view scripts/ode/ode15s.m @ 30564:796f54d4ddbf stable

update Octave Project Developers copyright for the new year In files that have the "Octave Project Developers" copyright notice, update for 2021. In all .txi and .texi files except gpl.txi and gpl.texi in the doc/liboctave and doc/interpreter directories, change the copyright to "Octave Project Developers", the same as used for other source files. Update copyright notices for 2022 (not done since 2019). For gpl.txi and gpl.texi, change the copyright notice to be "Free Software Foundation, Inc." and leave the date at 2007 only because this file only contains the text of the GPL, not anything created by the Octave Project Developers. Add Paul Thomas to contributors.in.
author John W. Eaton <jwe@octave.org>
date Tue, 28 Dec 2021 18:22:40 -0500
parents 363fb10055df
children e1788b1a315f
line wrap: on
line source

########################################################################
##
## Copyright (C) 2016-2022 The Octave Project Developers
##
## See the file COPYRIGHT.md in the top-level directory of this
## distribution or <https://octave.org/copyright/>.
##
## This file is part of Octave.
##
## Octave is free software: you can redistribute it and/or modify it
## under the terms of the GNU General Public License as published by
## the Free Software Foundation, either version 3 of the License, or
## (at your option) any later version.
##
## Octave is distributed in the hope that it will be useful, but
## WITHOUT ANY WARRANTY; without even the implied warranty of
## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
## GNU General Public License for more details.
##
## You should have received a copy of the GNU General Public License
## along with Octave; see the file COPYING.  If not, see
## <https://www.gnu.org/licenses/>.
##
########################################################################

## -*- texinfo -*-
## @deftypefn  {} {[@var{t}, @var{y}] =} ode15s (@var{fun}, @var{trange}, @var{y0})
## @deftypefnx {} {[@var{t}, @var{y}] =} ode15s (@var{fun}, @var{trange}, @var{y0}, @var{ode_opt})
## @deftypefnx {} {[@var{t}, @var{y}, @var{te}, @var{ye}, @var{ie}] =} ode15s (@dots{})
## @deftypefnx {} {@var{solution} =} ode15s (@dots{})
## @deftypefnx {} {} ode15s (@dots{})
## Solve a set of stiff Ordinary Differential Equations (ODEs) or stiff
## semi-explicit index 1 Differential Algebraic Equations (DAEs).
##
## @code{ode15s} uses a variable step, variable order BDF (Backward
## Differentiation Formula) method that ranges from order 1 to 5.
##
## @var{fun} is a function handle, inline function, or string containing the
## name of the function that defines the ODE: @code{y' = f(t,y)}.  The function
## must accept two inputs where the first is time @var{t} and the second is a
## column vector of unknowns @var{y}.
##
## @var{trange} specifies the time interval over which the ODE will be
## evaluated.  Typically, it is a two-element vector specifying the initial and
## final times (@code{[tinit, tfinal]}).  If there are more than two elements
## then the solution will also be evaluated at these intermediate time
## instances.
##
## @var{init} contains the initial value for the unknowns.  If it is a row
## vector then the solution @var{y} will be a matrix in which each column is
## the solution for the corresponding initial value in @var{init}.
##
## The optional fourth argument @var{ode_opt} specifies non-default options to
## the ODE solver.  It is a structure generated by @code{odeset}.
##
## The function typically returns two outputs.  Variable @var{t} is a
## column vector and contains the times where the solution was found.  The
## output @var{y} is a matrix in which each column refers to a different
## unknown of the problem and each row corresponds to a time in @var{t}.
##
## The output can also be returned as a structure @var{solution} which has a
## field @var{x} containing a row vector of times where the solution was
## evaluated and a field @var{y} containing the solution matrix such that each
## column corresponds to a time in @var{x}.  Use
## @w{@code{fieldnames (@var{solution})}} to see the other fields and
## additional information returned.
##
## If no output arguments are requested, and no @qcode{"OutputFcn"} is
## specified in @var{ode_opt}, then the @qcode{"OutputFcn"} is set to
## @code{odeplot} and the results of the solver are plotted immediately.
##
## If using the @qcode{"Events"} option then three additional outputs may be
## returned.  @var{te} holds the time when an Event function returned a zero.
## @var{ye} holds the value of the solution at time @var{te}.  @var{ie}
## contains an index indicating which Event function was triggered in the case
## of multiple Event functions.
##
## Example: Solve @nospell{Robertson's} equations:
##
## @smallexample
## @group
## function r = robertson_dae (@var{t}, @var{y})
##   r = [ -0.04*@var{y}(1) + 1e4*@var{y}(2)*@var{y}(3)
##          0.04*@var{y}(1) - 1e4*@var{y}(2)*@var{y}(3) - 3e7*@var{y}(2)^2
##               @var{y}(1) + @var{y}(2) + @var{y}(3) - 1 ];
## endfunction
## opt = odeset ("Mass", [1 0 0; 0 1 0; 0 0 0], "MStateDependence", "none");
## [@var{t},@var{y}] = ode15s (@@robertson_dae, [0, 1e3], [1; 0; 0], opt);
## @end group
## @end smallexample
## @seealso{decic, odeset, odeget, ode23, ode45}
## @end deftypefn

