There is an error in theorem 2.1 of this paper, which shows a representation of the Fréchet derivative with respect to the boundary of the far field operator from scattering by obstacles with Robin boundary conditions. The boundary condition in equation (2.4) has to be replaced by
where H denotes the mean curvature of the boundary.
The error occurs in the derivation of equation (2.12). Therefore the proof of theorem 2.1 presented in this paper has to be modified on pages 376-7 along the following lines.
For the surface integral we obtain
This can be seen by considering a fixed point and a local parametrization , of with
The perturbed boundary is then locally described by . We compute at the point U0
with
and the orthonormality of the tangential vectors leads to
We obtain
This holds independently of the parametrization and uniformly on .
With the estimates (2.10)-(2.12) we first see from (2.8) by a perturbation argument that tends to u in . This is well known: the scattered field depends continuously on the boundary, or in our notation FR is a continuous operator (cf Kress and Zinn (1992) for the Dirichlet problem). Therefore, with equation (2.8) the proof is completed by showing
for all v.
Since u' is a radiating solution of (2.3)-(2.4) we obtain from the boundary condition
Now we proceed as in the paper with the only modification of adding the integral
in any following computation of the sesquilinear form . This rectifies the proof.
Reference
Kress R and Zinn A 1992 On the numerical solution of the three-dimensional inverse obstacle scattering problem J. Comput. Appl. Math.42 46-61