]> https://gitweb.dealii.org/ - dealii.git/commitdiff
Split out most of the parameter handling into a namespace and classes of their own.
authorWolfgang Bangerth <bangerth@math.tamu.edu>
Wed, 14 May 2008 16:09:27 +0000 (16:09 +0000)
committerWolfgang Bangerth <bangerth@math.tamu.edu>
Wed, 14 May 2008 16:09:27 +0000 (16:09 +0000)
git-svn-id: https://svn.dealii.org/trunk@16092 0785d39b-7218-0410-832d-ea1e28bc413d

deal.II/examples/step-33/step-33.cc

index 4878af89b8411e5bdd4ab89d0df29b67dd7c0b2d..4c182b359ffed5e798bd5be4a04356e31eb10ec0 100644 (file)
@@ -20,6 +20,7 @@
 #include <base/function.h>
 #include <base/parameter_handler.h>
 #include <base/function_parser.h>
+#include <base/utilities.h>
 
 #include <lac/vector.h>
 #include <lac/sparse_matrix.h>
@@ -286,6 +287,287 @@ template <int dim>
 const double EulerEquations<dim>::gas_gamma = 1.4;
 
 
+namespace Parameters
+{
+                                  // An object to store parameter information
+                                  // about the Aztec solver.
+  struct Solver 
+  {
+      int LIN_OUTPUT;
+      enum solver_type { GMRES = 0, DIRECT = 1};
+      solver_type SOLVER;
+      
+      enum  output_type { QUIET = 0, VERBOSE = 1 };
+      output_type OUTPUT;
+                                      // Linear residual tolerance.
+      double RES;
+      int MAX_ITERS;
+                                      // We use the ILUT preconditioner.
+                                      // This is similar to the ILU.  FILL is
+                                      // the number of extra entries to add
+                                      // when forming the ILU decomposition.
+      double ILUT_FILL;
+                                      // When forming the preconditioner, for
+                                      // certain problems bad conditioning
+                                      // (or just bad luck) can cause the
+                                      // preconditioner to be very poorly
+                                      // conditioned.  Hence it can help to
+                                      // add diagonal perturbations to the
+                                      // original matrix and form the
+                                      // preconditioner for this slightly
+                                      // better matrix.  ATOL is an absolute
+                                      // perturbation that is added to the
+                                      // diagonal before forming the prec,
+                                      // and RTOL is a scaling factor $rtol
+                                      // >= 1$.
+      double ILUT_ATOL;
+      double ILUT_RTOL;
+                                      // The ILUT will drop any values that
+                                      // have magnitude less than this value.
+                                      // This is a way to manage the amount
+                                      // of memory used by this
+                                      // preconditioner.
+      double ILUT_DROP;
+
+      static void declare_parameters (ParameterHandler &prm);
+      void parse_parameters (ParameterHandler &prm);
+  };
+
+
+
+  void Solver::declare_parameters (ParameterHandler &prm)
+  {
+    prm.enter_subsection("linear solver");
+    {
+      prm.declare_entry("output", "quiet",
+                       Patterns::Selection(
+                         "quiet|verbose"),
+                       "<quiet|verbose>");
+      prm.declare_entry("method", "gmres",
+                       Patterns::Selection(
+                         "gmres|direct"),
+                       "<gmres|direct>");
+      prm.declare_entry("residual", "1e-10",
+                       Patterns::Double(),
+                       "linear solver residual");
+      prm.declare_entry("max iters", "300",
+                       Patterns::Integer(),
+                       "maximum solver iterations");
+      prm.declare_entry("ilut fill", "2",
+                       Patterns::Double(),
+                       "ilut preconditioner fill");
+      prm.declare_entry("ilut absolute tolerance", "1e-9",
+                       Patterns::Double(),
+                       "ilut preconditioner tolerance");
+      prm.declare_entry("ilut relative tolerance", "1.1",
+                       Patterns::Double(),
+                       "rel tol");
+      prm.declare_entry("ilut drop tolerance", "1e-10",
+                       Patterns::Double(),
+                       "ilut drop tol");
+    }
+    prm.leave_subsection();
+  }
+  
+    
+
+  
+  void Solver::parse_parameters (ParameterHandler &prm)
+  {
+    prm.enter_subsection("linear solver");
+    {
+      const std::string op = prm.get("output");
+      if (op == "verbose")
+       OUTPUT = Parameters::Solver::VERBOSE;
+      if (op == "quiet")
+       OUTPUT = Parameters::Solver::QUIET;
+    
+      const std::string sv = prm.get("method");
+      if (sv == "direct") 
+       SOLVER = Parameters::Solver::DIRECT;
+      else if (sv == "gmres")
+       SOLVER = Parameters::Solver::GMRES;
+
+      RES       = prm.get_double("residual");
+      MAX_ITERS = prm.get_integer("max iters");
+      ILUT_FILL = prm.get_double("ilut fill");
