Nektar++
CoupledLinearNS.cpp
Go to the documentation of this file.
1///////////////////////////////////////////////////////////////////////////////
2//
3// File: CoupledLinearNS.cpp
4//
5// For more information, please see: http://www.nektar.info
6//
7// The MIT License
8//
9// Copyright (c) 2006 Division of Applied Mathematics, Brown University (USA),
10// Department of Aeronautics, Imperial College London (UK), and Scientific
11// Computing and Imaging Institute, University of Utah (USA).
12//
13// Permission is hereby granted, free of charge, to any person obtaining a
14// copy of this software and associated documentation files (the "Software"),
15// to deal in the Software without restriction, including without limitation
16// the rights to use, copy, modify, merge, publish, distribute, sublicense,
17// and/or sell copies of the Software, and to permit persons to whom the
18// Software is furnished to do so, subject to the following conditions:
19//
20// The above copyright notice and this permission notice shall be included
21// in all copies or substantial portions of the Software.
22//
23// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
24// OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
25// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
26// THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
27// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
28// FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
29// DEALINGS IN THE SOFTWARE.
30//
31// Description: Coupled Solver for the Linearised Incompressible
32// Navier Stokes equations
33///////////////////////////////////////////////////////////////////////////////
34
35#include <boost/algorithm/string.hpp>
36
42
43using namespace std;
44
45namespace Nektar
46{
47
50 "CoupledLinearisedNS", CoupledLinearNS::create);
51
54 "SolverType", "CoupledLinearisedNS", eCoupledLinearisedNS);
55
56/**
57 * @class CoupledLinearNS
58 *
59 * Set up expansion field for velocity and pressure, the local to
60 * global mapping arrays and the basic memory definitions for
61 * coupled matrix system
62 */
66 : UnsteadySystem(pSession, pGraph), IncNavierStokes(pSession, pGraph),
67 m_zeroMode(false)
68{
69}
70
71void CoupledLinearNS::v_InitObject(bool DeclareField)
72{
74
75 size_t i;
76 int expdim = m_graph->GetMeshDimension();
77
78 // Get Expansion list for orthogonal expansion at p-2
79 const SpatialDomains::ExpansionInfoMap &pressure_exp =
80 GenPressureExp(m_graph->GetExpansionInfo("u"));
81
83 if (boost::iequals(
84 m_boundaryConditions->GetVariable(m_nConvectiveFields - 1), "p"))
85 {
86 ASSERTL0(false, "Last field is defined as pressure but this is not "
87 "suitable for this solver, please remove this field as "
88 "it is implicitly defined");
89 }
90 // Decide how to declare explist for pressure.
91 if (expdim == 2)
92 {
93 size_t nz;
94
96 {
97 ASSERTL0(m_fields.size() > 2, "Expect to have three at least three "
98 "components of velocity variables");
99 LibUtilities::BasisKey Homo1DKey =
100 m_fields[0]->GetHomogeneousBasis()->GetBasisKey();
101
104 m_homogen_dealiasing, pressure_exp);
105
106 ASSERTL1(m_npointsZ % 2 == 0,
107 "Non binary number of planes have been specified");
108 nz = m_npointsZ / 2;
109 }
110 else
111 {
112 m_pressure =
114 m_session, pressure_exp);
115 nz = 1;
116 }
117
119 for (i = 0; i < m_velocity.size(); ++i)
120 {
122 }
123
124 // Set up Array of mappings
126
127 if (m_singleMode)
128 {
129
130 ASSERTL0(nz <= 2,
131 "For single mode calculation can only have nz <= 2");
132 if (m_session->DefinesSolverInfo("BetaZero"))
133 {
134 m_zeroMode = true;
135 }
136 int nz_loc = 2;
137 m_locToGloMap[0] =
140 m_pressure, nz_loc, m_zeroMode);
141 }
142 else
143 {
144 // base mode
145 int nz_loc = 1;
146 m_locToGloMap[0] =
149 m_pressure, nz_loc);
150
151 if (nz > 1)
152 {
153 nz_loc = 2;
154 // Assume all higher modes have the same boundary conditions and
155 // re-use mapping
156 m_locToGloMap[1] =
160 m_pressure->GetPlane(2), nz_loc, false);
161 // Note high order modes cannot be singular
162 for (i = 2; i < nz; ++i)
163 {
165 }
166 }
167 }
168 }
169 else if (expdim == 3)
170 {
171 // m_pressure =
172 // MemoryManager<MultiRegions::ExpList3D>::AllocateSharedPtr(pressure_exp);
173 ASSERTL0(false, "Setup mapping aray");
174 }
175 else
176 {
177 ASSERTL0(false, "Exp dimension not recognised");
178 }
179
180 // creation of the extrapolation object
182 {
183 std::string vExtrapolation = "Standard";
184
185 if (m_session->DefinesSolverInfo("Extrapolation"))
186 {
187 vExtrapolation = m_session->GetSolverInfo("Extrapolation");
188 }
189
191 vExtrapolation, m_session, m_fields, m_pressure, m_velocity,
193 }
194}
195
196/**
197 * Set up a coupled linearised Naviers-Stokes solve. Primarily a
198 * wrapper fuction around the full mode by mode version of
199 * SetUpCoupledMatrix.
200 *
201 */
203 const NekDouble lambda, const Array<OneD, Array<OneD, NekDouble>> &Advfield,
204 bool IsLinearNSEquation)
205{
206
207 int nz;
208 if (m_singleMode)
209 {
210
211 NekDouble lambda_imag;
212
213 // load imaginary component of any potential shift
214 // Probably should be called from DriverArpack but not yet
215 // clear how to do this
216 m_session->LoadParameter("imagShift", lambda_imag,
218 nz = 1;
220
221 ASSERTL1(m_npointsZ <= 2, "Only expected a maxmimum of two planes in "
222 "single mode linear NS solver");
223
224 if (m_zeroMode)
225 {
226 SetUpCoupledMatrix(lambda, Advfield, IsLinearNSEquation, 0,
227 m_mat[0], m_locToGloMap[0], lambda_imag);
228 }
229 else
230 {
231 NekDouble beta = 2 * M_PI / m_LhomZ;
232 NekDouble lam = lambda + m_kinvis * beta * beta;
233
234 SetUpCoupledMatrix(lam, Advfield, IsLinearNSEquation, 1, m_mat[0],
235 m_locToGloMap[0], lambda_imag);
236 }
237 }
238 else
239 {
240 int n;
241 if (m_npointsZ > 1)
242 {
243 nz = m_npointsZ / 2;
244 }
245 else
246 {
247 nz = 1;
248 }
249
251
252 // mean mode or 2D mode.
253 SetUpCoupledMatrix(lambda, Advfield, IsLinearNSEquation, 0, m_mat[0],
254 m_locToGloMap[0]);
255
256 for (n = 1; n < nz; ++n)
257 {
258 NekDouble beta = 2 * M_PI * n / m_LhomZ;
259
260 NekDouble lam = lambda + m_kinvis * beta * beta;
261
262 SetUpCoupledMatrix(lam, Advfield, IsLinearNSEquation, n, m_mat[n],
263 m_locToGloMap[n]);
264 }
265 }
266}
267
268/**
269 * Set up a coupled linearised Naviers-Stokes solve in the
270 * following manner:
271 *
272 * Consider the linearised Navier-Stokes element matrix denoted as
273 *
274 * \f[ \left [ \begin{array}{ccc} A
275 * & -D_{bnd}^T & B\\ -D_{bnd}& 0 & -D_{int}\\ \tilde{B}^T &
276 * -D_{int}^T & C \end{array} \right ] \left [ \begin{array}{c} {\bf
277 * v}_{bnd} \\ p\\ {\bf v}_{int} \end{array} \right ] = \left [
278 * \begin{array}{c} {\bf f}_{bnd} \\ 0\\ {\bf f}_{int} \end{array}
279 * \right ] \f]
280 *
281 * where \f${\bf v}_{bnd}, {\bf f}_{bnd}\f$ denote the degrees of
282 * freedom of the elemental velocities on the boundary of the
283 * element, \f${\bf v}_{int}, {\bf f}_{int}\f$ denote the degrees
284 * of freedom of the elemental velocities on the interior of the
285 * element and \f$p\f$ is the piecewise continuous pressure. The
286 * matrices have the interpretation
287 *
288 * \f[ A[n,m] = (\nabla \phi^b_n, \nu \nabla
289 * \phi^b_m) + (\phi^b_n,{\bf U \cdot \nabla} \phi^b_m) +
290 * (\phi^b_n \nabla^T {\bf U} \phi^b_m), \f]
291 *
292 * \f[ B[n,m] = (\nabla \phi^b_n, \nu \nabla \phi^i_m) +
293 * (\phi^b_n,{\bf U \cdot \nabla} \phi^i_m) + (\phi^b_n \nabla^T
294 * {\bf U} \phi^i_m),\f]
295 *
296 * \f[\tilde{B}^T[n,m] = (\nabla \phi^i_n, \nu \nabla \phi^b_m) +
297 * (\phi^i_n,{\bf U \cdot \nabla} \phi^b_m) + (\phi^i_n \nabla^T
298 * {\bf U} \phi^b_m) \f]
299 *
300 * \f[ C[n,m] = (\nabla \phi^i_n, \nu \nabla
301 * \phi^i_m) + (\phi^i_n,{\bf U \cdot \nabla} \phi^i_m) +
302 * (\phi^i_n \nabla^T {\bf U} \phi^i_m),\f]
303 *
304 * \f[ D_{bnd}[n,m] = (\psi_m,\nabla \phi^b),\f]
305 *
306 * \f[ D_{int}[n,m] = (\psi_m,\nabla \phi^i) \f]
307 *
308 * where \f$\psi\f$ is the space of pressures typically at order
309 * \f$P-2\f$ and \f$\phi\f$ is the velocity vector space of
310 * polynomial order \f$P\f$. (Note the above definitions for the
311 * transpose terms shoudl be interpreted with care since we have
312 * used a tensor differential for the \f$\nabla^T \f$ terms)
313 *
314 * Note \f$B = \tilde{B}^T\f$ if just considering the stokes
315 * operator and then \f$C\f$ is also symmetric. Also note that
316 * \f$A,B\f$ and \f$C\f$ are block diagonal in the Oseen equation
317 * when \f$\nabla^T {\bf U}\f$ are zero.
318 *
319 * Since \f$C\f$ is invertible we can premultiply the governing
320 * elemental equation by the following matrix:
321 *
322 * \f[ \left [ \begin{array}{ccc} I & 0 &
323 * -BC^{-1}\\ 0& I & D_{int}C^{-1}\\ 0 & 0 & I \end{array}
324 * \right ] \left \{ \left [ \begin{array}{ccc} A & -D_{bnd}^T &
325 * B\\ -D_{bnd}& 0 & -D_{int}\\ \tilde{B}^T & -D_{int}^T & C
326 * \end{array} \right ] \left [ \begin{array}{c} {\bf v}_{bnd} \\
327 * p\\ {\bf v_{int}} \end{array} \right ] = \left [
328 * \begin{array}{c} {\bf f}_{bnd} \\ 0\\ {\bf f_{int}} \end{array}
329 * \right ] \right \} \f]
330 *
331 * which if we multiply out the matrix equation we obtain
332 *
333 * \f[ \left [ \begin{array}{ccc} A - B
334 * C^{-1}\tilde{B}^T & -D_{bnd}^T +B C^{-1} D_{int}^T& 0\\
335 * -D_{bnd}+D_{int} C^{-1} \tilde{B}^T & -D_{int} C^{-1}
336 * D_{int}^T & 0\\ \tilde{B}^T & -D_{int}^T & C \end{array} \right ]
337 * \left [ \begin{array}{c} {\bf v}_{bnd} \\ p\\ {\bf v_{int}}
338 * \end{array} \right ] = \left [ \begin{array}{c} {\bf f}_{bnd}
339 * -B C^{-1} {\bf f}_{int}\\ f_p = D_{int} C^{-1} {\bf
340 * f}_{int}\\ {\bf f_{int}} \end{array} \right ] \f]
341 *
342 * In the above equation the \f${\bf v}_{int}\f$ degrees of freedom
343 * are decoupled and so we need to solve for the \f${\bf v}_{bnd},
344 * p\f$ degrees of freedom. The final step is to perform a second
345 * level of static condensation but where we will lump the mean
346 * pressure mode (or a pressure degree of freedom containing a
347 * mean component) with the velocity boundary degrees of
348 * freedom. To do we define \f${\bf b} = [{\bf v}_{bnd}, p_0]\f$ where
349 * \f$p_0\f$ is the mean pressure mode and \f$\hat{p}\f$ to be the
350 * remainder of the pressure space. We now repartition the top \f$2
351 * \times 2\f$ block of matrices of previous matrix equation as
352 *
353 * \f[ \left [ \begin{array}{cc} \hat{A} & \hat{B}\\ \hat{C} &
354 * \hat{D} \end{array} \right ] \left [ \begin{array}{c} {\bf b}
355 * \\ \hat{p} \end{array} \right ] = \left [ \begin{array}{c}
356 * \hat{\bf f}_{bnd} \\ \hat{f}_p \end{array} \right ]
357 * \label{eqn.linNS_fac2} \f]
358 *
359 * where
360 *
361 * \f[ \hat{A}_{ij} = \left [ \begin{array}{cc} A - B
362 * C^{-1}\tilde{B}^T & [-D_{bnd}^T +B C^{-1} D_{int}^T]_{i0}\\
363 * {[}-D_{bnd}+D_{int} C^{-1} \tilde{B}^T]_{0j} & -[D_{int}
364 * C^{-1} D_{int}^T ]_{00} \end{array} \right ] \f]
365 *
366 * \f[ \hat{B}_{ij} = \left [ \begin{array}{c} [-D_{bnd}^T +B
367 * C^{-1} D_{int}^T]_{i,j+1} \\ {[} -D_{int} C^{-1} D^T_{int}
368 * ]_{0j}\end{array} \right ] \f]
369 *
370 * \f[ \hat{C}_{ij} = \left [\begin{array}{cc} -D_{bnd} +
371 * D_{int} C^{-1} \tilde{B}^T, & {[} -D_{int} C^{-1} D^T_{int}
372 * ]_{i+1,0}\end{array} \right ] \f]
373 *
374 * \f[ \hat{D}_{ij} = \left [\begin{array}{c} {[} -D_{int}
375 * C^{-1} D^T_{int} ]_{i+1,j+1}\end{array} \right ] \f]
376 *
377 * and
378 *
379 * \f[ fh\_{bnd} = \left [ \begin{array}{c} {\bf
380 * f}_{bnd} -B C^{-1} {\bf f}_{int}\\ {[}D_{int} C^{-1} {\bf
381 * f}_{int}]_0 \end{array}\right ] \hspace{1cm} [fh\_p_{i} =
382 * \left [ \begin{array}{c} {[}D_{int} C^{-1} {\bf f}_{iint}]_{i+1}
383 * \end{array}\right ] \f]
384 *
385 * Since the \f$\hat{D}\f$ is decoupled and invertible we can now
386 * statically condense the previous matrix equationto decouple
387 * \f${\bf b}\f$ from \f$\hat{p}\f$ by solving the following system
388 *
389 * \f[ \left [ \begin{array}{cc} \hat{A} - \hat{B} \hat{D}^{-1}
390 * \hat{C} & 0 \\ \hat{C} & \hat{D} \end{array} \right ] \left
391 * [ \begin{array}{c} {\bf b} \\ \hat{p} \end{array} \right ] =
392 * \left [ \begin{array}{c} \hat{\bf f}_{bnd} - \hat{B}
393 * \hat{D}^{-1} \hat{f}_p\\ \hat{f}_p \end{array} \right ] \f]
394 *
395 * The matrix \f$\hat{A} - \hat{B} \hat{D}^{-1} \hat{C}\f$ has
396 * to be globally assembled and solved iteratively or
397 * directly. One we obtain the solution to \f${\bf b}\f$ we can use
398 * the second row of equation fourth matrix equation to solve for
399 * \f$\hat{p}\f$ and finally the last row of equation second
400 * matrix equation to solve for \f${\bf v}_{int}\f$.
