Zero-stability controls whether small perturbations are amplified uniformly as a fixed time interval is covered by more and more steps. For a constant-coefficient linear multistep method, the zero-step recurrence isZero-stability requires a bound , with independent of small and the starting perturbation. The root condition for a multistep method characterizes this uniform boundedness.[ref]
Fundamentals of Numerical Computation by Tobin A. Driscoll and others. Zero-stability of multistep methods, Theorem 74, root condition and recurrence-mode proof
For , every solution is , so perturbations remain bounded even though the alternating component does not decay. For , the mode grows like at fixed positive time and violates zero-stability. This is distinct from absolute stability, which tests nonzero and can constrain a particular step size.
The root condition for a multistep method says that every root of a polynomial of its first characteristic polynomial satisfies , and each root with is simple. For constant coefficients this is equivalent to zero-stability.
A root of multiplicity produces recurrence modes for . If , the exponential decay dominates every such polynomial factor, so repeated interior roots are allowed. At , any nonconstant polynomial factor grows, explaining the simplicity requirement. A root outside the unit disk gives exponential growth even when simple. Thus satisfies the condition, while does not.
An amplification root of a multistep method is parasitic when it represents an additional numerical mode rather than the principal mode approximating the exact differential-equation flow. For a consistent method the principal root starts at one as the step tends to zero, while other roots can start elsewhere. A simple parasitic root on the unit circle is compatible with zero-stability, but its motion outside the circle for a negative test parameter can destroy absolute stability. Fixed-time convergence and all-time damping must be distinguished.
Replacing the derivative in by a central finite difference gives . Its principal root is , while the parasitic amplification root is . On a fixed time interval, converges uniformly to exactly when and . A fixed nonzero parasitic amplitude alternates between two incompatible limits, but a vanishing starting error can converge despite the root's modulus exceeding one. This distinguishes fixed-time numerical convergence from long-time absolute stability.
The polynomial satisfies the root condition for a multistep method exactly for . For a root is below ; for the quadratic root product forces a root outside the unit disc. The endpoints have simple unit-modulus roots, and the repeated root at is strictly inside. A consistent method with this polynomial is convergent throughout the closed interval.
Articles by others on the same topic
There are currently no matching articles.