1. Consistency
A numerical scheme is consistent if its discrete operator (with finite differences) converges towards the continuous operator (with derivatives) of the PDE for Δ t , Δ x → 0 \Delta t, \Delta x\rightarrow 0 Δt,Δx→0 (vanishing truncation error).
2. Stability
“Noise” (from initial conditions, round-off errors,…) does not grow.
3. Convergence
The solution of the numerical scheme converges towards the real solution of the PDE for Δ t , Δ x → 0 \Delta t, \Delta x\rightarrow 0 Δt,Δx→0.
4. Relationships: Lax’s equivalence theorem
Lax’s equivalence theorem: Given a properly posed initial value problem and a finite-difference approximation to it that satisfies the consistency condition, stability is the necessary and sufficient condition for convergence.’’
一句话,也就是说,consistency和stability是 convergence的充分必要条件,即
consistency + stability
⇔
\Leftrightarrow
⇔ convergence。
这里再补充一下,识别性与一致性的关系:
识别性是一致性的必要条件。也就是,consistency
→
\to
→ identifiability.。
对于某一些特定的方法,一致性与识别性可能是等价的。
Reference: 【1】https://www.astro.uu.se/~bf/course/numhd_course/2_3_15Consistency_stability.html
【2】http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.515.3910&rep=rep1&type=pdf