Next: 4.1.4 Implicit Methods
Up: 4.1 Initial Value Problems
Previous: 4.1.2 Stability and Accuracy
4.1.3 Explicit Methods
- Euler-Cauchy (from DNGF; see above)
- Leapfrog algorithm (from DST):
This is an example of a multistep technique, as timesteps
and contribute .
Stability analysis for such algorithms is as follows:
Let the explicit multistep scheme be written as
Inserting a slightly deviating solution
and computing the difference, we have
We combine the errors at subsequent time steps to a vector
and define the quadratic matrix
Then
Stability is guaranteed if
EXAMPLE 1:
Leapfrog / Relaxation equation:
Therefore
which means that
, and , and
the matrix
is
with eigenvalues
For real
we find that always, meaning
that the leapfrog scheme is unstable for the relaxation (or growth)
equation.
EXAMPLE 2:
Leapfrog / Harmonic oscillator:
and
For the amplification matrix we find
(with
)
with eigenvalues
But the modulus of this is always .
The leapfrog
algorithm is marginally stable for the harmonic oscillator.
Next: 4.1.4 Implicit Methods
Up: 4.1 Initial Value Problems
Previous: 4.1.2 Stability and Accuracy
Franz J. Vesely Oct 2005
See also: "Computational Physics - An Introduction," Kluwer-Plenum 2001