Bug with PPPM and compute ?

Dear Lammps developers,

You will find attached 2 test cases showing a strange behavior : they are exactly the same calculation except one has a compute command (followed by the corresponding dump).

In the case with compute & dump, the calculation crashes at PPPM initialization of the second step, in the case without compute, it does not.

Regards,

Alexandre BARTHELEMY|R&D Engineer

Soitec | Parc Technologique des Fontaines, 38190 Bernin (France)

Ce message, ainsi que tous les fichiers joints, est exclusivement destiné aux personnes dont le nom figure ci-dessus. Il peut contenir des informations dont la divulgation est à ce titre rigoureusement interdite. Dans l’hypothèse où vous avez reçu ce mail par erreur, merci de le renvoyer à l’adresse email ci-dessus et de détruire toute copie.
This message and any of the attachments may contain proprietary material for the sole use of the intended recipient. Any review or distribution by others is strictly prohibited. If you are not the intended recipient please contact the sender and delete all copies.

without_compute.out (28.9 KB)

with_compute.in (3.97 KB)

with_compute.out (44.9 KB)

without_compute.in (3.89 KB)

There was a memory deallocation bug in PPPM for per-atom data, when the
restart file was written between runs. Will post a patch later today.

Thanks for posting the example script.

Steve