[lammps-users] rRESPA status in latest LAMMPS releases.

hi all,

can anybody here enlighten me on the status of rRESPA
in the latest release of LAMMPS (12Feb2007+patches)?

depending on the pair style i'm getting the weirdest
results (looks like some memory corruption somewhere)
or somewhat reasonable (but not too much) trajectories
and then a nice segfault/coredump when writing a restart
(here is an example stack trace on a x86 linux machine
running fedora core 6):
(gdb) where
#0 0x00b437bf in _int_free () from /lib/libc.so.6
#1 0x00b47550 in free () from /lib/libc.so.6
#2 0x080dac96 in LAMMPS_NS::Neighbor::init (this=0x8c6f0d0) at neighbor.cpp:498
#3 0x080ca78c in LAMMPS_NS::LAMMPS::init (this=0x8c6d008) at lammps.cpp:310
#4 0x08162ded in LAMMPS_NS::WriteRestart::command (this=0xbf9f7538, narg=1, arg=0x8ccb890) at write_restart.cpp:75
#5 0x080c9b0e in LAMMPS_NS::Input::execute_command (this=0x8c6d478) at style.h:71
#6 0x080c9edf in LAMMPS_NS::Input::file (this=0x8c6d478) at input.cpp:187
#7 0x080cf770 in main (argc=5, argv=0xbf9f8a04) at main.cpp:29

any suggestions, or some test inputs that _do_ work (i won't be
surprised if i ran into an 'undocumented feature' of the compiler)
are much appreciated.

thanks,
    axel.

Send an example (as simple as possible) of something
that doesn't work and we'll take a look.

Steve

ok. the stack trace was created by modifying
the melt example as follows.

it crashes on the write restart.

[[email protected] melt]$ diff -u in.melt in.melt-respa
--- in.melt 2007-02-07 19:41:31.000000000 -0500
+++ in.melt-respa 2007-02-19 18:58:00.000000000 -0500
@@ -20,6 +20,7 @@
fix 1 all nve

dump id all atom 10 dump.melt

Your first problem (crash on write_restart) is a memory bug with
rRESPA. I'll post a patch this AM. Your 2nd problem looks different,
I'll have to test further.

Steve