+      ILUT_ATOL = prm.get_double("ilut absolute tolerance");
+      ILUT_RTOL = prm.get_double("ilut relative tolerance");
+      ILUT_DROP = prm.get_double("ilut drop tolerance");
+      RES = prm.get_double("residual");
+    }
+    prm.leave_subsection();  
+  }
+  
+
+  
+  struct Refinement
+  {
+      enum refine_type { NONE = 0, FIXED_NUMBER = 1, SHOCK = 2};
+      double high_frac;
+      double low_frac;
+      refine_type refine;
+      double high_frac_sav;
+      double max_cells;
+      double shock_val;
+      double shock_levels;
+
+      static void declare_parameters (ParameterHandler &prm);
+      void parse_parameters (ParameterHandler &prm);
+  };
+
+
+
+  void Refinement::declare_parameters (ParameterHandler &prm)
+  {
+
+    prm.enter_subsection("refinement");
+    {
+      prm.declare_entry("refinement", "none",
+                       Patterns::Selection(
+                         "none|fixed number|shock"),
+                       "<on|off>");
+      prm.declare_entry("refinement fraction", "0.1",
+                       Patterns::Double(),
+                       "Fraction of high refinement");
+      prm.declare_entry("unrefinement fraction", "0.1",
+                       Patterns::Double(),
+                       "Fraction of low unrefinement");
+      prm.declare_entry("max elements", "1000000",
+                       Patterns::Double(),
+                       "maximum number of elements");
+      prm.declare_entry("shock value", "4.0",
+                       Patterns::Double(),
+                       "value for shock indicator");
+      prm.declare_entry("shock levels", "3.0",
+                       Patterns::Double(),
+                       "number of shock refinement levels");
+    }
+    prm.leave_subsection();
+  }
+  
+
+  void Refinement::parse_parameters (ParameterHandler &prm)
+  {
+  
+                                    // And refiement.
+    prm.enter_subsection("refinement");
+    {
+      const std::string ref = prm.get("refinement");
+      if (ref == "none") 
+       refine = NONE;
+      else if (ref == "fixed number") 
+       refine = FIXED_NUMBER;
+      else if (ref == "shock") 
+       refine = SHOCK;
+      else
+       high_frac = prm.get_double("refinement fraction");
+    
+      high_frac_sav = high_frac;
+      low_frac = prm.get_double("unrefinement fraction");
+      max_cells = prm.get_double("max elements");
+      shock_val = prm.get_double("shock value");
+      shock_levels = prm.get_double("shock levels");
+    }
+    prm.leave_subsection();
+  }
+  
+
+
+  struct Flux
+  {
+      typedef enum {CONSTANT=1,MESH=2} LF_stab_type;
+      LF_stab_type LF_stab;
+                                    // The user can set the stabilization
+                                    // parameter $\alpha$ in the
+                                    // Lax-Friedrich's flux.
+      double LF_stab_value;
+
+      static void declare_parameters (ParameterHandler &prm);
+      void parse_parameters (ParameterHandler &prm);
+  };
+
+
+  void Flux::declare_parameters (ParameterHandler &prm)
+  {
+    prm.enter_subsection("flux");
+    {
+      prm.declare_entry("stab", "alpha",
+                       Patterns::Selection(
+                         "alpha|constant|mesh"),
+                       "<alpha|constant|mesh>");
+      prm.declare_entry("stab value", "1",
+                       Patterns::Double(),
+                       "alpha stabilization");
+    }
+    prm.leave_subsection();  
+  }
+  
+  
+  void Flux::parse_parameters (ParameterHandler &prm)
+  {
+    prm.enter_subsection("flux");
+    {
+      const std::string stab = prm.get("stab");
+      if (stab == "constant") 
+       LF_stab = CONSTANT;
+      else if (stab == "mesh ") 
+       LF_stab = MESH;
+  
+      LF_stab_value = prm.get_double("stab value");
+    }
+    prm.leave_subsection();
+  }
+
+
+
+  struct Output
+  {
+                                      // If true, we output the squared
+                                      // gradient of the density instead of
+                                      // density.  Using this one can create
+                                      // shock plots.
+      bool schlieren_plot;
+                                      // How often to create an output file.
+      double output_step;
+
+      static void declare_parameters (ParameterHandler &prm);
+      void parse_parameters (ParameterHandler &prm);
+  };
+
+
+
+  void Output::declare_parameters (ParameterHandler &prm)
+  {
+    prm.enter_subsection("output");
+    {  
+      prm.declare_entry("density", "standard",
+                       Patterns::Selection(
+                         "standard|schlieren"),
+                       "<standard|schlieren>");
+      prm.declare_entry("step", "-1",
+                       Patterns::Double(),
+                       "output once per this period");
+    }
+    prm.leave_subsection();
+  }
+  
+
+
+  void Output::parse_parameters (ParameterHandler &prm)
+  {
+    prm.enter_subsection("output");
+    {
+      schlieren_plot = (prm.get("density") == "schlieren" ? true : false);
+      output_step = prm.get_double("step");
+    }
+    prm.leave_subsection();
+  }
+}
+
+  
+      
+
                                 // @sect3{Conservation Law class}
 
