Dear Users:
I was able to get T-REMD running in the NPT ensemble by commenting out the error message in src/pmemd/src/mdin_ctrl_dat.F90
< if (ntp .gt. 0) then
<
< write(mdout, '(a,a)') error_hdr, 'REMD cannot be run with ntp > 0!'
< inerr = 1
<
< end if
However, the job halts after about 70 ns/replica with the following message:
ERROR: Calculation halted. Periodic box dimensions have changed too much from their initial values.
Your system density has likely changed by a large amount, probably from
starting the simulation from a structure a long way from equilibrium.
>From trajectories, the system doesn't seem to be "blowing up" in an integration failure sense, so I presume that what is happening is simply that the though since temperatures are exchanged rather than coordinates (and hence the box is not exchanged) it is not clear to me how this would be effected by T-REMD. I see that the relevant error message is in src/pmemd/src/cuda/gpu.c and I am planning on simply commenting out this error message as well. However, I have some concerns about the following additional message that was printed to stderr/stdout as my job was halted:
The GPU code does not automatically reorganize grid cells and thus you
will need to restart the calculation from the previous restart file.
This will generate new grid cells and allow the calculation to continue.
It may be necessary to repeat this restarting multiple times if your system
is a long way from an equilibrated density.
It is not entirely clear to me what this means. I have no problem starting again from the beginning, so I am not in need of a way to generate new grid cells, but I am wondering if this will somehow effect a T-REMD run with pressure coupling (as in whether it will for e.g. mean that nonbonded interactions are missed rather than simply degrading performance). I would appreciate it if anybody has some ideas about the importance of (a) the PBC dimensions changing too much or (b) this comment about reorganizing grid cells.
The largest recorded positive change in volume: (18% increase) initial at t0 = 382379.4994 and at some point during the run, 450850.8417
The largest recorded negative change in volume: (13% decrease) initial at t0 = 442839.6928 and at some point during the run, 385643.6992
Note that I started each replica with a conformation (and box size) that had been equilibrated already at the starting temperature for that replica.
>From the logs that I have, I don't see any largest change immediately before the halt, though I also don't save the energies (and hence the volumes) every step. In the table that follows, first column is replica, second column is *max* volume difference from t to t+50 ps, and third column is the *last* such volume difference recorded before the halt.
$ for ((i=1;i<=60;i++)); do REP=$(printf "%03d" $i); grep VOLUME MD.mdout.$REP | awk '{if(NR!=1){d=$NF-last;if(d<0)d*=-1.0;print NR,d};last=$NF}'> tmp; max=$(cat tmp |sort -n -k 2 |tail -n 1 | awk '{print $2}'); last=$(cat tmp|tail -n 1 |awk '{print $2}'); echo $REP $max $last; done
001 5717.14 1331.61
002 7965.82 1732.08
003 5118.88 2178.76
004 5196.47 1858.21
005 5347.42 2924.22
006 5618.12 1357.75
007 4655.18 1243.8
008 7648.98 1953.42
009 7179.7 42.0076
010 5549.81 439.308
011 4893.99 594.135
012 8597.4 3007.45
013 5238.72 1295.94
014 7932.11 1348.97
015 5632.89 312.725
016 5628.1 718.705
017 5906.35 1480.63
018 5790.5 3248.41
019 6129.48 699.029
020 9237.54 858.447
021 7076.76 1222.76
022 8498.18 572.783
023 8951.82 5152.34
024 7662.97 32.6745
025 7074.56 2571
026 9152.38 533.203
027 6077.94 2180.5
028 5866.51 537.677
029 7006.05 4434.32
030 8895.76 2625.68
031 10193.8 1051.85
032 6939.52 733.162
033 5225.87 449.429
034 6320.66 53.5065
035 7712.99 3510.23
036 7006.88 223.376
037 7597.94 4189.95
038 8252.34 707.967
039 9834.48 3355.27
040 8664.68 2288.8
041 8506.97 978.02
042 7859.59 2638.47
043 7837.08 4372.9
044 9263.26 230.814
045 9651.46 2234.11
046 8071.91 1434.64
047 7312.67 3263.4
048 8233.13 927.043
049 8593.26 1719.09
050 10031.3 1951.99
051 7921.8 605.211
052 11189.6 2612.17
053 9959.93 539.968
054 7579.55 1072.76
055 8209.21 1236.75
056 9640.55 1377.32
057 10861.9 2306.08
058 8642.93 2376.92
059 9185.31 1910.51
060 9581.79 226.088
Thank you,
Chris.
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Dec 19 2016 - 09:30:02 PST
