Interface for Dynamics (ODEs)

DyCon Toolbox aims to group all the problems studied by the research team of the chair of mathematics. That is why it is necessary to create a common interface for you dissolve studied equations. To get an idea of the variety of equations involved we will name some:

  • Heat Equation

  • Population Dynamics

  • Collective Behavior

  • Schrodinger Equation

  • Burgers Equation

  • Waves Equation

Since these equations can be solved in different ways, we have opted to create a communication interface with external programs. All the equations that are defined in DyCon toolbox they will be represented by programming classes. So that the classes defined are compatible with the entire system must meet the following requirements.

  • They must have a property InitialCondition, which will be a vector double of $[n \times 1]$ dimensions
  • They must have a tspan property, indicating the integration intervals. This must be a vector double $[1 \times Nt]$ , where Nt is the number of points in time.
  • They must have a method ‘solve’, which resumes the dynamics for that initial condition and for that interval ‘tspan’, This method must accept an optional parameter ‘Control’, which allows solving the dynamics dependent on a function over time. This optional parameter must be an array of dimensions $[m \times Nt]$, where m is the dimension of the control vector

These simple requirements allow to make conenxiones to other specialized programs in the resolution of the different types of equations, previously mentioned. Dycon Toolbox has already implemented a general version to define ODEs. If we wanted to define the following ODE:

We could define as follows:

Symbolic State and Control Vectors

Y = sym('y',[2 1]); U = sym('u',[2 1]);
%% Dynamics Definition
F = @(t,Y,U,Params) [ U(1) + sin(Y(1)*Y(2)) +   (Y(1)*Y(2))     ; ...
                      U(2) +      Y(2)      +  cos(Y(1)*Y(2)) ] ;

dynamics = ode(F,Y,U);

The class ode allows you to store all the information related to the dynamics, and meets all requirements. If we look inside this function

dynamics

dynamics = 

  ode with properties:

         StateVector: [1x1 struct]
             Control: [1x1 struct]
     DynamicEquation: [1x1 SymNumFun]
              Params: [0x0 param]
         Derivatives: [1x1 odeDerivatives]
    InitialCondition: [2x1 double]
           FinalTime: 1
                  Nt: 10
          MassMatrix: [2x2 double]
               label: ''
              Solver: @eulere
    SolverParameters: {}
               tspan: [1x10 double]
    ControlDimension: 2
      StateDimension: 2
                  dt: 0.1000


You can see the InitialCondition property.

There is also the solver method that allows solving the dynamics.

[tspan,solution] = solve(dynamics);


Table of contents