid)];
424: PetscInt bs, NN;
425: PetscCall(DMSwarmGetField(sw, "DMSwarmPIC_coor", &bs, &dtype, (void **)&coords));
426: PetscCall(DMSwarmGetLocalSize(sw, &NN));
427: PetscCall(DMSwarmGetField(sw, "w_q", &bs, &dtype, (void **)&wp));
428: for (PetscInt pp = 0; pp < NN; pp++) {
429: PetscReal fact = (dim == 2) ? 2.0 * PETSC_PI * coords[pp * dim + 0] : 1, w = fact * wp[pp] * ctx->n_0 * ctx->masses[ctx->species_offset[grid]], ww = w * N_inv;
430: if (w > PETSC_REAL_MIN) {
431: moments[3] -= ww * PetscLogReal(ww);
432: } else moments[4] -= w;
433: }
434: PetscCall(DMSwarmRestoreField(sw, "DMSwarmPIC_coor", &bs, &dtype, (void **)&coords));
435: PetscCall(DMSwarmRestoreField(sw, "w_q", &bs, &dtype, (void **)&wp));
436: }
437: PetscCall(PetscInfo(X, "%4d) time %e, Landau particle moments: 0: %18.12e 1: %19.12e 2: %18.12e entropy: %e loss %e. energy = %e + %e + %e\n", (int)n, (double)t, (double)moments[0], (double)moments[1], (double)moments[2], (double)moments[3], (double)(moments[4] / moments[0]), (double)e_grid[0], (double)e_grid[1], (double)e_grid[2]));
438: }
439: if (printCtx->print && printCtx->g_target >= 0) {
440: PetscInt grid = printCtx->g_target, id;
441: static PetscReal last_t = -100000, period = .5;
442: if (last_t == -100000) last_t = -period + t;
443: if (t >= last_t + period) {
444: last_t = t;
445: PetscCall(DMGetOutputSequenceNumber(ctx->plex[grid], &id, NULL));
446: PetscCall(DMSetOutputSequenceNumber(ctx->plex[grid], id + 1, t));
447: PetscCall(VecViewFromOptions(printCtx->globXArray[LAND_PACK_IDX(v_id % ctx->batch_sz, grid)], NULL, "-ex30_vec_view"));
448: if (ctx->num_grids > grid + 1) {
449: PetscCall(DMSetOutputSequenceNumber(ctx->plex[grid + 1], id + 1, t));
450: PetscCall(VecViewFromOptions(printCtx->globXArray[LAND_PACK_IDX(v_id % ctx->batch_sz, grid + 1)], NULL, "-ex30_vec_view2"));
451: }
452: PetscCall(PetscInfo(X, "%4d) time %e View\n", (int)n, (double)t));
453: }
454: }
455: PetscCall(DMCompositeRestoreAccessArray(pack, X, nDMs, NULL, printCtx->globXArray));
456: PetscFunctionReturn(PETSC_SUCCESS);
457: }
459: PetscErrorCode go(TS ts, Vec X, const PetscInt num_vertices, const PetscInt a_Np, const PetscInt dim, const PetscInt v_target, const PetscInt g_target, PetscReal shift, PetscBool use_uniform_particle_grid)
460: {
461: DM pack, *globSwarmArray, grid_dm[LANDAU_MAX_GRIDS];
462: Mat *globMpArray, g_Mass[LANDAU_MAX_GRIDS];
463: KSP t_ksp[LANDAU_MAX_GRIDS][EX30_MAX_NUM_THRDS];
464: Vec t_fhat[LANDAU_MAX_GRIDS][EX30_MAX_NUM_THRDS];
465: PetscInt nDMs;
466: PetscErrorCode ierr = (PetscErrorCode)0; // used for inside thread loops
467: #if defined(PETSC_HAVE_OPENMP) && defined(PETSC_HAVE_THREADSAFETY)
468: PetscInt numthreads = PetscNumOMPThreads;
469: #else
470: PetscInt numthreads = 1;
471: #endif
472: LandauCtx *ctx;
473: Vec *globXArray;
474: PetscReal moments_0[5], moments_1a[5], moments_1b[5], dt_init;
475: PrintCtx *printCtx;
477: PetscFunctionBeginUser;
478: PetscCheck(numthreads <= EX30_MAX_NUM_THRDS, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "Too many threads %" PetscInt_FMT " > %d", numthreads, EX30_MAX_NUM_THRDS);
479: PetscCheck(numthreads > 0, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "Number threads %" PetscInt_FMT " > %d", numthreads, EX30_MAX_NUM_THRDS);
480: PetscCall(TSGetDM(ts, &pack));
481: PetscCall(DMGetApplicationContext(pack, &ctx));
482: PetscCheck(ctx->batch_sz % numthreads == 0, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "batch size (-dm_landau_batch_size) %" PetscInt_FMT " mod #threads %" PetscInt_FMT " must equal zero", ctx->batch_sz, numthreads);
483: PetscCall(DMCompositeGetNumberDM(pack, &nDMs)); // number of vertices * number of grids
484: PetscCall(PetscInfo(pack, "Have %" PetscInt_FMT " total grids, with %" PetscInt_FMT " Landau local batched and %" PetscInt_FMT " global items (vertices) %d DMs\n", ctx->num_grids, ctx->batch_sz, num_vertices, (int)nDMs));
485: PetscCall(PetscMalloc(sizeof(*globXArray) * nDMs, &globXArray));
486: PetscCall(PetscMalloc(sizeof(*globMpArray) * nDMs, &globMpArray));
487: PetscCall(PetscMalloc(sizeof(*globSwarmArray) * nDMs, &globSwarmArray));
488: // print ctx
489: PetscCall(PetscNew(&printCtx));
490: PetscCall(TSSetApplicationContext(ts, printCtx));
491: printCtx->v_target = v_target;
492: printCtx->g_target = g_target;
493: printCtx->ctx = ctx;
494: printCtx->globSwarmArray = globSwarmArray;
495: printCtx->grid_dm = grid_dm;
496: printCtx->globMpArray = globMpArray;
497: printCtx->g_Mass = g_Mass;
498: printCtx->globXArray = globXArray;
499: printCtx->print_entropy = PETSC_FALSE;
500: PetscOptionsBegin(PETSC_COMM_SELF, "", "Print Options", "DMPLEX");
501: PetscCall(PetscOptionsBool("-print_entropy", "Print entropy and moments at each time step", "ex30.c", printCtx->print_entropy, &printCtx->print_entropy, NULL));
502: PetscOptionsEnd();
503: // view
504: PetscCall(DMViewFromOptions(ctx->plex[g_target], NULL, "-ex30_dm_view"));
505: if (ctx->num_grids > g_target + 1) PetscCall(DMViewFromOptions(ctx->plex[g_target + 1], NULL, "-ex30_dm_view2"));
506: // create mesh mass matrices
507: PetscCall(VecZeroEntries(X));
508: PetscCall(DMCompositeGetAccessArray(pack, X, nDMs, NULL, globXArray)); // just to duplicate
509: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) { // add same particels for all grids
510: Vec subX = globXArray[LAND_PACK_IDX(0, grid)];
511: DM dm = ctx->plex[grid];
512: PetscSection s;
513: grid_dm[grid] = dm;
514: PetscCall(DMCreateMassMatrix(dm, dm, &g_Mass[grid]));
515: //
516: PetscCall(DMGetLocalSection(dm, &s));
517: PetscCall(DMPlexCreateClosureIndex(dm, s));
518: for (PetscInt tid = 0; tid < numthreads; tid++) {
519: PC pc;
520: PetscCall(VecDuplicate(subX, &t_fhat[grid][tid]));
521: PetscCall(KSPCreate(PETSC_COMM_SELF, &t_ksp[grid][tid]));
522: PetscCall(KSPSetType(t_ksp[grid][tid], KSPCG));
523: PetscCall(KSPGetPC(t_ksp[grid][tid], &pc));
