[AMBER] Possible memory leak with LMOD

From: Charo del Genio <the.paraw.gmail.com>
Date: Tue, 22 May 2018 17:39:08 +0100

Dear all,
        I'm trying to use LMOD to dock a ligand on a protein, but I'm running into some odd behaviour. After the initial TNCG minimization, which completes in quite a short time, the amount of memory used
keeps growing, until the system runs out of memory and the process is killed. The protocol I am trying to use computes 15 modes and selects 5 of them for the perturbations, doing a total of 100
iterations and restarting every 10, using a pool of 10 configurations and keeping up to 50 in conflib.dat. For the code, I am using NAB from AmberTools18, compiled with gcc 6.4.0. The same behaviour
occurs with NAB, MPINAB and NAB with OpenMP support.

Again, to be clear, the memory use does not increase during the calculation of the eigenvectors, but only during the actual LMOD perturbations, and it keeps increasing, slowly but steadily. Has anyone
come across a similar behaviour? Am I misunderstanding something?



Our new book on Synchronization phenomena is out: http://www.cambridge.org/9781107056268
Dr. Charo I. del Genio
University of Warwick
Gibbet Hill Road
Coventry CV4 7AL
Tel: +44 (0)759 311 9487
AMBER mailing list
Received on Tue May 22 2018 - 10:00:03 PDT
Custom Search