segfault when using fix nve on group with intel

I'm interested in applying a thermostat to the outer atoms of a box.
So I made an outer and inner group of atoms. I can integrate the inner
group of atoms with NVE, but when I turn on the intel suffix, I get a
segfault. (I hadn't got around to applying the thermostat before I hit
this)

This appears to reproduce the problem using a simple benchmark input script:

diff --git a/bench/in.eam b/bench/in.eam
index 1dc0e1a64..842ee8696 100644
--- a/bench/in.eam
+++ b/bench/in.eam
@@ -24,7 +24,18 @@ velocity all create 1600.0 376847 loop geom
neighbor 1.0 bin
neigh_modify every 1 delay 5 check yes

-fix 1 all nve
+variable edge equal 2.0
+variable xleft equal xlo+\{edge\} \+variable xright equal xhi\-{edge}
+variable yleft equal ylo+\{edge\} \+variable yright equal yhi\-{edge}
+variable zleft equal ylo+\{edge\} \+variable zright equal yhi\-{edge}
+region outer_edges block \{xleft\} {xright} \{yleft\} {yright}
\{zleft\} {zright} units lattice side out
+group outer_atoms region outer_edges
+group inner_atoms subtract all outer_atoms

thanks for reporting this, jesse and providing an input example.
the changes in pull request #1561 in the LAMMPS github project should correct the issue.
https://github.com/lammps/lammps/pull/1561

the change should be included in the next LAMMPS patch.

axel.

Thanks Axel for the quick fix!