function varargout = ode15s (fun, trange, y0, varargin)

  if (nargin < 3)
    print_usage ();
  endif

  solver = "ode15s";
  ## Check fun, trange, y0, yp0
  fun = check_default_input (fun, trange, solver, y0);

  n = numel (y0);

  if (nargin > 3)
   options = varargin{1};
  else
   options = odeset ();
  endif

  if (! isempty (options.Mass))
    if (ischar (options.Mass))
      if (! exist (options.Mass))
        error ("Octave:invalid-input-arg",
               ['ode15s: "Mass" function "' options.Mass '" not found']);
      endif
      options.Mass = str2func (options.Mass);
    endif
    if (! is_function_handle (options.Mass) && ! isnumeric (options.Mass))
      error ("Octave:invalid-input-arg",
             'ode15s: "Mass" field must be a function handle or square matrix');
    endif
  endif

  if (! isempty (options.Jacobian))
    if (ischar (options.Jacobian))
      if (! exist (options.Jacobian))
        error ("Octave:invalid-input-arg",
               ['ode15s: "Jacobian" function "' options.Jacobian '" not found']);
      endif
      options.Jacobian = str2func (options.Jacobian);
    endif
    if (! is_function_handle (options.Jacobian)
        && ! isnumeric (options.Jacobian))
      error ("Octave:invalid-input-arg",
             'ode15s: "Jacobian" field must be a function handle or square matrix');
    endif
  endif

  if (! isempty (options.OutputFcn))
    if (ischar (options.OutputFcn))
      if (! exist (options.OutputFcn))
        error ("Octave:invalid-input-arg",
               ['ode15s: "OutputFcn" function "' options.OutputFcn '" not found']);
      endif
      options.OutputFcn = str2func (options.OutputFcn);
    endif
    if (! is_function_handle (options.OutputFcn))
      error ("Octave:invalid-input-arg",
             'ode15s: "OutputFcn" must be a valid function handle');
    endif
  endif

  if (! isempty (options.Events))
    if (ischar (options.Events))
      if (! exist (options.Events))
        error ("Octave:invalid-input-arg",
               ['ode15s: "Events" function "' options.Events '" not found']);
      endif
      options.Events = str2func (options.Events);
    endif
    if (! is_function_handle (options.Events))
      error ("Octave:invalid-input-arg",
             'ode15s: "Events" must be a valid function handle');
    endif
  endif

  [defaults, classes, attributes] = odedefaults (n, trange(1), trange(end));

  classes    = odeset (classes, "Vectorized", {});
  attributes = odeset (attributes, "Jacobian", {}, "Vectorized", {});

  options = odemergeopts ("ode15s", options, defaults,
                          classes, attributes, solver);

