Re: negative/complex epsilon

On Wed, 28 Nov 2007, Xiaoyang Huang wrote:
>  The second question is about using anistropic dielectric constant. I 
> wanted to see the band structures of complex epslion material. I want to 
> set off-diagonal epsilon a complex number. I have configured 
> "--with-hermitian-eps". But i got error as shown in the output file 
> (petra.out) attached here. Also attached is the comp-eps.ctl file.

Complex epsilon matrices are allowed, but they must be hermitian and 
positive definite -- the epsilon you have supplied violates this condition 
(it has a negative eigenvalue).

Regards,
Steven G. Johnson