401 *
402 */
403
405 const NekDouble lambda, const Array<OneD, Array<OneD, NekDouble>> &Advfield,
406 bool IsLinearNSEquation, const int HomogeneousMode,
409 const NekDouble lambda_imag)
410{
411 size_t n, i, j, k;
412 size_t nel = m_fields[m_velocity[0]]->GetNumElmts();
413 size_t nvel = m_velocity.size();
414
415 // if Advfield is defined can assume it is an Oseen or LinearNS equation
416 bool AddAdvectionTerms =
417 (Advfield == NullNekDoubleArrayOfArray) ? false : true;
418 // bool AddAdvectionTerms = true; // Temporary debugging trip
419
420 // call velocity space Static condensation and fill block
421 // matrices. Need to set this up differently for Oseen and
422 // Lin NS. Ideally should make block diagonal for Stokes and
423 // Oseen problems.
424 DNekScalMatSharedPtr loc_mat;
426 NekDouble one = 1.0;
427 size_t nint, nbndry;
428 size_t rows, cols;
429 NekDouble zero = 0.0;
430 Array<OneD, unsigned int> bmap, imap;
431
432 Array<OneD, unsigned int> nsize_bndry(nel);
433 Array<OneD, unsigned int> nsize_bndry_p1(nel);
434 Array<OneD, unsigned int> nsize_int(nel);
435 Array<OneD, unsigned int> nsize_p(nel);
436 Array<OneD, unsigned int> nsize_p_m1(nel);
437
438 size_t nz_loc;
439
440 if (HomogeneousMode) // Homogeneous mode flag
441 {
442 nz_loc = 2;
443 }
444 else
445 {
446 if (m_singleMode)
447 {
448 nz_loc = 2;
449 }
450 else
451 {
452 nz_loc = 1;
453 }
454 }
455
456 // Set up block matrix sizes -
457 for (n = 0; n < nel; ++n)
458 {
459 nsize_bndry[n] = nvel *
460 m_fields[m_velocity[0]]->GetExp(n)->NumBndryCoeffs() *
461 nz_loc;
462 nsize_bndry_p1[n] = nsize_bndry[n] + nz_loc;
463 nsize_int[n] =
464 (nvel * m_fields[m_velocity[0]]->GetExp(n)->GetNcoeffs() * nz_loc -
465 nsize_bndry[n]);
466 nsize_p[n] = m_pressure->GetExp(n)->GetNcoeffs() * nz_loc;
467 nsize_p_m1[n] = nsize_p[n] - nz_loc;
468 }
469
470 MatrixStorage blkmatStorage = eDIAGONAL;
473 nsize_bndry_p1, nsize_bndry_p1, blkmatStorage);
475 nsize_bndry, nsize_int, blkmatStorage);
477 nsize_bndry, nsize_int, blkmatStorage);
479 nsize_int, nsize_int, blkmatStorage);
480
482 nsize_p, nsize_bndry, blkmatStorage);
483
485 nsize_p, nsize_int, blkmatStorage);
486
487 // Final level static condensation matrices.
490 nsize_bndry_p1, nsize_p_m1, blkmatStorage);
493 nsize_p_m1, nsize_bndry_p1, blkmatStorage);
496 blkmatStorage);
497
499 timer.Start();
500
501 for (n = 0; n < nel; ++n)
502 {
503 nbndry = nsize_bndry[n];
504 nint = nsize_int[n];
505 k = nsize_bndry_p1[n];
508 Array<OneD, NekDouble> Ah_data = Ah->GetPtr();
509 int AhRows = k;
512 Array<OneD, NekDouble> B_data = B->GetPtr();
515 Array<OneD, NekDouble> C_data = C->GetPtr();
518 Array<OneD, NekDouble> D_data = D->GetPtr();
519
521 nsize_p[n], nsize_bndry[n], zero);
523 nsize_p[n], nsize_int[n], zero);
524
525 locExp = m_fields[m_velocity[0]]->GetExp(n);
526 locExp->GetBoundaryMap(bmap);
527 locExp->GetInteriorMap(imap);
531 StdRegions::eHelmholtz, locExp->DetShapeType(), *locExp, factors);
532
533 size_t ncoeffs = m_fields[m_velocity[0]]->GetExp(n)->GetNcoeffs();
534 size_t nphys = m_fields[m_velocity[0]]->GetExp(n)->GetTotPoints();
535 size_t nbmap = bmap.size();
536 size_t nimap = imap.size();
537
538 Array<OneD, NekDouble> coeffs(ncoeffs);
539 Array<OneD, NekDouble> phys(nphys);
540 size_t psize = m_pressure->GetExp(n)->GetNcoeffs();
541 size_t pqsize = m_pressure->GetExp(n)->GetTotPoints();
542
543 Array<OneD, NekDouble> deriv(pqsize);
544 Array<OneD, NekDouble> pcoeffs(psize);
545
546 if (AddAdvectionTerms == false) // use static condensed managed matrices
547 {
548 // construct velocity matrices using statically
549 // condensed elemental matrices and then construct
550 // pressure matrix systems
552 CondMat = locExp->GetLocStaticCondMatrix(helmkey);
553
554 for (k = 0; k < nvel * nz_loc; ++k)
555 {
556 DNekScalMat &SubBlock = *CondMat->GetBlock(0, 0);
557 rows = SubBlock.GetRows();
558 cols = SubBlock.GetColumns();
559 for (i = 0; i < rows; ++i)
560 {
561 for (j = 0; j < cols; ++j)
562 {
563 (*Ah)(i + k * rows, j + k * cols) =
564 m_kinvis * SubBlock(i, j);
565 }
566 }
567 }
568
569 for (k = 0; k < nvel * nz_loc; ++k)
570 {
571 DNekScalMat &SubBlock = *CondMat->GetBlock(0, 1);
572 DNekScalMat &SubBlock1 = *CondMat->GetBlock(1, 0);
573 rows = SubBlock.GetRows();
574 cols = SubBlock.GetColumns();
575 for (i = 0; i < rows; ++i)
576 {
577 for (j = 0; j < cols; ++j)
578 {
579 (*B)(i + k * rows, j + k * cols) = SubBlock(i, j);
580 (*C)(i + k * rows, j + k * cols) =
581 m_kinvis * SubBlock1(j, i);
582 }
583 }
584 }
585
586 for (k = 0; k < nvel * nz_loc; ++k)
587 {
588 DNekScalMat &SubBlock = *CondMat->GetBlock(1, 1);
589 NekDouble inv_kinvis = 1.0 / m_kinvis;
590 rows = SubBlock.GetRows();
591 cols = SubBlock.GetColumns();
592 for (i = 0; i < rows; ++i)
593 {
594 for (j = 0; j < cols; ++j)
595 {
596 (*D)(i + k * rows, j + k * cols) =
597 inv_kinvis * SubBlock(i, j);
598 }
599 }
600 }
601
602 // Loop over pressure space and construct boundary block matrices.
603 for (i = 0; i < bmap.size(); ++i)
604 {
605 // Fill element with mode
606 Vmath::Zero(ncoeffs, coeffs, 1);
607 coeffs[bmap[i]] = 1.0;
608 m_fields[m_velocity[0]]->GetExp(n)->BwdTrans(coeffs, phys);
609
610 // Differentiation & Inner product wrt base.
611 for (j = 0; j < nvel; ++j)
612 {
613 if ((nz_loc == 2) && (j == 2)) // handle d/dz derivative
614 {
615 NekDouble beta = -2 * M_PI * HomogeneousMode / m_LhomZ;
616
618 m_fields[m_velocity[0]]->GetExp(n)->GetTotPoints(),
619 beta, phys, 1, deriv, 1);
620
621 m_pressure->GetExp(n)->IProductWRTBase(deriv, pcoeffs);
622
623 Blas::Dcopy(psize, &(pcoeffs)[0], 1,
624 Dbnd->GetRawPtr() +
625 ((nz_loc * j + 1) * bmap.size() + i) *
626 nsize_p[n],
627 1);
628
629 Vmath::Neg(psize, pcoeffs, 1);
630 Blas::Dcopy(psize, &(pcoeffs)[0], 1,
631 Dbnd->GetRawPtr() +
632 ((nz_loc * j) * bmap.size() + i) *
633 nsize_p[n] +
634 psize,
635 1);
636 }
637 else
638 {
639 if (j < 2) // required for mean mode of homogeneous
640 // expansion
641 {
642 locExp->PhysDeriv(MultiRegions::DirCartesianMap[j],
643 phys, deriv);
644 m_pressure->GetExp(n)->IProductWRTBase(deriv,
645 pcoeffs);
646 // copy into column major storage.
647 for (k = 0; k < nz_loc; ++k)
648 {
650 psize, &(pcoeffs)[0], 1,
651 Dbnd->GetRawPtr() +
652 ((nz_loc * j + k) * bmap.size() + i) *
653 nsize_p[n] +
654 k * psize,
655 1);
656 }
657 }
658 }
659 }
660 }
661
662 for (i = 0; i < imap.size(); ++i)
663 {
664 // Fill element with mode
665 Vmath::Zero(ncoeffs, coeffs, 1);
666 coeffs[imap[i]] = 1.0;
667 m_fields[m_velocity[0]]->GetExp(n)->BwdTrans(coeffs, phys);
668
669 // Differentiation & Inner product wrt base.
670 for (j = 0; j < nvel; ++j)
671 {
672 if ((nz_loc == 2) && (j == 2)) // handle d/dz derivative
673 {
674 NekDouble beta = -2 * M_PI * HomogeneousMode / m_LhomZ;
675
677 m_fields[m_velocity[0]]->GetExp(n)->GetTotPoints(),
678 beta, phys, 1, deriv, 1);
679
680 m_pressure->GetExp(n)->IProductWRTBase(deriv, pcoeffs);
681
682 Blas::Dcopy(psize, &(pcoeffs)[0], 1,
683 Dint->GetRawPtr() +
684 ((nz_loc * j + 1) * imap.size() + i) *
685 nsize_p[n],
686 1);
687
688 Vmath::Neg(psize, pcoeffs, 1);
689 Blas::Dcopy(psize, &(pcoeffs)[0], 1,
690 Dint->GetRawPtr() +
691 ((nz_loc * j) * imap.size() + i) *
692 nsize_p[n] +
693 psize,
694 1);
695 }
696 else
697 {
698 if (j < 2) // required for mean mode of homogeneous
699 // expansion
700 {
701 // m_fields[m_velocity[0]]->GetExp(n)->PhysDeriv(j,phys,
702 // deriv);
703 locExp->PhysDeriv(MultiRegions::DirCartesianMap[j],
704 phys, deriv);
705
706 m_pressure->GetExp(n)->IProductWRTBase(deriv,
707 pcoeffs);
708
709 // copy into column major storage.