  ## Mass
  options.havemassfun    = false;
  options.havestatedep   = false;
  options.havetimedep    = false;
  options.havemasssparse = false;

  if (! isempty (options.Mass))
    if (is_function_handle (options.Mass))
      options.havemassfun = true;
      if (nargin (options.Mass) == 2)
        options.havestatedep = true;
        M = options.Mass (trange(1), y0);
        if (! issquare (M) || rows (M) != n || ! isnumeric (M) || ! isreal (M))
          error ("Octave:invalid-input-arg",
                 'ode15s: "Mass" function must evaluate to a real square matrix');
        endif
        options.havemasssparse = issparse (M);
      elseif (nargin (options.Mass) == 1)
        options.havetimedep = true;
        M = options.Mass (trange(1));
        if (! issquare (M) || rows (M) != n || ! isnumeric (M) || ! isreal (M))
          error ("Octave:invalid-input-arg",
                 'ode15s: "Mass" function must evaluate to a real square matrix');
        endif
        options.havemasssparse = issparse (M);
      else
        error ("Octave:invalid-input-arg",
               'ode15s: invalid value assigned to field "Mass"');
      endif
    else    # matrix Mass input
      if (! issquare (options.Mass) || rows (options.Mass) != n
          || ! isnumeric (options.Mass) || ! isreal (options.Mass))
        error ("Octave:invalid-input-arg",
               'ode15s: "Mass" matrix must be a real square matrix');
      endif
      options.havemasssparse = issparse (options.Mass);
    endif
  endif

  ## Jacobian
  options.havejac       = false;
  options.havejacsparse = false;
  options.havejacfun    = false;

  if (! isempty (options.Jacobian))
    options.havejac = true;
    if (is_function_handle (options.Jacobian))
      options.havejacfun = true;
      if (nargin (options.Jacobian) == 2)
        A = options.Jacobian (trange(1), y0);
        if (! issquare (A) || rows (A) != n || ! isnumeric (A) || ! isreal (A))
          error ("Octave:invalid-input-arg",
                 'ode15s: "Jacobian" function must evaluate to a real square matrix');
        endif
        options.havejacsparse = issparse (A);  # Jac is sparse fun
      else
        error ("Octave:invalid-input-arg",
               'ode15s: invalid value assigned to field "Jacobian"');
      endif
    else  # matrix input
      if (! issquare (options.Jacobian) || rows (options.Jacobian) != n
          || ! isnumeric (options.Jacobian) || ! isreal (options.Jacobian))
        error ("Octave:invalid-input-arg",
               'ode15s: "Jacobian" matrix must be a real square matrix');
      endif
      options.havejacsparse = issparse (options.Jacobian);
    endif
  endif

  ## Derivative of M(t,y) for implicit problem not implemented yet
  if (! isempty (options.Mass) && ! isempty (options.Jacobian))
    if (options.MStateDependence != "none" || options.havestatedep == true)
      options.havejac = false;
      options.Jacobian = [];
      warning ("ode15s:mass_state_dependent_provided",
              ["with MStateDependence != 'none' an internal", ...
               " approximation of the Jacobian Matrix will be used.", ...
               "  Set MStateDependence equal to 'none' if you want", ...
               " to provide a constant or time-dependent Jacobian"]);
    endif
  endif

  ## Use sparse methods only if all matrices are sparse
  if (! isempty (options.Mass)) && (! options.havemasssparse)
    options.havejacsparse = false;
  endif

  ## If Mass or Jacobian is fun, then new Jacobian is fun
  if (options.havejac)
    if (options.havejacfun || options.havetimedep)
      options.Jacobian = @ (t, y, yp) wrapjacfun (t, y, yp,
                                                  options.Jacobian,
                                                  options.Mass,
                                                  options.havetimedep,
                                                  options.havejacfun);
      options.havejacfun = true;
    else   # All matrices are constant
      if (! isempty (options.Mass))
        options.Jacobian = {[- options.Jacobian], [options.Mass]};
      else
        if (options.havejacsparse)
          options.Jacobian = {[- options.Jacobian], speye(n)};
        else
          options.Jacobian = {[- options.Jacobian], eye(n)};
        endif
      endif

    endif
  endif

  ## Abstol and Reltol
  options.haveabstolvec = false;

  if (numel (options.AbsTol) != 1 && numel (options.AbsTol) != n)
    error ("Octave:invalid-input-arg",
           'ode15s: invalid value assigned to field "AbsTol"');
  elseif (numel (options.AbsTol) == n)
    options.haveabstolvec = true;
  endif

