1*149fb536SJames Wright // Copyright (c) 2017-2024, Lawrence Livermore National Security, LLC and other CEED contributors. 2*149fb536SJames Wright // All Rights Reserved. See the top-level LICENSE and NOTICE files for details. 3*149fb536SJames Wright // 4*149fb536SJames Wright // SPDX-License-Identifier: BSD-2-Clause 5*149fb536SJames Wright // 6*149fb536SJames Wright // This file is part of CEED: http://github.com/ceed 7*149fb536SJames Wright 8*149fb536SJames Wright // libCEED + PETSc Example: Navier-Stokes 9*149fb536SJames Wright // 10*149fb536SJames Wright // This example demonstrates a simple usage of libCEED with PETSc to solve a Navier-Stokes problem. 11*149fb536SJames Wright // 12*149fb536SJames Wright // Build with: 13*149fb536SJames Wright // 14*149fb536SJames Wright // make [PETSC_DIR=</path/to/petsc>] [CEED_DIR=</path/to/libceed>] navierstokes 15*149fb536SJames Wright // 16*149fb536SJames Wright // Sample runs: 17*149fb536SJames Wright // 18*149fb536SJames Wright // ./navierstokes -ceed /cpu/self -options_file gaussianwave.yml 19*149fb536SJames Wright // ./navierstokes -ceed /gpu/cuda -problem advection -degree 1 20*149fb536SJames Wright // 21*149fb536SJames Wright //TESTARGS(name="Newtonian and Riemann Solver Unit Tests",only="cpu") -ceed {ceed_resource} -test_type solver -options_file examples/gaussianwave.yaml -compare_final_state_atol 1e100 -compare_final_state_filename tests/output/fluids-navierstokes-gaussianwave-IDL-entropy.bin -dm_plex_box_faces 5,5,1 -ts_max_steps 0 -newtonian_unit_tests -riemann_solver_unit_tests 22*149fb536SJames Wright //TESTARGS(name="Gaussian Wave, IDL and Entropy variables") -ceed {ceed_resource} -test_type solver -options_file examples/gaussianwave.yaml -compare_final_state_atol 2e-11 -compare_final_state_filename tests/output/fluids-navierstokes-gaussianwave-IDL-entropy.bin -state_var entropy -dm_plex_box_faces 5,5,1 -ts_max_steps 5 -idl_decay_time 2e-3 -idl_length 0.25 -idl_start 0 -idl_pressure 70 23*149fb536SJames Wright //TESTARGS(name="Blasius, SGS DataDriven Sequential Torch",only="torch") -ceed {ceed_resource} -options_file tests/blasius_stgtest.yaml -sgs_model_type data_driven -sgs_model_dd_leakyrelu_alpha 0.3 -sgs_model_dd_parameter_dir dd_sgs_data -ts_dt 2e-9 -state_var primitive -ksp_rtol 1e-12 -snes_rtol 1e-12 -stg_mean_only -stg_fluctuating_IC -test_type solver -compare_final_state_atol 1e-10 -compare_final_state_filename tests/output/fluids-navierstokes-blasius-sgs-data-driven.bin -sgs_model_dd_implementation sequential_torch -sgs_model_dd_torch_model_path ./tests/createPyTorchModel/NNModel_HIT_fp64_jit.pt 24*149fb536SJames Wright //TESTARGS(name="Blasius, SGS DataDriven Sequential Ceed") -ceed {ceed_resource} -options_file tests/blasius_stgtest.yaml -sgs_model_type data_driven -sgs_model_dd_leakyrelu_alpha 0.3 -sgs_model_dd_parameter_dir dd_sgs_data -ts_dt 2e-9 -state_var primitive -ksp_rtol 1e-12 -snes_rtol 1e-12 -stg_mean_only -stg_fluctuating_IC -test_type solver -compare_final_state_atol 1e-10 -compare_final_state_filename tests/output/fluids-navierstokes-blasius-sgs-data-driven.bin -sgs_model_dd_implementation sequential_ceed 25*149fb536SJames Wright //TESTARGS(name="Gaussian Wave, explicit, supg, IDL") -ceed {ceed_resource} -test_type solver -options_file examples/gaussianwave.yaml -compare_final_state_atol 1e-8 -compare_final_state_filename tests/output/fluids-navierstokes-gaussianwave-explicit.bin -dm_plex_box_faces 2,2,1 -ts_max_steps 5 -degree 3 -implicit false -ts_type rk -stab supg -state_var conservative -mass_ksp_type gmres -mass_pc_jacobi_type diagonal -idl_decay_time 2e-3 -idl_length 0.25 -idl_start 0 -idl_pressure 70 26*149fb536SJames Wright //TESTARGS(name="Advection 2D, rotation, explicit, supg, consistent mass") -ceed {ceed_resource} -test_type solver -problem advection -degree 3 -dm_plex_box_faces 2,2 -dm_plex_box_lower 0,0 -dm_plex_box_upper 125,125 -bc_wall 1,2,3,4 -wall_comps 4 -units_kilogram 1e-9 -rc 100. -ts_dt 1e-3 -ts_max_steps 10 -stab supg -Ctaus 0.5 -mass_ksp_type gmres -mass_pc_type vpbjacobi -compare_final_state_atol 1e-10 -compare_final_state_filename tests/output/fluids-navierstokes-adv2d-rotation-explicit-stab-supg-consistent-mass.bin 27*149fb536SJames Wright //TESTARGS(name="Advection, skew") -ceed {ceed_resource} -test_type solver -options_file examples/advection.yaml -ts_max_steps 