计算化学公社

 找回密码 Forget password
 注册 Register
Views: 9539|回复 Reply: 3
打印 Print 上一主题 Last thread 下一主题 Next thread

[Amber] 求助:使用Amber的ABMD计算自由能但是出现很多0?

[复制链接 Copy URL]

31

帖子

0

威望

95

eV
积分
126

Level 2 能力者

跳转到指定楼层 Go to specific reply
楼主
使用Amber的ABMD计算自由能。使用二维来推演一个C-O键的断裂与另一个的形成,colva如下
&colvar
cv_type = 'DISTANCE'
cv_ni = 2, cv_i = 8335, 8317
cv_min = 1.5, cv_max = 4 ! min is not needed for DISTANCE
resolution = 0.2 ! required for mode = FLOODING
&end
&colvar
cv_type = 'DISTANCE'
cv_ni = 2, cv_i = 8335, 8322
cv_min = 1.5, cv_max = 4 ! min is not needed for DISTANCE
resolution = 0.2 ! required for mode = FLOODING
&end

脚本控制文件如下
&cntrl
imin=0,           ! Perform MD
ntx=5,            ! Read coordinates and velocities from ASCII formatted rst7 coordinate file
irest=1           ! restart simulation
nstlim=50000,     ! Number of MD steps
dt=0.002,         ! Timestep (ps)

ntp=1,            ! Use the Berendsen barostat for constant pressure simulation
ntc=2,            ! Enable SHAKE to constrain all bonds involving hydrogen
ntf=2,            ! Setting to not calculate force for SHAKE constrained bonds
ntt=3,            ! Langevin thermostat
gamma_ln=2.0      ! Collision Frequency for thermostat
ig=-1,            ! Random seed for thermostat
temp0=300         ! Simulation temperature (K)

............................略中间一些读写参数
&qmmm
qmmask=':509-513', ! Residue 509-512 should be treated using QM
qm_theory='DFTB3', ! Use the DFTB3 semi-empirical Hamiltonian
qmcut=14.0         ! Use 14 angstrom cut off for QM region
qmcharge= 3        ! The charge of the quantum section
qmshake=1          ! Shake QM H atoms

&end
&abmd
mode = 'FLOODING'              ! sets the execution mode
monitor_file= 'FW_1step.out'   ! sets the name of the file to which value(s) of reaction coordinate(s)
monitor_freq=50                ! the frequency of the output to the monitor_file
timescale=1                    ! τF, the flooding timescale in picoseconds
umbrella_file='basic_1step.nc' ! biasing potential file name
selection_freq=50              ! positive integer number that sets the frequency of the resampling algorithm (in MD steps)
selection_constant=0.001       ! Parameter C is to determine how strong the selection mechanism is. If C is too large, all the walkers will be replaced with the most dominant one.If C is too small, there will be no killing/duplicating of walkers
selection_epsilon=0            ! positive real number (typically less than unity) that sets the stopping criterion parameter ε
wt_temperature=10000           ! The smaller the T’; the smoother/slower the convergence
wt_umbrella_file='wt_basic_1step.nc' !the file name of true biasing potential after modification by 1 + ( T /T’ ) in which T is thereference temperature of the system (temp0)
cv_file = 'cv.rst'
&end

运行完后我发现在我的初始结构还能得到能量值
cv#1                      cv#2                  E
......................
1.500000e+00   2.950000e+00  -4.077634e+00
   1.500000e+00   3.000000e+00  -3.342485e+00
   1.500000e+00   3.050000e+00  -2.698778e+00
   1.500000e+00   3.100000e+00  -2.157272e+00
   1.500000e+00   3.150000e+00  -1.817898e+00
   1.500000e+00   3.200000e+00  -1.734833e+00

--------------------
但是随着cv#1 增大及 cv#2减少所有的值都是0
...........................
1.950000e+00   3.750000e+00  -9.039146e-49
   1.950000e+00   3.800000e+00  -3.118966e-48
   1.950000e+00   3.850000e+00  -3.550809e-48
   1.950000e+00   3.900000e+00  -1.548028e-48
   1.950000e+00   3.950000e+00  -2.153079e-49
   1.950000e+00   4.000000e+00   0.000000e+00
   2.000000e+00   1.500000e+00   0.000000e+00
   2.000000e+00   1.550000e+00   0.000000e+00
   2.000000e+00   1.600000e+00   0.000000e+00
   2.000000e+00   1.650000e+00   0.000000e+00
   2.000000e+00   1.700000e+00   0.000000e+00
   2.000000e+00   1.750000e+00   0.000000e+00

请问是采样不充分吗?
还是我的参数设置有错误?
谢谢


156

帖子

0

威望

1663

eV
积分
1819

Level 5 (御坂)

2#
发表于 Post on 2019-5-27 23:51:04 | 只看该作者 Only view this author
100 ps 取样时间太少了, 另外你可以看看轨迹,有没有多次取样键的断裂与形成呢?
你同时取样的是2个CVs, 这样会增加取样的难度。
建议改为一个CV, d=cv1-cv2 (或者cv2-cv1), amber里可以这样设置

1

帖子

0

威望

101

eV
积分
102

Level 2 能力者

3#
发表于 Post on 2021-9-23 17:02:15 | 只看该作者 Only view this author
请问有没有amber ABMD的教程呀

39

帖子

0

威望

517

eV
积分
556

Level 4 (黑子)

4#
发表于 Post on 2022-12-4 00:58:51 | 只看该作者 Only view this author
本帖最后由 hangmint 于 2022-12-4 01:09 编辑
beyond 发表于 2019-5-27 23:51
100 ps 取样时间太少了, 另外你可以看看轨迹,有没有多次取样键的断裂与形成呢?
你同时取样的是2个CVs, ...
您好,我同样用的ABMD模拟了键的断裂,我用nfe-umbrella-slice 获取PMF之后,最开始的CV坐标对应的能量是不随时间变化的,请问有什么好的建议吗?
-1.800000e+00,0.000000e+00-1.750000e+00,0.000000e+00
-1.700000e+00,0.000000e+00
-1.650000e+00,0.000000e+00
...
-8.000000e-01,0.000000e+00

cv.in

&colvar
cv_type ='LCOD'
cv_min=-2.0, cv max = 4, resolution = 0.2
cv_ni = 4.cv nr =2
cv_i = 12194,12104,12100,12126
cv_r =-1.0,1.0
其它文件跟楼主的类似,只有cv_type修改为LCOD,我的采样时间只有5ps

本版积分规则 Credits rule

手机版 Mobile version|北京科音自然科学研究中心 Beijing Kein Research Center for Natural Sciences|京公网安备 11010502035419号|计算化学公社 — 北京科音旗下高水平计算化学交流论坛 ( 京ICP备14038949号-1 )|网站地图

GMT+8, 2026-2-25 02:17 , Processed in 0.173557 second(s), 26 queries , Gzip On.

快速回复 返回顶部 返回列表 Return to list