[vsipl++] [patch] Toeplitz system solver

[Stefan Seefeld]

Jules Bergmann wrote:

> It may be worthwhile to make 'conj' have the same behavior as 
> 'impl_conj'.  This would let users write generic code that works for 
> both complex and real values.

I second that. As an extension, make herm() call trans() if its argument
is real. (I note that the code already contains a trans_or_herm() extension,
which we may merge with the herm() function.)

Regards,
		Stefan