5 -wind_type translation -wind_translation -0.5547002,0.83205029,0 -advection_ic_type skew -dm_plex_box_faces 2,1,1 -degree 2 -stab supg -stab_tau advdiff_shakib -Ctau_a 4 -ksp_type gmres -diffusion_coeff 5e-4 -compare_final_state_atol 7e-10 -compare_final_state_filename tests/output/fluids-navierstokes-adv-skew.bin 28*149fb536SJames Wright //TESTARGS(name="Blasius, bc_slip") -ceed {ceed_resource} -test_type solver -options_file examples/blasius.yaml -ts_max_steps 5 -dm_plex_box_faces 3,20,1 -platemesh_nDelta 10 -platemesh_growth 1.2 -bc_outflow 5 -bc_slip 4 -compare_final_state_atol 2E-11 -compare_final_state_filename tests/output/fluids-navierstokes-blasius-bc_slip.bin 29*149fb536SJames Wright //TESTARGS(name="Advection, rotation, cosine") -ceed {ceed_resource} -test_type solver -options_file examples/advection.yaml -ts_max_steps 0 -advection_ic_type cosine_hill -dm_plex_box_faces 2,1,1 -compare_final_state_atol 1e-10 -compare_final_state_filename tests/output/fluids-navierstokes-adv-rotation-cosine.bin 30*149fb536SJames Wright //TESTARGS(name="Gaussian Wave, using MatShell") -ceed {ceed_resource} -test_type solver -options_file examples/gaussianwave.yaml -compare_final_state_atol 1e-8 -compare_final_state_filename tests/output/fluids-navierstokes-gaussianwave-shell.bin -dm_plex_box_faces 2,2,1 -ts_max_steps 5 -degree 3 -amat_type shell -pc_type vpbjacobi -ts_alpha_radius 0.5 31*149fb536SJames Wright //TESTARGS(name="Taylor-Green Vortex IC") -ceed {ceed_resource} -problem taylor_green -test_type solver -dm_plex_dim 3 -dm_plex_box_faces 6,6,6 -ts_max_steps 0 -compare_final_state_atol 1e-12 -compare_final_state_filename tests/output/fluids-navierstokes-taylor-green-IC.bin 32*149fb536SJames Wright //TESTARGS(name="Blasius, SGS DataDriven Fused") -ceed {ceed_resource} -options_file tests/blasius_stgtest.yaml -sgs_model_type data_driven -sgs_model_dd_leakyrelu_alpha 0.3 -sgs_model_dd_parameter_dir dd_sgs_data -ts_dt 2e-9 -state_var primitive -ksp_rtol 1e-12 -snes_rtol 1e-12 -stg_mean_only -stg_fluctuating_IC -test_type solver -compare_final_state_atol 1e-10 -compare_final_state_filename tests/output/fluids-navierstokes-blasius-sgs-data-driven.bin 33*149fb536SJames Wright //TESTARGS(name="Blasius, Anisotropic Differential Filter") -ceed {ceed_resource} -test_type diff_filter -options_file tests/blasius_test.yaml -compare_final_state_atol 5e-10 -compare_final_state_filename tests/output/fluids-navierstokes-blasius_diff_filter_aniso_vandriest.bin -diff_filter_monitor -ts_max_steps 0 -state_var primitive -diff_filter_friction_length 1e-5 -diff_filter_wall_damping_function van_driest -diff_filter_ksp_rtol 1e-8 -diff_filter_grid_based_width -diff_filter_width_scaling 1,0.7,1 34*149fb536SJames Wright //TESTARGS(name="Blasius, Isotropic Differential Filter") -ceed {ceed_resource} -test_type diff_filter -options_file tests/blasius_test.yaml -compare_final_state_atol 2e-12 -compare_final_state_filename tests/output/fluids-navierstokes-blasius_diff_filter_iso.bin -diff_filter_monitor -ts_max_steps 0 -diff_filter_width_scaling 4.2e-5,4.2e-5,4.2e-5 -diff_filter_ksp_atol 1e-14 -diff_filter_ksp_rtol 1e-16 35*149fb536SJames Wright //TESTARGS(name="Gaussian Wave, with IDL") -ceed {ceed_resource} -test_type solver -options_file examples/gaussianwave.yaml -compare_final_state_atol 2e-11 -compare_final_state_filename tests/output/fluids-navierstokes-gaussianwave-IDL.bin -dm_plex_box_faces 5,5,1 -ts_max_steps 5 -idl_decay_time 2e-3 -idl_length 0.25 -idl_start 0 -ts_alpha_radius 0.5 -idl_pressure 70 36*149fb536SJames Wright //TESTARGS(name="Spanwise Turbulence Statistics") -ceed {ceed_resource} -test_type turb_spanstats -options_file tests/stats_test.yaml -compare_final_state_atol 1E-11 -compare_final_state_filename tests/output/fluids-navierstokes-turb-spanstats-stats.bin 37*149fb536SJames Wright //TESTARGS(name="Blasius") -ceed {ceed_resource} -test_type solver -options_file tests/blasius_test.yaml -compare_final_state_atol 2E-11 -compare_final_state_filename tests/output/fluids-navierstokes-blasius.bin 38*149fb536SJames Wright //TESTARGS(name="Blasius, STG Inflow") -ceed {ceed_resource} -test_type solver -options_file tests/blasius_stgtest.yaml -compare_final_state_atol 2E-11 -compare_final_state_filename tests/output/fluids-navierstokes-blasius_STG.bin 39*149fb536SJames Wright //TESTARGS(name="Blasius, STG