Skip to content

practical numerical optimization Heidelberg ( Matlab integration of SUNDIALS )

License

Notifications You must be signed in to change notification settings

nashmit/SUNDIALS2Matlab

Repository files navigation

SUNDIALS2Matlab

practical numerical optimization Heidelberg ( Matlab integration of SUNDIALS )

Instalation!

This project requires CasADi framework which can be taken from here: https://web.casadi.org/get/ The CasADi folder must be placed inside SUNDIALS2Matlab folder.

NOTE: This is an initial description of the interface. I will extend the number of needed functions over time. But i guess the description below enables you to start the implementation. The names of the functions are not very precise; maybe you have a better idea...

Dynamical System Description

$$ \dot{x}(t) = f(t, x(t), q, p) \\ x(t_0) =x_0 $$

  • $t$ time
  • $x$ differential states
  • $q$ control( constant)
  • $p$ parameter

Calling Convention for Matlab Functions

All functions are called with on input argument inp and return one output argument outp. So all function calls have the header function outp = functionname(inp). Not all input attributes are always needed and not alle output attributes are always computed

The input has the following attributes:

  • inp.thoriz -- integration horizon in the form [$t_0$ $t_f$]
  • inp.sd -- initial value for differential states $x_0$
  • inp.sa -- initial value for algebraic stats (not used right now)
  • inp.q -- constant control for integration horizon
  • inp.p -- local parameters
  • inp.sensdirs -- sensitivity directions in form of matrix
  • inp.lambda -- adjoint sensitivity direction

The output has the following attributes:

  • outp.x -- solution of ode at $t_f$
  • outp.G -- sensitivities with respect to $x_0$ and $q$

About

practical numerical optimization Heidelberg ( Matlab integration of SUNDIALS )

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published