Error analysis of stabilized semi-implicit method of Allen-Cahn equation (Q1030565): Difference between revisions

From MaRDI portal
Set OpenAlex properties.
Import241208061232 (talk | contribs)
Normalize DOI.
 
Property / DOI
 
Property / DOI: 10.3934/dcdsb.2009.11.1057 / rank
Normal rank
 
Property / DOI
 
Property / DOI: 10.3934/DCDSB.2009.11.1057 / rank
 
Normal rank

Latest revision as of 13:54, 10 December 2024

scientific article
Language Label Description Also known as
English
Error analysis of stabilized semi-implicit method of Allen-Cahn equation
scientific article

    Statements

    Error analysis of stabilized semi-implicit method of Allen-Cahn equation (English)
    0 references
    2 July 2009
    0 references
    The author considers a stabilized semi-implicit (in time) scheme and a splitting scheme for the Allen-Cahn equation \[ \begin{cases} \phi_{t}-\triangle\phi+\frac{1}{\varepsilon^{2}}f(\phi)=0, \quad (x, t)\in\Omega\times(0, T),\\ \frac{\partial\phi}{\partial n}|_{\partial\Omega}=0, \quad \phi(t_{0})=\phi_{0}.\end{cases}\tag{1} \] where \(\Omega\subset\mathbb R^N\), \(N=1,2,3\), is a bounded domain having \(C^{1,1}\) boundary \(\partial\Omega\), \(n\) is the outward normal, \(f(\phi)=F'(\phi)\) and \(F(\phi)=\frac{1}{4}(\phi^{2}-1)^{2}\). This nonlinear problem is treated numerically via two different time-splitting approaches based on the finite-difference methodology. The author starts with a stabilized semi-implicit method in the form \[ \begin{cases}\frac{\phi_{m}-\phi_{m-1}}{k}-\triangle\phi_{m}+\frac{1}{\varepsilon^{2}}(f(\phi_{m-1})+s(\phi_{m}-\phi_{m-1}))=0,\\ \frac{\partial\phi_{m}}{\partial n}|_{\partial\Omega}=0;\end{cases}\tag{2} \] of first order and \[ \begin{cases} \frac{3\phi_{m+1}-4\phi_{m}+\phi_{m-1}}{2\delta t}-\triangle\phi_{m+1}+\frac{1}{\varepsilon^{2}}(2f(\phi_{m})-f(\phi_{m-1})\\ +s(\phi_{m+1}-2\phi_{m}+\phi_{m-1}))=0,\\ \frac{\partial\phi_{m+1}}{\partial n}|_{\partial\Omega}=0; \end{cases} \] of second order schemes applied to (1). He then establishes an a priori error estimate for the first-order scheme (2) only, and further improves it employing several spectrum estimates. The final error bound for the solutions of (1) via (2) then takes the form \[ \|\phi(t_{m})-\phi_{m}\|_{0} \lesssim k\varepsilon^{\min(-\sigma_{1}-2,-\sigma_{3})}, \] where \(k\displaystyle \lesssim \min(\varepsilon^{2}, \varepsilon^{\alpha_{1}}, \varepsilon^{\alpha_{2}})\) and the initial value \(\phi_0\) of (1) satisfies some additional assumptions with parameters \(\sigma_{1}\), \(\sigma_{2}\). Such a priori error estimates are numerically justified by comparison with the operator splitting schemes of the corresponding order.
    0 references
    finite-difference schemes
    0 references
    Allen-Cahn equation
    0 references
    semi-implicit methods
    0 references
    error analysis
    0 references
    stability
    0 references
    operator splitting schemes
    0 references
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references