710 for (k = 0; k < nz_loc; ++k)
711 {
713 psize, &(pcoeffs)[0], 1,
714 Dint->GetRawPtr() +
715 ((nz_loc * j + k) * imap.size() + i) *
716 nsize_p[n] +
717 k * psize,
718 1);
719 }
720 }
721 }
722 }
723 }
724 }
725 else
726 {
727 // construct velocity matrices and pressure systems at
728 // the same time resusing differential of velocity
729 // space
730
731 DNekScalMat &HelmMat =
732 *(locExp->as<LocalRegions::Expansion>()->GetLocMatrix(helmkey));
733 DNekScalMatSharedPtr MassMat;
734
735 Array<OneD, const NekDouble> HelmMat_data =
736 HelmMat.GetOwnedMatrix()->GetPtr();
737 NekDouble HelmMatScale = HelmMat.Scale();
738 int HelmMatRows = HelmMat.GetRows();
739
740 if ((lambda_imag != NekConstants::kNekUnsetDouble) && (nz_loc == 2))
741 {
743 StdRegions::eMass, locExp->DetShapeType(), *locExp);
744 MassMat = locExp->as<LocalRegions::Expansion>()->GetLocMatrix(
745 masskey);
746 }
747
749 Array<OneD, Array<OneD, NekDouble>> AdvDeriv(nvel * nvel);
750 // Use ExpList phys array for temporaary storage
751 Array<OneD, NekDouble> tmpphys = m_fields[0]->UpdatePhys();
752 int phys_offset = m_fields[m_velocity[0]]->GetPhys_Offset(n);
753 size_t nv;
754 int npoints = locExp->GetTotPoints();
755
756 // Calculate derivative of base flow
757 if (IsLinearNSEquation)
758 {
759 size_t nv1;
760 size_t cnt = 0;
761 AdvDeriv[0] = Array<OneD, NekDouble>(nvel * nvel * npoints);
762 for (nv = 0; nv < nvel; ++nv)
763 {
764 for (nv1 = 0; nv1 < nvel; ++nv1)
765 {
766 if (cnt < nvel * nvel - 1)
767 {
768 AdvDeriv[cnt + 1] = AdvDeriv[cnt] + npoints;
769 ++cnt;
770 }
771
772 if ((nv1 == 2) && (m_HomogeneousType == eHomogeneous1D))
773 {
774 Vmath::Zero(npoints, AdvDeriv[nv * nvel + nv1],
775 1); // dU/dz = 0
776 }
777 else
778 {
779 locExp->PhysDeriv(
781 Advfield[nv] + phys_offset,
782 AdvDeriv[nv * nvel + nv1]);
783 }
784 }
785 }
786 }
787
788 for (i = 0; i < nbmap; ++i)
789 {
790
791 // Fill element with mode
792 Vmath::Zero(ncoeffs, coeffs, 1);
793 coeffs[bmap[i]] = 1.0;
794 locExp->BwdTrans(coeffs, phys);
795
796 for (k = 0; k < nvel * nz_loc; ++k)
797 {
798 for (j = 0; j < nbmap; ++j)
799 {
800 // Ah_data[i+k*nbmap +
801 // (j+k*nbmap)*AhRows] +=
802 // m_kinvis*HelmMat(bmap[i],bmap[j]);
803 Ah_data[i + k * nbmap + (j + k * nbmap) * AhRows] +=
804 m_kinvis * HelmMatScale *
805 HelmMat_data[bmap[i] + HelmMatRows * bmap[j]];
806 }
807
808 for (j = 0; j < nimap; ++j)
809 {
810 B_data[i + k * nbmap + (j + k * nimap) * nbndry] +=
811 m_kinvis * HelmMatScale *
812 HelmMat_data[bmap[i] + HelmMatRows * imap[j]];
813 }
814 }
815
816 if ((lambda_imag != NekConstants::kNekUnsetDouble) &&
817 (nz_loc == 2))
818 {
819 for (k = 0; k < nvel; ++k)
820 {
821 for (j = 0; j < nbmap; ++j)
822 {
823 Ah_data[i + 2 * k * nbmap +
824 (j + (2 * k + 1) * nbmap) * AhRows] -=
825 lambda_imag * (*MassMat)(bmap[i], bmap[j]);
826 }
827
828 for (j = 0; j < nbmap; ++j)
829 {
830 Ah_data[i + (2 * k + 1) * nbmap +
831 (j + 2 * k * nbmap) * AhRows] +=
832 lambda_imag * (*MassMat)(bmap[i], bmap[j]);
833 }
834
835 for (j = 0; j < nimap; ++j)
836 {
837 B_data[i + 2 * k * nbmap +
838 (j + (2 * k + 1) * nimap) * nbndry] -=
839 lambda_imag * (*MassMat)(bmap[i], imap[j]);
840 }
841
842 for (j = 0; j < nimap; ++j)
843 {
844 B_data[i + (2 * k + 1) * nbmap +
845 (j + 2 * k * nimap) * nbndry] +=
846 lambda_imag * (*MassMat)(bmap[i], imap[j]);
847 }
848 }
849 }
850
851 for (k = 0; k < nvel; ++k)
852 {
853 if ((nz_loc == 2) && (k == 2)) // handle d/dz derivative
854 {
855 NekDouble beta = -2 * M_PI * HomogeneousMode / m_LhomZ;
856
857 // Real Component
858 Vmath::Smul(npoints, beta, phys, 1, deriv, 1);
859
860 m_pressure->GetExp(n)->IProductWRTBase(deriv, pcoeffs);
861 Blas::Dcopy(psize, &(pcoeffs)[0], 1,
862 Dbnd->GetRawPtr() +
863 ((nz_loc * k + 1) * bmap.size() + i) *
864 nsize_p[n],
865 1);
866
867 // Imaginary Component
868 Vmath::Neg(psize, pcoeffs, 1);
869 Blas::Dcopy(psize, &(pcoeffs)[0], 1,
870 Dbnd->GetRawPtr() +
871 ((nz_loc * k) * bmap.size() + i) *
872 nsize_p[n] +
873 psize,
874 1);
875
876 // now do advection terms
877 Vmath::Vmul(npoints, Advtmp = Advfield[k] + phys_offset,
878 1, deriv, 1, tmpphys, 1);
879
880 locExp->IProductWRTBase(tmpphys, coeffs);
881
882 // real contribution
883 for (nv = 0; nv < nvel; ++nv)
884 {
885 for (j = 0; j < nbmap; ++j)
886 {
887 Ah_data[j + 2 * nv * nbmap +
888 (i + (2 * nv + 1) * nbmap) * AhRows] +=
889 coeffs[bmap[j]];
890 }
891
892 for (j = 0; j < nimap; ++j)
893 {
894 C_data[i + (2 * nv + 1) * nbmap +
895 (j + 2 * nv * nimap) * nbndry] +=
896 coeffs[imap[j]];
897 }
898 }
899
900 Vmath::Neg(ncoeffs, coeffs, 1);
901 // imaginary contribution
902 for (nv = 0; nv < nvel; ++nv)
903 {
904 for (j = 0; j < nbmap; ++j)
905 {
906 Ah_data[j + (2 * nv + 1) * nbmap +
907 (i + 2 * nv * nbmap) * AhRows] +=
908 coeffs[bmap[j]];
909 }
910
911 for (j = 0; j < nimap; ++j)
912 {
913 C_data[i + 2 * nv * nbmap +
914 (j + (2 * nv + 1) * nimap) * nbndry] +=
915 coeffs[imap[j]];
916 }
917 }
918 }
919 else
920 {
921 if (k < 2)
922 {
923 locExp->PhysDeriv(MultiRegions::DirCartesianMap[k],
924 phys, deriv);
925 Vmath::Vmul(npoints,
926 Advtmp = Advfield[k] + phys_offset, 1,
927 deriv, 1, tmpphys, 1);
928 locExp->IProductWRTBase(tmpphys, coeffs);
929
930 for (nv = 0; nv < nvel * nz_loc; ++nv)
931 {
932 for (j = 0; j < nbmap; ++j)
933 {
934 Ah_data[j + nv * nbmap +
935 (i + nv * nbmap) * AhRows] +=
936 coeffs[bmap[j]];
937 }
938
939 for (j = 0; j < nimap; ++j)
940 {
941 C_data[i + nv * nbmap +
942 (j + nv * nimap) * nbndry] +=
943 coeffs[imap[j]];
944 }
945 }
946
947 // copy into column major storage.
948 m_pressure->GetExp(n)->IProductWRTBase(deriv,
949 pcoeffs);
950 for (j = 0; j < nz_loc; ++j)
951 {
953 psize, &(pcoeffs)[0], 1,
954 Dbnd->GetRawPtr() +
955 ((nz_loc * k + j) * bmap.size() + i) *
956 nsize_p[n] +
957 j * psize,
958 1);
959 }
960 }
961 }
962
963 if (IsLinearNSEquation)
964 {
965 for (nv = 0; nv < nvel; ++nv)
966 {
967 // u' . Grad U terms
968 Vmath::Vmul(npoints, phys, 1,
969 AdvDeriv[k * nvel + nv], 1, tmpphys, 1);
970 locExp->IProductWRTBase(tmpphys, coeffs);
971
972 for (size_t n1 = 0; n1 < nz_loc; ++n1)
973 {
974 for (j = 0; j < nbmap; ++j)
975 {
976 Ah_data[j + (k * nz_loc + n1) * nbmap +
977 (i + (nv * nz_loc + n1) * nbmap) *
978 AhRows] += coeffs[bmap[j]];
979 }
980
981 for (j = 0; j < nimap; ++j)
982 {
983 C_data[i + (nv * nz_loc + n1) * nbmap +
984 (j + (k * nz_loc + n1) * nimap) *
985 nbndry] += coeffs[imap[j]];
986 }
987 }
988 }
989 }
990 }
991 }
992
993 for (i = 0; i < nimap; ++i)
994 {
995 // Fill element with mode
996 Vmath::Zero(ncoeffs, coeffs, 1);
997 coeffs[imap[i]] = 1.0;
998 locExp->BwdTrans(coeffs, phys);
999
1000 for (k = 0; k < nvel * nz_loc; ++k)
1001 {
1002 for (j = 0; j < nbmap; ++j) // C set up as transpose
1003 {
1004 C_data[j + k * nbmap + (i + k * nimap) * nbndry] +=
1005 m_kinvis * HelmMatScale *
1006 HelmMat_data[imap[i] + HelmMatRows * bmap[j]];
1007 }
1008
1009 for (j = 0; j < nimap; ++j)
1010 {
1011 D_data[i + k * nimap + (j + k * nimap) * nint] +=
1012 m_kinvis * HelmMatScale *
1013 HelmMat_data[imap[i] + HelmMatRows * imap[j]];
1014 }
1015 }
1016
1017 if ((lambda_imag != NekConstants::kNekUnsetDouble) &&
1018 (nz_loc == 2))
1019 {
1020 for (k = 0; k < nvel; ++k)
1021 {
1022 for (j = 0; j < nbmap; ++j) // C set up as transpose
1023 {
1024 C_data[j + 2 * k * nbmap +
1025 (i + (2 * k + 1) * nimap) * nbndry] +=
1026 lambda_imag * (*MassMat)(bmap[j], imap[i]);
1027 }
1028
1029 for (j = 0; j < nbmap; ++j) // C set up as transpose
1030 {
1031 C_data[j + (2 * k + 1) * nbmap +
1032 (i + 2 * k * nimap) * nbndry] -=
1033 lambda_imag * (*MassMat)(bmap[j], imap[i]);
1034 }
1035
1036 for (j = 0; j < nimap; ++j)
1037 {
1038 D_data[i + 2 * k * nimap +
1039 (j + (2 * k + 1) * nimap) * nint] -=
1040 lambda_imag * (*MassMat)(imap[i], imap[j]);
1041 }
1042
1043 for (j = 0; j < nimap; ++j)
1044 {
1045 D_data[i + (2 * k + 1) * nimap +
1046 (j + 2 * k * nimap) * nint] +=
1047 lambda_imag * (*MassMat)(imap[i], imap[j]);
1048 }
1049 }
1050 }
1051
1052 for (k = 0; k < nvel; ++k)
1053 {
1054 if ((nz_loc == 2) && (k == 2)) // handle d/dz derivative
1055 {
1056 NekDouble beta = -2 * M_PI * HomogeneousMode / m_LhomZ;
1057
1058 // Real Component
1059 Vmath::Smul(npoints, beta, phys, 1, deriv, 1);
1060 m_pressure->GetExp(n)->IProductWRTBase(deriv, pcoeffs);
1061 Blas::Dcopy(psize, &(pcoeffs)[0], 1,
1062 Dint->GetRawPtr() +
1063 ((nz_loc * k + 1) * imap.size() + i) *
1064 nsize_p[n],
1065 1);
1066 // Imaginary Component
1067 Vmath::Neg(psize, pcoeffs, 1);
1068 Blas::Dcopy(psize, &(pcoeffs)[0], 1,
1069 Dint->GetRawPtr() +
1070 ((nz_loc * k) * imap.size() + i) *
1071 nsize_p[n] +
1072 psize,
1073 1);
1074
1075 // Advfield[k] *d/dx_k to all velocity
1076 // components on diagonal
1077 Vmath::Vmul(npoints, Advtmp = Advfield[k] + phys_offset,
1078 1, deriv, 1, tmpphys, 1);
1079 locExp->IProductWRTBase(tmpphys, coeffs);
1080
1081 // Real Components
1082 for (nv = 0; nv < nvel; ++nv)
1083 {
1084 for (j = 0; j < nbmap; ++j)
1085 {
1086 B_data[j + 2 * nv * nbmap +
1087 (i + (2 * nv + 1) * nimap) * nbndry] +=
1088 coeffs[bmap[j]];
1089 }
1090
1091 for (j = 0; j < nimap; ++j)
1092 {
1093 D_data[j + 2 * nv * nimap +
1094 (i + (2 * nv + 1) * nimap) * nint] +=
1095 coeffs[imap[j]];
1096 }
1097 }
1098 Vmath::Neg(ncoeffs, coeffs, 1);
1099 // Imaginary
1100 for (nv = 0; nv < nvel; ++nv)
1101 {
1102 for (j = 0; j < nbmap; ++j)
1103 {
1104 B_data[j + (2 * nv + 1) * nbmap +
1105 (i + 2 * nv * nimap) * nbndry] +=
1106 coeffs[bmap[j]];
1107 }
1108
1109 for (j = 0; j < nimap; ++j)
1110 {
1111 D_data[j + (2 * nv + 1) * nimap +
1112 (i + 2 * nv * nimap) * nint] +=
1113 coeffs[imap[j]];
1114 }
1115 }
1116 }
1117 else
1118 {
1119 if (k < 2)
1120 {
1121 // Differentiation & Inner product wrt base.