524: PetscCall(PCSetType(pc, PCJACOBI));
525: PetscCall(KSPSetOptionsPrefix(t_ksp[grid][tid], "ptof_"));
526: PetscCall(KSPSetOperators(t_ksp[grid][tid], g_Mass[grid], g_Mass[grid]));
527: PetscCall(KSPSetFromOptions(t_ksp[grid][tid]));
528: }
529: }
530: PetscCall(DMCompositeRestoreAccessArray(pack, X, nDMs, NULL, globXArray));
531: PetscCall(TSGetTimeStep(ts, &dt_init)); // we could have an adaptive time stepper
532: // loop over all vertices in chucks that are batched for TSSolve
533: for (PetscInt i = 0; i < 5; i++) moments_0[i] = moments_1a[i] = moments_1b[i] = 0;
534: for (PetscInt global_vertex_id_0 = 0; global_vertex_id_0 < num_vertices; global_vertex_id_0 += ctx->batch_sz, shift /= 2) { // outer vertex loop
535: PetscCall(TSSetTime(ts, 0));
536: PetscCall(TSSetStepNumber(ts, 0));
537: PetscCall(TSSetTimeStep(ts, dt_init));
538: PetscCall(DMCompositeGetAccessArray(pack, X, nDMs, NULL, globXArray));
539: printCtx->global_vertex_id_0 = global_vertex_id_0;
540: if (v_target >= global_vertex_id_0 && v_target < global_vertex_id_0 + ctx->batch_sz) {
541: PetscCall(PetscObjectSetName((PetscObject)globXArray[LAND_PACK_IDX(v_target % ctx->batch_sz, g_target)], "rho"));
542: printCtx->print = PETSC_TRUE;
543: } else printCtx->print = PETSC_FALSE;
544: // create fake particles in batches with threads
545: for (PetscInt v_id_0 = 0; v_id_0 < ctx->batch_sz; v_id_0 += numthreads) {
546: PetscReal *xx_t[LANDAU_MAX_GRIDS][EX30_MAX_NUM_THRDS], *yy_t[LANDAU_MAX_GRIDS][EX30_MAX_NUM_THRDS], *zz_t[LANDAU_MAX_GRIDS][EX30_MAX_NUM_THRDS], *wp_t[LANDAU_MAX_GRIDS][EX30_MAX_NUM_THRDS] /* , radiuses[80000] */;
547: PetscInt Np_t[LANDAU_MAX_GRIDS][EX30_MAX_NUM_THRDS];
548: // make particles
549: for (PetscInt tid = 0; tid < numthreads; tid++) {
550: const PetscInt v_id = v_id_0 + tid, glb_v_id = global_vertex_id_0 + v_id;
551: if (glb_v_id < num_vertices) { // the ragged edge (in last batch)
552: PetscInt Npp0 = a_Np + (glb_v_id % (a_Np / 10 + 1)), nTargetP[LANDAU_MAX_GRIDS]; // n of particels in each dim with load imbalance
553: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) { // add same particels for all grids
554: // for (PetscInt sp = ctx->species_offset[grid], i0 = 0; sp < ctx->species_offset[grid + 1]; sp++, i0++) {
555: const PetscReal kT_m = ctx->k * ctx->thermal_temps[ctx->species_offset[grid]] / ctx->masses[ctx->species_offset[grid]] / (ctx->v_0 * ctx->v_0); /* theta = 2kT/mc^2 per species */
556: PetscReal lo[3] = {-ctx->radius[grid], -ctx->radius[grid], -ctx->radius[grid]}, hi[3] = {ctx->radius[grid], ctx->radius[grid], ctx->radius[grid]}, hp[3], vole; // would be nice to get box from DM
557: PetscInt Npi = Npp0, Npj = 2 * Npp0, Npk = 1;
558: PetscRandom rand;
559: PetscReal sigma = ctx->thermal_speed[grid] / ctx->thermal_speed[0], p2_shift = grid == 0 ? shift : -shift; // symmetric shift of e vs ions
560: PetscCall(PetscRandomCreate(PETSC_COMM_SELF, &rand));
561: PetscCall(PetscRandomSetInterval(rand, 0., 1.));
562: PetscCall(PetscRandomSetFromOptions(rand));
563: if (dim == 2) lo[0] = 0; // Landau coordinate (r,z)
564: else Npi = Npj = Npk = Npp0;
565: // User: use glb_v_id to index into your data
566: const PetscInt NNreal = Npi * Npj * Npk, NN = NNreal + (dim == 2 ? 3 : 6); // make room for bounding box
567: Np_t[grid][tid] = NN;
568: if (glb_v_id == v_target) nTargetP[grid] = NN;
569: PetscCall(PetscMalloc4(NN, &xx_t[grid][tid], NN, &yy_t[grid][tid], NN, &wp_t[grid][tid], dim == 2 ? 1 : NN, &zz_t[grid][tid]));
570: hp[0] = (hi[0] - lo[0]) / Npi;
571: hp[1] = (hi[1] - lo[1]) / Npj;
572: hp[2] = (hi[2] - lo[2]) / Npk;
573: if (dim == 2) hp[2] = 1;
574: PetscCall(PetscInfo(pack, " lo = %14.7e, hi = %14.7e; hp = %14.7e, %14.7e; kT_m = %g; \n", (double)lo[1], (double)hi[1], (double)hp[0], (double)hp[1], (double)kT_m)); // temp
575: vole = hp[0] * hp[1] * hp[2] * ctx->n[grid]; // fix for multi-species
576: PetscCall(PetscInfo(pack, "Vertex %" PetscInt_FMT ", grid %" PetscInt_FMT " with %" PetscInt_FMT " particles (diagnostic target = %" PetscInt_FMT ")\n", glb_v_id, grid, NN, v_target));
577: for (PetscInt pj = 0, pp = 0; pj < Npj; pj++) {
578: for (PetscInt pk = 0; pk < Npk; pk++) {
579: for (PetscInt pi = 0; pi < Npi; pi++, pp++) {
580: PetscReal p_shift = p2_shift;
581: wp_t[grid][tid][pp] = 0;
582: if (use_uniform_particle_grid) {
583: xx_t[grid][tid][pp] = lo[0] + hp[0] / 2.0 + pi * hp[0];
584: yy_t[grid][tid][pp] = lo[1] + hp[1] / 2.0 + pj * hp[1];
585: if (dim == 3) zz_t[grid][tid][pp] = lo[2] + hp[2] / 2.0 + pk * hp[2];
586: PetscReal x[] = {xx_t[grid][tid][pp], yy_t[grid][tid][pp], dim == 2 ? 0 : zz_t[grid][tid][pp]};
587: p_shift *= ctx->thermal_speed[grid] / ctx->v_0;
588: if (ctx->sphere && PetscSqrtReal(PetscSqr(xx_t[grid][tid][pp]) + PetscSqr(yy_t[grid][tid][pp])) > 0.92 * hi[0]) {
589: wp_t[grid][tid][pp] = 0;
590: } else {
591: maxwellian(dim, x, kT_m, vole, p_shift, &wp_t[grid][tid][pp]);
592: if (ctx->num_grids == 1 && shift != 0) { // bi-maxwellian, electron plasma
593: maxwellian(dim, x, kT_m, vole, -p_shift, &wp_t[grid][tid][pp]); // symmetric shift of electron plasma
594: }
595: }
596: } else {
597: PetscReal u1, u2;
598: do {
599: do {
600: PetscCall(PetscRandomGetValueReal(rand, &u1));
601: } while (u1 == 0);
602: PetscCall(PetscRandomGetValueReal(rand, &u2));
603: //compute z0 and z1
604: PetscReal mag = sigma * PetscSqrtReal(-2.0 * PetscLogReal(u1)); // is this the same scale grid Maxwellian? t_therm = sigma
605: xx_t[grid][tid][pp] = mag * PetscCosReal(2.0 * PETSC_PI * u2);
606: yy_t[grid][tid][pp] = mag * PetscSinReal(2.0 * PETSC_PI * u2);
607: if (dim == 2 && xx_t[grid][tid][pp] < lo[0]) xx_t[grid][tid][pp] = -xx_t[grid][tid][pp];
608: if (dim == 3) zz_t[grid][tid][pp] = lo[2] + hp[2] / 2.0 + pk * hp[2];
609: if (!ctx->sphere) {
610: if (dim == 2 && xx_t[grid][tid][pp] < 0) xx_t[grid][tid][pp] = -xx_t[grid][tid][pp]; // ???