  ## Stats
  options.havestats = strcmpi (options.Stats, "on");

  ## Don't use Refine when the output is a structure
  if (nargout == 1)
    options.Refine = 1;
  endif

  ## OutputFcn and OutputSel
  if (isempty (options.OutputFcn) && nargout == 0)
    options.OutputFcn = @odeplot;
    options.haveoutputfunction = true;
  else
    options.haveoutputfunction = ! isempty (options.OutputFcn);
  endif

  options.haveoutputselection = ! isempty (options.OutputSel);
  if (options.haveoutputselection)
    options.OutputSel = options.OutputSel - 1;
  endif

  ## Events
  options.haveeventfunction = ! isempty (options.Events);

  yp0 = options.InitialSlope;

  ## 2 arguments in the event callback of ode15s
  [t, y, te, ye, ie] = __ode15__ (@ (t, y, yp) wrap (t, y, yp, options.Mass,
                                                     options.havetimedep,
                                                     options.havestatedep,
                                                     fun),
                                  trange, y0, yp0, options, 2);

  if (nargout == 2)
    varargout{1} = t;
    varargout{2} = y;
  elseif (nargout == 1)
    varargout{1}.x = t.';  # Time stamps saved in field x (row vector)
    varargout{1}.y = y.';  # Results are saved in field y (row vector)
    varargout{1}.solver = solver;
    if (options.haveeventfunction)
      varargout{1}.xe = te.';  # Time info when an event occurred
      varargout{1}.ye = ye.';  # Results when an event occurred
      varargout{1}.ie = ie.';  # Index info which event occurred
    endif
  elseif (nargout > 2)
    varargout = cell (1,5);
    varargout{1} = t;
    varargout{2} = y;
    if (options.haveeventfunction)
      varargout{3} = te;  # Time info when an event occurred
      varargout{4} = ye;  # Results when an event occurred
      varargout{5} = ie;  # Index info which event occurred
    endif
  endif

endfunction

function res = wrap (t, y, yp, Mass, havetimedep, havestatedep, fun)

  if (! isempty (Mass) && havestatedep)
    res = Mass (t, y) * yp - fun (t, y);
  elseif (! isempty (Mass) && havetimedep)
    res = Mass (t) * yp - fun (t, y);
  elseif (! isempty (Mass))
    res = Mass * yp - fun (t, y);
  else
    res = yp - fun (t, y);
  endif

endfunction

function [jac, jact] = wrapjacfun (t, y, yp, Jac, Mass,
                                   havetimedep, havejacfun)

  if (havejacfun)
    jac = - Jac (t, y);
  else
    jac = - Jac;
  endif

  if (! isempty (Mass) && havetimedep)
    jact = Mass (t);
  elseif (! isempty (Mass))
    jact = Mass;
  else
    jact = speye (numel (y));
  endif

endfunction


%!demo
%! ## Solve Robertson's equations with ode15s
%! fun = @ (t, y) [-0.04*y(1) + 1e4*y(2).*y(3);
%!                  0.04*y(1) - 1e4*y(2).*y(3) - 3e7*y(2).^2;
%!                  y(1) + y(2) + y(3) - 1];
%!
%! y0 = [1; 0; 0];
%! tspan = [0, 4*logspace(-6, 6)];
%! M = [1, 0, 0; 0, 1, 0; 0, 0, 0];
%!
%! options = odeset ("RelTol", 1e-4, "AbsTol", [1e-6, 1e-10, 1e-6],
%!                   "MStateDependence", "none", "Mass", M);
%!
%! [t, y] = ode15s (fun, tspan, y0, options);
%!
%! y(:,2) = 1e4 * y(:,2);
%! figure (2);
%! semilogx (t, y, "o");
%! xlabel ("time");
%! ylabel ("species concentration");
%! title ("Robertson DAE problem with a Conservation Law");
%! legend ("y1", "y2", "y3");

