[AMBER] TI simulation blowing up with pmemd.CUDA

From: Delwakkada Liyanage, Senal Dinuka <sdd313.msstate.edu>
Date: Sun, 10 Jan 2021 18:07:39 +0000

Dear Amber users,

I am trying to do TI in Amber18 with pmemd.cuda. The system is a small porphyrin ligand (

C44H38N8),

prepared in a 16 A TIP3P water box. When

 using softcore potentials in the two-step transformation for removal of vdW interaction, the simulation crashes. I noticed that mdout resulted in excessive VDWAAL interaction energies.
TI region 1


 NSTEP = 2997000 TIME(PS) = 243616.999 TEMP(K) =********* PRESS = 0.0
 Etot = ************** EKtot = ************** EPtot = -22631739.3346
 BOND = ************** ANGLE = -35.9490 DIHED = 0.0001
 1-4 NB = 0.0000 1-4 EEL = 0.0000 VDWAALS = **************
 EELEC = -28649.8232 EHBOND = 0.0000 RESTRAINT = 0.0000
 EKCMT = 0.0000 VIRIAL = 0.0000 VOLUME = 84490.2572
                                                    Density = 1.0313
 DV/DL = **************
 ------------------------------------------------------------------------------

 Softcore part of the system: 90 atoms, TEMP(K) = 7735744996.28
 SC_Etot= *********** SC_EKtot= *********** SC_EPtot = ***********

But when I do the calculation using pmemd.MPI (with 16 threads) with the same starting coordinates and velocities the calculation completes successfully.
It must be mentioned that before sampling, system is equilibrated for 6 ns at each lambda value, (16-point gaussian quadrature) and all simulations are started from the previous 'restart.'

I would really appreciate your thoughts on handling this setback.

Thank you

Senal Liyanage
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun Jan 10 2021 - 10:30:02 PST
Custom Search