User-adjustable parameters

The CTDE domain has several unique simulation parameters that can be edited in the System Properties window.

The ODE solver

The CTDE domain provides several different mathematical methods for solving the ordinary differential equations (ODEs) used to simulate continuous or mixed-signal systems. (These methods are commonly called ODE solvers.) The ODE solver choice is user-selectable at execution time. Currently MLDesigner provides the six ODE solvers below.

Short name Long name Description
DOPRI5 Dormand Prince A Dormand-Prince method with order 5(4), embedded error estimation, and variable step size control.
DOPRI5D Dormand Prince Dense A Dormand-Prince method with order 5(4), embedded error estimation, variable step size control and dense output.
FE Forward Euler One-step Euler-Cauchy Method
RK2 Runge-Kutta 2 Second order Runge-Kutta method
RK4 Runge-Kutta 4 Classical Runge-Kutta method of order 4
ROS4 Rosenbrock-Wanner 4 Rosenbrock-Wanner method of order 4

At the moment, the short name of the ODE solver has to be supplied as text to the Solver property.

Solver parameters

Several parameters can be used to control the operation of the ODE solver. These parameters are listed below. (We assume that the reader is familiar with basic terms of ODE solvers shown in this table.)

Type Parameter Description
float StepSize Initial step size of the ODE solver. For fixed-step solvers, this step size is used throughout the whole simulation.
float MinStep Smallest allowed step size. If the error bounds cannot be met using this step size, the simulation is aborted.
float MaxStep The largest allowed step size. MinStep and MaxStep are used only in variable-step solvers.
int MaxNum Maximum number of iteration per step. This parameter applies only to implicit solvers.
float RelTol Allowed relative local truncation error of the solver. This value is useful when the absolute value of different states differs significantly (that is, by several orders of magnitude).
float AbsTol Maximum absolute local truncation error. Note that RelTol and AbsTol are used in conjunction if both values are non-zero. In this case, a weighted sum of these parameters is used as the maximum allowed local truncation error.