Stan Math Library  2.14.0
reverse mode automatic differentiation
cvodes_ode_data.hpp
Go to the documentation of this file.
1 #ifndef STAN_MATH_REV_MAT_FUNCTOR_CVODES_ODE_DATA_HPP
2 #define STAN_MATH_REV_MAT_FUNCTOR_CVODES_ODE_DATA_HPP
3 
6 #include <cvodes/cvodes.h>
7 #include <cvodes/cvodes_band.h>
8 #include <cvodes/cvodes_dense.h>
9 #include <nvector/nvector_serial.h>
10 #include <algorithm>
11 #include <vector>
12 
13 namespace stan {
14  namespace math {
15 
25  template<typename F, typename T_initial, typename T_param>
27  const std::vector<T_initial>& y0_;
28  const std::vector<T_param>& theta_;
29  const size_t N_;
30  const size_t M_;
31  const size_t param_var_ind_;
32  const ode_system<F> ode_system_;
33 
36 
37  public:
52  cvodes_ode_data(const F& f,
53  const std::vector<T_initial>& y0,
54  const std::vector<T_param>& theta,
55  const std::vector<double>& x,
56  const std::vector<int>& x_int,
57  std::ostream* msgs)
58  : y0_(y0),
59  theta_(theta),
60  N_(y0.size()),
61  M_(theta.size()),
62  param_var_ind_(initial_var::value ? N_ : 0),
63  ode_system_(f, value_of(theta), x, x_int, msgs) { }
64 
65  static int ode_rhs(double t, N_Vector y, N_Vector ydot, void* user_data) {
66  const ode_data* explicit_ode
67  = static_cast<const ode_data*>(user_data);
68  explicit_ode->rhs(NV_DATA_S(y), NV_DATA_S(ydot), t);
69  return 0;
70  }
71 
72  static int ode_rhs_sens(int Ns, realtype t,
73  N_Vector y, N_Vector ydot,
74  N_Vector *yS, N_Vector *ySdot, void *user_data,
75  N_Vector tmp1, N_Vector tmp2) {
76  const ode_data* explicit_ode = static_cast<const ode_data*>(user_data);
77  const std::vector<double> y_vec(NV_DATA_S(y),
78  NV_DATA_S(y) + explicit_ode->N_);
79  explicit_ode->rhs_sens(explicit_ode->y0_, explicit_ode->theta_,
80  t, y_vec, yS, ySdot);
81  return 0;
82  }
83 
84  static int dense_jacobian(long int N, // NOLINT(runtime/int)
85  realtype t, N_Vector y, N_Vector fy,
86  DlsMat J, void *user_data,
87  N_Vector tmp1, N_Vector tmp2, N_Vector tmp3) {
88  const ode_data* explicit_ode = static_cast<const ode_data*>(user_data);
89  return explicit_ode->dense_jacobian(NV_DATA_S(y), J, t);
90  }
91 
92  private:
93  void rhs(const double y[], double dy_dt[], double t) const {
94  const std::vector<double> y_vec(y, y + N_);
95  std::vector<double> dy_dt_vec(N_);
96  ode_system_(t, y_vec, dy_dt_vec);
97  std::copy(dy_dt_vec.begin(), dy_dt_vec.end(), dy_dt);
98  }
99 
100  int dense_jacobian(const double* y, DlsMat J, double t) const {
101  const std::vector<double> y_vec(y, y + N_);
102  Eigen::VectorXd fy(N_);
103  Eigen::Map<Eigen::MatrixXd> Jy_map(J->data, N_, N_);
104  ode_system_.jacobian(t, y_vec, fy, Jy_map);
105  return 0;
106  }
107 
108  inline void rhs_sens_initial(const Eigen::MatrixXd& Jy,
109  N_Vector *yS, N_Vector *ySdot) const {
110  for (size_t m = 0; m < N_; ++m) {
111  Eigen::Map<Eigen::VectorXd> yS_eig(NV_DATA_S(yS[m]), N_);
112  Eigen::Map<Eigen::VectorXd> ySdot_eig(NV_DATA_S(ySdot[m]), N_);
113  ySdot_eig = Jy * yS_eig;
114  }
115  }
116 
117  inline void rhs_sens_param(const Eigen::MatrixXd& Jy,
