recentering option for isolated system calculations

[Axel]

hi everybody,

recently i've been doing some test runs of isolated systems and
noticed
that - unlike with CPMD - the atomic positions are not recentered into
the box. to make the 'normal' code consistent with QM/MM it would be
nice to have an 'recenter at first step only' and an 'recenter every
step'
setup, with potentially also a C.O.M. and/or rotational momentum
removal.

any opinions?

while on the topic: hats off for the wavelet poisson solver!
i've been testing a system where the errors of the martyna/tuckerman
are visible pretty soon after going beyond the theoretically
'allowed'
maximum of the size of the charge distribution w.r.t. the box size
and
the wavelet solver beats it by a far margin. impressive!!

cheers,
    axel.