611: else if (dim == 3) {
612: while (zz_t[grid][tid][pp] >= hi[2] || zz_t[grid][tid][pp] <= lo[2]) zz_t[grid][tid][pp] *= .9;
613: }
614: while (xx_t[grid][tid][pp] >= hi[0] || xx_t[grid][tid][pp] <= lo[0]) xx_t[grid][tid][pp] *= .9;
615: while (yy_t[grid][tid][pp] >= hi[1] || yy_t[grid][tid][pp] <= lo[1]) yy_t[grid][tid][pp] *= .9;
616: } else { // 2D
617: //if (glb_v_id == v_target && pp < 80000) radiuses[pp] = PetscSqrtReal(PetscSqr(xx_t[grid][tid][pp]) + PetscSqr(yy_t[grid][tid][pp]));
618: while (PetscSqrtReal(PetscSqr(xx_t[grid][tid][pp]) + PetscSqr(yy_t[grid][tid][pp])) > 0.92 * hi[0]) { // safety factor for facets of sphere
619: xx_t[grid][tid][pp] *= .9;
620: yy_t[grid][tid][pp] *= .9;
621: }
622: }
623: if (ctx->num_grids == 1 && pp % 2 == 0) p_shift = 0; // one species, split bi-max
624: p_shift *= ctx->thermal_speed[grid] / ctx->v_0;
625: if (dim == 3) zz_t[grid][tid][pp] += p_shift;
626: else yy_t[grid][tid][pp] += p_shift;
627: wp_t[grid][tid][pp] += ctx->n[grid] / NNreal * PetscSqrtReal(ctx->masses[ctx->species_offset[grid]] / ctx->masses[0]);
628: if (p_shift <= 0) break; // add bi-max for electron plasma only
629: p_shift = -p_shift;
630: } while (ctx->num_grids == 1); // add bi-max for electron plasma only
631: }
632: {
633: if (glb_v_id == v_target) {
634: PetscReal x[] = {xx_t[grid][tid][pp], yy_t[grid][tid][pp], dim == 2 ? 0 : zz_t[grid][tid][pp]};
635: PetscReal v2 = 0, fact = dim == 2 ? 2.0 * PETSC_PI * x[0] : 1, w = fact * wp_t[grid][tid][pp] * ctx->n_0 * ctx->masses[ctx->species_offset[grid]];
636: for (PetscInt i = 0; i < dim; ++i) v2 += PetscSqr(x[i]);
637: moments_0[0] += w; // not thread safe
638: moments_0[1] += w * ctx->v_0 * x[1]; // z-momentum
639: moments_0[2] += w * 0.5 * ctx->v_0 * ctx->v_0 * v2;
640: }
641: }
642: }
643: }
644: }
645: if (dim == 2) { // fix bounding box
646: PetscInt pp = NNreal;
647: wp_t[grid][tid][pp] = 0;
648: xx_t[grid][tid][pp] = 1.e-7;
649: yy_t[grid][tid][pp++] = hi[1] - 5.e-7;
650: wp_t[grid][tid][pp] = 0;
651: xx_t[grid][tid][pp] = hi[0] - 5.e-7;
652: yy_t[grid][tid][pp++] = 0;
653: wp_t[grid][tid][pp] = 0;
654: xx_t[grid][tid][pp] = 1.e-7;
655: yy_t[grid][tid][pp++] = lo[1] + 5.e-7;
656: } else {
657: const PetscInt p0 = NNreal;
658: for (PetscInt pj = 0; pj < 6; pj++) xx_t[grid][tid][p0 + pj] = yy_t[grid][tid][p0 + pj] = zz_t[grid][tid][p0 + pj] = wp_t[grid][tid][p0 + pj] = 0;
659: xx_t[grid][tid][p0 + 0] = lo[0];
660: xx_t[grid][tid][p0 + 1] = hi[0];
661: yy_t[grid][tid][p0 + 2] = lo[1];
662: yy_t[grid][tid][p0 + 3] = hi[1];
663: zz_t[grid][tid][p0 + 4] = lo[2];
664: zz_t[grid][tid][p0 + 5] = hi[2];
665: }
666: PetscCall(PetscRandomDestroy(&rand));
667: }
668: // entropy init, need global n
669: if (glb_v_id == v_target) {
670: const PetscReal N_inv = 1 / moments_0[0];
671: PetscCall(PetscInfo(pack, "Target %" PetscInt_FMT " with %" PetscInt_FMT " particels\n", glb_v_id, nTargetP[0]));
672: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) {
673: const PetscInt NN = nTargetP[grid];
674: for (PetscInt pp = 0; pp < NN; pp++) {
675: const PetscReal fact = dim == 2 ? 2.0 * PETSC_PI * xx_t[grid][tid][pp] : 1, w = fact * ctx->n_0 * ctx->masses[ctx->species_offset[grid]] * wp_t[grid][tid][pp], ww = w * N_inv;
676: if (w > PETSC_REAL_MIN) {
677: moments_0[3] -= ww * PetscLogReal(ww);
678: PetscCheck(ww < 1 - PETSC_MACHINE_EPSILON, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "ww (%g) > 1", (double)ww);
679: } else moments_0[4] -= w;
680: }
681: } // grid
682: } // target
683: } // active
684: } // threads
685: /* Create particle swarm */
686: for (PetscInt tid = 0; tid < numthreads; tid++) {
687: const PetscInt v_id = v_id_0 + tid, glb_v_id = global_vertex_id_0 + v_id;
688: if (glb_v_id < num_vertices) { // the ragged edge of the last batch
689: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) { // add same particels for all grids
690: PetscSection section;
691: PetscInt Nf;
692: DM dm = grid_dm[grid];
693: PetscCall(DMGetLocalSection(dm, §ion));
694: PetscCall(PetscSectionGetNumFields(section, &Nf));
695: PetscCheck(Nf == 1, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "Only one species per grid supported -- todo");
696: PetscCall(DMViewFromOptions(dm, NULL, "-dm_view"));
697: PetscCall(PetscInfo(pack, "call createSwarm [%" PetscInt_FMT ".%" PetscInt_FMT "] local block index %" PetscInt_FMT "\n", v_id, grid, LAND_PACK_IDX(v_id, grid)));
