You can safely delete, rather than compress, those large databases.

For MD, there's really no reason to save the fingerprints as the odds of encountering an atom in the exact same relative position is almost zero. The databases are really for saving duplicate calculations in training, and we should ideally make an option to not save database files during force calls. We have this on our to-do list, but I'm not sure when it's going to rise to the top.

For MD, I'd strongly recommend trying the experimental LAMMPS interface put together by Efrem Braun and Levi Lentz. For now, you can find instructions on this mailing list  -- there is an email from April 13 from Efrem that describes it. In our experience, it takes a couple of hours to get it all set up correctly, but once you do you can run MD much faster, parallelize the force calls, and it won't write any database files.


On Sun, Jul 22, 2018 at 11:51 AM, Khorshidi, Alireza <[log in to unmask]> wrote:
Hi Eyal,

Do you overwrite fingerprints of images? If yes, then there is a single file that occupies almost 400 GB, right? If not, can you give us an idea of how many images are there in a 25 pico-seconds MD run? How much space on disk each image fingerprint prime takes?


On Sun, Jul 22, 2018 at 4:38 AM, Eyal Oren <[log in to unmask]> wrote:
Hi all,

How much of disk space should I expect the fingerprints data-bases to require? The training phase was fine in this regard but afterwards when I ran MD my storage got eaten as is, i.e. 25 pico-seconds of 1024 atoms MD run expanded the directory "/amp-fingerprint-primes.ampdb" to almost 400 GB. Now I can't even use "amp-compress" because there is not enough space in the disk left for the operation. Do I miss here something? (I'll admit I don't quite sure why should an MD run would create fingerprints in the first place).


Andrew Peterson
Associate Professor
Brown University School of Engineering
Barus & Holley 247
184 Hope Street
Providence, RI 02912
(401) 863-2153