%!function ydot = fpol (t, y)  # Van der Pol equation
%!  ydot = [y(2); (1 - y(1)^2) * y(2) - y(1)];
%!endfunction
%!
%!function ref = fref ()       # The computed reference sol
%!  ref = [0.32331666704577, -1.83297456798624];
%!endfunction
%!
%!function jac = fjac (t, y)   # its Jacobian
%!  jac = [0, 1; -1 - 2 * y(1) * y(2), 1 - y(1)^2];
%!endfunction
%!
%!function jac = fjcc (t, y)   # sparse type
%!  jac = sparse ([0, 1; -1 - 2 * y(1) * y(2), 1 - y(1)^2]);
%!endfunction
%!
%!function mas = fmas (t, y)
%!  mas = [1, 0; 0, 1];           # Dummy mass matrix for tests
%!endfunction
%!
%!function mas = fmsa (t, y)
%!  mas = sparse ([1, 0; 0, 1]);  # A sparse dummy matrix
%!endfunction
%!
%!function res = rob (t, y)
%!  res = [-0.04*y(1) + 1e4*y(2).*y(3);
%!          0.04*y(1) - 1e4*y(2).*y(3) - 3e7*y(2).^2;
%!          y(1) + y(2) + y(3) - 1];
%!endfunction
%!
%!function refrob = frefrob ()
%!  refrob = [100, 0.617234887614937, 0.000006153591397, 0.382758958793666];
%!endfunction
%!
%!function [val, isterminal, direction] = feve (t, y)
%!  isterminal = [0, 1];
%!  if (t < 1e1)
%!    val = [-1, -2];
%!  else
%!    val = [1, 3];
%!  endif
%!
%!  direction = [1, 0];
%!endfunction
%!
%!function masrob = massdensefunstate (t, y)
%!  masrob = [1, 0, 0; 0, 1, 0; 0, 0, 0];
%!endfunction
%!
%!function masrob = masssparsefunstate (t, y)
%!  masrob = sparse ([1, 0, 0; 0, 1, 0; 0, 0, 0]);
%!endfunction
%!
%!function masrob = massdensefuntime (t)
%!  masrob = [1, 0, 0; 0, 1, 0; 0, 0, 0];
%!endfunction
%!
%!function masrob = masssparsefuntime (t)
%!  masrob = sparse ([1, 0, 0; 0, 1, 0; 0, 0, 0]);
%!endfunction
%!
%!function jac = jacfundense (t, y)
%!  jac = [-0.04,           1e4*y(3),  1e4*y(2);
%!          0.04, -1e4*y(3)-6e7*y(2), -1e4*y(2);
%!             1,                  1,         1];
%!endfunction
%!
%!function jac = jacfunsparse (t, y)
%!  jac = sparse ([-0.04,           1e4*y(3),  1e4*y(2);
%!                  0.04, -1e4*y(3)-6e7*y(2), -1e4*y(2);
%!                     1,                  1,         1]);
%!endfunction

%!testif HAVE_SUNDIALS
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", [1, 0, 0; 0, 1, 0; 0, 0, 0]);
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", sparse ([1, 0, 0; 0, 1, 0; 0, 0, 0]));
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", @massdensefunstate);
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", @masssparsefunstate);
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", "massdensefuntime");
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", [1, 0, 0; 0, 1, 0; 0, 0, 0],
%!               "Jacobian", "jacfundense");
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", sparse ([1, 0, 0; 0, 1, 0; 0, 0, 0]),
%!               "Jacobian", @jacfundense);
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS
%! warning ("off", "ode15s:mass_state_dependent_provided", "local");
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", @massdensefunstate,
%!               "Jacobian", @jacfundense);
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS
%! warning ("off", "ode15s:mass_state_dependent_provided", "local");
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", @masssparsefunstate,
%!               "Jacobian", @jacfundense);
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", @massdensefuntime,
%!               "Jacobian", @jacfundense);
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", "masssparsefuntime",
%!               "Jacobian", "jacfundense");
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS_SUNLINSOL_KLU
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", [1, 0, 0; 0, 1, 0; 0, 0, 0],
%!               "Jacobian", @jacfunsparse);
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS_SUNLINSOL_KLU
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", sparse ([1, 0, 0; 0, 1, 0; 0, 0, 0]),
%!               "Jacobian", @jacfunsparse);
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS_SUNLINSOL_KLU
%! warning ("off", "ode15s:mass_state_dependent_provided", "local");
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", @massdensefunstate,
%!               "Jacobian", @jacfunsparse);
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS
%! warning ("off", "ode15s:mass_state_dependent_provided", "local");
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", @masssparsefunstate,
%!               "Jacobian", @jacfunsparse);
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS_SUNLINSOL_KLU
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", @massdensefuntime,
%!               "Jacobian", @jacfunsparse);
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