698: PetscCall(createSwarm(dm, dim, &globSwarmArray[LAND_PACK_IDX(v_id, grid)]));
699: }
700: } // active
701: } // threads
702: PetscCheck(ierr != 9999, PETSC_COMM_WORLD, PETSC_ERR_PLIB, "Only support one species per grid");
703: // make globMpArray
704: PetscPragmaOMP(parallel for)
705: for (PetscInt tid = 0; tid < numthreads; tid++) {
706: const PetscInt v_id = v_id_0 + tid, glb_v_id = global_vertex_id_0 + v_id;
707: if (glb_v_id < num_vertices) {
708: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) { // add same particels for all grids
709: // for (PetscInt sp = ctx->species_offset[grid], i0 = 0; sp < ctx->species_offset[grid + 1]; sp++, i0++) -- loop over species for Nf > 1 -- TODO
710: PetscErrorCode ierr_t;
711: DM sw = globSwarmArray[LAND_PACK_IDX(v_id, grid)];
712: ierr_t = PetscInfo(pack, "makeSwarm %" PetscInt_FMT ".%" PetscInt_FMT ") for block %" PetscInt_FMT "\n", v_id, grid, LAND_PACK_IDX(v_id, grid));
713: ierr_t = makeSwarm(sw, dim, Np_t[grid][tid], xx_t[grid][tid], yy_t[grid][tid], zz_t[grid][tid]);
714: if (ierr_t) ierr = ierr_t;
715: }
716: }
717: }
718: for (PetscInt tid = 0; tid < numthreads; tid++) {
719: const PetscInt v_id = v_id_0 + tid, glb_v_id = global_vertex_id_0 + v_id;
720: if (glb_v_id < num_vertices) {
721: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) { // add same particels for all grids
722: DM dm = grid_dm[grid];
723: DM sw = globSwarmArray[LAND_PACK_IDX(v_id, grid)];
724: PetscCall(PetscInfo(pack, "createMp %" PetscInt_FMT ".%" PetscInt_FMT ") for block %" PetscInt_FMT "\n", v_id, grid, LAND_PACK_IDX(v_id, grid)));
725: PetscCall(createMp(dm, sw, &globMpArray[LAND_PACK_IDX(v_id, grid)]));
726: PetscCall(MatViewFromOptions(globMpArray[LAND_PACK_IDX(v_id, grid)], NULL, "-mp_mat_view"));
727: }
728: }
729: }
730: // p --> g: set X
731: // PetscPragmaOMP(parallel for)
732: for (PetscInt tid = 0; tid < numthreads; tid++) {
733: const PetscInt v_id = v_id_0 + tid, glb_v_id = global_vertex_id_0 + v_id;
734: if (glb_v_id < num_vertices) {
735: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) { // add same particels for all grids
736: PetscErrorCode ierr_t;
737: DM dm = grid_dm[grid];
738: DM sw = globSwarmArray[LAND_PACK_IDX(v_id, grid)];
739: Vec subX = globXArray[LAND_PACK_IDX(v_id, grid)], work = t_fhat[grid][tid];
740: ierr_t = PetscInfo(pack, "particlesToGrid %" PetscInt_FMT ".%" PetscInt_FMT ") for block %" PetscInt_FMT "\n", v_id, grid, LAND_PACK_IDX(v_id, grid));
741: ierr_t = particlesToGrid(dm, sw, tid, dim, wp_t[grid][tid], subX, globMpArray[LAND_PACK_IDX(v_id, grid)]);
742: if (ierr_t) ierr = ierr_t;
743: // u = M^_1 f_w
744: ierr_t = VecCopy(subX, work);
745: ierr_t = KSPSolve(t_ksp[grid][tid], work, subX);
746: if (ierr_t) ierr = ierr_t;
747: }
748: }
749: }
750: PetscCheck(!ierr, PETSC_COMM_WORLD, PETSC_ERR_PLIB, "Error in OMP loop. ierr = %d", (int)ierr);
751: /* Cleanup */
752: for (PetscInt tid = 0; tid < numthreads; tid++) {
753: const PetscInt v_id = v_id_0 + tid, glb_v_id = global_vertex_id_0 + v_id;
754: if (glb_v_id < num_vertices) {
755: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) { // add same particels for all grids
756: PetscCall(PetscFree4(xx_t[grid][tid], yy_t[grid][tid], wp_t[grid][tid], zz_t[grid][tid]));
757: }
758: } // active
759: } // threads
760: } // (fake) particle loop
761: // standard view of initial conditions
762: if (v_target >= global_vertex_id_0 && v_target < global_vertex_id_0 + ctx->batch_sz) {
763: PetscCall(DMSetOutputSequenceNumber(ctx->plex[g_target], 0, 0.0));
764: PetscCall(VecViewFromOptions(globXArray[LAND_PACK_IDX(v_target % ctx->batch_sz, g_target)], NULL, "-ex30_vec_view"));
765: if (ctx->num_grids > g_target + 1) {
766: PetscCall(DMSetOutputSequenceNumber(ctx->plex[g_target + 1], 0, 0.0));
767: PetscCall(VecViewFromOptions(globXArray[LAND_PACK_IDX(v_target % ctx->batch_sz, g_target + 1)], NULL, "-ex30_vec_view2"));
768: }
769: PetscCall(MatViewFromOptions(globMpArray[LAND_PACK_IDX(v_target % ctx->batch_sz, g_target)], NULL, "-ex30_mass_mat_view"));
770: PetscCall(DMViewFromOptions(globSwarmArray[LAND_PACK_IDX(v_target % ctx->batch_sz, g_target)], NULL, "-ex30_sw_view"));
771: PetscCall(DMSwarmViewXDMF(globSwarmArray[LAND_PACK_IDX(v_target % ctx->batch_sz, g_target)], "initial_swarm.xmf")); // writes a file by default!!!
