[AMBER] differences in EGB energies sander vs. pmemd

From: Niransha Kumarachchi <nkumarachchi2019.fau.edu>
Date: Thu, 14 May 2020 20:20:28 -0300

Dear all,

I am trying run verify energies calculated by sander, sander.MPI, pmemd,
pmemd.MPI, and pmemd.cuda. What I am observing is differences in energies
calculated when pmemd.cuda is utilized. I have run 4 single-point energy
calculations for each of these cases where the initial structures are all
the same. Specifically, the EGB energies are different. Is this expected?

#
# PMEMD.MPI
#
#CASE Etot EKtot EPtot BOND ANGLE DIHED 1-4_NB 1-4_EEL
VDWAALS EELEC EGB REST
M1 -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
M2 -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
M3 -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
M4 -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000 
----
#
# PMEMD
#
#CASE     Etot  EKtot     EPtot   BOND   ANGLE   DIHED  1-4_NB   1-4_EEL
VDWAALS    EELEC       EGB   REST
P1   -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
P2   -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
P3   -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
P4   -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
----
#
# PMEMD.CUDA
#
#CASE     Etot  EKtot     EPtot   BOND   ANGLE   DIHED  1-4_NB   1-4_EEL
VDWAALS    EELEC       EGB   REST
R1   -419.9414 0.0000 -419.9414 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4369 0.0000
R2   -419.9414 0.0000 -419.9414 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4369 0.0000
R3   -419.9414 0.0000 -419.9414 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4369 0.0000
R4   -419.9414 0.0000 -419.9414 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4369 0.0000
----
#
# Sander MPI
#
#CASE     Etot  EKtot     EPtot   BOND   ANGLE   DIHED  1-4_NB   1-4_EEL
VDWAALS    EELEC       EGB   REST
S1   -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
S2   -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
S3   -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
S4   -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
----
#
# Sander
#
#CASE     Etot  EKtot     EPtot   BOND   ANGLE   DIHED  1-4_NB   1-4_EEL
VDWAALS    EELEC       EGB   REST
T1   -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
T2   -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
T3   -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
T4   -419.9416 0.0000 -419.9416 4.6826 97.7724 69.2333 43.9284 -326.5097
-11.5807 -25.0308 -272.4371 0.0000
----
The input file I use is as follows:
0 step MD for 100
 &cntrl
  imin=0,ntb=0,cut=999,nstlim=0,dt=0.001,nscm=5000,
  nrespa=1,ntt=3,gamma_ln=1,tempi=0,temp0=0,
  igb=5, saltcon=0.3, ig=666,
 /
One thing I noted is that minimization done via the CPU or GPU versions of
pmemd display differences. Is this expected? Thanks.
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu May 14 2020 - 13:30:02 PDT
Custom Search