1122 // locExp->PhysDeriv(k,phys, deriv);
1123 locExp->PhysDeriv(MultiRegions::DirCartesianMap[k],
1124 phys, deriv);
1125 Vmath::Vmul(npoints,
1126 Advtmp = Advfield[k] + phys_offset, 1,
1127 deriv, 1, tmpphys, 1);
1128 locExp->IProductWRTBase(tmpphys, coeffs);
1129
1130 for (nv = 0; nv < nvel * nz_loc; ++nv)
1131 {
1132 for (j = 0; j < nbmap; ++j)
1133 {
1134 B_data[j + nv * nbmap +
1135 (i + nv * nimap) * nbndry] +=
1136 coeffs[bmap[j]];
1137 }
1138
1139 for (j = 0; j < nimap; ++j)
1140 {
1141 D_data[j + nv * nimap +
1142 (i + nv * nimap) * nint] +=
1143 coeffs[imap[j]];
1144 }
1145 }
1146 // copy into column major storage.
1147 m_pressure->GetExp(n)->IProductWRTBase(deriv,
1148 pcoeffs);
1149 for (j = 0; j < nz_loc; ++j)
1150 {
1152 psize, &(pcoeffs)[0], 1,
1153 Dint->GetRawPtr() +
1154 ((nz_loc * k + j) * imap.size() + i) *
1155 nsize_p[n] +
1156 j * psize,
1157 1);
1158 }
1159 }
1160 }
1161
1162 if (IsLinearNSEquation)
1163 {
1164 size_t n1;
1165 for (nv = 0; nv < nvel; ++nv)
1166 {
1167 // u'.Grad U terms
1168 Vmath::Vmul(npoints, phys, 1,
1169 AdvDeriv[k * nvel + nv], 1, tmpphys, 1);
1170 locExp->IProductWRTBase(tmpphys, coeffs);
1171
1172 for (n1 = 0; n1 < nz_loc; ++n1)
1173 {
1174 for (j = 0; j < nbmap; ++j)
1175 {
1176 B_data[j + (k * nz_loc + n1) * nbmap +
1177 (i + (nv * nz_loc + n1) * nimap) *
1178 nbndry] += coeffs[bmap[j]];
1179 }
1180
1181 for (j = 0; j < nimap; ++j)
1182 {
1183 D_data[j + (k * nz_loc + n1) * nimap +
1184 (i + (nv * nz_loc + n1) * nimap) *
1185 nint] += coeffs[imap[j]];
1186 }
1187 }
1188 }
1189 }
1190 }
1191 }
1192
1193 D->Invert();
1194 (*B) = (*B) * (*D);
1195
1196 // perform (*Ah) = (*Ah) - (*B)*(*C) but since size of
1197 // Ah is larger than (*B)*(*C) easier to call blas
1198 // directly
1199 Blas::Dgemm('N', 'T', B->GetRows(), C->GetRows(), B->GetColumns(),
1200 -1.0, B->GetRawPtr(), B->GetRows(), C->GetRawPtr(),
1201 C->GetRows(), 1.0, Ah->GetRawPtr(), Ah->GetRows());
1202 }
1203
1204 mat.m_BCinv->SetBlock(
1205 n, n,
1207 mat.m_Btilde->SetBlock(
1208 n, n,
1210 mat.m_Cinv->SetBlock(
1211 n, n,
1213 mat.m_D_bnd->SetBlock(
1214 n, n,
1216 mat.m_D_int->SetBlock(
1217 n, n,
1219
1220 // Do matrix manipulations and get final set of block matries
1221 // reset boundary to put mean mode into boundary system.
1222
1223 DNekMatSharedPtr Cinv, BCinv, Btilde;
1224 DNekMat DintCinvDTint, BCinvDTint_m_DTbnd, DintCinvBTtilde_m_Dbnd;
1225
1226 Cinv = D;
1227 BCinv = B;
1228 Btilde = C;
1229
1230 DintCinvDTint = (*Dint) * (*Cinv) * Transpose(*Dint);
1231 BCinvDTint_m_DTbnd = (*BCinv) * Transpose(*Dint) - Transpose(*Dbnd);
1232
1233 // This could be transpose of BCinvDint in some cases
1234 DintCinvBTtilde_m_Dbnd =
1235 (*Dint) * (*Cinv) * Transpose(*Btilde) - (*Dbnd);
1236
1237 // Set up final set of matrices.
1239 nsize_bndry_p1[n], nsize_p_m1[n], zero);
1241 nsize_p_m1[n], nsize_bndry_p1[n], zero);
1243 nsize_p_m1[n], nsize_p_m1[n], zero);
1244 Array<OneD, NekDouble> Dh_data = Dh->GetPtr();
1245
1246 // Copy matrices into final structures.
1247 size_t n1, n2;
1248 for (n1 = 0; n1 < nz_loc; ++n1)
1249 {
1250 for (i = 0; i < psize - 1; ++i)
1251 {
1252 for (n2 = 0; n2 < nz_loc; ++n2)
1253 {
1254 for (j = 0; j < psize - 1; ++j)
1255 {
1256 //(*Dh)(i+n1*(psize-1),j+n2*(psize-1)) =
1257 //-DintCinvDTint(i+1+n1*psize,j+1+n2*psize);
1258 Dh_data[(i + n1 * (psize - 1)) +
1259 (j + n2 * (psize - 1)) * Dh->GetRows()] =
1260 -DintCinvDTint(i + 1 + n1 * psize,
1261 j + 1 + n2 * psize);
1262 }
1263 }
1264 }
1265 }
1266
1267 for (n1 = 0; n1 < nz_loc; ++n1)
1268 {
1269 for (i = 0; i < nsize_bndry_p1[n] - nz_loc; ++i)
1270 {
1271 (*Ah)(i, nsize_bndry_p1[n] - nz_loc + n1) =
1272 BCinvDTint_m_DTbnd(i, n1 * psize);
1273 (*Ah)(nsize_bndry_p1[n] - nz_loc + n1, i) =
1274 DintCinvBTtilde_m_Dbnd(n1 * psize, i);
1275 }
1276 }
1277
1278 for (n1 = 0; n1 < nz_loc; ++n1)
1279 {
1280 for (n2 = 0; n2 < nz_loc; ++n2)
1281 {
1282 (*Ah)(nsize_bndry_p1[n] - nz_loc + n1,
1283 nsize_bndry_p1[n] - nz_loc + n2) =
1284 -DintCinvDTint(n1 * psize, n2 * psize);
1285 }
1286 }
1287
1288 for (n1 = 0; n1 < nz_loc; ++n1)
1289 {
1290 for (j = 0; j < psize - 1; ++j)
1291 {
1292 for (i = 0; i < nsize_bndry_p1[n] - nz_loc; ++i)
1293 {
1294 (*Bh)(i, j + n1 * (psize - 1)) =
1295 BCinvDTint_m_DTbnd(i, j + 1 + n1 * psize);
1296 (*Ch)(j + n1 * (psize - 1), i) =
1297 DintCinvBTtilde_m_Dbnd(j + 1 + n1 * psize, i);
1298 }
1299 }
1300 }
1301
1302 for (n1 = 0; n1 < nz_loc; ++n1)
1303 {
1304 for (n2 = 0; n2 < nz_loc; ++n2)
1305 {
1306 for (j = 0; j < psize - 1; ++j)
1307 {
1308 (*Bh)(nsize_bndry_p1[n] - nz_loc + n1,
1309 j + n2 * (psize - 1)) =
1310 -DintCinvDTint(n1 * psize, j + 1 + n2 * psize);
1311 (*Ch)(j + n2 * (psize - 1),
1312 nsize_bndry_p1[n] - nz_loc + n1) =
1313 -DintCinvDTint(j + 1 + n2 * psize, n1 * psize);
1314 }
1315 }
1316 }
1317
1318 // Do static condensation
1319 Dh->Invert();
1320 (*Bh) = (*Bh) * (*Dh);
1321 //(*Ah) = (*Ah) - (*Bh)*(*Ch);
1322 Blas::Dgemm('N', 'N', Bh->GetRows(), Ch->GetColumns(), Bh->GetColumns(),
1323 -1.0, Bh->GetRawPtr(), Bh->GetRows(), Ch->GetRawPtr(),
1324 Ch->GetRows(), 1.0, Ah->GetRawPtr(), Ah->GetRows());
1325
1326 // Set matrices for later inversion. Probably do not need to be
1327 // attached to class
1328 pAh->SetBlock(
1329 n, n,
1331 pBh->SetBlock(
1332 n, n,
1334 pCh->SetBlock(
1335 n, n,
1337 pDh->SetBlock(
1338 n, n,
1340 }
1341 timer.Stop();
1342 cout << "Matrix Setup Costs: " << timer.TimePerTest(1) << endl;
1343
1344 timer.Start();
1345 // Set up global coupled boundary solver.
1346 // This is a key to define the solution matrix type
1347 // currently we are giving it a argument of eLInearAdvectionReaction
1348 // since this then makes the matrix storage of type eFull
1350 locToGloMap);
1351 mat.m_CoupledBndSys =
1353 AllocateSharedPtr(key, m_fields[0], pAh, pBh, pCh, pDh,
1354 locToGloMap);
1355 mat.m_CoupledBndSys->Initialise(locToGloMap);
1356}
1357
1359{
1360 SolverUtils::AddSummaryItem(s, "Solver Type", "Coupled Linearised NS");
1361}
1362
1363void CoupledLinearNS::v_DoInitialise(bool dumpInitialConditions)
1364{
1365 switch (m_equationType)
1366 {
1367 case eUnsteadyStokes:
1369 {
1370 // LibUtilities::TimeIntegrationMethod intMethod;
1371 // std::string TimeIntStr =
1372 // m_session->GetSolverInfo("TIMEINTEGRATIONMETHOD");
1373 // int i;
1374 // for(i = 0; i < (int)
1375 // LibUtilities::SIZE_TimeIntegrationMethod; ++i)
1376 // {
1377 // if(boost::iequals(LibUtilities::TimeIntegrationMethodMap[i],TimeIntStr))
1378 // {
1379 // intMethod =
1380 // (LibUtilities::TimeIntegrationMethod)i;
1381 // break;
1382 // }
1383 // }
1384 //
1385 // ASSERTL0(i != (int)
1386 // LibUtilities::SIZE_TimeIntegrationMethod, "Invalid
1387 // time integration type.");
1388 //
1389 // m_integrationScheme =
1390 // LibUtilities::GetTimeIntegrationWrapperFactory().CreateInstance(LibUtilities::TimeIntegrationMethodMap[intMethod]);
1391
1392 // Could defind this from IncNavierStokes class?