772: }
773: // coarse graining moments_1a, bring f back from grid before advance
774: if (v_target >= global_vertex_id_0 && v_target < global_vertex_id_0 + ctx->batch_sz && printCtx->print_entropy) {
775: const PetscInt v_id = v_target % ctx->batch_sz;
776: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) {
777: PetscDataType dtype;
778: PetscReal *wp, *coords;
779: DM sw = globSwarmArray[LAND_PACK_IDX(v_id, grid)];
780: Vec work, subX = globXArray[LAND_PACK_IDX(v_id, grid)];
781: PetscInt bs, NN;
782: // C-G moments
783: PetscCall(VecDuplicate(subX, &work));
784: PetscCall(gridToParticles(grid_dm[grid], sw, subX, work, globMpArray[LAND_PACK_IDX(v_id, grid)], g_Mass[grid]));
785: PetscCall(VecDestroy(&work));
786: // moments
787: PetscCall(DMSwarmGetField(sw, "DMSwarmPIC_coor", &bs, &dtype, (void **)&coords));
788: PetscCall(DMSwarmGetLocalSize(sw, &NN));
789: PetscCall(DMSwarmGetField(sw, "w_q", &bs, &dtype, (void **)&wp));
790: for (PetscInt pp = 0; pp < NN; pp++) {
791: PetscReal v2 = 0, fact = (dim == 2) ? 2.0 * PETSC_PI * coords[pp * dim + 0] : 1, w = fact * wp[pp] * ctx->n_0 * ctx->masses[ctx->species_offset[grid]];
792: for (PetscInt i = 0; i < dim; ++i) v2 += PetscSqr(coords[pp * dim + i]);
793: moments_1a[0] += w;
794: moments_1a[1] += w * ctx->v_0 * coords[pp * dim + 1]; // z-momentum
795: moments_1a[2] += w * 0.5 * ctx->v_0 * ctx->v_0 * v2;
796: }
797: PetscCall(DMSwarmRestoreField(sw, "DMSwarmPIC_coor", &bs, &dtype, (void **)&coords));
798: PetscCall(DMSwarmRestoreField(sw, "w_q", &bs, &dtype, (void **)&wp));
799: }
800: // entropy
801: const PetscReal N_inv = 1 / moments_1a[0];
802: PetscCall(PetscInfo(pack, "Entropy batch %" PetscInt_FMT " of %" PetscInt_FMT ", n = %g\n", v_target, num_vertices, (double)(1 / N_inv)));
803: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) {
804: PetscDataType dtype;
805: PetscReal *wp, *coords;
806: DM sw = globSwarmArray[LAND_PACK_IDX(v_id, grid)];
807: PetscInt bs, NN;
808: PetscCall(DMSwarmGetLocalSize(sw, &NN));
809: PetscCall(DMSwarmGetField(sw, "w_q", &bs, &dtype, (void **)&wp));
810: PetscCall(DMSwarmGetField(sw, "DMSwarmPIC_coor", &bs, &dtype, (void **)&coords));
811: for (PetscInt pp = 0; pp < NN; pp++) {
812: PetscReal fact = (dim == 2) ? 2.0 * PETSC_PI * coords[pp * dim + 0] : 1, w = fact * wp[pp] * ctx->n_0 * ctx->masses[ctx->species_offset[grid]], ww = w * N_inv;
813: if (w > PETSC_REAL_MIN) {
814: moments_1a[3] -= ww * PetscLogReal(ww);
815: PetscCheck(ww < 1 - PETSC_MACHINE_EPSILON, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "ww (%g) > 1", (double)ww);
816: } else moments_1a[4] -= w;
817: }
818: PetscCall(DMSwarmRestoreField(sw, "w_q", &bs, &dtype, (void **)&wp));
819: PetscCall(DMSwarmRestoreField(sw, "DMSwarmPIC_coor", &bs, &dtype, (void **)&coords));
820: }
821: }
822: // restore vector
823: PetscCall(DMCompositeRestoreAccessArray(pack, X, nDMs, NULL, globXArray));
824: // view initial grid
825: if (v_target >= global_vertex_id_0 && v_target < global_vertex_id_0 + ctx->batch_sz) PetscCall(DMPlexLandauPrintNorms(X, 0));
826: // advance
827: PetscCall(TSSetSolution(ts, X));
828: PetscCall(PetscInfo(pack, "Advance vertex %" PetscInt_FMT " to %" PetscInt_FMT "\n", global_vertex_id_0, global_vertex_id_0 + ctx->batch_sz));
829: PetscCall(TSSetPostStep(ts, PostStep));
830: PetscCall(PostStep(ts));
831: PetscCall(TSSolve(ts, X));
832: // view
833: PetscCall(DMCompositeGetAccessArray(pack, X, nDMs, NULL, globXArray));
834: if (v_target >= global_vertex_id_0 && v_target < global_vertex_id_0 + ctx->batch_sz) {
835: /* Visualize original particle field */
836: DM sw = globSwarmArray[LAND_PACK_IDX(v_target % ctx->batch_sz, g_target)];
837: Vec f;
838: PetscCall(DMSetOutputSequenceNumber(sw, 0, 0.0));
839: PetscCall(DMViewFromOptions(grid_dm[g_target], NULL, "-weights_dm_view"));
840: PetscCall(DMViewFromOptions(sw, NULL, "-weights_sw_view"));
841: PetscCall(DMSwarmCreateGlobalVectorFromField(sw, "w_q", &f));
842: PetscCall(PetscObjectSetName((PetscObject)f, "weights"));
843: PetscCall(VecViewFromOptions(f, NULL, "-weights_vec_view"));
844: PetscCall(DMSwarmDestroyGlobalVectorFromField(sw, "w_q", &f));
845: //
846: PetscCall(DMPlexLandauPrintNorms(X, 1));
847: }
848: if (!use_uniform_particle_grid) { // resample to uniform grid
849: for (PetscInt v_id_0 = 0; v_id_0 < ctx->batch_sz; v_id_0 += numthreads) {
850: PetscReal *xx_t[LANDAU_MAX_GRIDS][EX30_MAX_NUM_THRDS], *yy_t[LANDAU_MAX_GRIDS][EX30_MAX_NUM_THRDS], *zz_t[LANDAU_MAX_GRIDS][EX30_MAX_NUM_THRDS], *wp_t[LANDAU_MAX_GRIDS][EX30_MAX_NUM_THRDS];
851: PetscInt Np_t[LANDAU_MAX_GRIDS][EX30_MAX_NUM_THRDS];
852: for (PetscInt tid = 0; tid < numthreads; tid++) {
853: const PetscInt v_id = v_id_0 + tid, glb_v_id = global_vertex_id_0 + v_id;
854: if (glb_v_id < num_vertices) {
855: // create uniform grid w/o weights & smaller
856: PetscInt Npp0 = (a_Np + (glb_v_id % (a_Np / 10 + 1))) / 2, Nv; // 1/2 of uniform particle grid size
857: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) {
858: // for (PetscInt sp = ctx->species_offset[grid], i0 = 0; sp < ctx->species_offset[grid + 1]; sp++, i0++)
859: PetscReal lo[3] = {-ctx->radius[grid], -ctx->radius[grid], -ctx->radius[grid]}, hi[3] = {ctx->radius[grid], ctx->radius[grid], ctx->radius[grid]}, hp[3];
860: PetscInt Npi = Npp0, Npj = 2 * Npp0, Npk = 1, NN;
861: // delete old particles and particle mass matrix
862: PetscCall(DMDestroy(&globSwarmArray[LAND_PACK_IDX(v_id, grid)]));
863: PetscCall(MatDestroy(&globMpArray[LAND_PACK_IDX(v_id, grid)]));