                                 // Here we define a Conservation Law
@@ -313,10 +595,11 @@ class ConsLaw
     void refine_grid ();
     void output_results (const unsigned int cycle) const;
     void initialize();
-    void zero_matrix();
     void estimate();
     void postprocess();
     void compute_predictor();
+
+    static const unsigned int max_n_boundaries = 10;
     
     Triangulation<dim>   triangulation;
     const MappingQ1<dim> mapping;
@@ -380,75 +663,16 @@ class ConsLaw
     typedef typename std::map<unsigned int, std::pair<std::vector<bc_type>, Function<dim>*> > bdry_map_type;
     bdry_map_type bdry_map;
 
-    void add_boundary(unsigned int bd, std::vector<bc_type>& flags, Function<dim> *bf);
-
-                                    // An object to store parameter information about the Aztec solver.
-    typedef struct {
-       int LIN_OUTPUT;
-       typedef enum { GMRES = 0, DIRECT = 1} solver_type;
-       solver_type SOLVER;
-       typedef enum { QUIET = 0, VERBOSE = 1 } output_type;
-       output_type OUTPUT;
-                                        // Linear residual tolerance.
-       double RES;
-       int MAX_ITERS;
-                                        // We use the ILUT preconditioner.  This is similar
-                                        // to the ILU.  FILL is the number of extra entries
-                                        // to add when forming the ILU decomposition.
-       double ILUT_FILL;
-                                        // When forming the preconditioner, for certain problems
-                                        // bad conditioning (or just bad luck) can cause the
-                                        // preconditioner to be very poorly conditioned.  Hence
-                                        // it can help to add diagonal perturbations to the
-                                        // original matrix and form the preconditioner for this
-                                        // slightly better matrix.  ATOL is an absolute perturbation
-                                        // that is added to the diagonal before forming the
-                                        // prec, and RTOL is a scaling factor $rtol >= 1$.
-       double ILUT_ATOL;
-       double ILUT_RTOL;
-                                        // The ILUT will drop any values that have magnitude less
-                                        // than this value.  This is a way to
-                                        // manage the amount of memory used by this preconditioner.
-       double ILUT_DROP;
-    } solver_params_type;
-
-    solver_params_type solver_params;
-
-                                    // Some refinement parameters.
-    typedef struct {
-       typedef enum { NONE = 0, FIXED_NUMBER = 1, SHOCK = 2} refine_type;
-       double high_frac;
-       double low_frac;
-       refine_type refine;
-       double high_frac_sav;
-       double max_cells;
-       double shock_val;
-       double shock_levels;
-    } refinement_params_type;
-
-    refinement_params_type refinement_params;
-
-                                    // The user can set the stabilization parameter $\alpha$ 
-                                    // in the Lax-Friedrich's flux.  
-    typedef struct {
-       typedef enum {CONSTANT=1,MESH=2} LF_stab_type;
-       LF_stab_type LF_stab;
-       double LF_stab_value;
-    } flux_params_type;
-
-    flux_params_type flux_params;
+    Parameters::Solver     solver_params;
+    Parameters::Refinement refinement_params;
+    Parameters::Flux       flux_params;
+    Parameters::Output     output_params;
 
     bool is_stationary;
 