1394
1397
1398 // Set initial condition using time t=0
1399
1400 SetInitialConditions(0.0, dumpInitialConditions);
1401 break;
1402 }
1403 case eSteadyStokes:
1404 SetUpCoupledMatrix(0.0);
1405 break;
1406 case eSteadyOseen:
1407 {
1409 for (size_t i = 0; i < m_velocity.size(); ++i)
1410 {
1411 AdvField[i] = Array<OneD, NekDouble>(
1412 m_fields[m_velocity[i]]->GetTotPoints(), 0.0);
1413 }
1414
1415 ASSERTL0(m_session->DefinesFunction("AdvectionVelocity"),
1416 "Advection Velocity section must be defined in "
1417 "session file.");
1418
1419 std::vector<std::string> fieldStr;
1420 for (size_t i = 0; i < m_velocity.size(); ++i)
1421 {
1422 fieldStr.push_back(
1423 m_boundaryConditions->GetVariable(m_velocity[i]));
1424 }
1425 GetFunction("AdvectionVelocity")->Evaluate(fieldStr, AdvField);
1426
1427 SetUpCoupledMatrix(0.0, AdvField, false);
1428 }
1429 break;
1431 {
1432 m_session->LoadParameter("KinvisMin", m_kinvisMin);
1433 m_session->LoadParameter("KinvisPercentage", m_KinvisPercentage);
1434 m_session->LoadParameter("Tolerence", m_tol);
1435 m_session->LoadParameter("MaxIteration", m_maxIt);
1436 m_session->LoadParameter("MatrixSetUpStep", m_MatrixSetUpStep);
1437 m_session->LoadParameter("Restart", m_Restart);
1438
1440
1441 if (m_Restart == 1)
1442 {
1443 ASSERTL0(m_session->DefinesFunction("Restart"),
1444 "Restart section must be defined in session file.");
1445
1447 for (size_t i = 0; i < m_velocity.size(); ++i)
1448 {
1449 Restart[i] = Array<OneD, NekDouble>(
1450 m_fields[m_velocity[i]]->GetTotPoints(), 0.0);
1451 }
1452 std::vector<std::string> fieldStr;
1453 for (size_t i = 0; i < m_velocity.size(); ++i)
1454 {
1455 fieldStr.push_back(
1456 m_boundaryConditions->GetVariable(m_velocity[i]));
1457 }
1458 GetFunction("Restart")->Evaluate(fieldStr, Restart);
1459
1460 for (size_t i = 0; i < m_velocity.size(); ++i)
1461 {
1462 m_fields[m_velocity[i]]->FwdTransLocalElmt(
1463 Restart[i], m_fields[m_velocity[i]]->UpdateCoeffs());
1464 }
1465 cout << "Saving the RESTART file for m_kinvis = " << m_kinvis
1466 << " (<=> Re = " << 1 / m_kinvis << ")" << endl;
1467 }
1468 else // We solve the Stokes Problem
1469 {
1470
1471 SetUpCoupledMatrix(0.0);
1472 m_initialStep = true;
1473 m_counter = 1;
1474 // SolveLinearNS(m_ForcingTerm_Coeffs);
1475 Solve();
1476 m_initialStep = false;
1477 cout << "Saving the Stokes Flow for m_kinvis = " << m_kinvis
1478 << " (<=> Re = " << 1 / m_kinvis << ")" << endl;
1479 }
1480 }
1481 break;
1483 {
1484 SetInitialConditions(0.0, dumpInitialConditions);
1485
1487 for (size_t i = 0; i < m_velocity.size(); ++i)
1488 {
1489 AdvField[i] = Array<OneD, NekDouble>(
1490 m_fields[m_velocity[i]]->GetTotPoints(), 0.0);
1491 }
1492
1493 ASSERTL0(m_session->DefinesFunction("AdvectionVelocity"),
1494 "Advection Velocity section must be defined in "
1495 "session file.");
1496
1497 std::vector<std::string> fieldStr;
1498 for (size_t i = 0; i < m_velocity.size(); ++i)
1499 {
1500 fieldStr.push_back(
1501 m_boundaryConditions->GetVariable(m_velocity[i]));
1502 }
1503 GetFunction("AdvectionVelocity")->Evaluate(fieldStr, AdvField);
1504
1505 SetUpCoupledMatrix(m_lambda, AdvField, true);
1506 }
1507 break;
1508 case eNoEquationType:
1509 default:
1510 ASSERTL0(false,
1511 "Unknown or undefined equation type for CoupledLinearNS");
1512 }
1513}
1514
1516 const Array<OneD, const Array<OneD, NekDouble>> &inarray,
1517 Array<OneD, Array<OneD, NekDouble>> &outarray, const NekDouble time)
1518{
1519 // evaluate convection terms
1520 EvaluateAdvectionTerms(inarray, outarray, time);
1521
1522 for (auto &x : m_forcing)
1523 {
1524 x->Apply(m_fields, outarray, outarray, time);
1525 }
1526}
1527
1529 const Array<OneD, const Array<OneD, NekDouble>> &inarray,
1530 Array<OneD, Array<OneD, NekDouble>> &outarray,
1531 [[maybe_unused]] const NekDouble time, const NekDouble aii_Dt)
1532{
1533 size_t i;
1535 NekDouble lambda = 1.0 / aii_Dt;
1536 static NekDouble lambda_store;
1538 // Matrix solution
1539 if (fabs(lambda_store - lambda) > 1e-10)
1540 {
1541 SetUpCoupledMatrix(lambda);
1542 lambda_store = lambda;
1543 }
1544
1545 // Forcing for advection solve
1546 for (i = 0; i < m_velocity.size(); ++i)
1547 {
1548 bool waveSpace = m_fields[m_velocity[i]]->GetWaveSpace();
1549 m_fields[m_velocity[i]]->SetWaveSpace(true);
1550 m_fields[m_velocity[i]]->IProductWRTBase(
1551 inarray[i], m_fields[m_velocity[i]]->UpdateCoeffs());
1552 m_fields[m_velocity[i]]->SetWaveSpace(waveSpace);
1554 m_fields[m_velocity[i]]->GetCoeffs(), 1,
1555 m_fields[m_velocity[i]]->UpdateCoeffs(), 1);
1556 forcing[i] = m_fields[m_velocity[i]]->GetCoeffs();
1557 }
1558
1559 SolveLinearNS(forcing);
1560
1561 for (i = 0; i < m_velocity.size(); ++i)
1562 {
1563 m_fields[m_velocity[i]]->BwdTrans(m_fields[m_velocity[i]]->GetCoeffs(),
1564 outarray[i]);
1565 }
1566}
1567
1569{
1570 size_t nfields = m_fields.size();
1571 for (size_t k = 0; k < nfields; ++k)
1572 {
1573 // Backward Transformation in physical space for time evolution
1574 m_fields[k]->BwdTrans(m_fields[k]->GetCoeffs(),
1575 m_fields[k]->UpdatePhys());
1576 }
1577}
1578
1580{
1581 size_t nfields = m_fields.size();
1582 for (size_t k = 0; k < nfields; ++k)
1583 {
1584 // Forward Transformation in physical space for time evolution
1585 m_fields[k]->FwdTransLocalElmt(m_fields[k]->GetPhys(),
1586 m_fields[k]->UpdateCoeffs());
1587 }
1588}
1589
1591{
1592 switch (m_equationType)
1593 {
1594 case eUnsteadyStokes:
1596 // AdvanceInTime(m_steps);
1598 break;
1599 case eSteadyStokes:
1600 case eSteadyOseen:
1602 {
1603 Solve();
1604 break;
1605 }
1607 {
1608 LibUtilities::Timer Generaltimer;
1609 Generaltimer.Start();
1610
1611 int Check(0);
1612
1613 // Saving the init datas
1614 Checkpoint_Output(Check);
1615 Check++;
1616
1617 cout << "We execute INITIALLY SolveSteadyNavierStokes for m_kinvis "
1618 "= "
1619 << m_kinvis << " (<=> Re = " << 1 / m_kinvis << ")" << endl;
1621
1622 while (m_kinvis > m_kinvisMin)
1623 {
1624 if (Check == 1)
1625 {
1626 cout << "We execute SolveSteadyNavierStokes for m_kinvis = "
1627 << m_kinvis << " (<=> Re = " << 1 / m_kinvis << ")"
1628 << endl;
1630 Checkpoint_Output(Check);
1631 Check++;
1632 }
1633
1634 Continuation();
1635
1636 if (m_kinvis > m_kinvisMin)
1637 {
1638 cout << "We execute SolveSteadyNavierStokes for m_kinvis = "
1639 << m_kinvis << " (<=> Re = " << 1 / m_kinvis << ")"
1640 << endl;
1642 Checkpoint_Output(Check);
1643 Check++;
1644 }
1645 }
1646
1647 Generaltimer.Stop();
1648 cout << "\nThe total calculation time is : "
1649 << Generaltimer.TimePerTest(1) / 60 << " minute(s). \n\n";
1650
1651 break;
1652 }
1653 case eNoEquationType:
1654 default:
1655 ASSERTL0(false,
1656 "Unknown or undefined equation type for CoupledLinearNS");
1657 }
1658}
1659
1660/** Virtual function to define if operator in DoSolve is negated
1661 * with regard to the strong form. This is currently only used in
1662 * Arnoldi solves. For Coupledd solver this is true since Stokes
1663 * operator is set up as a LHS rather than RHS operation
1664 */
1666{
1667 return true;
1668}
1669
1671{
1672 const size_t ncmpt = m_velocity.size();
1673 Array<OneD, Array<OneD, NekDouble>> forcing_phys(ncmpt);
1674 Array<OneD, Array<OneD, NekDouble>> forcing(ncmpt);
1675
1676 for (size_t i = 0; i < ncmpt; ++i)
1677 {
1678 forcing_phys[i] =
1680 forcing[i] =
1682 }
1683
1684 for (auto &x : m_forcing)
1685 {
1686 const NekDouble time = 0;
1687 x->Apply(m_fields, forcing_phys, forcing_phys, time);
1688 }
1689 for (size_t i = 0; i < ncmpt; ++i)
1690 {
1691 bool waveSpace = m_fields[m_velocity[i]]->GetWaveSpace();
1692 m_fields[i]->SetWaveSpace(true);
1693 m_fields[i]->IProductWRTBase(forcing_phys[i], forcing[i]);
1694 m_fields[i]->SetWaveSpace(waveSpace);
1695 }
1696
1697 SolveLinearNS(forcing);
1698}
1699
1701{
1705
1706 for (size_t i = 0; i < m_velocity.size(); ++i)
1707 {
1709 m_fields[m_velocity[i]]->GetTotPoints(), 0.0);
1712 }
1713
1714 if (m_session->DefinesFunction("ForcingTerm"))
1715 {
1716 std::vector<std::string> fieldStr;
1717 for (size_t i = 0; i < m_velocity.size(); ++i)
1718 {
1719 fieldStr.push_back(
1720 m_boundaryConditions->GetVariable(m_velocity[i]));
1721 }
1722 GetFunction("ForcingTerm")->Evaluate(fieldStr, m_ForcingTerm);
1723 for (size_t i = 0; i < m_velocity.size(); ++i)
1724 {
1725 m_fields[m_velocity[i]]->FwdTransLocalElmt(m_ForcingTerm[i],
1727 }
1728 }
1729 else
1730 {
1731 cout << "'ForcingTerm' section has not been defined in the input file "
1732 "=> forcing=0"
1733 << endl;
1734 }
1735}
1736
1738{
1739 LibUtilities::Timer Newtontimer;
1740 Newtontimer.Start();
1741
1744 Array<OneD, Array<OneD, NekDouble>> delta_velocity_Phys(m_velocity.size());
1745 Array<OneD, Array<OneD, NekDouble>> Velocity_Phys(m_velocity.size());
1746 Array<OneD, NekDouble> L2_norm(m_velocity.size(), 1.0);
1747 Array<OneD, NekDouble> Inf_norm(m_velocity.size(), 1.0);
1748
1749 for (size_t i = 0; i < m_velocity.size(); ++i)
1750 {
1751 delta_velocity_Phys[i] = Array<OneD, NekDouble>(
1752 m_fields[m_velocity[i]]->GetTotPoints(), 1.0);
1753 Velocity_Phys[i] = Array<OneD, NekDouble>(
1754 m_fields[m_velocity[i]]->GetTotPoints(), 0.0);
1755 }
1756
1757 m_counter = 1;
1758
1759 L2Norm(delta_velocity_Phys, L2_norm);
1760
1761 // while(max(Inf_norm[0], Inf_norm[1]) > m_tol)
1762 while (max(L2_norm[0], L2_norm[1]) > m_tol)
1763 {
1764 if (m_counter == 1)
1765 // At the first Newton step, we use the solution of the
1766 // Stokes problem (if Restart=0 in input file) Or the
1767 // solution of the .rst file (if Restart=1 in input
1768 // file)
1769 {
1770 for (size_t i = 0; i < m_velocity.size(); ++i)
1771 {
1772 RHS_Coeffs[i] = Array<OneD, NekDouble>(
1773 m_fields[m_velocity[i]]->GetNcoeffs(), 0.0);
1774 RHS_Phys[i] = Array<OneD, NekDouble>(
1775 m_fields[m_velocity[i]]->GetTotPoints(), 0.0);
1776 }
1777
1778 for (size_t i = 0; i < m_velocity.size(); ++i)
1779 {
1780 m_fields[m_velocity[i]]->BwdTrans(
1781 m_fields[m_velocity[i]]->GetCoeffs(), Velocity_Phys[i]);
1782 }
1783
1784 m_initialStep = true;
1785 EvaluateNewtonRHS(Velocity_Phys, RHS_Coeffs);
1786 SetUpCoupledMatrix(0.0, Velocity_Phys, true);
1787 SolveLinearNS(RHS_Coeffs);
1788 m_initialStep = false;
1789 }
1790 if (m_counter > 1)
1791 {
1792 EvaluateNewtonRHS(Velocity_Phys, RHS_Coeffs);
1793
1795 0) // Setting Up the matrix is expensive. We do it at each
1796 // "m_MatrixSetUpStep" step.