864: // create fake particles in batches with threads
865: PetscCall(MatGetLocalSize(g_Mass[grid], &Nv, NULL));
866: if (dim == 2) lo[0] = 0;
867: else Npi = Npj = Npk = Npp0;
868: NN = Npi * Npj * Npk + (dim == 2 ? 3 : 6); // make a regular grid of particles Npp x Npp
869: while (Npi * Npj * Npk < Nv) { // make stable - no LS
870: Npi++;
871: Npj++;
872: Npk++;
873: NN = Npi * Npj * Npk + (dim == 2 ? 3 : 6);
874: }
875: Np_t[grid][tid] = NN;
876: PetscCall(PetscMalloc4(NN, &xx_t[grid][tid], NN, &yy_t[grid][tid], NN, &wp_t[grid][tid], dim == 2 ? 1 : NN, &zz_t[grid][tid]));
877: hp[0] = (hi[0] - lo[0]) / Npi;
878: hp[1] = (hi[1] - lo[1]) / Npj;
879: hp[2] = (hi[2] - lo[2]) / Npk;
880: if (dim == 2) hp[2] = 1;
881: PetscCall(PetscInfo(pack, "Resampling %d particles, %d vertices\n", (int)NN, (int)Nv)); // temp
882: for (PetscInt pj = 0, pp = 0; pj < Npj; pj++) {
883: for (PetscInt pk = 0; pk < Npk; pk++) {
884: for (PetscInt pi = 0; pi < Npi; pi++, pp++) {
885: wp_t[grid][tid][pp] = 0;
886: xx_t[grid][tid][pp] = lo[0] + hp[0] / 2.0 + pi * hp[0];
887: yy_t[grid][tid][pp] = lo[1] + hp[1] / 2.0 + pj * hp[1];
888: if (dim == 3) zz_t[grid][tid][pp] = lo[2] + hp[2] / 2.0 + pk * hp[2];
889: }
890: }
891: }
892: if (dim == 2) { // fix bounding box
893: PetscInt pp = NN - 3;
894: wp_t[grid][tid][pp] = 0;
895: xx_t[grid][tid][pp] = 1.e-7;
896: yy_t[grid][tid][pp++] = hi[1] - 5.e-7;
897: wp_t[grid][tid][pp] = 0;
898: xx_t[grid][tid][pp] = hi[0] - 5.e-7;
899: yy_t[grid][tid][pp++] = 0;
900: wp_t[grid][tid][pp] = 0;
901: xx_t[grid][tid][pp] = 1.e-7;
902: yy_t[grid][tid][pp++] = lo[1] + 5.e-7;
903: } else {
904: const PetscInt p0 = NN - 6;
905: for (PetscInt pj = 0; pj < 6; pj++) xx_t[grid][tid][p0 + pj] = yy_t[grid][tid][p0 + pj] = zz_t[grid][tid][p0 + pj] = wp_t[grid][tid][p0 + pj] = 0;
906: xx_t[grid][tid][p0 + 0] = lo[0];
907: xx_t[grid][tid][p0 + 1] = hi[0];
908: yy_t[grid][tid][p0 + 2] = lo[1];
909: yy_t[grid][tid][p0 + 3] = hi[1];
910: zz_t[grid][tid][p0 + 4] = lo[2];
911: zz_t[grid][tid][p0 + 5] = hi[2];
912: }
913: }
914: } // active
915: } // threads
916: /* Create particle swarm */
917: for (PetscInt tid = 0; tid < numthreads; tid++) {
918: const PetscInt v_id = v_id_0 + tid, glb_v_id = global_vertex_id_0 + v_id;
919: if (glb_v_id < num_vertices) { // the ragged edge of the last batch
920: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) { // add same particels for all grids
921: // for (PetscInt sp = ctx->species_offset[grid], i0 = 0; sp < ctx->species_offset[grid + 1]; sp++, i0++) -- loop over species for Nf > 1 -- TODO
922: PetscErrorCode ierr_t;
923: PetscSection section;
924: PetscInt Nf;
925: DM dm = grid_dm[grid];
926: ierr_t = DMGetLocalSection(dm, §ion);
927: ierr_t = PetscSectionGetNumFields(section, &Nf);
928: if (Nf != 1) ierr_t = (PetscErrorCode)9999;
929: else {
930: ierr_t = DMViewFromOptions(dm, NULL, "-dm_view");
931: ierr_t = PetscInfo(pack, "call createSwarm [%" PetscInt_FMT ".%" PetscInt_FMT "] local block index %" PetscInt_FMT "\n", v_id, grid, LAND_PACK_IDX(v_id, grid));
932: ierr_t = createSwarm(dm, dim, &globSwarmArray[LAND_PACK_IDX(v_id, grid)]);
933: }
934: if (ierr_t) ierr = ierr_t;
935: }
936: } // active
937: } // threads
938: PetscCheck(ierr != 9999, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "Only support one species per grid");
939: PetscCheck(!ierr, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "Error in OMP loop. ierr = %d", (int)ierr);
940: // make globMpArray
941: PetscPragmaOMP(parallel for)
942: for (PetscInt tid = 0; tid < numthreads; tid++) {
943: const PetscInt v_id = v_id_0 + tid, glb_v_id = global_vertex_id_0 + v_id;
944: if (glb_v_id < num_vertices) {
945: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) { // add same particels for all grids
946: // for (PetscInt sp = ctx->species_offset[grid], i0 = 0; sp < ctx->species_offset[grid + 1]; sp++, i0++) -- loop over species for Nf > 1 -- TODO
947: PetscErrorCode ierr_t;
948: DM sw = globSwarmArray[LAND_PACK_IDX(v_id, grid)];
949: ierr_t = PetscInfo(pack, "makeSwarm %" PetscInt_FMT ".%" PetscInt_FMT ") for block %" PetscInt_FMT "\n", v_id, grid, LAND_PACK_IDX(v_id, grid));
950: ierr_t = makeSwarm(sw, dim, Np_t[grid][tid], xx_t[grid][tid], yy_t[grid][tid], zz_t[grid][tid]);
951: if (ierr_t) ierr = ierr_t;
952: }
953: } // active
954: } // threads
955: // create particle mass matrices
956: //PetscPragmaOMP(parallel for)
957: for (PetscInt tid = 0; tid < numthreads; tid++) {
958: const PetscInt v_id = v_id_0 + tid, glb_v_id = global_vertex_id_0 + v_id;
959: if (glb_v_id < num_vertices) {
960: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) { // add same particels for all grids
961: PetscErrorCode ierr_t;
962: DM dm = grid_dm[grid];
963: DM sw = globSwarmArray[LAND_PACK_IDX(v_id, grid)];
964: ierr_t = PetscInfo(pack, "createMp %" PetscInt_FMT ".%" PetscInt_FMT ") for block %" PetscInt_FMT "\n", v_id, grid, LAND_PACK_IDX(v_id, grid));
965: ierr_t = createMp(dm, sw, &globMpArray[LAND_PACK_IDX(v_id, grid)]);
966: if (ierr_t) ierr = ierr_t;
967: }
968: } // active
969: } // threads
970: PetscCheck(!ierr, PETSC_COMM_WORLD, PETSC_ERR_PLIB, "Error in OMP loop. ierr = %d", (int)ierr);
971: /* Cleanup */
972: for (PetscInt tid = 0; tid < numthreads; tid++) {
973: const PetscInt v_id = v_id_0 + tid, glb_v_id = global_vertex_id_0 + v_id;
974: if (glb_v_id < num_vertices) {
975: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) { // add same particels for all grids
