[makedoc]

[makedoc]

[makedoc]

shuffledAtoms was not properly set with particle decomposition

Lammps always shuffles the atom indices even when it is not running
on multiple processors.  As a context it is always important to use
the gatindex when collecting atom positions.  Previously this was
not done but now a check is added to ensure this is done if it is
needed.

for some reason clang stopped working with LD_LIBRARY_PATH was set to
a blank variable. Now the -rpath stuff is only used when LD_LIBRARY_PATH is set

[makedoc]

[makedoc]

[makedoc]

[makedoc]

Using numgroup as suggested by @swails on amber mailing list

(cherry picked from commit 200f4cb42744926a027e30393e56d315f24cd535)

Before this fix, amber was working correctly with multiple
replicas but not working correctly with a single replica.

Now both situations should be ok.

Notice that I used a hack to detect if replicas are present
or not in sander. Probably this should be cleaned up in
some way

(cherry picked from commit 0b7dc44635708e821e14ed6f1156317807e6db7d)

It should now work properly

(cherry picked from commit 6f96353a8a7975f7429426bce3777fb555deed65)

[makedoc]

[makedoc]

They were giving warnings with cppcheck

repeated assignment

initialization list

check for empty container

[almost]

[almost]