1797 {
1798 SetUpCoupledMatrix(0.0, Velocity_Phys, true);
1799 }
1800 SolveLinearNS(RHS_Coeffs);
1801 }
1802
1803 for (size_t i = 0; i < m_velocity.size(); ++i)
1804 {
1805 m_fields[m_velocity[i]]->BwdTrans(RHS_Coeffs[i], RHS_Phys[i]);
1806 m_fields[m_velocity[i]]->BwdTrans(
1807 m_fields[m_velocity[i]]->GetCoeffs(), delta_velocity_Phys[i]);
1808 }
1809
1810 for (size_t i = 0; i < m_velocity.size(); ++i)
1811 {
1812 Vmath::Vadd(Velocity_Phys[i].size(), Velocity_Phys[i], 1,
1813 delta_velocity_Phys[i], 1, Velocity_Phys[i], 1);
1814 }
1815
1816 // InfNorm(delta_velocity_Phys, Inf_norm);
1817 L2Norm(delta_velocity_Phys, L2_norm);
1818
1819 if (max(Inf_norm[0], Inf_norm[1]) > 100)
1820 {
1821 cout << "\nThe Newton method has failed at m_kinvis = " << m_kinvis
1822 << " (<=> Re = " << 1 / m_kinvis << ")" << endl;
1823 ASSERTL0(0, "The Newton method has failed... \n");
1824 }
1825
1826 cout << "\n";
1827 m_counter++;
1828 }
1829
1830 if (m_counter > 1) // We save u:=u+\delta u in u->Coeffs
1831 {
1832 for (size_t i = 0; i < m_velocity.size(); ++i)
1833 {
1834 m_fields[m_velocity[i]]->FwdTrans(
1835 Velocity_Phys[i], m_fields[m_velocity[i]]->UpdateCoeffs());
1836 }
1837 }
1838
1839 Newtontimer.Stop();
1840 cout << "We have done " << m_counter - 1 << " iteration(s) in "
1841 << Newtontimer.TimePerTest(1) / 60 << " minute(s). \n\n";
1842}
1843
1845{
1850
1851 cout << "We apply the continuation method: " << endl;
1852
1853 for (size_t i = 0; i < m_velocity.size(); ++i)
1854 {
1856 0.0);
1857 m_fields[m_velocity[i]]->BwdTrans(m_fields[m_velocity[i]]->GetCoeffs(),
1858 u_N[i]);
1859
1861 0.0);
1862 tmp_RHS[i] = Array<OneD, NekDouble>(
1863 m_fields[m_velocity[i]]->GetTotPoints(), 0.0);
1864
1865 m_fields[m_velocity[i]]->PhysDeriv(i, u_N[i], tmp_RHS[i]);
1866 Vmath::Smul(tmp_RHS[i].size(), m_kinvis, tmp_RHS[i], 1, tmp_RHS[i], 1);
1867
1868 bool waveSpace = m_fields[m_velocity[i]]->GetWaveSpace();
1869 m_fields[m_velocity[i]]->SetWaveSpace(true);
1870 m_fields[m_velocity[i]]->IProductWRTDerivBase(i, tmp_RHS[i], RHS[i]);
1871 m_fields[m_velocity[i]]->SetWaveSpace(waveSpace);
1872 }
1873
1874 SetUpCoupledMatrix(0.0, u_N, true);
1875 SolveLinearNS(RHS);
1876
1877 for (size_t i = 0; i < m_velocity.size(); ++i)
1878 {
1879 u_star[i] = Array<OneD, NekDouble>(
1880 m_fields[m_velocity[i]]->GetTotPoints(), 0.0);
1881 m_fields[m_velocity[i]]->BwdTrans(m_fields[m_velocity[i]]->GetCoeffs(),
1882 u_star[i]);
1883
1884 // u_star(k+1) = u_N(k) + DeltaKinvis * u_star(k)
1885 Vmath::Smul(u_star[i].size(), m_kinvis, u_star[i], 1, u_star[i], 1);
1886 Vmath::Vadd(u_star[i].size(), u_star[i], 1, u_N[i], 1, u_star[i], 1);
1887
1888 m_fields[m_velocity[i]]->FwdTrans(
1889 u_star[i], m_fields[m_velocity[i]]->UpdateCoeffs());
1890 }
1891
1893}
1894
1896 Array<OneD, NekDouble> &outarray)
1897{
1898 for (size_t i = 0; i < m_velocity.size(); ++i)
1899 {
1900 outarray[i] = 0.0;
1901 for (size_t j = 0; j < inarray[i].size(); ++j)
1902 {
1903 if (inarray[i][j] > outarray[i])
1904 {
1905 outarray[i] = inarray[i][j];
1906 }
1907 }
1908 cout << "InfNorm[" << i << "] = " << outarray[i] << endl;
1909 }
1910}
1911
1913 Array<OneD, NekDouble> &outarray)
1914{
1915 for (size_t i = 0; i < m_velocity.size(); ++i)
1916 {
1917 outarray[i] = 0.0;
1918 for (size_t j = 0; j < inarray[i].size(); ++j)
1919 {
1920 outarray[i] += inarray[i][j] * inarray[i][j];
1921 }
1922 outarray[i] = sqrt(outarray[i]);
1923 cout << "L2Norm[" << i << "] = " << outarray[i] << endl;
1924 }
1925}
1926
1928 Array<OneD, Array<OneD, NekDouble>> &Velocity,
1929 Array<OneD, Array<OneD, NekDouble>> &outarray)
1930{
1936
1937 for (size_t i = 0; i < m_velocity.size(); ++i)
1938 {
1939 Eval_Adv[i] = Array<OneD, NekDouble>(
1940 m_fields[m_velocity[i]]->GetTotPoints(), 0.0);
1941 tmp_DerVel[i] = Array<OneD, NekDouble>(
1942 m_fields[m_velocity[i]]->GetTotPoints(), 0.0);
1943
1944 AdvTerm[i] = Array<OneD, NekDouble>(
1945 m_fields[m_velocity[i]]->GetTotPoints(), 0.0);
1946 ViscTerm[i] = Array<OneD, NekDouble>(
1947 m_fields[m_velocity[i]]->GetTotPoints(), 0.0);
1948 Forc[i] = Array<OneD, NekDouble>(
1949 m_fields[m_velocity[i]]->GetTotPoints(), 0.0);
1950 outarray[i] = Array<OneD, NekDouble>(
1951 m_fields[m_velocity[i]]->GetTotPoints(), 0.0);
1952
1953 m_fields[m_velocity[i]]->PhysDeriv(i, Velocity[i], tmp_DerVel[i]);
1954
1955 Vmath::Smul(tmp_DerVel[i].size(), m_kinvis, tmp_DerVel[i], 1,
1956 tmp_DerVel[i], 1);
1957 }
1958
1959 EvaluateAdvectionTerms(Velocity, Eval_Adv, m_time);
1960
1961 for (size_t i = 0; i < m_velocity.size(); ++i)
1962 {
1963 bool waveSpace = m_fields[m_velocity[i]]->GetWaveSpace();
1964 m_fields[m_velocity[i]]->SetWaveSpace(true);
1965 m_fields[m_velocity[i]]->IProductWRTBase(Eval_Adv[i],
1966 AdvTerm[i]); //(w, (u.grad)u)
1967 m_fields[m_velocity[i]]->IProductWRTDerivBase(
1968 i, tmp_DerVel[i], ViscTerm[i]); //(grad w, grad u)
1969 m_fields[m_velocity[i]]->IProductWRTBase(m_ForcingTerm[i],
1970 Forc[i]); //(w, f)
1971 m_fields[m_velocity[i]]->SetWaveSpace(waveSpace);
1972
1973 Vmath::Vsub(outarray[i].size(), outarray[i], 1, AdvTerm[i], 1,
1974 outarray[i], 1);
1975 Vmath::Vsub(outarray[i].size(), outarray[i], 1, ViscTerm[i], 1,
1976 outarray[i], 1);
1977
1978 Vmath::Vadd(outarray[i].size(), outarray[i], 1, Forc[i], 1, outarray[i],
1979 1);
1980 }
1981}
1982
1985{
1987 returnval =
1989
1990 int nummodes;
1991
1992 for (auto &expMapIter : VelExp)
1993 {
1995
1996 for (size_t i = 0; i < expMapIter.second->m_basisKeyVector.size(); ++i)
1997 {
1998 LibUtilities::BasisKey B = expMapIter.second->m_basisKeyVector[i];
1999 nummodes = B.GetNumModes();
2000 ASSERTL0(nummodes > 3,
2001 "Velocity polynomial space not sufficiently high (>= 4)");
2002 // Should probably set to be an orthogonal basis.
2003 LibUtilities::BasisKey newB(B.GetBasisType(), nummodes - 2,
2004 B.GetPointsKey());
2005 BasisVec.push_back(newB);
2006 }
2007
2008 // Put new expansion into list.
2009 SpatialDomains::ExpansionInfoShPtr expansionElementShPtr =
2011 expMapIter.second->m_geomShPtr, BasisVec);
2012 (*returnval)[expMapIter.first] = expansionElementShPtr;
2013 }
2014
2015 // Save expansion into graph.
2016 m_graph->SetExpansionInfo("p", returnval);
2017
2018 return *returnval;
2019}
2020
2021/**
2022 * @param forcing A list of forcing functions for each velocity
2023 * component
2024 *
2025 * The routine involves two levels of static
2026 * condensations. Initially we require a statically condensed
2027 * forcing function which requires the following manipulation
2028 *
2029 * \f[ {F\_bnd} = {\bf f}_{bnd} -m\_B \,m\_Cinv\, {\bf f}_{int},
2030 * \hspace{1cm} F\_p = m\_D\_{int}\, m\_Cinv\, {\bf f}_{int} \f]
2031 *
2032 * Where \f${\bf f}_{bnd}\f$ denote the forcing degrees of
2033 * freedom of the elemental velocities on the boundary of the
2034 * element, \f${\bf f}_{int}\f$ denote the forcing degrees of
2035 * freedom of the elemental velocities on the interior of the
2036 * element. (see detailed description for more details).
2037 *
2038 * This vector is further manipulated into
2039 *
2040 * \f[ Fh\_{bnd} = \left [ \begin{array}{c} f\_{bnd} -m\_B \,
2041 * m\_Cinv\, {\bf f}_{int}\\ \left [m\_D\_{int} \, m\_Cinv \,{\bf
2042 * f}_{int} \right]_0 \end{array}\right ] \hspace{1cm} [Fh\_p]_{i} =
2043 * \begin{array}{c} [m\_D\_{int} \, m\_Cinv \, {\bf
2044 * f}_{int}]_{i+1} \end{array} \f]
2045 *
2046 * where \f$-{[}m\_D\_{int}^T\, m\_Cinv \,{\bf f}_{int}]_0\f$
2047 * which is corresponds to the mean mode of the pressure degrees
2048 * of freedom and is now added to the boundary system and the
2049 * remainder of the block becomes the interior forcing for the
2050 * inner static condensation (see detailed description for more
2051 * details) which is set up in a GlobalLinSysDirectStaticCond
2052 * class.
2053 *
2054 * Finally we perform the final maniplation of the forcing to
2055 * using hte
2056 * \f[ Fh\_{bnd} = Fh\_{bnd} - m\_Bh \,m\_Chinv \, Fh\_p \f]
2057 *
2058 * We can now call the solver to the global coupled boundary
2059 * system (through the call to #m_CoupledBndSys->Solve) to obtain
2060 * the velocity boundary solution as the mean pressure solution,
2061 * i.e.
2062 *
2063 * \f[ {\cal A}^T(\hat{A} - \hat{C}^T \hat{D}^{-1} \hat{B} ){\cal
2064 * A} \, Bnd = Fh\_{bnd} \f]
2065 *
2066 * Once we know the solution to the above the rest of the pressure
2067 * modes are recoverable thorugh
2068 *
2069 * \f[ Ph = m\_Dhinv\, (Bnd - m\_Ch^T \, Fh_{bnd}) \f]
2070 *
2071 * We can now unpack \f$ Fh\_{bnd} \f$ (last elemental mode) and
2072 * \f$ Ph \f$ into #m_pressure and \f$ F_p\f$ and \f$ Fh\_{bnd}\f$
2073 * into a closed pack list of boundary velocoity degrees of
2074 * freedom stored in \f$ F\_bnd\f$.
2075 *
2076 * Finally the interior velocity degrees of freedom are then
2077 * obtained through the relationship
2078 *
2079 * \f[ F\_{int} = m\_Cinv\ ( F\_{int} + m\_D\_{int}^T\,
2080 * F\_p - m\_Btilde^T\, Bnd) \f]
2081 *
2082 * We then unpack the solution back to the MultiRegion structures
2083 * #m_velocity and #m_pressure
2084 */
2086 const Array<OneD, Array<OneD, NekDouble>> &forcing)
2087{
2088 size_t i;
2091
2092 // Impose Dirichlet conditions on velocity fields
2093 for (i = 0; i < m_velocity.size(); ++i)
2094 {
2095 Vmath::Zero(m_fields[i]->GetNcoeffs(), m_fields[i]->UpdateCoeffs(), 1);
2096 m_fields[i]->ImposeDirichletConditions(m_fields[i]->UpdateCoeffs());
2097 }
2098
2100 {
2101 int ncoeffsplane = m_fields[m_velocity[0]]->GetPlane(0)->GetNcoeffs();
2102 for (int n = 0; n < m_npointsZ / 2; ++n)
2103 {
2104 // Get the a Fourier mode of velocity and forcing components.
