This is an archived course. A more recent version may be available at ocw.mit.edu.

1.5 Zero Stability and the Dahlquist Equivalence Theorem

1.5.3 Dahlquist Equivalence Theorem

Measurable Outcome 1.7, Measurable Outcome 1.8, Measurable Outcome 1.9, Measurable Outcome 1.10

In order for a multi-step method to be convergent (as described in Section 1.4.2), two conditions must be met:

Consistency: In the limit of \({\Delta t}\rightarrow 0\), the method must be a consistent discretization of the ordinary differential equation.

Stability In the limit of \({\Delta t}\rightarrow 0\), the method must not have solutions that can grow unbounded as \(n = T/{\Delta t}\rightarrow \infty\).

The Dahlquist Equivalence Theorem in fact guarantees that a consistent and stable multi-step method is convergent, and vice-versa:

Dahlquist Equivalence Theorem

A multi-step method is convergent if and only if it is consistent and stable.

Exercise The numerical scheme defined by the equation \(v^{n+1} + 4v^ n -5v^{n-1} = 4\Delta t f^ n + 2\Delta t f^{n-1}\) is

Exercise 1

Answer: The method is consistent, but not zero stable. Since it is not zero stable, the Dahlquist equivalence theorem tells us that the method is not convergent.