77d9547e9560c67e.tex
1: \begin{abstract}
2:   Although high-order Maxwell integral equation solvers provide
3:   significant advantages in terms of speed and accuracy over
4:   corresponding low-order integral methods, their performance
5:   significantly degrades in presence of non-smooth geometries---owing
6:   to field enhancement and singularities that arise at sharp edges and
7:   corners which, if left untreated, give rise to significant accuracy
8:   losses. The problem is particularly challenging in cases in which
9:   the ``density'' (i.e., the solution of the integral equation) tends
10:   to infinity at corners and edges---a difficulty that can be bypassed
11:   for 2D configurations, but which is unavoidable in 3D Maxwell
12:   integral formulations, wherein the component tangential to an edge
13:   of the electrical-current integral density vector tends to infinity
14:   at the edge. In order to tackle the problem this paper restricts
15:   attention to the simplest context in which the unbounded-density
16:   difficulty arises, namely, integral formulations in 2D space whose
17:   integral density blows up at corners; the strategies proposed,
18:   however, generalize directly to the 3D context. The novel
19:   methodologies presented in this paper yield high-order convergence
20:   for such challenging equations and achieve highly accurate solutions
21:   (even near edges and corners) without requiring a priori analysis of
22:   the geometry or use of singular bases.
23: \end{abstract}
24: