Not that I really know how it works, but here's an absolutely filthy fixif you have limits ( 0 64 )
that seems to work, at least for the spacegroups which I tried. If you
specify GRID SAMPLE in FFT, then there's a line in the log file that says
"Map limits in grid points on xyz"
and some numbers. Add 1 (one) to those numbers and give them to EXTEND as
grid, then ARP is happy.
I wish I knew why.
BUT, again, the easiest is simply to define the AU limits
in fractional coors in mapmask ... just take care in trigonal and cubic sg's
to use 0.334 instead of 0.333333 and 0.0834 instead of 0.08333333
There are consistent inconsistencies with real space asymmetric units between the various programs :-(
Ah, yes, this is an ellegant description of the situation.
I actually think that only ARP is 'inconsistent', but its very
polite on telling you what it really needs as AU limits.