                                     // Power for the mesh stabilization term.
     double diffusion_power;
     double gravity;
-                                    // If true, we output the squared gradient of the
-                                    // density instead of density.  Using this one can
-                                    // create shock plots.
-    bool schlieren_plot;
-                                    // How often to create an output file.
-    double output_step;
 
     Epetra_Map         *Map;
     Epetra_CrsMatrix   *Matrix;
@@ -460,14 +684,31 @@ class ConsLaw
 };
 
 
-                                // Asign a row of the conservation law a specified
-                                // boundary type and (possibly) function.
+                                // Create a conservation law with some defaults.
 template <int dim>
-void ConsLaw<dim>::add_boundary(unsigned int bd,
-                               std::vector<bc_type> &flags, Function<dim> *bf) {
+ConsLaw<dim>::ConsLaw ()
+               :
+               mapping (),
+                fe (FE_Q<dim>(1), EulerEquations<dim>::n_components),
+               dof_handler (triangulation),
+               quadrature (2),
+               face_quadrature (2),
+                T(0),
+                dT(0.05),
+                TF(10),
+               initial_conditions (EulerEquations<dim>::n_components),
+                is_stationary(false),
+                Map(NULL),
+                Matrix(NULL),
+                theta(0.5) 
+{}
+
 
-  std::pair<std::vector<bc_type>, Function<dim> *> entry(flags, bf);
-  bdry_map[bd] = entry;
+                                // Bye bye Conservation law.
+template <int dim>
+ConsLaw<dim>::~ConsLaw () 
+{
+  dof_handler.clear ();
 }
 
 
@@ -818,10 +1059,10 @@ void ConsLaw<dim>::assemble_face_term(
   double alpha = 1;
 
   switch(flux_params.LF_stab) {
-    case flux_params_type::CONSTANT:
+    case Parameters::Flux::CONSTANT:
          alpha = flux_params.LF_stab_value;
          break;
-    case flux_params_type::MESH:
+    case Parameters::Flux::MESH:
          alpha = face_diameter/(2.0*dT);
          break;
   }
@@ -1105,33 +1346,6 @@ void ConsLaw<dim>::assemble_system (double &res_norm)
                                   // Compute the nonlinear residual.
   res_norm = right_hand_side.l2_norm();
     
-}
-
-                                // Create a conservation law with some defaults.
-template <int dim>
-ConsLaw<dim>::ConsLaw ()
-               :
-               mapping (),
-                fe (FE_Q<dim>(1), EulerEquations<dim>::n_components),
-               dof_handler (triangulation),
-               quadrature (2),
-               face_quadrature (2),
-                T(0),
-                dT(0.05),
-                TF(10),
-               initial_conditions (EulerEquations<dim>::n_components),
-                is_stationary(false),
-                Map(NULL),
-                Matrix(NULL),
-                theta(0.5) 
-{}
-
-
-                                // Bye bye Conservation law.
-template <int dim>
-ConsLaw<dim>::~ConsLaw () 
-{
-  dof_handler.clear ();
 }
 
                                 // @sect3{Initialize System}
@@ -1250,7 +1464,7 @@ void ConsLaw<dim>::solve (Vector<double> &dsolution, int &niter, double &lin_res
   Epetra_Vector b(View, *Map, right_hand_side.begin());
 
                                   // The Direct option selects the Amesos solver.
-  if (solver_params.SOLVER == solver_params_type::DIRECT) {
+  if (solver_params.SOLVER == Parameters::Solver::DIRECT) {
    
                                     // Setup for solving with
                                     // Amesos. Other solvers are
@@ -1268,12 +1482,12 @@ void ConsLaw<dim>::solve (Vector<double> &dsolution, int &niter, double &lin_res
                                     // out the sparsity patterns, and then the
                                     // numerical part actually performs Gaussian
                                     // elimination or whatever the approach is.
-    if (solver_params.OUTPUT == solver_params_type::VERBOSE)
+    if (solver_params.OUTPUT == Parameters::Solver::VERBOSE)
       std::cout << "Starting Symbolic fact\n" << std::flush;
 
     solver->SymbolicFactorization();
 