%!testif HAVE_SUNDIALS_SUNLINSOL_KLU
%! opt = odeset ("MStateDependence", "none",
%!               "Mass", @masssparsefuntime,
%!               "Jacobian", @jacfunsparse);
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert ([t(end), y(end,:)], frefrob, 1e-3);

## Jacobian as const matrix
%!testif HAVE_SUNDIALS
%! opt = odeset ("RelTol", 1e-4, "AbsTol", 1e-5,
%!               "Jacobian", [98, 198; -99, -199]);
%! [t, y] = ode15s (@(t, y)[98, 198; -99, -199] * (y - [1; 0]),
%!                 [0, 5], [2; 0], opt);
%! y1xct = @(t) 2 * exp (-t) - exp (-100 * t) + 1;
%! y2xct = @(t) - exp (-t) + exp (-100 * t);
%! assert ([y1xct(t), y2xct(t)], y, 1e-3);

## two output arguments
%!testif HAVE_SUNDIALS
%! [t, y] = ode15s (@fpol, [0, 2], [2, 0]);
%! assert ([t(end), y(end,:)], [2, fref], 1e-2);

## anonymous function instead of real function
%!testif HAVE_SUNDIALS
%! fvdb = @(t,y) [y(2); (1 - y(1)^2) * y(2) - y(1)];
%! [t, y] = ode15s (fvdb, [0, 2], [2, 0]);
%! assert ([t(end), y(end,:)], [2, fref], 1e-2);

## Solve another anonymous function below zero
%!testif HAVE_SUNDIALS
%! ref = [0, 14.77810590694212];
%! [t, y] = ode15s (@(t,y) y, [-2, 0], 2);
%! assert ([t(end), y(end,:)], ref, 5e-2);

## InitialStep option
%!testif HAVE_SUNDIALS
%! opt = odeset ("InitialStep", 1e-8);
%! [t, y] = ode15s (@fpol, [0, 0.2], [2, 0], opt);
%! assert (t(2)-t(1), 1e-8, 1e-9);

## MaxStep option
%!testif HAVE_SUNDIALS
%! opt = odeset ("MaxStep", 1e-3);
%! sol = ode15s (@fpol, [0, 0.2], [2, 0], opt);
%! assert (sol.x(5)-sol.x(4), 1e-3, 1e-3);

## Solve in backward direction starting at t=0
%!testif HAVE_SUNDIALS
%! ref = [-1.205364552835178; 0.951542399860817];
%! sol = ode15s (@fpol, [0 -2], [2, 0]);
%! assert ([sol.x(end); sol.y(:,end)], [-2; ref], 5e-3);

## Solve in backward direction starting at t=2
%!testif HAVE_SUNDIALS
%! ref = [-1.205364552835178; 0.951542399860817];
%! sol = ode15s (@fpol, [2, 0 -2], fref);
%! assert ([sol.x(end); sol.y(:,end)], [-2; ref], 3e-2);

## Solve another anonymous function in backward direction
%!testif HAVE_SUNDIALS
%! ref = [-1; 0.367879437558975];
%! sol = ode15s (@(t,y) y, [0 -1], 1);
%! assert ([sol.x(end); sol.y(:,end)], ref, 1e-2);

## Solve another anonymous function below zero
%!testif HAVE_SUNDIALS
%! ref = [0; 14.77810590694212];
%! sol = ode15s (@(t,y) y, [-2, 0], 2);
%! assert ([sol.x(end); sol.y(:,end)], ref, 5e-2);

## Solve in backward direction starting at t=0 with MaxStep option
%!testif HAVE_SUNDIALS
%! ref = [-1.205364552835178; 0.951542399860817];
%! opt = odeset ("MaxStep", 1e-3);
%! sol = ode15s (@fpol, [0 -2], [2, 0], opt);
%! assert (abs (sol.x(8)-sol.x(7)), 1e-3, 1e-3);
%! assert ([sol.x(end); sol.y(:,end)], [-2; ref], 1e-3);

## AbsTol option
%!testif HAVE_SUNDIALS
%! opt = odeset ("AbsTol", 1e-5);
%! sol = ode15s (@fpol, [0, 2], [2, 0], opt);
%! assert ([sol.x(end); sol.y(:,end)], [2, fref].', 4e-3);