2105 for (i = 0; i < m_velocity.size(); ++i)
2106 {
2107 vel_fields[i] = m_fields[m_velocity[i]]->GetPlane(2 * n);
2108 // Note this needs to correlate with how we pass forcing
2109 force[i] = forcing[i] + 2 * n * ncoeffsplane;
2110 }
2111
2112 SolveLinearNS(force, vel_fields, m_pressure->GetPlane(2 * n), n);
2113 }
2114 for (i = 0; i < m_velocity.size(); ++i)
2115 {
2116 m_fields[m_velocity[i]]->SetPhysState(false);
2117 }
2118 m_pressure->SetPhysState(false);
2119 }
2120 else
2121 {
2122 for (i = 0; i < m_velocity.size(); ++i)
2123 {
2124 vel_fields[i] = m_fields[m_velocity[i]];
2125 // Note this needs to correlate with how we pass forcing
2126 force[i] = forcing[i];
2127 }
2128 SolveLinearNS(force, vel_fields, m_pressure);
2129 }
2130}
2131
2136{
2137 size_t i, j, k, n, cnt, cnt1;
2138 size_t nbnd, nint, offset;
2139 size_t nvel = m_velocity.size();
2140 size_t nel = fields[0]->GetNumElmts();
2141 Array<OneD, unsigned int> bmap, imap;
2142
2143 Array<OneD, NekDouble> f_bnd(m_mat[mode].m_BCinv->GetRows());
2144 NekVector<NekDouble> F_bnd(f_bnd.size(), f_bnd, eWrapper);
2145 Array<OneD, NekDouble> f_int(m_mat[mode].m_BCinv->GetColumns());
2146 NekVector<NekDouble> F_int(f_int.size(), f_int, eWrapper);
2147
2148 size_t nz_loc;
2149 int nplanecoeffs =
2150 fields[m_velocity[0]]
2151 ->GetNcoeffs(); // this is fine since we pass the nplane coeff data.
2152
2153 if (mode) // Homogeneous mode flag
2154 {
2155 nz_loc = 2;
2156 }
2157 else
2158 {
2159 if (m_singleMode)
2160 {
2161 nz_loc = 2;
2162 }
2163 else
2164 {
2165 nz_loc = 1;
2167 {
2169 // Zero fields to set complex mode to zero;
2170 for (i = 0; i < fields.size(); ++i)
2171 {
2172 Vmath::Zero(nplanecoeffs,
2173 tmp = fields[i]->UpdateCoeffs() + nplanecoeffs,
2174 1);
2175 }
2176 Vmath::Zero(2 * pressure->GetNcoeffs(),
2177 pressure->UpdateCoeffs(), 1);
2178 }
2179 }
2180 }
2181
2182 for (k = 0; k < nvel; ++k)
2183 {
2185 std::dynamic_pointer_cast<MultiRegions::ContField>(fields[k]);
2186
2188 cfield->GetLocalToGlobalMap()->GetBndCondCoeffsToLocalCoeffsSign();
2189 const Array<OneD, const int> map =
2190 cfield->GetLocalToGlobalMap()->GetBndCondCoeffsToLocalCoeffsMap();
2191
2192 // Add weak boundary conditions to forcing
2194 fields[k]->GetBndConditions();
2196
2198 {
2199 bndCondExp =
2200 m_fields[k]->GetPlane(2 * mode)->GetBndCondExpansions();
2201 }
2202 else
2203 {
2204 bndCondExp = m_fields[k]->GetBndCondExpansions();
2205 }
2206
2207 for (n = 0; n < nz_loc; ++n)
2208 {
2209 int bndcnt = 0;
2210 for (i = 0; i < bndCondExp.size(); ++i)
2211 {
2212 const Array<OneD, const NekDouble> bndcoeffs =
2213 bndCondExp[i]->GetCoeffs();
2214 size_t nCoeffs = (bndCondExp[i])->GetNcoeffs();
2215
2216 cnt = 0;
2217 if (bndConds[i]->GetBoundaryConditionType() ==
2219 bndConds[i]->GetBoundaryConditionType() ==
2221 {
2222 if (m_locToGloMap[mode]->GetSignChange())
2223 {
2224 for (j = 0; j < nCoeffs; j++)
2225 {
2226 forcing[k][n * nplanecoeffs + map[bndcnt + j]] +=
2227 sign[bndcnt + j] * bndcoeffs[j];
2228 }
2229 }
2230 else
2231 {
2232 for (j = 0; j < nCoeffs; j++)
2233 {
2234 forcing[k][n * nplanecoeffs + map[bndcnt + j]] +=
2235 bndcoeffs[j];
2236 }
2237 }
2238 }
2239
2240 bndcnt += bndCondExp[i]->GetNcoeffs();
2241 }
2242 }
2243 }
2244
2245 Array<OneD, NekDouble> bnd(m_locToGloMap[mode]->GetNumLocalCoeffs(), 0.0);
2246
2247 // Construct f_bnd and f_int and fill in bnd from inarray
2248 // (with Dirichlet BCs imposed)
2249 int bndoffset = 0;
2250 cnt = cnt1 = 0;
2251 for (i = 0; i < nel; ++i) // loop over elements
2252 {
2253 fields[m_velocity[0]]->GetExp(i)->GetBoundaryMap(bmap);
2254 fields[m_velocity[0]]->GetExp(i)->GetInteriorMap(imap);
2255 nbnd = bmap.size();
2256 nint = imap.size();
2257 offset = fields[m_velocity[0]]->GetCoeff_Offset(i);
2258
2259 for (j = 0; j < nvel; ++j) // loop over velocity fields
2260 {
2261 Array<OneD, NekDouble> incoeffs = fields[j]->UpdateCoeffs();
2262
2263 for (n = 0; n < nz_loc; ++n)
2264 {
2265 for (k = 0; k < nbnd; ++k)
2266 {
2267 f_bnd[cnt + k] =
2268 forcing[j][n * nplanecoeffs + offset + bmap[k]];
2269
2270 bnd[bndoffset + (n + j * nz_loc) * nbnd + k] =
2271 incoeffs[n * nplanecoeffs + offset + bmap[k]];
2272 }
2273 for (k = 0; k < nint; ++k)
2274 {
2275 f_int[cnt1 + k] =
2276 forcing[j][n * nplanecoeffs + offset + imap[k]];
2277 }
2278
2279 cnt += nbnd;
2280 cnt1 += nint;
2281 }
2282 }
2283 bndoffset +=
2284 nvel * nz_loc * nbnd + nz_loc * (pressure->GetExp(i)->GetNcoeffs());
2285 }
2286
2287 Array<OneD, NekDouble> f_p(m_mat[mode].m_D_int->GetRows());
2288 NekVector<NekDouble> F_p(f_p.size(), f_p, eWrapper);
2289 NekVector<NekDouble> F_p_tmp(m_mat[mode].m_Cinv->GetRows());
2290
2291 // fbnd does not currently hold the pressure mean
2292 F_bnd = F_bnd - (*m_mat[mode].m_BCinv) * F_int;
2293 F_p_tmp = (*m_mat[mode].m_Cinv) * F_int;
2294 F_p = (*m_mat[mode].m_D_int) * F_p_tmp;
2295
2296 // construct inner forcing
2297 Array<OneD, NekDouble> fh_bnd(m_locToGloMap[mode]->GetNumLocalCoeffs(),
2298 0.0);
2299
2300 offset = cnt = 0;
2301 for (i = 0; i < nel; ++i)
2302 {
2303 nbnd = nz_loc * fields[0]->GetExp(i)->NumBndryCoeffs();
2304
2305 for (j = 0; j < nvel; ++j)
2306 {
2307 for (k = 0; k < nbnd; ++k)
2308 {
2309 fh_bnd[offset + j * nbnd + k] = f_bnd[cnt + k];
2310 }
2311 cnt += nbnd;
2312 }
2313
2314 nint = pressure->GetExp(i)->GetNcoeffs();
2315 offset += nvel * nbnd + nint * nz_loc;
2316 }
2317
2318 offset = cnt1 = 0;
2319 for (i = 0; i < nel; ++i)
2320 {
2321 nbnd = nz_loc * fields[0]->GetExp(i)->NumBndryCoeffs();
2322 nint = pressure->GetExp(i)->GetNcoeffs();
2323
2324 for (n = 0; n < nz_loc; ++n)
2325 {
2326 for (j = 0; j < nint; ++j)
2327 {
2328 fh_bnd[offset + nvel * nbnd + n * nint + j] = f_p[cnt1 + j];
2329 }
2330 cnt1 += nint;
2331 }
2332 offset += nvel * nbnd + nz_loc * nint;
2333 }
2334 m_mat[mode].m_CoupledBndSys->Solve(fh_bnd, bnd, m_locToGloMap[mode]);
2335
2336 // unpack pressure and velocity boundary systems.
2337 offset = cnt = 0;
2338 int totpcoeffs = pressure->GetNcoeffs();
2339 Array<OneD, NekDouble> p_coeffs = pressure->UpdateCoeffs();
2340 for (i = 0; i < nel; ++i)
2341 {
2342 nbnd = nz_loc * fields[0]->GetExp(i)->NumBndryCoeffs();
2343 nint = pressure->GetExp(i)->GetNcoeffs();
2344 for (j = 0; j < nvel; ++j)
2345 {
2346 for (k = 0; k < nbnd; ++k)
2347 {
2348 f_bnd[cnt + k] = bnd[offset + j * nbnd + k];
2349 }
2350 cnt += nbnd;
2351 }
2352 offset += nvel * nbnd + nint * nz_loc;
2353 }
2354
2355 pressure->SetPhysState(false);
2356
2357 offset = cnt = cnt1 = 0;
2358 for (i = 0; i < nel; ++i)
2359 {
2360 nint = pressure->GetExp(i)->GetNcoeffs();
2361 nbnd = fields[0]->GetExp(i)->NumBndryCoeffs();
2362 cnt1 = pressure->GetCoeff_Offset(i);
2363
2364 for (n = 0; n < nz_loc; ++n)
2365 {
2366 for (j = 0; j < nint; ++j)
2367 {
2368 p_coeffs[n * totpcoeffs + cnt1 + j] = f_p[cnt + j] =
2369 bnd[offset + nvel * nz_loc * nbnd + n * nint + j];
2370 }
2371 cnt += nint;
2372 }
2373 offset += (nvel * nbnd + nint) * nz_loc;
2374 }
2375
2376 // Back solve first level of static condensation for interior
2377 // velocity space and store in F_int
2378 F_int = F_int + Transpose(*m_mat[mode].m_D_int) * F_p -
2379 Transpose(*m_mat[mode].m_Btilde) * F_bnd;
2380 F_int = (*m_mat[mode].m_Cinv) * F_int;
2381
2382 // Unpack solution from Bnd and F_int to v_coeffs
2383 cnt = cnt1 = 0;
2384 for (i = 0; i < nel; ++i) // loop over elements
2385 {
2386 fields[0]->GetExp(i)->GetBoundaryMap(bmap);
2387 fields[0]->GetExp(i)->GetInteriorMap(imap);
2388 nbnd = bmap.size();
2389 nint = imap.size();
2390 offset = fields[0]->GetCoeff_Offset(i);
2391
2392 for (j = 0; j < nvel; ++j) // loop over velocity fields
2393 {
2394 for (n = 0; n < nz_loc; ++n)
2395 {
2396 for (k = 0; k < nbnd; ++k)
2397 {
2398 fields[j]->SetCoeff(n * nplanecoeffs + offset + bmap[k],
2399 f_bnd[cnt + k]);
2400 }
2401
2402 for (k = 0; k < nint; ++k)
2403 {
2404 fields[j]->SetCoeff(n * nplanecoeffs + offset + imap[k],
2405 f_int[cnt1 + k]);
2406 }
2407 cnt += nbnd;
2408 cnt1 += nint;
2409 }
2410 }
2411 }
2412
2413 for (j = 0; j < nvel; ++j)
2414 {
2415 fields[j]->SetPhysState(false);
2416 }
2417}
2418
2420{
2421 std::vector<Array<OneD, NekDouble>> fieldcoeffs(m_fields.size() + 1);
2422 std::vector<std::string> variables(m_fields.size() + 1);
2423 size_t i;
2424
2425 for (i = 0; i < m_fields.size(); ++i)
2426 {
2427 fieldcoeffs[i] = m_fields[i]->UpdateCoeffs();
2428 variables[i] = m_boundaryConditions->GetVariable(i);
2429 }
2430
2431 fieldcoeffs[i] = Array<OneD, NekDouble>(m_fields[0]->GetNcoeffs());
2432 // project pressure field to velocity space
2433 if (m_singleMode == true)
2434 {
2435 Array<OneD, NekDouble> tmpfieldcoeffs(m_fields[0]->GetNcoeffs() / 2);
2436 m_pressure->GetPlane(0)->BwdTrans(
2437 m_pressure->GetPlane(0)->GetCoeffs(),
2438 m_pressure->GetPlane(0)->UpdatePhys());
2439 m_pressure->GetPlane(1)->BwdTrans(
2440 m_pressure->GetPlane(1)->GetCoeffs(),
2441 m_pressure->GetPlane(1)->UpdatePhys());
2442 m_fields[0]->GetPlane(0)->FwdTransLocalElmt(
2443 m_pressure->GetPlane(0)->GetPhys(), fieldcoeffs[i]);
2444 m_fields[0]->GetPlane(1)->FwdTransLocalElmt(
2445 m_pressure->GetPlane(1)->GetPhys(), tmpfieldcoeffs);
2446 for (int e = 0; e < m_fields[0]->GetNcoeffs() / 2; e++)
2447 {
2448 fieldcoeffs[i][e + m_fields[0]->GetNcoeffs() / 2] =
2449 tmpfieldcoeffs[e];
2450 }
2451 }
2452 else
2453 {
2454 m_pressure->BwdTrans(m_pressure->GetCoeffs(), m_pressure->UpdatePhys());
2455 m_fields[0]->FwdTransLocalElmt(m_pressure->GetPhys(), fieldcoeffs[i]);
2456 }
2457 variables[i] = "p";
2458
2459 if (!m_comm->IsParallelInTime())
2460 {
2461 WriteFld(m_sessionName + ".fld", m_fields[0], fieldcoeffs, variables);
2462 }
2463 else
2464 {
2465 std::string newdir = m_sessionName + ".pit";
2466 if (!fs::is_directory(newdir))
2467 {
2468 fs::create_directory(newdir);
2469 }
2470 WriteFld(
2471 newdir + "/" + m_sessionName + "_" +
2472 std::to_string(m_windowPIT * m_comm->GetTimeComm()->GetSize() +
2473 m_comm->GetTimeComm()->GetRank() + 1) +
2474 ".fld",
2475 m_fields[0], fieldcoeffs, variables);
2476 }
2477}
2478
2480{
2481 return m_session->GetVariables().size();
2482}
2483} // namespace Nektar
#define ASSERTL0(condition, msg)
Definition: ErrorUtil.hpp:208
#define ASSERTL1(condition, msg)
Assert Level 1 – Debugging which is used whether in FULLDEBUG or DEBUG compilation mode....