Inflow, Weak Temperature") -ceed {ceed_resource} -test_type solver -options_file tests/blasius_stgtest.yaml -compare_final_state_atol 1E-11 -compare_final_state_filename tests/output/fluids-navierstokes-blasius_STG_weakT.bin -weakT 40*149fb536SJames Wright //TESTARGS(name="Blasius, Strong STG Inflow") -ceed {ceed_resource} -test_type solver -options_file tests/blasius_stgtest.yaml -compare_final_state_atol 1E-10 -compare_final_state_filename tests/output/fluids-navierstokes-blasius_STG_strongBC.bin -stg_strong true 41*149fb536SJames Wright //TESTARGS(name="Channel") -ceed {ceed_resource} -test_type solver -options_file examples/channel.yaml -compare_final_state_atol 2e-11 -compare_final_state_filename tests/output/fluids-navierstokes-channel.bin -dm_plex_box_faces 5,5,1 -ts_max_steps 5 42*149fb536SJames Wright //TESTARGS(name="Channel, Primitive") -ceed {ceed_resource} -test_type solver -options_file examples/channel.yaml -compare_final_state_atol 2e-11 -compare_final_state_filename tests/output/fluids-navierstokes-channel-prim.bin -dm_plex_box_faces 5,5,1 -ts_max_steps 5 -state_var primitive 43*149fb536SJames Wright //TESTARGS(name="Density Current, explicit") -ceed {ceed_resource} -test_type solver -degree 3 -q_extra 2 -dm_plex_box_faces 1,1,2 -dm_plex_box_lower 0,0,0 -dm_plex_box_upper 125,125,250 -dm_plex_dim 3 -bc_symmetry_x 5,6 -bc_symmetry_y 3,4 -bc_symmetry_z 1,2 -units_kilogram 1e-9 -center 62.5,62.5,187.5 -rc 100. -thetaC -35. -mu 75 -gravity 0,0,-9.81 -ts_dt 1e-3 -units_meter 1e-2 -units_second 1e-2 -ts_max_steps 10 -compare_final_state_atol 1E-11 -compare_final_state_filename tests/output/fluids-navierstokes-dc-explicit.bin 44*149fb536SJames Wright //TESTARGS(name="Density Current, implicit, no stabilization") -ceed {ceed_resource} -test_type solver -degree 3 -dm_plex_box_faces 1,1,2 -dm_plex_box_lower 0,0,0 -dm_plex_box_upper 125,125,250 -dm_plex_dim 3 -bc_symmetry_x 5,6 -bc_symmetry_y 3,4 -bc_symmetry_z 1,2 -units_kilogram 1e-9 -center 62.5,62.5,187.5 -rc 100. -thetaC -35. -mu 75 -gravity 0,0,-9.81 -units_meter 1e-2 -units_second 1e-2 -ksp_atol 1e-4 -ksp_rtol 1e-3 -ksp_type bcgs -snes_atol 1e-3 -snes_lag_jacobian 100 -snes_lag_jacobian_persists -snes_mf_operator -ts_dt 1e-3 -implicit -ts_type alpha -ts_max_steps 10 -compare_final_state_atol 5E-4 -compare_final_state_filename tests/output/fluids-navierstokes-dc-implicit-stab-none.bin 45*149fb536SJames Wright //TESTARGS(name="Advection, rotation, implicit, SUPG stabilization") -ceed {ceed_resource} -test_type solver -problem advection -CtauS .3 -stab supg -degree 3 -dm_plex_box_faces 1,1,2 -dm_plex_box_lower 0,0,0 -dm_plex_box_upper 125,125,250 -dm_plex_dim 3 -bc_wall 1,2,3,4,5,6 -wall_comps 4 -units_kilogram 1e-9 -rc 100. -ksp_atol 1e-4 -ksp_rtol 1e-3 -ksp_type bcgs -snes_atol 1e-3 -snes_lag_jacobian 100 -snes_lag_jacobian_persists -snes_mf_operator -ts_dt 1e-3 -implicit -dm_mat_preallocate_skip 0 -ts_type alpha -compare_final_state_atol 5E-4 -ts_max_steps 10 -compare_final_state_filename tests/output/fluids-navierstokes-adv-rotation-implicit-stab-supg.bin 46*149fb536SJames Wright //TESTARGS(name="Advection, translation, implicit, SU stabilization") -ceed {ceed_resource} -test_type solver -problem advection -CtauS .3 -stab su -degree 3 -dm_plex_box_faces 1,1,2 -dm_plex_box_lower 0,0,0 -dm_plex_box_upper 125,125,250 -dm_plex_dim 3 -units_kilogram 1e-9 -rc 100. -ksp_atol 1e-4 -ksp_rtol 1e-3 -ksp_type bcgs -snes_atol 1e-3 -snes_lag_jacobian 100 -snes_lag_jacobian_persists -snes_mf_operator -ts_dt 1e-3 -implicit -dm_mat_preallocate_skip 0 -ts_type alpha -wind_type translation -wind_translation .53,-1.33,-2.65 -bc_inflow 1,2,3,4,5,6 -ts_max_steps 10 -compare_final_state_atol 5E-4 -compare_final_state_filename tests/output/fluids-navierstokes-adv-translation-implicit-stab-su.bin 47*149fb536SJames Wright //TESTARGS(name="Advection 2D, rotation, explicit, strong form") -ceed {ceed_resource} -test_type solver -problem advection -strong_form 1 -degree 3 -dm_plex_box_faces 2,2 -dm_plex_box_lower 0,0 -dm_plex_box_upper 125,125 -bc_wall 1,2,3,4 -wall_comps 4 -units_kilogram 1e-9 -rc 100. -ts_dt 1e-3 -compare_final_state_atol 5E-11 -ts_max_steps 10 -compare_final_state_filename tests/output/fluids-navierstokes-adv2d-rotation-explicit-strong.bin 48*149fb536SJames Wright //TESTARGS(name="Advection 2D, rotation, implicit, SUPG