## AbsTol and RelTol option
%!testif HAVE_SUNDIALS
%! opt = odeset ("AbsTol", 1e-8, "RelTol", 1e-8);
%! sol = ode15s (@fpol, [0, 2], [2, 0], opt);
%! assert ([sol.x(end); sol.y(:,end)], [2, fref].', 1e-3);

## RelTol option -- higher accuracy
%!testif HAVE_SUNDIALS
%! opt = odeset ("RelTol", 1e-8);
%! sol = ode15s (@fpol, [0, 2], [2, 0], opt);
%! assert ([sol.x(end); sol.y(:,end)], [2, fref].', 1e-4);

## Mass option as function
%!testif HAVE_SUNDIALS
%! opt = odeset ("Mass", @fmas, "MStateDependence", "none");
%! sol = ode15s (@fpol, [0, 2], [2, 0], opt);
%! assert ([sol.x(end); sol.y(:,end)], [2, fref].', 3e-3);

## Mass option as matrix
%!testif HAVE_SUNDIALS
%! opt = odeset ("Mass", eye (2,2), "MStateDependence", "none");
%! sol = ode15s (@fpol, [0, 2], [2, 0], opt);
%! assert ([sol.x(end); sol.y(:,end)], [2, fref].', 3e-3);

## Mass option as sparse matrix
%!testif HAVE_SUNDIALS
%! opt = odeset ("Mass", speye (2), "MStateDependence", "none");
%! sol = ode15s (@fpol, [0, 2], [2, 0], opt);
%! assert ([sol.x(end); sol.y(:,end)], [2, fref].', 3e-3);

## Mass option as function and sparse matrix
%!testif HAVE_SUNDIALS
%! opt = odeset ("Mass", "fmsa", "MStateDependence", "none");
%! sol = ode15s (@fpol, [0, 2], [2, 0], opt);
%! assert ([sol.x(end); sol.y(:,end)], [2, fref].', 3e-3);

## Refine
%!testif HAVE_SUNDIALS
%! opt2 = odeset ("Refine", 3, "Mass", @massdensefunstate,
%!                "MStateDependence", "none");
%! opt1 = odeset ("Mass", @massdensefunstate, "MStateDependence", "none");
%! [t, y] = ode15s (@rob, [0, 100], [1; 0; 0], opt1);
%! [t2, y2] = ode15s (@rob, [0, 100], [1; 0; 0], opt2);
%! assert (numel (t2), numel (t) * 3, 3);

## Refine ignored if numel (trange) > 2
%!testif HAVE_SUNDIALS
%! opt2 = odeset ("Refine", 3, "Mass", "massdensefunstate",
%!                "MStateDependence", "none");
%! opt1 = odeset ("Mass", @massdensefunstate, "MStateDependence", "none");
%! [t, y] = ode15s ("rob", [0, 10, 100], [1; 0; 0], opt1);
%! [t2, y2] = ode15s ("rob", [0, 10, 100], [1; 0; 0], opt2);
%! assert (numel (t2), numel (t));

## Events option add further elements in sol
%!testif HAVE_SUNDIALS
%! opt = odeset ("Events", @feve, "Mass", @massdensefunstate,
%!               "MStateDependence", "none");
%! sol = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert (isfield (sol, "ie"));
%! assert (sol.ie, [1, 2]);
%! assert (isfield (sol, "xe"));
%! assert (isfield (sol, "ye"));
%! assert (sol.x(end), 10, 1);

## Events option, five output arguments
%!testif HAVE_SUNDIALS
%! opt = odeset ("Events", @feve, "Mass", @massdensefunstate,
%!               "MStateDependence", "none");
%! [t, y, te, ye, ie] = ode15s (@rob, [0, 100], [1; 0; 0], opt);
%! assert (t(end), 10, 1);
%! assert (te, [10; 10], 0.5);
%! assert (ie, [1; 2]);

## Initial solution as row vector
%!testif HAVE_SUNDIALS
%! A = zeros (2);
%! [tout, yout] = ode15s (@(t, y) A * y, [0, 1], [1, 1]);
%! assert (yout, ones (18, 2));

%!testif HAVE_SUNDIALS
%! A = zeros (2);
%! fail ("ode15s (@(t, y) A * y, [0, 1], eye (2))",
%!       "ode15s: Y0 must be a numeric vector");