|
|
本帖最后由 JAM452 于 2025-10-21 23:07 编辑
各位老师好,我想进行200ns的“底物+抑制剂+蛋白质”的三元复合物的MD模拟,NVT平衡阶段顺利结束,但正式的MD重复了几次都在70-100ns之间非正常结束,出现的提示有两种情况,有时是 “ Can not fix pbc. ”,有时是(更多)“[1]+ Segmentation fault (core dumped)”。MD阶段没有产生gro文件,使用NVT平衡产生的gro文件和从md.xtc截取的最后几ns轨迹,通过VMD可视化检查没有发现明显不合理。
想请各位老师帮忙分析异常终止的可能原因,麻烦了!
GROMACS version: 2021.4
cpu( Intel (R) Xeon (R) Silver 4310)有12核,使用命令是: nohup gmx mdrun -v -deffnm md -nt 6 -update gpu -gpu_id 0 -pin on -pinoffset 0 -pinstride 2 &
gpu(RTX 3090)显存24G
文件无法上传,详细的模拟文件放在了网盘:通过网盘分享的文件:模拟文件.rar
链接: https://pan.baidu.com/s/18CFMuXGy0SzJ_IGnYunFWA?pwd=dvge 提取码: dvge
运行时的cpu占用情况:
nohup.out末尾:
$ tail -f nohup.out
100000000 steps, 200000.0 ps.
step 16900: timed with pme grid 64 64 64, coulomb cutoff 1.000: 63.2 M-cycles
step 17100: timed with pme grid 56 56 56, coulomb cutoff 1.097: 69.9 M-cycles
step 17300: timed with pme grid 48 48 48, coulomb cutoff 1.280: 77.1 M-cycles
step 17500: timed with pme grid 52 52 52, coulomb cutoff 1.182: 78.8 M-cycles
step 17700: timed with pme grid 56 56 56, coulomb cutoff 1.097: 70.1 M-cycles
step 17900: timed with pme grid 60 60 60, coulomb cutoff 1.024: 63.5 M-cycles
step 18100: timed with pme grid 64 64 64, coulomb cutoff 1.000: 67.5 M-cycles
optimal pme grid 64 64 64, coulomb cutoff 1.000
step 36648700, will finish Tue Oct 21 20:30:49 2025
md.log末尾:
$ tail -f md.log
COM: 3.75357 3.73217 3.95918
P: -209.80145 147.17165 -313.60504
V: -0.00638 0.00447 -0.00953
J: -536.54285 -730.00024 173.38251
w: -0.00768 -0.01064 0.00303
Inertia tensor (3x3):
Inertia tensor[ 0]={ 4.29591e+04, 6.90597e+03, 4.77303e+03}
Inertia tensor[ 1]={ 6.90597e+03, 4.35239e+04, -1.18188e+04}
Inertia tensor[ 2]={ 4.77303e+03, -1.18188e+04, 3.40181e+04}
Group Protein with mass 3.29075e+
|
|