stabilization") -ceed {ceed_resource} -test_type solver -problem advection -CtauS .3 -stab supg -degree 3 -dm_plex_box_faces 1,1,2 -dm_plex_box_lower 0,0 -dm_plex_box_upper 125,125 -bc_wall 1,2,3,4 -wall_comps 4 -units_kilogram 1e-9 -rc 100. -ksp_atol 1e-4 -ksp_rtol 1e-3 -ksp_type bcgs -snes_atol 1e-3 -snes_lag_jacobian 100 -snes_lag_jacobian_persists -snes_mf_operator -ts_dt 1e-3 -implicit -dm_mat_preallocate_skip 0 -ts_type alpha -ts_max_steps 10 -compare_final_state_atol 5E-4 -compare_final_state_filename tests/output/fluids-navierstokes-adv2d-rotation-implicit-stab-supg.bin 49*149fb536SJames Wright //TESTARGS(name="Euler, implicit") -ceed {ceed_resource} -test_type solver -problem euler_vortex -degree 3 -dm_plex_box_faces 1,1,2 -dm_plex_box_lower 0,0,0 -dm_plex_box_upper 125,125,250 -dm_plex_dim 3 -units_meter 1e-4 -units_second 1e-4 -mean_velocity 1.4,-2.,0 -bc_inflow 4,6 -bc_outflow 3,5 -bc_symmetry_z 1,2 -vortex_strength 2 -ksp_atol 1e-4 -ksp_rtol 1e-3 -ksp_type bcgs -snes_atol 1e-3 -snes_lag_jacobian 100 -snes_lag_jacobian_persists -snes_mf_operator -ts_dt 1e-3 -implicit -dm_mat_preallocate_skip 0 -ts_type alpha -ts_max_steps 10 -compare_final_state_atol 5E-4 -compare_final_state_filename tests/output/fluids-navierstokes-euler-implicit.bin 50*149fb536SJames Wright //TESTARGS(name="Euler, explicit") -ceed {ceed_resource} -test_type solver -problem euler_vortex -degree 3 -q_extra 2 -dm_plex_box_faces 2,2,1 -dm_plex_box_lower 0,0,0 -dm_plex_box_upper 125,125,250 -dm_plex_dim 3 -units_meter 1e-4 -units_second 1e-4 -mean_velocity 1.4,-2.,0 -bc_inflow 4,6 -bc_outflow 3,5 -bc_symmetry_z 1,2 -vortex_strength 2 -ts_dt 1e-7 -ts_rk_type 5bs -ts_rtol 1e-10 -ts_atol 1e-10 -ts_max_steps 10 -compare_final_state_atol 1E-7 -compare_final_state_filename tests/output/fluids-navierstokes-euler-explicit.bin 51*149fb536SJames Wright //TESTARGS(name="Sod Shocktube, explicit, SU stabilization, y-z-beta shock capturing") -ceed {ceed_resource} -test_type solver -problem shocktube -degree 1 -q_extra 2 -dm_plex_box_faces 50,1,1 -units_meter 1e-2 units_second 1e-2 -dm_plex_box_lower 0,0,0 -dm_plex_box_upper 1000,20,20 -dm_plex_dim 3 -bc_symmetry_x 5,6 -bc_symmetry_y 3,4 -bc_symmetry_z 1,2 -yzb -stab su -ts_max_steps 10 -compare_final_state_atol 1E-11 -compare_final_state_filename tests/output/fluids-navierstokes-shocktube-explicit-su-yzb.bin 52*149fb536SJames Wright 53*149fb536SJames Wright /// @file 54*149fb536SJames Wright /// Navier-Stokes example using PETSc 55*149fb536SJames Wright 56*149fb536SJames Wright const char help[] = "Solve Navier-Stokes using PETSc and libCEED\n"; 57*149fb536SJames Wright 58*149fb536SJames Wright #include <navierstokes.h> 59*149fb536SJames Wright #include <petscdevice.h> 60*149fb536SJames Wright 61*149fb536SJames Wright #include <ceed.h> 62*149fb536SJames Wright #include <petscdmplex.h> 63*149fb536SJames Wright #include <petscts.h> 64*149fb536SJames Wright 65*149fb536SJames Wright int main(int argc, char **argv) { 66*149fb536SJames Wright // --------------------------------------------------------------------------- 67*149fb536SJames Wright // Initialize PETSc 68*149fb536SJames Wright // --------------------------------------------------------------------------- 69*149fb536SJames Wright PetscCall(PetscInitialize(&argc, &argv, NULL, help)); 70*149fb536SJames Wright 71*149fb536SJames Wright // --------------------------------------------------------------------------- 72*149fb536SJames Wright // Create structs 73*149fb536SJames Wright // --------------------------------------------------------------------------- 74*149fb536SJames Wright AppCtx app_ctx; 75*149fb536SJames Wright PetscCall(PetscCalloc1(1, &app_ctx)); 76*149fb536SJames Wright 77*149fb536SJames Wright ProblemData problem; 78*149fb536SJames Wright PetscCall(PetscCalloc1(1, &problem)); 79*149fb536SJames Wright 80*149fb536SJames Wright User user; 81*149fb536SJames Wright PetscCall(PetscCalloc1(1, &user)); 82*149fb536SJames Wright 83*149fb536SJames Wright CeedData ceed_data; 84*149fb536SJames Wright PetscCall(PetscCalloc1(1, &ceed_data)); 85*149fb536SJames Wright 86*149fb536SJames Wright SimpleBC bc; 87*149fb536SJames Wright PetscCall(PetscCalloc1(1, &bc)); 88*149fb536SJames Wright 89*149fb536SJames Wright Physics phys_ctx; 90*149fb536SJames Wright PetscCall(PetscCalloc1(1, &phys_ctx)); 91*149fb536SJames