976: PetscCall(PetscFree4(xx_t[grid][tid], yy_t[grid][tid], wp_t[grid][tid], zz_t[grid][tid]));
977: }
978: } // active
979: } // threads
980: } // batch
981: // view
982: if (v_target >= global_vertex_id_0 && v_target < global_vertex_id_0 + ctx->batch_sz) {
983: /* Visualize particle field */
984: DM sw = globSwarmArray[LAND_PACK_IDX(v_target % ctx->batch_sz, g_target)];
985: Vec f;
986: PetscCall(DMSetOutputSequenceNumber(sw, 0, 0.0));
987: PetscCall(DMViewFromOptions(sw, NULL, "-resampled_weights_sw_view"));
988: PetscCall(DMSwarmCreateGlobalVectorFromField(sw, "w_q", &f));
989: PetscCall(PetscObjectSetName((PetscObject)f, "resampled_weights"));
990: PetscCall(VecViewFromOptions(f, NULL, "-resampled_weights_vec_view"));
991: PetscCall(DMSwarmDestroyGlobalVectorFromField(sw, "w_q", &f));
992: PetscCall(DMSwarmViewXDMF(sw, "resampled.xmf"));
993: }
994: } // !uniform
995: // particles to grid, compute moments and entropy, for target vertex only
996: if (v_target >= global_vertex_id_0 && v_target < global_vertex_id_0 + ctx->batch_sz && printCtx->print_entropy) {
997: PetscReal energy_error_rel;
998: PetscCall(gridToParticles_private(grid_dm, globSwarmArray, dim, v_target, numthreads, num_vertices, global_vertex_id_0, globMpArray, g_Mass, t_fhat, moments_1b, globXArray, ctx));
999: energy_error_rel = PetscAbsReal(moments_1b[2] - moments_0[2]) / moments_0[2];
1000: PetscCall(PetscPrintf(PETSC_COMM_WORLD, "Particle Moments:\t number density momentum (par) energy entropy negative weights : # OMP threads %g\n", (double)numthreads));
1001: PetscCall(PetscPrintf(PETSC_COMM_WORLD, "\tInitial: %18.12e %19.12e %18.12e %18.12e %g %%\n", (double)moments_0[0], (double)moments_0[1], (double)moments_0[2], (double)moments_0[3], 100 * (double)(moments_0[4] / moments_0[0])));
1002: PetscCall(PetscPrintf(PETSC_COMM_WORLD, "\tCoarse-graining: %18.12e %19.12e %18.12e %18.12e %g %%\n", (double)moments_1a[0], (double)moments_1a[1], (double)moments_1a[2], (double)moments_1a[3], 100 * (double)(moments_1a[4] / moments_0[0])));
1003: PetscCall(PetscPrintf(PETSC_COMM_WORLD, "\tLandau: %18.12e %19.12e %18.12e %18.12e %g %%\n", (double)moments_1b[0], (double)moments_1b[1], (double)moments_1b[2], (double)moments_1b[3], 100 * (double)(moments_1b[4] / moments_0[0])));
1004: PetscCall(PetscPrintf(PETSC_COMM_WORLD, "Coarse-graining entropy generation = %e ; Landau entropy generation = %e\n", (double)(moments_1a[3] - moments_0[3]), (double)(moments_1b[3] - moments_0[3])));
1005: PetscCall(PetscPrintf(PETSC_COMM_WORLD, "(relative) energy conservation: Coarse-graining = %e, Landau = %e (%g %d)\n", (double)(moments_1a[2] - moments_0[2]) / (double)moments_0[2], (double)energy_error_rel, (double)PetscLog10Real(energy_error_rel), (int)(PetscLog10Real(energy_error_rel) + .5)));
1006: }
1007: // restore vector
1008: PetscCall(DMCompositeRestoreAccessArray(pack, X, nDMs, NULL, globXArray));
1009: // cleanup
1010: for (PetscInt v_id_0 = 0; v_id_0 < ctx->batch_sz; v_id_0 += numthreads) {
1011: for (PetscInt tid = 0; tid < numthreads; tid++) {
1012: const PetscInt v_id = v_id_0 + tid, glb_v_id = global_vertex_id_0 + v_id;
1013: if (glb_v_id < num_vertices) {
1014: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) {
1015: PetscCall(DMDestroy(&globSwarmArray[LAND_PACK_IDX(v_id, grid)]));
1016: PetscCall(MatDestroy(&globMpArray[LAND_PACK_IDX(v_id, grid)]));
1017: }
1018: }
1019: }
1020: }
1021: } // user batch, not used
1022: /* Cleanup */
1023: PetscCall(PetscFree(globXArray));
1024: PetscCall(PetscFree(globSwarmArray));
1025: PetscCall(PetscFree(globMpArray));
1026: PetscCall(PetscFree(printCtx));
1027: // clean up mass matrices
1028: for (PetscInt grid = 0; grid < ctx->num_grids; grid++) { // add same particels for all grids
1029: PetscCall(MatDestroy(&g_Mass[grid]));
1030: for (PetscInt tid = 0; tid < numthreads; tid++) {
1031: PetscCall(VecDestroy(&t_fhat[grid][tid]));
1032: PetscCall(KSPDestroy(&t_ksp[grid][tid]));
1033: }
1034: }
1035: PetscFunctionReturn(PETSC_SUCCESS);
1036: }
1038: int main(int argc, char **argv)
1039: {
1040: DM pack;
1041: Vec X;
1042: PetscInt dim = 2, num_vertices = 1, Np = 10, v_target = 0, g_target = 0;
1043: TS ts;
1044: Mat J;
1045: LandauCtx *ctx;
1046: PetscReal shift = 0;
1047: PetscBool use_uniform_particle_grid = PETSC_TRUE;
1049: PetscFunctionBeginUser;
1050: PetscCall(PetscInitialize(&argc, &argv, NULL, help));
1051: // process args
1052: PetscOptionsBegin(PETSC_COMM_SELF, "", "Collision Options", "DMPLEX");
1053: PetscCall(PetscOptionsInt("-dim", "Velocity space dimension", "ex30.c", dim, &dim, NULL));
1054: PetscCall(PetscOptionsInt("-number_spatial_vertices", "Number of user spatial vertices to be batched for Landau", "ex30.c", num_vertices, &num_vertices, NULL));
1055: PetscCall(PetscOptionsInt("-number_particles_per_dimension", "Number of particles per grid, with slight modification per spatial vertex, in each dimension of base Cartesian grid", "ex30.c", Np, &Np, NULL));
1056: PetscCall(PetscOptionsBool("-use_uniform_particle_grid", "Use uniform particle grid", "ex30.c", use_uniform_particle_grid, &use_uniform_particle_grid, NULL));
1057: PetscCall(PetscOptionsInt("-vertex_view_target", "Global vertex for diagnostics", "ex30.c", v_target, &v_target, NULL));
1058: PetscCall(PetscOptionsReal("-e_shift", "Bi-Maxwellian shift", "ex30.c", shift, &shift, NULL));
1059: PetscCheck(v_target < num_vertices, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "Batch to view %" PetscInt_FMT " should be < number of vertices %" PetscInt_FMT, v_target, num_vertices);