Definition: ErrorUtil.hpp:242
#define sign(a, b)
return the sign(b)*a
Definition: Polylib.cpp:47
void L2Norm(Array< OneD, Array< OneD, NekDouble > > &inarray, Array< OneD, NekDouble > &outarray)
CoupledLinearNS(const LibUtilities::SessionReaderSharedPtr &pSesssion, const SpatialDomains::MeshGraphSharedPtr &pGraph)
void v_GenerateSummary(SolverUtils::SummaryList &s) override
Print a summary of time stepping parameters.
void EvaluateNewtonRHS(Array< OneD, Array< OneD, NekDouble > > &Velocity, Array< OneD, Array< OneD, NekDouble > > &outarray)
void v_DoSolve(void) override
Solves an unsteady problem.
Array< OneD, CoupledLocalToGlobalC0ContMapSharedPtr > m_locToGloMap
void SolveUnsteadyStokesSystem(const Array< OneD, const Array< OneD, NekDouble > > &inarray, Array< OneD, Array< OneD, NekDouble > > &outarray, const NekDouble time, const NekDouble a_iixDt)
void SolveLinearNS(const Array< OneD, Array< OneD, NekDouble > > &forcing)
void v_Output(void) override
void v_DoInitialise(bool dumpInitialConditions=true) override
Sets up initial conditions.
const SpatialDomains::ExpansionInfoMap & GenPressureExp(const SpatialDomains::ExpansionInfoMap &VelExp)
bool v_NegatedOp(void) override
void EvaluateAdvection(const Array< OneD, const Array< OneD, NekDouble > > &inarray, Array< OneD, Array< OneD, NekDouble > > &outarray, const NekDouble time)
static SolverUtils::EquationSystemSharedPtr create(const LibUtilities::SessionReaderSharedPtr &pSession, const SpatialDomains::MeshGraphSharedPtr &pGraph)
Creates an instance of this class.
void v_InitObject(bool DeclareField=true) override
Initialisation object for EquationSystem.
static std::string className
Name of class.
bool m_zeroMode
Id to identify when single mode is mean mode (i.e. beta=0);.
Array< OneD, Array< OneD, NekDouble > > m_ForcingTerm_Coeffs
Array< OneD, Array< OneD, NekDouble > > m_ForcingTerm
void v_TransPhysToCoeff(void) override
Virtual function for transformation to coefficient space.
void SetUpCoupledMatrix(const NekDouble lambda=0.0, const Array< OneD, Array< OneD, NekDouble > > &Advfield=NullNekDoubleArrayOfArray, bool IsLinearNSEquation=true)
int v_GetForceDimension(void) override
void InfNorm(Array< OneD, Array< OneD, NekDouble > > &inarray, Array< OneD, NekDouble > &outarray)
void v_TransCoeffToPhys(void) override
Virtual function for transformation to physical space.
Array< OneD, CoupledSolverMatrices > m_mat
static std::string solverTypeLookupId
This class is the base class for Navier Stokes problems.
MultiRegions::ExpListSharedPtr m_pressure
Pointer to field holding pressure field.
NekDouble m_kinvis
Kinematic viscosity.
void v_InitObject(bool DeclareField=true) override
Initialisation object for EquationSystem.
ExtrapolateSharedPtr m_extrapolation
Array< OneD, int > m_velocity
int which identifies which components of m_fields contains the velocity (u,v,w);
EquationType m_equationType
equation type;
int m_nConvectiveFields
Number of fields to be convected;.
std::vector< SolverUtils::ForcingSharedPtr > m_forcing
Forcing terms.
void EvaluateAdvectionTerms(const Array< OneD, const Array< OneD, NekDouble > > &inarray, Array< OneD, Array< OneD, NekDouble > > &outarray, const NekDouble time)
Describes the specification for a Basis.
Definition: Basis.h:45
BasisType GetBasisType() const
Return type of expansion basis.
Definition: Basis.h:131
PointsKey GetPointsKey() const
Return distribution of points.
Definition: Basis.h:137
int GetNumModes() const
Returns the order of the basis.
Definition: Basis.h:74
tKey RegisterCreatorFunction(tKey idKey, CreatorFunction classCreator, std::string pDesc="")
Register a class with the factory.
Definition: NekFactory.hpp:197
tBaseSharedPtr CreateInstance(tKey idKey, tParam... args)
Create an instance of the class referred to by idKey.
Definition: NekFactory.hpp:143
static std::string RegisterEnumValue(std::string pEnum, std::string pString, int pEnumValue)
Registers an enumeration value.
void DefineOdeRhs(FuncPointerT func, ObjectPointerT obj)
void DefineImplicitSolve(FuncPointerT func, ObjectPointerT obj)
NekDouble TimePerTest(unsigned int n)
Returns amount of seconds per iteration in a test with n iterations.
Definition: Timer.cpp:65
General purpose memory allocation routines with the ability to allocate from thread specific memory p...
static std::shared_ptr< DataType > AllocateSharedPtr(const Args &...args)
Allocate a shared pointer from the memory pool.
SolverUtils::AdvectionSharedPtr m_advObject
Advection term.
bool m_useFFT
Flag to determine if FFT is used for homogeneous transform.
SpatialDomains::MeshGraphSharedPtr m_graph
Pointer to graph defining mesh.
LibUtilities::CommSharedPtr m_comm
Communicator.
NekDouble m_time
Current time of simulation.
Array< OneD, MultiRegions::ExpListSharedPtr > m_fields
Array holding all dependent variables.
int m_windowPIT
Index of windows for parallel-in-time time iteration.
NekDouble m_lambda
Lambda constant in real system if one required.
int m_npointsZ
number of points in Z direction (if homogeneous)
SOLVER_UTILS_EXPORT void Checkpoint_Output(const int n)
Write checkpoint file of m_fields.
SOLVER_UTILS_EXPORT void WriteFld(const std::string &outname)
Write field data to the given filename.
std::string m_sessionName
Name of the session.
LibUtilities::SessionReaderSharedPtr m_session
The session reader.
bool m_singleMode
Flag to determine if single homogeneous mode is used.
enum HomogeneousType m_HomogeneousType
SOLVER_UTILS_EXPORT int GetNpoints()
bool m_homogen_dealiasing
Flag to determine if dealiasing is used for homogeneous simulations.
SOLVER_UTILS_EXPORT int GetNcoeffs()
SpatialDomains::BoundaryConditionsSharedPtr m_boundaryConditions
Pointer to boundary conditions object.
NekDouble m_LhomZ
physical length in Z direction (if homogeneous)
SOLVER_UTILS_EXPORT void SetInitialConditions(NekDouble initialtime=0.0, bool dumpInitialConditions=true, const int domain=0)
Initialise the data in the dependent fields.
SOLVER_UTILS_EXPORT SessionFunctionSharedPtr GetFunction(std::string name, const MultiRegions::ExpListSharedPtr &field=MultiRegions::NullExpListSharedPtr, bool cache=false)
Get a SessionFunction by name.
SOLVER_UTILS_EXPORT int GetTotPoints()
Base class for unsteady solvers.
LibUtilities::TimeIntegrationSchemeOperators m_ode
The time integration scheme operators to use.
SOLVER_UTILS_EXPORT void v_DoSolve() override
Solves an unsteady problem.
static void Dcopy(const int &n, const double *x, const int &incx, double *y, const int &incy)
BLAS level 1: Copy x to y.
Definition: Blas.hpp:128
static void Dgemm(const char &transa, const char &transb, const int &m, const int &n, const int &k, const double &alpha, const double *a, const int &lda, const double *b, const int &ldb, const double &beta, double *c, const int &ldc)
BLAS level 3: Matrix-matrix multiply C = A x B where op(A)[m x k], op(B)[k x n], C[m x n] DGEMM perfo...
Definition: Blas.hpp:383
std::vector< BasisKey > BasisKeyVector
Name for a vector of BasisKeys.
std::shared_ptr< SessionReader > SessionReaderSharedPtr
@ beta
Gauss Radau pinned at x=-1,.
Definition: PointsType.h:59
const std::vector< NekDouble > velocity
MultiRegions::Direction const DirCartesianMap[]
Definition: ExpList.h:86
std::shared_ptr< ExpList > ExpListSharedPtr
Shared pointer to an ExpList object.
std::shared_ptr< ContField > ContFieldSharedPtr
Definition: ContField.h:268
static const NekDouble kNekUnsetDouble
std::vector< std::pair< std::string, std::string > > SummaryList
Definition: Misc.h:46
EquationSystemFactory & GetEquationSystemFactory()
void AddSummaryItem(SummaryList &l, const std::string &name, const std::string &value)
Adds a summary item to the summary info list.
Definition: Misc.cpp:47
std::shared_ptr< ExpansionInfoMap > ExpansionInfoMapShPtr
Definition: MeshGraph.h:143
std::shared_ptr< ExpansionInfo > ExpansionInfoShPtr
Definition: MeshGraph.h:140
std::shared_ptr< MeshGraph > MeshGraphSharedPtr
Definition: MeshGraph.h:174
std::map< int, ExpansionInfoShPtr > ExpansionInfoMap
Definition: MeshGraph.h:141
std::shared_ptr< StdExpansion > StdExpansionSharedPtr
std::map< ConstFactorType, NekDouble > ConstFactorMap
Definition: StdRegions.hpp:402
StdRegions::ConstFactorMap factors
std::shared_ptr< DNekScalMat > DNekScalMatSharedPtr
@ eSteadyNavierStokes
@ eUnsteadyStokes
@ eUnsteadyNavierStokes
@ eSteadyLinearisedNS
@ eNoEquationType
@ eCoupledLinearisedNS
static Array< OneD, Array< OneD, NekDouble > > NullNekDoubleArrayOfArray
std::shared_ptr< DNekScalBlkMat > DNekScalBlkMatSharedPtr
Definition: NekTypeDefs.hpp:79
ExtrapolateFactory & GetExtrapolateFactory()
Definition: Extrapolate.cpp:48
NekMatrix< InnerMatrixType, BlockMatrixTag > Transpose(NekMatrix< InnerMatrixType, BlockMatrixTag > &rhs)
std::shared_ptr< CoupledLocalToGlobalC0ContMap > CoupledLocalToGlobalC0ContMapSharedPtr
std::shared_ptr< DNekMat > DNekMatSharedPtr
Definition: NekTypeDefs.hpp:75
double NekDouble
void Vmul(int n, const T *x, const int incx, const T *y, const int incy, T *z, const int incz)
Multiply vector z = x*y.
Definition: Vmath.hpp:72
void Neg(int n, T *x, const int incx)
Negate x = -x.
Definition: Vmath.hpp:292
void Vadd(int n, const T *x, const int incx, const T *y, const int incy, T *z, const int incz)
Add vector z = x+y.
Definition: Vmath.hpp:180
void Smul(int n, const T alpha, const T *x, const int incx, T *y, const int incy)
Scalar multiply y = alpha*x.
Definition: Vmath.hpp:100
void Zero(int n, T *x, const int incx)
Zero vector.
Definition: Vmath.hpp:273
void Vsub(int n, const T *x, const int incx, const T *y, const int incy, T *z, const int incz)
Subtract vector z = x-y.
Definition: Vmath.hpp:220
scalarT< T > sqrt(scalarT< T > in)
Definition: scalar.hpp:294
DNekScalBlkMatSharedPtr m_D_int
Inner product of pressure system with divergence of the interior velocity space .
DNekScalBlkMatSharedPtr m_D_bnd
Inner product of pressure system with divergence of the boundary velocity space .
DNekScalBlkMatSharedPtr m_Cinv
Interior-Interior Laplaican plus linearised convective terms inverted, i.e. the inverse of .
DNekScalBlkMatSharedPtr m_Btilde
Interior-boundary Laplacian plus linearised convective terms .
DNekScalBlkMatSharedPtr m_BCinv
Boundary-interior Laplacian plus linearised convective terms pre-multiplying Cinv: .
MultiRegions::GlobalLinSysSharedPtr m_CoupledBndSys