Wright 92*149fb536SJames Wright Units units; 93*149fb536SJames Wright PetscCall(PetscCalloc1(1, &units)); 94*149fb536SJames Wright 95*149fb536SJames Wright user->app_ctx = app_ctx; 96*149fb536SJames Wright user->units = units; 97*149fb536SJames Wright user->phys = phys_ctx; 98*149fb536SJames Wright problem->set_bc_from_ics = PETSC_TRUE; 99*149fb536SJames Wright 100*149fb536SJames Wright PetscCall(RegisterLogEvents()); 101*149fb536SJames Wright 102*149fb536SJames Wright // --------------------------------------------------------------------------- 103*149fb536SJames Wright // Process command line options 104*149fb536SJames Wright // --------------------------------------------------------------------------- 105*149fb536SJames Wright // -- Register problems to be available on the command line 106*149fb536SJames Wright PetscCall(RegisterProblems_NS(app_ctx)); 107*149fb536SJames Wright 108*149fb536SJames Wright // -- Process general command line options 109*149fb536SJames Wright MPI_Comm comm = PETSC_COMM_WORLD; 110*149fb536SJames Wright user->comm = comm; 111*149fb536SJames Wright PetscCall(ProcessCommandLineOptions(comm, app_ctx, bc)); 112*149fb536SJames Wright PetscCall(BoundaryConditionSetUp(user, problem, app_ctx, bc)); 113*149fb536SJames Wright 114*149fb536SJames Wright // --------------------------------------------------------------------------- 115*149fb536SJames Wright // Initialize libCEED 116*149fb536SJames Wright // --------------------------------------------------------------------------- 117*149fb536SJames Wright // -- Initialize backend 118*149fb536SJames Wright Ceed ceed; 119*149fb536SJames Wright PetscCheck(CeedInit(app_ctx->ceed_resource, &ceed) == CEED_ERROR_SUCCESS, comm, PETSC_ERR_LIB, "Ceed initialization failed"); 120*149fb536SJames Wright user->ceed = ceed; 121*149fb536SJames Wright 122*149fb536SJames Wright PetscCheck(CeedSetErrorHandler(ceed, CeedErrorStore) == CEED_ERROR_SUCCESS, comm, PETSC_ERR_LIB, "Setting libCEED error handler failed"); 123*149fb536SJames Wright 124*149fb536SJames Wright // -- Check preferred MemType 125*149fb536SJames Wright CeedMemType mem_type_backend; 126*149fb536SJames Wright PetscCallCeed(ceed, CeedGetPreferredMemType(ceed, &mem_type_backend)); 127*149fb536SJames Wright 128*149fb536SJames Wright { 129*149fb536SJames Wright const char *resource; 130*149fb536SJames Wright PetscCallCeed(ceed, CeedGetResource(ceed, &resource)); 131*149fb536SJames Wright if (strstr(resource, "/gpu/sycl")) { 132*149fb536SJames Wright PetscDeviceContext dctx; 133*149fb536SJames Wright PetscCall(PetscDeviceContextGetCurrentContext(&dctx)); 134*149fb536SJames Wright void *stream_handle; 135*149fb536SJames Wright PetscCall(PetscDeviceContextGetStreamHandle(dctx, &stream_handle)); 136*149fb536SJames Wright PetscCallCeed(ceed, CeedSetStream(ceed, stream_handle)); 137*149fb536SJames Wright } 138*149fb536SJames Wright } 139*149fb536SJames Wright 140*149fb536SJames Wright // --------------------------------------------------------------------------- 141*149fb536SJames Wright // Set up global mesh 142*149fb536SJames Wright // --------------------------------------------------------------------------- 143*149fb536SJames Wright // -- Create DM 144*149fb536SJames Wright DM dm; 145*149fb536SJames Wright VecType vec_type = NULL; 146*149fb536SJames Wright MatType mat_type = NULL; 147*149fb536SJames Wright switch (mem_type_backend) { 148*149fb536SJames Wright case CEED_MEM_HOST: 149*149fb536SJames Wright vec_type = VECSTANDARD; 150*149fb536SJames Wright break; 151*149fb536SJames Wright case CEED_MEM_DEVICE: { 152*149fb536SJames Wright const char *resolved; 153*149fb536SJames Wright PetscCallCeed(ceed, CeedGetResource(ceed, &resolved)); 154*149fb536SJames Wright if (strstr(resolved, "/gpu/cuda")) vec_type = VECCUDA; 155*149fb536SJames Wright else if (strstr(resolved, "/gpu/hip")) vec_type = VECKOKKOS; 156*149fb536SJames Wright else if (strstr(resolved, "/gpu/sycl")) vec_type = VECKOKKOS; 157*149fb536SJames Wright else vec_type = VECSTANDARD; 158*149fb536SJames Wright } 159*149fb536SJames Wright } 160*149fb536SJames Wright if (strstr(vec_type, VECCUDA)) mat_type = MATAIJCUSPARSE; 161*149fb536SJames Wright else if (strstr(vec_type, VECKOKKOS)) mat_type = MATAIJKOKKOS; 162*149fb536SJames Wright else mat_type = MATAIJ; 163*149fb536SJames Wright