We have decided to discontinue the publication of preprints on our preprint server as of 1 March 2024. The publication culture within mathematics has changed so much due to the rise of repositories such as ArXiV (www.arxiv.org) that we are encouraging all institute members to make their preprints available there. An institute's repository in its previous form is, therefore, unnecessary. The preprints published to date will remain available here, but we will not add any new preprints here.

MiS Preprint

46/2007

The regularisation of the $N$-well problem by finite elements and by singular perturbation are scaling equivalent

Andrew Lorent

Abstract

Let $K:=SO\left(2\right)A_1\cup SO\left(2\right)A_2\dots SO\left(2\right)A_{N}$ where $A_1,A_2,\dots A_{N}$ are matrices of non-zero determinant. We establish a sharp relation between the following two minimisation problems.

Firstly the $N$-well problem with surface energy. Let $p\in\left[1,2\right]$. Let $$ I^p_{\epsilon}\left(u\right)=\int_{\Omega} d^p\left(Du\left(z\right),K\right)+\epsilon \left|D^2 u\left(z\right)\right|^2 dL^2 z $$ and let $A_F$ denote the subspace of functions in $W^{2,2}\left(\Omega\right)$ that satisfy the affine boundary condition $Du=F$ on $\partial \Omega$ (in the sense of trace), where $F\not\in K$. We consider the scaling (with respect to $\epsilon$) of $$ m^p_{\epsilon}:=\inf_{u\in A_F} I^p_{\epsilon}\left(u\right). $$ Secondly the finite element approximation to the $N$-well problem without surface energy.

We will show there exists a space of functions $\mathcal{D}_F^{h}$ where each function $v\in \mathcal{D}_F^{h}$ is piecewise affine on a regular (non-degenerate) $h$-triangulation and satisfies the affine boundary condition $v=l_F$ on $\partial \Omega$ (where $l_F$ is affine with $Dl_F=F$) such that for $$ \alpha_p\left(h\right):=\inf_{v\in \mathcal{D}_F^{h}} \int_{\Omega}d^p\left(Dv\left(z\right),K\right) dL^2 z $$ there exists positive constants $\mathcal{C}_1<1<\mathcal{C}_2$ (depending on $A_1,\dots A_{N}$, $\varsigma$, $p$) for which the following holds true $$ \mathcal{C}_1\alpha_p\left(\sqrt{\epsilon}\right)\leq m^p_{\epsilon}\leq \mathcal{C}_2\alpha_p\left(\sqrt{\epsilon}\right)\;\;\;\;\;\text{for}\text{all}\;\epsilon>0. $$