1060: PetscCall(PetscOptionsInt("-grid_view_target", "Grid to view with diagnostics", "ex30.c", g_target, &g_target, NULL));
1061: PetscOptionsEnd();
1062: /* Create a mesh */
1063: PetscCall(DMPlexLandauCreateVelocitySpace(PETSC_COMM_SELF, dim, "", &X, &J, &pack));
1064: PetscCall(DMGetApplicationContext(pack, &ctx));
1065: PetscCall(DMSetUp(pack));
1066: PetscCall(DMSetOutputSequenceNumber(pack, 0, 0.0));
1067: PetscCheck(g_target < ctx->num_grids, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "Grid to view %" PetscInt_FMT " should be < number of grids %" PetscInt_FMT, g_target, ctx->num_grids);
1068: PetscCheck(ctx->batch_view_idx == v_target % ctx->batch_sz, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "Global view index %" PetscInt_FMT " mode batch size %" PetscInt_FMT " != ctx->batch_view_idx %" PetscInt_FMT, v_target, ctx->batch_sz, ctx->batch_view_idx);
1069: // PetscCheck(!use_uniform_particle_grid || !ctx->sphere, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "Can not use -use_uniform_particle_grid and -dm_landau_sphere");
1070: /* Create timestepping solver context */
1071: PetscCall(TSCreate(PETSC_COMM_SELF, &ts));
1072: PetscCall(TSSetDM(ts, pack));
1073: PetscCall(TSSetIFunction(ts, NULL, DMPlexLandauIFunction, NULL));
1074: PetscCall(TSSetIJacobian(ts, J, J, DMPlexLandauIJacobian, NULL));
1075: PetscCall(TSSetExactFinalTime(ts, TS_EXACTFINALTIME_STEPOVER));
1076: PetscCall(TSSetFromOptions(ts));
1077: PetscCall(PetscObjectSetName((PetscObject)X, "X"));
1078: // do particle advance
1079: PetscCall(go(ts, X, num_vertices, Np, dim, v_target, g_target, shift, use_uniform_particle_grid));
1080: PetscCall(MatZeroEntries(J)); // need to zero out so as to not reuse it in Landau's logic
1081: /* clean up */
1082: PetscCall(DMPlexLandauDestroyVelocitySpace(&pack));
1083: PetscCall(TSDestroy(&ts));
1084: PetscCall(VecDestroy(&X));
1085: PetscCall(PetscFinalize());
1086: return 0;
1087: }
1089: /*TEST
1091: build:
1092: requires: !complex
1094: testset:
1095: requires: double defined(PETSC_USE_DMLANDAU_2D)
1096: output_file: output/ex30_0.out
1097: args: -dim 2 -petscspace_degree 3 -dm_landau_num_species_grid 1,1,1 -dm_refine 1 -number_particles_per_dimension 20 \
1098: -dm_landau_batch_size 4 -number_spatial_vertices 6 -vertex_view_target 5 -grid_view_target 1 -dm_landau_batch_view_idx 1 \
1099: -dm_landau_n 1.000018,1,1e-6 -dm_landau_thermal_temps 2,1,1 -dm_landau_ion_masses 2,180 -dm_landau_ion_charges 1,18 \
1100: -ftop_ksp_rtol 1e-10 -ftop_ksp_type lsqr -ftop_pc_type bjacobi -ftop_sub_pc_factor_shift_type nonzero -ftop_sub_pc_type lu -ftop_ksp_error_if_not_converged \
1101: -ksp_type gmres -ksp_error_if_not_converged -dm_landau_verbose 4 -print_entropy \
1102: -ptof_ksp_type cg -ptof_pc_type jacobi -ptof_ksp_rtol 1e-12 -ptof_ksp_error_if_not_converged\
1103: -snes_converged_reason -snes_monitor -snes_rtol 1e-12 -snes_stol 1e-12 \
1104: -ts_dt 0.01 -ts_rtol 1e-1 -ts_exact_final_time stepover -ts_max_snes_failures -1 -ts_max_steps 1 -ts_monitor -ts_type beuler
1105: test:
1106: suffix: cpu
1107: args: -dm_landau_device_type cpu -pc_type jacobi
1108: test:
1109: suffix: kokkos
1110: # failed on Sunspot@ALCF with sycl
1111: requires: kokkos_kernels !openmp !sycl
1112: args: -dm_landau_device_type kokkos -dm_mat_type aijkokkos -dm_vec_type kokkos -pc_type bjkokkos -pc_bjkokkos_ksp_type tfqmr -pc_bjkokkos_pc_type jacobi
1114: testset:
1115: requires: double !defined(PETSC_USE_DMLANDAU_2D)
1116: output_file: output/ex30_3d.out
1117: args: -dim 3 -petscspace_degree 2 -dm_landau_num_species_grid 1,1 -dm_refine 0 -number_particles_per_dimension 10 -dm_plex_hash_location \
1118: -dm_landau_batch_size 1 -number_spatial_vertices 1 -vertex_view_target 0 -grid_view_target 0 -dm_landau_batch_view_idx 0 \
1119: -dm_landau_n 1.000018,1 -dm_landau_thermal_temps 2,1 -dm_landau_ion_masses 2 -dm_landau_ion_charges 1 \
1120: -ftop_ksp_type cg -ftop_pc_type jacobi -ftop_ksp_rtol 1e-12 -ftop_ksp_error_if_not_converged -ksp_type preonly -pc_type lu -ksp_error_if_not_converged \
1121: -ptof_ksp_type cg -ptof_pc_type jacobi -ptof_ksp_rtol 1e-12 -ptof_ksp_error_if_not_converged \
1122: -snes_converged_reason -snes_monitor -snes_rtol 1e-12 -snes_stol 1e-12 \
1123: -ts_dt 0.1 -ts_exact_final_time stepover -ts_max_snes_failures -1 -ts_max_steps 1 -ts_monitor -ts_type beuler -print_entropy
1124: test:
1125: suffix: cpu_3d
1126: args: -dm_landau_device_type cpu
1127: test:
1128: suffix: kokkos_3d
1129: requires: kokkos_kernels !openmp
1130: args: -dm_landau_device_type kokkos -dm_mat_type aijkokkos -dm_vec_type kokkos -pc_type bjkokkos -pc_bjkokkos_ksp_type tfqmr -pc_bjkokkos_pc_type jacobi
1132: test:
1133: suffix: conserve
1134: requires: !complex double defined(PETSC_USE_DMLANDAU_2D) !cuda
1135: args: -dm_landau_batch_size 4 -dm_refine 0 -dm_landau_num_species_grid 1 -dm_landau_thermal_temps 1 -petscspace_degree 3 -snes_converged_reason -ts_type beuler -ts_dt .1 \
1136: -ts_max_steps 1 -ksp_type preonly -ksp_error_if_not_converged -snes_rtol 1e-14 -snes_stol 1e-14 -dm_landau_device_type cpu -number_particles_per_dimension 20 \
1137: -ptof_ksp_type cg -ptof_pc_type jacobi -ptof_ksp_rtol 1e-14 -ptof_ksp_error_if_not_converged -pc_type lu -dm_landau_simplex 1 -use_uniform_particle_grid false -dm_landau_sphere -print_entropy -number_particles_per_dimension 50 -ftop_ksp_type cg -ftop_pc_type jacobi -ftop_ksp_rtol 1e-14
1139: TEST*/