PetscCall(CreateDM(comm, problem, mat_type, vec_type, &dm)); 164*149fb536SJames Wright user->dm = dm; 165*149fb536SJames Wright PetscCall(DMSetApplicationContext(dm, user)); 166*149fb536SJames Wright 167*149fb536SJames Wright // --------------------------------------------------------------------------- 168*149fb536SJames Wright // Choose the problem from the list of registered problems 169*149fb536SJames Wright // --------------------------------------------------------------------------- 170*149fb536SJames Wright { 171*149fb536SJames Wright PetscErrorCode (*p)(ProblemData, DM, void *, SimpleBC); 172*149fb536SJames Wright PetscCall(PetscFunctionListFind(app_ctx->problems, app_ctx->problem_name, &p)); 173*149fb536SJames Wright PetscCheck(p, PETSC_COMM_SELF, 1, "Problem '%s' not found", app_ctx->problem_name); 174*149fb536SJames Wright PetscCall((*p)(problem, dm, &user, bc)); 175*149fb536SJames Wright } 176*149fb536SJames Wright 177*149fb536SJames Wright // -- Set up DM 178*149fb536SJames Wright PetscCall(SetUpDM(dm, problem, app_ctx->degree, app_ctx->q_extra, bc, phys_ctx)); 179*149fb536SJames Wright 180*149fb536SJames Wright // -- Refine DM for high-order viz 181*149fb536SJames Wright if (app_ctx->viz_refine) PetscCall(VizRefineDM(dm, user, problem, bc, phys_ctx)); 182*149fb536SJames Wright 183*149fb536SJames Wright // --------------------------------------------------------------------------- 184*149fb536SJames Wright // Create solution vectors 185*149fb536SJames Wright // --------------------------------------------------------------------------- 186*149fb536SJames Wright // -- Set up global state vector Q 187*149fb536SJames Wright Vec Q; 188*149fb536SJames Wright PetscCall(DMCreateGlobalVector(dm, &Q)); 189*149fb536SJames Wright PetscCall(VecZeroEntries(Q)); 190*149fb536SJames Wright 191*149fb536SJames Wright // -- Set up local state vectors Q_loc, Q_dot_loc 192*149fb536SJames Wright PetscCall(DMCreateLocalVector(dm, &user->Q_loc)); 193*149fb536SJames Wright PetscCall(DMCreateLocalVector(dm, &user->Q_dot_loc)); 194*149fb536SJames Wright PetscCall(VecZeroEntries(user->Q_dot_loc)); 195*149fb536SJames Wright 196*149fb536SJames Wright // --------------------------------------------------------------------------- 197*149fb536SJames Wright // Set up libCEED 198*149fb536SJames Wright // --------------------------------------------------------------------------- 199*149fb536SJames Wright // -- Set up libCEED objects 200*149fb536SJames Wright PetscCall(SetupLibceed(ceed, ceed_data, dm, user, app_ctx, problem, bc)); 201*149fb536SJames Wright 202*149fb536SJames Wright // --------------------------------------------------------------------------- 203*149fb536SJames Wright // Set up ICs 204*149fb536SJames Wright // --------------------------------------------------------------------------- 205*149fb536SJames Wright // -- Fix multiplicity for ICs 206*149fb536SJames Wright PetscCall(ICs_FixMultiplicity(dm, ceed_data, user, user->Q_loc, Q, 0.0)); 207*149fb536SJames Wright 208*149fb536SJames Wright // --------------------------------------------------------------------------- 209*149fb536SJames Wright // Record boundary values from initial condition 210*149fb536SJames Wright // --------------------------------------------------------------------------- 211*149fb536SJames Wright // -- This overrides DMPlexInsertBoundaryValues(). 212*149fb536SJames Wright // We use this for the main simulation DM because the reference DMPlexInsertBoundaryValues() is very slow on the GPU due to extra device-to-host 213*149fb536SJames Wright // communication. If we disable this, we should still get the same results due to the problem->bc function, but with potentially much slower 214*149fb536SJames Wright // execution. 