-    if (solver_params.OUTPUT == solver_params_type::VERBOSE)
+    if (solver_params.OUTPUT == Parameters::Solver::VERBOSE)
       std::cout << "Starting Numeric fact\n" << std::flush;
 
     solver->NumericFactorization();
@@ -1284,7 +1498,7 @@ void ConsLaw<dim>::solve (Vector<double> &dsolution, int &niter, double &lin_res
     prob.SetRHS(&b);
     prob.SetLHS(&x);
                                     // And finally solve the problem.
-    if (solver_params.OUTPUT == solver_params_type::VERBOSE)
+    if (solver_params.OUTPUT == Parameters::Solver::VERBOSE)
       std::cout << "Starting solve\n" << std::flush;
     solver->Solve();
     niter = 0;
@@ -1294,16 +1508,16 @@ void ConsLaw<dim>::solve (Vector<double> &dsolution, int &niter, double &lin_res
                                     // for us.
     delete solver;
 
-  } else if (solver_params.SOLVER == solver_params_type::GMRES) {
+  } else if (solver_params.SOLVER == Parameters::Solver::GMRES) {
 
                                     // For the iterative solvers, we use Aztec.
     AztecOO Solver;
 
                                     // Select the appropriate level of verbosity.
-    if (solver_params.OUTPUT == solver_params_type::QUIET)
+    if (solver_params.OUTPUT == Parameters::Solver::QUIET)
       Solver.SetAztecOption(AZ_output, AZ_none);
 
-    if (solver_params.OUTPUT == solver_params_type::VERBOSE)
+    if (solver_params.OUTPUT == Parameters::Solver::VERBOSE)
       Solver.SetAztecOption(AZ_output, AZ_all);
 
                                     // Select gmres.  Other solvers are available.
@@ -1402,16 +1616,18 @@ void ConsLaw<dim>::postprocess() {
                                       // Pressure
       ppsolution(dofs[eidx]) = (EulerEquations<dim>::gas_gamma-1.0)*(solution(dofs[eidx]) - 0.5*rho_normVsqr);
 
-                                      // Either output density or gradient squared of density,
-                                      // depending on what the user wants.
-      if (!schlieren_plot) {
+                                      // Either output density or gradient
+                                      // squared of density, depending on
+                                      // what the user wants.
+      if (output_params.schlieren_plot == false)
         ppsolution(dofs[didx]) = solution(dofs[didx]);
-      } else {
-        double ng = 0;
-        for (unsigned int i = 0; i < dim; i++) ng += dU[q][EulerEquations<dim>::density_component][i]*dU[q][EulerEquations<dim>::density_component][i];
-        ng = std::sqrt(ng);
-        ppsolution(dofs[didx]) = ng;
-      }
+      else
+       {
+         double ng = 0;
+         for (unsigned int i = 0; i < dim; i++) ng += dU[q][EulerEquations<dim>::density_component][i]*dU[q][EulerEquations<dim>::density_component][i];
+         ng = std::sqrt(ng);
+         ppsolution(dofs[didx]) = ng;
+       }
     }
 
   } // cell
@@ -1579,11 +1795,10 @@ void ConsLaw<dim>::output_results (const unsigned int cycle) const
                                 // leave a detailed explanation of these 
                                 // parameters to our description of the input
                                 // sample file.
-const unsigned int MAX_BD = 10;
 template <int dim>
 void ConsLaw<dim>::declare_parameters() {
 
-                                   // Global scope parameters/
+                                   // Global scope parameters
   prm.declare_entry("mesh", "grid.inp",
                     Patterns::Anything(),
                     "intput file");
@@ -1608,7 +1823,7 @@ void ConsLaw<dim>::declare_parameters() {
 
 
                                   // Declare the boundary parameters
-  for (unsigned int b = 0; b < MAX_BD; b++) {
+  for (unsigned int b = 0; b < max_n_boundaries; b++) {
     char bd[512];
     std::sprintf(bd, "boundary_%d", b);
     prm.enter_subsection(bd);
@@ -1650,88 +1865,18 @@ void ConsLaw<dim>::declare_parameters() {
   prm.leave_subsection();
 
