Asymptotic expansions for approximate solutions of boundary integral equations

Akshay Rane and Kunalkumar Shelar

arXiv: 2411.00060v1 · PDF · 31 October 2024
Categories: math.NA, cs.NA
Licence: CC BY 4.0 The site shows the text.

Abstract

This paper uses the Modified Projection Method to examine the errors in solving the boundary integral equation from Laplace equation. The analysis uses weighted norms, and parallel algorithms help solve the independent linear systems. By applying the method developed by Kulkarni, the study shows how the approximate solution behaves in polygonal domains. It also explores computational techniques using the double layer potential kernel to solve Laplace equation in these domains. The iterated Galerkin method provides an approximation of order 2r+2 in smooth domains. However, the corners in polygonal domains cause singularities that reduce the accuracy. Adjusting the mesh near these corners can almost restore accuracy when the error is measured using the uniform norm. This paper builds on the work of Rude et al. By using modified operator suggested by Kulkarni, superconvergence in iterated solutions is observed. This leads to an asymptotic error expansion, with the leading term being O(h^4) and the remaining error term O(h^6), resulting in a method with similar accuracy.

Items 5

Proposition

Proposition 1

T4RF

If 0<\alpha<\alpha^{*}, u \in C_\alpha^1{(\Gamma)}, \beta=(\beta_1,...,\beta_r) satisfying \beta_j=\alpha_j-2/q_j and q_j>2/\alpha_j then

\left\lVertT(I-\pi_h)u\right\rVert_{0,\infty}\leq Ch^2 \left\lVertu\right\rVert_{1,\alpha}

and

\left\lVertT(I-\pi_h)u\right\rVert_{1,\beta}\leq Ch^2 \left\lVertu\right\rVert_{1,\alpha}

Lemma

Lemma 1

67G3

If u\in C^4(\tau), g\in C^3(\tau), then

\int_\tau ((u-\pi_h u)g)(t) dt=\frac{1}{12}h_\tau^2\int_\tau Du(t)Dg(t)dt+\gamma_\tau,\\ \text{ where } h_\tau=|\tau| \text{ and }

|\gamma_\tau|\leq Ch^4_\tau\sum_{i=0}^{3} \left\lVertD^{4-i}u\right\rVert_{0,\infty,\tau} \left\lVertD^i g\right\rVert_{0,1,\tau}.

Proposition

Proposition 2

0UBI

If u \in C_\alpha^4{(\Gamma)} , q_j >4/\alpha_j, j=1,2,...r then

(T(I-\pi_h)u)(s)=\frac{1}{12} \sum_{j=1}^{r} \sum_{\tau\in T^h, s_j \notin \tau} h_\tau^2 \int \partial_t {k(s,t)} Du(t) dt + O(h^4) \left\lVertu\right\rVert_{4,\alpha_j}

Proposition

Proposition 3

9DG4

If u\in C^4_\alpha(\Gamma), q_j>4/\alpha_j for j=1,2,...r \alpha_j=a_j-2/q_j and a_j>2/q_j then

\begin{split} &(T(I-\pi_h)T(I-\pi_h)u)(s)\\ &=\frac{1}{144} \sum_{j_1=1}^r \sum_{j_2=1}^r \sum_{\tau_1\in T_1^h, s_j\notin \tau_1} \sum_{\tau_2\in T_2^h, s_j\notin \tau_1} h_{\tau^*}^{4} \int_{\tau_1}\partial_t k(s,t) \left( \int_{\tau_2}\partial_t\partial_s k(s,t) Du(t)dt\right)dt + \mathcal{O}(h^6_j)\left\lVertu\right\rVert_{8,\alpha_j} \end{split}

Proof:

Theorem

Theorem 1

U0MD

\begin{gathered} If u \in C^{4}_{\alpha}(\Gamma), \alpha < \alpha^{*}, \alpha_j=2-q_j, q_j>2/a_j, for j=1,2,...,r ,then \\ \left\lVert\frac{1}{3}\biggl(4 \sum_{j=1}^r u_j^*-(4r-3)u_0^*\right\rVert_{0,\infty} \leq Ch^6 \end{gathered}

Source and licence

The text on this page is from arXiv:2411.00060v1 by Akshay Rane and Kunalkumar Shelar, under the licence CC BY 4.0. The corpus made these changes: it read the items out of the LaTeX source, it expanded the macros of the paper, it put each proof after its statement, and it converted the text to HTML. The formulas are the TeX of the paper, and KaTeX renders them. A formula or a reference can look different from the PDF. The page is published under CC BY 4.0.

The record of each item is in analysis/objects/2411/2411.00060.jsonl.