215*149fb536SJames Wright if (problem->set_bc_from_ics) { 216*149fb536SJames Wright PetscCall(SetBCsFromICs(dm, Q, user->Q_loc)); 217*149fb536SJames Wright } 218*149fb536SJames Wright 219*149fb536SJames Wright // --------------------------------------------------------------------------- 220*149fb536SJames Wright // Create output directory 221*149fb536SJames Wright // --------------------------------------------------------------------------- 222*149fb536SJames Wright PetscMPIInt rank; 223*149fb536SJames Wright MPI_Comm_rank(comm, &rank); 224*149fb536SJames Wright if (!rank) { 225*149fb536SJames Wright PetscCall(PetscMkdir(app_ctx->output_dir)); 226*149fb536SJames Wright } 227*149fb536SJames Wright 228*149fb536SJames Wright // --------------------------------------------------------------------------- 229*149fb536SJames Wright // Gather initial Q values in case of continuation of simulation 230*149fb536SJames Wright // --------------------------------------------------------------------------- 231*149fb536SJames Wright // -- Set up initial values from binary file 232*149fb536SJames Wright if (app_ctx->cont_steps) { 233*149fb536SJames Wright PetscCall(SetupICsFromBinary(comm, app_ctx, Q)); 234*149fb536SJames Wright } 235*149fb536SJames Wright 236*149fb536SJames Wright // -- Zero Q_loc 237*149fb536SJames Wright PetscCall(VecZeroEntries(user->Q_loc)); 238*149fb536SJames Wright 239*149fb536SJames Wright // --------------------------------------------------------------------------- 240*149fb536SJames Wright // TS: Create, setup, and solve 241*149fb536SJames Wright // --------------------------------------------------------------------------- 242*149fb536SJames Wright TS ts; 243*149fb536SJames Wright PetscScalar final_time; 244*149fb536SJames Wright PetscCall(TSSolve_NS(dm, user, app_ctx, phys_ctx, problem, &Q, &final_time, &ts)); 245*149fb536SJames Wright 246*149fb536SJames Wright // --------------------------------------------------------------------------- 247*149fb536SJames Wright // Post-processing 248*149fb536SJames Wright // --------------------------------------------------------------------------- 249*149fb536SJames Wright PetscCall(PostProcess(ts, ceed_data, dm, problem, user, Q, final_time)); 250*149fb536SJames Wright 251*149fb536SJames Wright // --------------------------------------------------------------------------- 252*149fb536SJames Wright // Destroy libCEED objects 253*149fb536SJames Wright // --------------------------------------------------------------------------- 254*149fb536SJames Wright 255*149fb536SJames Wright PetscCall(TurbulenceStatisticsDestroy(user, ceed_data)); 256*149fb536SJames Wright PetscCall(NodalProjectionDataDestroy(user->grad_velo_proj)); 257*149fb536SJames Wright PetscCall(SgsDDDataDestroy(user->sgs_dd_data)); 258*149fb536SJames Wright PetscCall(DifferentialFilterDataDestroy(user->diff_filter)); 259*149fb536SJames Wright PetscCall(SGS_DD_TrainingDataDestroy(user->sgs_dd_train)); 260*149fb536SJames Wright PetscCall(SmartSimDataDestroy(user->smartsim)); 261*149fb536SJames Wright 262*149fb536SJames Wright // -- Vectors 263*149fb536SJames Wright PetscCallCeed(ceed, CeedVectorDestroy(&ceed_data->x_coord)); 264*149fb536SJames Wright PetscCallCeed(ceed, CeedVectorDestroy(&ceed_data->q_data)); 265*149fb536SJames Wright PetscCallCeed(ceed, CeedVectorDestroy(&user->q_ceed)); 266*149fb536SJames Wright PetscCallCeed(ceed, CeedVectorDestroy(&user->q_dot_ceed)); 267*149fb536SJames Wright PetscCallCeed(ceed, CeedVectorDestroy(&user->g_ceed)); 268*149fb536SJames Wright 269*149fb536SJames Wright // -- Bases 270*149fb536SJames Wright PetscCallCeed(ceed, CeedBasisDestroy(&ceed_data->basis_q)); 271*149fb536SJames Wright PetscCallCeed(ceed, CeedBasisDestroy(&ceed_data->basis_x)); 272*149fb536SJames Wright 273*149fb536SJames Wright // -- Restrictions 274*149fb536SJames Wright PetscCallCeed(ceed, CeedElemRestrictionDestroy(&ceed_data->elem_restr_q)); 275*149fb536SJames Wright PetscCallCeed(ceed, CeedElemRestrictionDestroy(&ceed_data->elem_restr_x)); 276*149fb536SJames Wright PetscCallCeed(ceed, CeedElemRestrictionDestroy(&ceed_data->elem_restr_qd_i)); 277*149fb536SJames Wright 278*149fb536SJames Wright // Destroy QFunction contexts after using 279*149fb536SJames Wright // ToDo: Simplify tracked libCEED objects, smaller struct 280*149fb536SJames Wright { 281*149fb536SJames Wright PetscCallCeed(ceed, CeedQFunctionContextDestroy(&problem->apply_inflow.qfunction_context)); 282*149fb536SJames Wright PetscCallCeed(ceed, CeedQFunctionContextDestroy(&problem->apply_inflow_jacobian.qfunction_context)); 283*149fb536SJames Wright PetscCallCeed(ceed, CeedQFunctionContextDestroy(&problem->apply_outflow.qfunction_context)); 284*149fb536SJames Wright PetscCallCeed(ceed, CeedQFunctionContextDestroy(&problem->apply_outflow_jacobian.qfunction_context)); 285*149fb536SJames Wright PetscCallCeed(ceed, CeedQFunctionContextDestroy(&problem->apply_freestream.qfunction_context)); 286*149fb536SJames Wright