                                   // The linear solver block.
-  prm.enter_subsection("linear solver");
-  prm.declare_entry("output", "quiet",
-                   Patterns::Selection(
-                     "quiet|verbose"),
-                   "<quiet|verbose>");
-  prm.declare_entry("method", "gmres",
-                   Patterns::Selection(
-                     "gmres|direct"),
-                   "<gmres|direct>");
-  prm.declare_entry("residual", "1e-10",
-                   Patterns::Double(),
-                   "linear solver residual");
-  prm.declare_entry("max iters", "300",
-                   Patterns::Double(),
-                   "maximum solver iterations");
-  prm.declare_entry("ilut fill", "2",
-                   Patterns::Double(),
-                   "ilut preconditioner fill");
-  prm.declare_entry("ilut absolute tolerance", "1e-9",
-                   Patterns::Double(),
-                   "ilut preconditioner tolerance");
-  prm.declare_entry("ilut relative tolerance", "1.1",
-                   Patterns::Double(),
-                   "rel tol");
-  prm.declare_entry("ilut drop tolerance", "1e-10",
-                   Patterns::Double(),
-                   "ilut drop tol");
-  prm.leave_subsection();
-
-
-                                  // A refinement controller block.
-  prm.enter_subsection("refinement");
-  prm.declare_entry("refinement", "none",
-                   Patterns::Selection(
-                     "none|fixed number|shock"),
-                   "<on|off>");
-  prm.declare_entry("refinement fraction", "0.1",
-                   Patterns::Double(),
-                   "Fraction of high refinement");
-  prm.declare_entry("unrefinement fraction", "0.1",
-                   Patterns::Double(),
-                   "Fraction of low unrefinement");
-  prm.declare_entry("max elements", "1000000",
-                   Patterns::Double(),
-                   "maximum number of elements");
-  prm.declare_entry("shock value", "4.0",
-                   Patterns::Double(),
-                   "value for shock indicator");
-  prm.declare_entry("shock levels", "3.0",
-                   Patterns::Double(),
-                   "number of shock refinement levels");
-  prm.leave_subsection();
-
-                                  // Output control.
-  prm.enter_subsection("output");
-  prm.declare_entry("density", "standard",
-                   Patterns::Selection(
-                     "standard|schlieren"),
-                   "<standard|schlieren>");
-  prm.declare_entry("step", "-1",
-                   Patterns::Double(),
-                   "output once per this period");
-  prm.leave_subsection();
-
-                                  // Flux control
-  prm.enter_subsection("flux");
-  prm.declare_entry("stab", "alpha",
-                   Patterns::Selection(
-                     "alpha|constant|mesh"),
-                   "<alpha|constant|mesh>");
-  prm.declare_entry("stab value", "1",
-                   Patterns::Double(),
-                   "alpha stabilization");
-  prm.leave_subsection();
-
-
+  Parameters::Solver::declare_parameters (prm);
+  Parameters::Refinement::declare_parameters (prm);
+  Parameters::Flux::declare_parameters (prm);
+  Parameters::Output::declare_parameters (prm);
 }
 
-                                // Code to actually parse an input file.  This function
-                                // matches the declarations above.
+                                // Code to actually parse an input file.
+                                // This function matches the declarations
+                                // above.
 template <int dim>
-void ConsLaw<dim>::load_parameters(const char *infile){
+void ConsLaw<dim>::load_parameters(const char *infile)
+{
 
   prm.read_input(infile);
 
@@ -1744,60 +1889,74 @@ void ConsLaw<dim>::load_parameters(const char *infile){
 
                                   // The time stepping.
   prm.enter_subsection("time stepping");
-  dT = prm.get_double("time step");
-  std::cout << "dT=" << dT << std::endl;
-  if (dT == 0) {
-    is_stationary = true;
-    dT = 1.0;
-    TF = 1.0;
-    std::cout << "Stationary mode" << std::endl;
+  {
+    dT = prm.get_double("time step");
+    if (dT == 0)
+      {
+       is_stationary = true;
+       dT = 1.0;
+       TF = 1.0;
+       std::cout << "Stationary mode" << std::endl;
+      }
+    TF = prm.get_double("final time");
+
+    std::cout << "dT=" << dT << std::endl;
+    std::cout << "TF=" << TF << std::endl;
   }
-  TF = prm.get_double("final time");
-  std::cout << "TF=" << TF << std::endl;
   prm.leave_subsection();
 
                                   // The boundary info
-  for (unsigned int b = 0; b < MAX_BD; b++) {
-    std::vector<bc_type> flags(EulerEquations<dim>::n_components, OUTFLOW_BC);
+  for (unsigned int b = 0; b < max_n_boundaries; ++b)
+    {
+      prm.enter_subsection("boundary_" + Utilities::int_to_string(b));
+      {
+       std::vector<bc_type> flags(EulerEquations<dim>::n_components,
+                                  OUTFLOW_BC);
 
-                                    // Define a parser for every boundary, though it may be
-                                    // unused.
-    FunctionParser<dim> *sd = new FunctionParser<dim>(EulerEquations<dim>::n_components);
+                                        // Define a parser for every boundary,
+                                        // though it may be unused.
+       FunctionParser<dim> *sd
+         = new FunctionParser<dim>(EulerEquations<dim>::n_components);
 