118  const Eigen::MatrixXd& Jtheta,
119  N_Vector *yS, N_Vector *ySdot) const {
120  using Eigen::Map;
121  using Eigen::VectorXd;
122  for (size_t m = 0; m < M_; ++m) {
123  Map<VectorXd> yS_eig(NV_DATA_S(yS[param_var_ind_ + m]), N_);
124  Map<VectorXd> ySdot_eig(NV_DATA_S(ySdot[param_var_ind_ + m]), N_);
125  ySdot_eig = Jy * yS_eig + Jtheta.col(m);
126  }
127  }
128 
140  void rhs_sens(const std::vector<var>& initial,
141  const std::vector<var>& param,
142  double t, const std::vector<double>& y,
143  N_Vector *yS, N_Vector *ySdot) const {
144  Eigen::VectorXd dy_dt(N_);
145  Eigen::MatrixXd Jy(N_, N_);
146  Eigen::MatrixXd Jtheta(N_, M_);
147  ode_system_.jacobian(t, y, dy_dt, Jy, Jtheta);
148  rhs_sens_initial(Jy, yS, ySdot);
149  rhs_sens_param(Jy, Jtheta, yS, ySdot);
150  }
151 
163  void rhs_sens(const std::vector<double>& initial,
164  const std::vector<var>& param,
165  double t, const std::vector<double>& y,
166  N_Vector *yS, N_Vector *ySdot) const {
167  Eigen::VectorXd dy_dt(N_);
168  Eigen::MatrixXd Jy(N_, N_);
169  Eigen::MatrixXd Jtheta(N_, M_);
170  ode_system_.jacobian(t, y, dy_dt, Jy, Jtheta);
171  rhs_sens_param(Jy, Jtheta, yS, ySdot);
172  }
173 
185  void rhs_sens(const std::vector<var>& initial,
186  const std::vector<double>& param,
187  double t, const std::vector<double>& y,
188  N_Vector *yS, N_Vector *ySdot) const {
189  Eigen::VectorXd dy_dt(N_);
190  Eigen::MatrixXd Jy(N_, N_);
191  ode_system_.jacobian(t, y, dy_dt, Jy);
192  rhs_sens_initial(Jy, yS, ySdot);
193  }
194 
206  void rhs_sens(const std::vector<double>& initial,
207  const std::vector<double>& param,
208  double t, const std::vector<double>& y,
209  N_Vector *yS, N_Vector *ySdot) const {
210  }
211  };
212 
213  }
214 }
215 #endif
void jacobian(double t, const std::vector< double > &y, Eigen::MatrixBase< Derived1 > &dy_dt, Eigen::MatrixBase< Derived2 > &Jy) const
Calculate the Jacobian of the ODE RHS wrt to states y.
Definition: ode_system.hpp:67
T value_of(const fvar< T > &v)
Return the value of the specified variable.
Definition: value_of.hpp:16
static int ode_rhs_sens(int Ns, realtype t, N_Vector y, N_Vector ydot, N_Vector *yS, N_Vector *ySdot, void *user_data, N_Vector tmp1, N_Vector tmp2)
CVODES ode data holder object which is used during CVODES integration for CVODES callbacks.
static int dense_jacobian(long int N, realtype t, N_Vector y, N_Vector fy, DlsMat J, void *user_data, N_Vector tmp1, N_Vector tmp2, N_Vector tmp3)
Internal representation of an ODE model object which provides convenient Jacobian functions to obtain...
Definition: ode_system.hpp:21
cvodes_ode_data(const F &f, const std::vector< T_initial > &y0, const std::vector< T_param > &theta, const std::vector< double > &x, const std::vector< int > &x_int, std::ostream *msgs)
Construct CVODES ode data object to enable callbacks from CVODES during ODE integration.
int size(const std::vector< T > &x)
Return the size of the specified standard vector.
Definition: size.hpp:17
static int ode_rhs(double t, N_Vector y, N_Vector ydot, void *user_data)

     [ Stan Home Page ] © 2011–2016, Stan Development Team.