PetscCallCeed(ceed, CeedQFunctionContextDestroy(&problem->apply_freestream_jacobian.qfunction_context)); 287*149fb536SJames Wright PetscCallCeed(ceed, CeedQFunctionContextDestroy(&problem->apply_slip.qfunction_context)); 288*149fb536SJames Wright PetscCallCeed(ceed, CeedQFunctionContextDestroy(&problem->apply_slip_jacobian.qfunction_context)); 289*149fb536SJames Wright PetscCallCeed(ceed, CeedQFunctionContextDestroy(&problem->ics.qfunction_context)); 290*149fb536SJames Wright PetscCallCeed(ceed, CeedQFunctionContextDestroy(&problem->apply_vol_rhs.qfunction_context)); 291*149fb536SJames Wright PetscCallCeed(ceed, CeedQFunctionContextDestroy(&problem->apply_vol_ifunction.qfunction_context)); 292*149fb536SJames Wright PetscCallCeed(ceed, CeedQFunctionContextDestroy(&problem->apply_vol_ijacobian.qfunction_context)); 293*149fb536SJames Wright } 294*149fb536SJames Wright 295*149fb536SJames Wright // -- Operators 296*149fb536SJames Wright PetscCall(OperatorApplyContextDestroy(ceed_data->op_ics_ctx)); 297*149fb536SJames Wright PetscCall(OperatorApplyContextDestroy(user->op_rhs_ctx)); 298*149fb536SJames Wright PetscCall(OperatorApplyContextDestroy(user->op_strong_bc_ctx)); 299*149fb536SJames Wright PetscCallCeed(ceed, CeedOperatorDestroy(&user->op_ifunction)); 300*149fb536SJames Wright 301*149fb536SJames Wright // -- Ceed 302*149fb536SJames Wright PetscCheck(CeedDestroy(&ceed) == CEED_ERROR_SUCCESS, comm, PETSC_ERR_LIB, "Destroying Ceed object failed"); 303*149fb536SJames Wright 304*149fb536SJames Wright if (app_ctx->test_type != TESTTYPE_NONE) { 305*149fb536SJames Wright PetscInt num_options_left = 0; 306*149fb536SJames Wright PetscCall(PetscOptionsLeftGet(NULL, &num_options_left, NULL, NULL)); 307*149fb536SJames Wright PetscCheck(num_options_left == 0, PETSC_COMM_WORLD, -1, 308*149fb536SJames Wright "There are unused options. This is not allowed. See error message for the unused options (or use -options_left directly)"); 309*149fb536SJames Wright } 310*149fb536SJames Wright 311*149fb536SJames Wright // --------------------------------------------------------------------------- 312*149fb536SJames Wright // Clean up PETSc 313*149fb536SJames Wright // --------------------------------------------------------------------------- 314*149fb536SJames Wright // -- Vectors 315*149fb536SJames Wright PetscCall(VecDestroy(&Q)); 316*149fb536SJames Wright PetscCall(VecDestroy(&user->Q_loc)); 317*149fb536SJames Wright PetscCall(VecDestroy(&user->Q_dot_loc)); 318*149fb536SJames Wright 319*149fb536SJames Wright PetscCall(KSPDestroy(&user->mass_ksp)); 320*149fb536SJames Wright 321*149fb536SJames Wright // -- Matrices 322*149fb536SJames Wright PetscCall(MatDestroy(&user->interp_viz)); 323*149fb536SJames Wright PetscCall(MatDestroy(&user->mat_ijacobian)); 324*149fb536SJames Wright 325*149fb536SJames Wright // -- DM 326*149fb536SJames Wright PetscCall(DMDestroy(&dm)); 327*149fb536SJames Wright PetscCall(DMDestroy(&user->dm_viz)); 328*149fb536SJames Wright 329*149fb536SJames Wright // -- TS 330*149fb536SJames Wright PetscCall(TSDestroy(&ts)); 331*149fb536SJames Wright 332*149fb536SJames Wright // -- Function list 333*149fb536SJames Wright PetscCall(PetscFunctionListDestroy(&app_ctx->problems)); 334*149fb536SJames Wright 335*149fb536SJames Wright PetscCall(PetscFree(app_ctx->amat_type)); 336*149fb536SJames Wright PetscCall(PetscFree(app_ctx->wall_forces.walls)); 337*149fb536SJames Wright PetscCall(PetscViewerDestroy(&app_ctx->wall_forces.viewer)); 338*149fb536SJames Wright 339*149fb536SJames Wright // -- Structs 340*149fb536SJames Wright for (PetscInt i = 0; i < problem->num_bc_defs; i++) { 341*149fb536SJames Wright PetscCall(BCDefinitionDestroy(&problem->bc_defs[i])); 342*149fb536SJames Wright } 343*149fb536SJames Wright PetscCall(PetscFree(problem->bc_defs)); 344*149fb536SJames Wright PetscCall(PetscFree(units)); 345*149fb536SJames Wright PetscCall(PetscFree(user)); 346*149fb536SJames Wright PetscCall(PetscFree(problem)); 347*149fb536SJames Wright PetscCall(PetscFree(bc)); 348*149fb536SJames Wright PetscCall(PetscFree(phys_ctx)); 349*149fb536SJames Wright PetscCall(PetscFree(app_ctx)); 350*149fb536SJames Wright PetscCall(PetscFree(ceed_data)); 351*149fb536SJames Wright PetscCall(PetscFree(problem)); 352*149fb536SJames Wright 353*149fb536SJames Wright return PetscFinalize(); 354*149fb536SJames Wright } 355