-    std::vector<std::string> expressions(EulerEquations<dim>::n_components, "0.0");
+       std::vector<std::string> expressions(EulerEquations<dim>::n_components,
+                                            "0.0");
     
-    char bd[512];
-    std::sprintf(bd, "boundary_%d", b);
-    prm.enter_subsection(bd);
+       const std::string nopen = prm.get("no penetration");
 
-    const std::string nopen = prm.get("no penetration");
+                                        // Determine how each component is
+                                        // handled.
+       for (unsigned int di=0; di<EulerEquations<dim>::n_components; ++di)
+         {
+           const std::string btype
+             = prm.get("w_" + Utilities::int_to_string(di));
+           const std::string var_value
+             = prm.get("w_" + Utilities::int_to_string(di) +
+                       " value");
+
+           if (di < dim && nopen == "true")
+             flags[di] = NO_PENETRATION_BC;
+           else if (btype == "inflow")
+             {
+               flags[di] = INFLOW_BC;
+               expressions[di] = var_value;
+             }
+           else if (btype == "pressure")
+             {
+               flags[di] = PRESSURE_BC;
+               expressions[di] = var_value;
+             }
+         }
 
-                                    // Determine how each component is handled.
-    for (unsigned int di = 0; di < EulerEquations<dim>::n_components; di++) {
-      char var[512];
-      std::sprintf(var, "w_%d", di);
-      std::string btype = prm.get(var);
-      std::sprintf(var, "w_%d value", di);
-      std::string var_value = prm.get(var);
-
-      if (di < dim && nopen == "true") {
-        flags[di] = NO_PENETRATION_BC;
-      } else if (btype == "inflow") {
-        flags[di] = INFLOW_BC;
-        expressions[di] = var_value;
-      } else if (btype == "pressure") {
-        flags[di] = PRESSURE_BC;
-        expressions[di] = var_value;
-      }
-    } 
-    prm.leave_subsection();
 
-                                    // Add the boundary condition to the law.
-    sd->initialize (FunctionParser<dim>::default_variable_names(),
-                   expressions,
-                   std::map<std::string, double>());
-    add_boundary(b, flags, sd);
-  }
+                                        // Add the boundary condition to the
+                                        // law.
+       sd->initialize (FunctionParser<dim>::default_variable_names(),
+                       expressions,
+                       std::map<std::string, double>());
+       bdry_map[b] = std::make_pair (flags, sd);
+      }
+      prm.leave_subsection();
+    }
 
                                   // Initial conditions.
   prm.enter_subsection("initial condition");
@@ -1805,87 +1964,28 @@ void ConsLaw<dim>::load_parameters(const char *infile){
     std::vector<std::string> expressions (EulerEquations<dim>::n_components,
                                          "0.0");
     for (unsigned int di = 0; di < EulerEquations<dim>::n_components; di++)
-      {
-       char var[512];
-       std::sprintf(var, "w_%d value", di);
-       expressions[di] = prm.get(var);
-      }
+      expressions[di] = prm.get("w_" + Utilities::int_to_string(di) +
+                               " value");
     initial_conditions.initialize (FunctionParser<dim>::default_variable_names(),
                                   expressions,
                                   std::map<std::string, double>());
   }
   prm.leave_subsection();
 
