Next: About this document
Up: DynaSys
Previous: Numerical Details
There are several parameters that the user may set to help control the solution of equations. In order to set these, use the mouse or the keyboard to select the Numerics menu, and the Parameters item in it.
- Time Duration: This allows the user to choose for how long the computations should be carried out.
- Computational Bound: If floating point numbers become too large, the computer will reward you with an overflow error, which kills the program. To try to prevent this, the program keeps track of the size of the numbers computed, and if they grow larger than this computational bound, then it terminates computations on the orbit in question.
- Error Tolerance: The Runge-Kutta-Fehlberg scheme estimates the truncation error at each step and tries to control it by changing the step size. This parameter sets the tolerance below which the RKF routine tries to keep the error. A lower tolerance means a more accurate, but slower scheme. This same tolerance level is used to control the variable step Modified Euler and Modified Runge-Kutta methods: with a finer tolerance corresponding to a shorter spatial step length.
- Step Size: This is the most important parameter governing computations for fixed step size methods. When it is set to zero, the program computes the step size automatically. If this is unsatisfactory, you may set it manually.
- Minimum Step Size: If variable step size methods are allowed to change the step size at will, then they can sometimes stagnate. This parameter forces the program to keep moving. This is again used by the other variable step methods. At times it may be necessary to reduce the minimum step size in order to follow an orbit closely.
- Maximum Step Size: Again, this is to keep variable step size methods from becoming unreasonable.
Next: About this document
Up: DynaSys
Previous: Numerical Details
Kevin Cooper
Wed Sep 30 15:21:52 PDT 1998