No label defined (Q4249784)

About 15 years have passed since the claim made by \textit{D. Colton} and \textit{R. Kress} in their book [\((*)\) Integral equation methods in scattering theory (1983; Zbl 0522.35001)] that the development of the integral equation methods for the direct scattering problem is close to completion. However up till the present it remains incomplete, at least concerning the vector 3D problems for screens and nonsmooth scatterers. When solving the monochromatic scattering problem on surfaces of arbitrary shape, one usually applies the so-called electric field integral equation. Its theory is given in \((*)\) only for closed surfaces. And even for them, in the 3D case this theory is not sufficient to prove the convergence of the Galerkin method (the basic means to solve such equations numerically). For screens we had no rigorous background till 1994 [cf. \((**)\) \textit{Yu. G. Smirnov}, Zh. Vychisl. Mat. Mat. Fiz. 34, No. 10, 1461-1475 (1994; Zbl 0831.45009)]. It can be argued that we still have some algorithms that work quite successfully. However, we cannot prove this. Moreover, we cannot produce any reliable estimate of how accurate is the solution so obtained. The splitting theory \((**)\) is the only basis visible at present to advance these questions. In discussions of Galerkin schemes, a common first concern is to establish the approximation property of the basis functions. However, the requirement to prove the stability property is equally important. The latter is taken for granted for many problems, since it follows immediately from the strong ellipticity property. But this is not the case for the vector 3D electric field equation. Due to the splitting theory we know that the principal part of the electric field operator can be split into two parts, one being ``positive definite'' and the other ``negative definite'' when considered on relevant functional spaces. Consequently, the state-of-the-art algorithms should somehow match this splitting. In this paper, we discuss both properties, that of stability and of approximation. For the entire collection see [Zbl 0902.00041].