-                                  // The linear solver.
-  prm.enter_subsection("linear solver");
-  const std::string &op = prm.get("output");
-  if (op == "verbose") solver_params.OUTPUT = solver_params_type::VERBOSE;
-  if (op == "quiet") solver_params.OUTPUT = solver_params_type::QUIET;
-  const std::string &sv = prm.get("method");
-  if (sv == "direct") {
-    solver_params.SOLVER = solver_params_type::DIRECT;
-  } else if (sv == "gmres") {
-    solver_params.SOLVER = solver_params_type::GMRES;
-  } 
-
-  solver_params.RES = prm.get_double("residual");
-  solver_params.MAX_ITERS = (int) prm.get_double("max iters");
-  solver_params.ILUT_FILL = prm.get_double("ilut fill");
-  solver_params.ILUT_ATOL = prm.get_double("ilut absolute tolerance");
-  solver_params.ILUT_RTOL = prm.get_double("ilut relative tolerance");
-  solver_params.ILUT_DROP = prm.get_double("ilut drop tolerance");
-  solver_params.RES = prm.get_double("residual");
-  prm.leave_subsection();
-
-
-                                  // And refiement.
-  prm.enter_subsection("refinement");
-  const std::string &ref = prm.get("refinement");
-  if (ref == "none") {
-    refinement_params.refine = refinement_params_type::NONE;
-  } else if (ref == "fixed number") {
-    refinement_params.refine = refinement_params_type::FIXED_NUMBER;
-  } else if (ref == "shock") {
-    refinement_params.refine = refinement_params_type::SHOCK;
-  } else
-    refinement_params.high_frac = prm.get_double("refinement fraction");
-  refinement_params.high_frac_sav = refinement_params.high_frac;
-  refinement_params.low_frac = prm.get_double("unrefinement fraction");
-  refinement_params.max_cells = prm.get_double("max elements");
-  refinement_params.shock_val = prm.get_double("shock value");
-  refinement_params.shock_levels = prm.get_double("shock levels");
-  prm.leave_subsection();
-    
-                                  // Output control.
-  prm.enter_subsection("output");
-  const std::string &dens = prm.get("density");
-  schlieren_plot = dens == "schlieren" ? true : false;
-  output_step = prm.get_double("step");
-  prm.leave_subsection();
-
-                                  // Flux control.
-  prm.enter_subsection("flux");
-  const std::string &stab = prm.get("stab");
-  if (stab == "constant") {
-    flux_params.LF_stab = flux_params_type::CONSTANT;
-  } else if (stab == "mesh ") {
-    flux_params.LF_stab = flux_params_type::MESH;
-  }
-  flux_params.LF_stab_value = prm.get_double("stab value");
-  prm.leave_subsection();
+  solver_params.parse_parameters (prm);
+  refinement_params.parse_parameters (prm);
+  flux_params.parse_parameters (prm);
+  output_params.parse_parameters (prm);
+}
 
 
-}
 
-template<int dim>
-void ConsLaw<dim>::zero_matrix() {
-  Matrix->PutScalar(0); Matrix->FillComplete();
-}
 
-                                // We use a predictor to try and make adaptivity
-                                // work better.  The idea is to try and refine ahead
-                                // of a front, rather than stepping into a coarse
-                                // set of elements and smearing the solution.  This
+                                // We use a predictor to try and make
+                                // adaptivity work better.  The idea is to
+                                // try and refine ahead of a front, rather
+                                // than stepping into a coarse set of
+                                // elements and smearing the solution.  This
                                 // simple time extrapolator does the job.
 template<int dim>
 void ConsLaw<dim>::compute_predictor() {
@@ -1922,7 +2022,7 @@ void ConsLaw<dim>::run ()
                                   // Initial refinement.  We apply the ic,
                                   // estimate, refine, and repeat until
                                   // happy.
-  if (refinement_params.refine != refinement_params_type::NONE)
+  if (refinement_params.refine != Parameters::Refinement::NONE)
     for (unsigned int i = 0; i < refinement_params.shock_levels; i++) {
       estimate();
       refine_grid();
@@ -1934,7 +2034,7 @@ void ConsLaw<dim>::run ()
   output_results (nstep);
 
                                   // Determine when we will output next.
-  double next_output = T + output_step;
+  double next_output = T + output_params.output_step;
 
                                   // @sect4{Main time stepping loop}
   predictor = solution;
@@ -1969,7 +2069,10 @@ void ConsLaw<dim>::run ()
       nlsolution = predictor;
       while (!nonlin_done) {
         lin_iter = 0;
-        zero_matrix();
+
+       Matrix->PutScalar(0);
+       Matrix->FillComplete();
+       
         right_hand_side = 0;
         assemble_system (res_norm);
                                         // Flash a star to the screen so one can
@@ -2014,15 +2117,15 @@ void ConsLaw<dim>::run ()
       T += dT;
 
                                       // Output if it is time.
-      if (output_step < 0) {
+      if (output_params.output_step < 0) {
         output_results (++nstep);
       } else if (T >= next_output) {
         output_results (++nstep);
-        next_output += output_step;
+        next_output += output_params.output_step;
       }
 
                                       // Refine, if refinement is selected.
-      if (refinement_params.refine != refinement_params_type::NONE) {
+      if (refinement_params.refine != Parameters::Refinement::NONE) {
         refine_grid();
         setup_system();
       }

In the beginning the Universe was created. This has made a lot of people very angry and has been widely regarded as a bad move.

Douglas Adams


Typeset in Trocchi and Trocchi Bold Sans Serif.