Re: AccelMD + GBIS

From: Thomas Evangelidis (tevang3_at_gmail.com)
Date: Sat Dec 01 2012 - 12:00:44 CST

I am quoting below my whole communication with Yi Wang (in separate
messages to be more readable), which might be helpful for other people as
well:

Hi Thomas,
    the dihedral mode of aMD shouldn't touch anything related with ELEC and
VDW -- the script doesn't 'calculate' these two energies --- it merely ask
the system to return the current step's energy and print them out. Only the
dihedral part is in fact manipulated.

    I'm quite puzzled by your result --- you have posted 4 outputs, 1 and 2
for aMD and 3 and 4 for regular energy output --- is 3 from the same file
as 1 and 4 from 2? Or, If so, this is really odd for NAMD-CPU and likely
indicates something very seriously wrong. We've tested the code thoroughly
and never saw this, so if this is indeed the case, please let me know the
NAMD version you used and I'll need to further debug it. In any case, it'll
be very helpful if you could post more output of your log --- say, from
step 0 to step 100 without any cutout.

Yi

On Thu, Nov 29, 2012 at 6:24 AM, Thomas Evangelidis
<tevang_at_pharm.uoa.gr<https://webmail02.uoa.gr/src/compose.php?send_to=tevang@pharm.uoa.gr>
> wrote:

> HI Yi,
>
> I have a question about the output of aMD. I have the feeling that the
> boost on the dihedrals works well on NAMD-CUDA, unlike the boost on the
> total energy. However, I find differences in the ELECT, VDW and POTENTIAL
> values that are printed in every line starting with "ACCELERATED MD:"? How
> are these 3 values calculated?
>
> Below are the first 3 such lines from my NAMD-CPU and NAMD-CUDA output:
>
> NAMD-CPU
> ACCELERATED MD: STEP 0 dV 511.291 dVAVG 511.291 BOND 589.719 ANGLE 1512.07
> DIHED 2195.83 IMPRP 0 ELECT -5463.64 VDW -665.55 POTENTIAL -1831.56
> ACCELERATED MD: STEP 100 dV 25.2511 dVAVG 65.6174 BOND 568.137 ANGLE
> 1553.34 DIHED 2743.06 IMPRP 0 ELECT -5495.64 VDW -656.323 POTENTIAL
> -1287.42
> ACCELERATED MD: STEP 200 dV 26.5659 dVAVG 34.4785 BOND 604.916 ANGLE
> 1573.26 DIHED 2740.83 IMPRP 0 ELECT -5554.53 VDW -681.267 POTENTIAL
> -1316.79
>
> NAMD-CUDA
> ACCELERATED MD: STEP 0 dV 511.291 dVAVG 511.291 BOND 589.719 ANGLE 1512.07
> DIHED 2195.83 IMPRP 0 ELECT -641.836 VDW -557.026 POTENTIAL 3098.76
> ACCELERATED MD: STEP 100 dV 25.2552 dVAVG 65.6171 BOND 568.14 ANGLE
> 1553.34 DIHED 2743.05 IMPRP 0 ELECT -624.136 VDW -574.191 POTENTIAL
> 3666.21
> ACCELERATED MD: STEP 200 dV 26.5747 dVAVG 34.481 BOND 604.895 ANGLE
> 1573.22 DIHED 2740.81 IMPRP 0 ELECT -635.562 VDW -563.641 POTENTIAL
> 3719.72
>
> As you can see these are the only values that differ. Yet the Energies
> seem to be identical. How can this happen? What should I trust eventually?
>
> NAMD-CPU
> ENERGY: 0 589.7187 1512.0739 2195.8329
> 0.0000 -10082.7644 -665.5502 0.0000 0.0000
> 2089.2282 -4361.4609 301.9812 -6450.6891
> -4334.2257 301.9812
> ENERGY: 100 568.1366 1553.3445 2743.0617
> 0.0000 -10146.9652 -656.3233 0.0000 0.0000
> 2115.1457 -3823.6001 305.7273 -5938.7458
> -3794.1522 300.1026
> ENERGY: 200 604.9161 1573.2592 2740.8257
> 0.0000 -10168.3852 -681.2667 0.0000 0.0000
> 2131.6608 -3798.9901 308.1145 -5930.6508
> -3767.1282 304.1386
>
> NAMD-CUDA
> ENERGY: 0 589.7187 1512.0739 2195.8329
> 0.0000 -10082.7801 -665.5344 0.0000 0.0000
> 2089.2282 -4361.4607 301.9812 -6450.6889
> -4334.2255 301.9812
> ENERGY: 100 568.1395 1553.3445 2743.0545
> 0.0000 -10146.9683 -656.3144 0.0000 0.0000
> 2115.1508 -3823.5933 305.7281 -5938.7441
> -3794.1445 300.1028
> ENERGY: 200 604.8949 1573.2175 2740.8110
> 0.0000 -10168.3465 -681.2638 0.0000 0.0000
> 2131.6969 -3798.9900 308.1197 -5930.6868
> -3767.1274 304.1387
>
>
> thanks,
> Thomas

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:22:19 CST