计算化学公社

标题: 一模一样的初始文件，跑rerun，两次结果不一样 [打印本页]

作者
Author: xxx127 时间: 2024-12-3 15:18
标题: 一模一样的初始文件，跑rerun，两次结果不一样
本帖最后由 xxx127 于 2024-12-4 10:59 编辑

对比两个log文件，发现只有ld-seed这个参数有差别，有没有大佬知道这是什么原因导致的 (, 下载次数 Times of downloads: 1)
附我认为错误的模拟(ld-seed = 123604042)：
@ s0 legend "Coul-SR:leftin-leftin"
@ s1 legend "LJ-SR:leftin-leftin"
@ s2 legend "Coul-SR:rightin-rightin"
@ s3 legend "LJ-SR:rightin-rightin"
   0.000000  -478.628479  -118.852997 0.000000 0.000000
1.000000  -474.392029  -103.202522 0.000000 0.000000
2.000000  -471.920166  -122.774239 0.000000 0.000000
3.000000  -473.729126  -126.067245 0.000000 0.000000
4.000000  -470.662170  -121.280731 0.000000 0.000000
5.000000  -475.543701  -107.152908 0.000000 0.000000
6.000000  -472.631958  -117.776604 0.000000 0.000000
7.000000  -475.125946  -117.924507 0.000000 0.000000
8.000000  -472.335083  -97.736244 0.000000 0.000000
9.000000  -473.706116  -100.887009 0.000000 0.000000
10.000000  -477.310669  -91.141815 0.000000 0.000000
11.000000  -474.757172  -92.951813 0.000000 0.000000
12.000000  -476.350311  -84.839539 0.000000 0.000000
13.000000  -480.117004  -86.455193 0.000000 0.000000
14.000000  -471.353333  -86.339157 0.000000 0.000000
15.000000  -478.203552  -71.673683 0.000000 0.000000
16.000000  -479.165161  -85.107513 0.000000 0.000000
17.000000  -478.340668  -75.278351 0.000000 0.000000
18.000000  -473.911987  -83.797638 0.000000 0.000000
19.000000  -476.786499  -91.984390 0.000000 0.000000
20.000000  -478.927826  -100.777863 0.000000 0.000000
21.000000  -480.460876  -96.634804 0.000000 0.000000
22.000000  -474.064606  -105.588707 0.000000 0.000000
23.000000  -477.911804  -97.156754 0.000000 0.000000
24.000000  -478.648193  -99.559036 0.000000 0.000000
25.000000  -472.125824  -104.962997 0.000000 0.000000
26.000000  -485.988861  -85.459961 0.000000 0.000000
27.000000  -484.931335  -108.146156 0.000000 0.000000
28.000000  -478.451111  -104.143417 0.000000 0.000000
29.000000  -475.137848  -90.729675 0.000000 0.000000
30.000000  -483.564087  -85.601746 0.000000 0.000000
31.000000  -482.912079  -94.166672 0.000000 0.000000
32.000000  -478.855347  -90.254486 0.000000 0.000000
33.000000  -484.706238  -104.702026 0.000000 0.000000
34.000000  -486.294495  -94.498283 0.000000 0.000000
35.000000  -487.990967  -81.701225 0.000000 0.000000
36.000000  -484.663788  -101.429947 0.000000 0.000000
37.000000  -490.228119  -89.068947 0.000000 0.000000
38.000000  -481.703430  -101.721565 0.000000 0.000000
39.000000  -482.488403  -88.403488 0.000000 0.000000
40.000000  -484.626770  -102.168518 0.000000 0.000000
41.000000  -484.468872  -104.910881 0.000000 0.000000
42.000000  -483.734802  -96.607147 0.000000 0.000000
43.000000  -481.612854  -84.228065 0.000000 0.000000
44.000000  -479.892548  -85.868683 0.000000 0.000000
45.000000  -487.144745  -72.582977 0.000000 0.000000
46.000000  -482.595276  -88.166504 0.000000 0.000000
47.000000  -488.102997  -87.397392 0.000000 0.000000
48.000000  -487.386230  -93.904411 0.000000 0.000000
49.000000  -484.514435  -102.363945 0.000000 0.000000
50.000000  -483.913025  -102.513077 0.000000 0.000000

附我认为正确的模拟(ld-seed  = -139688839)：
@ s0 legend "Coul-SR:leftin-leftin"
@ s1 legend "LJ-SR:leftin-leftin"
@ s2 legend "Coul-SR:rightin-rightin"
@ s3 legend "LJ-SR:rightin-rightin"
0.000000  -478.628479  -118.852997  -467.359863  -342.902283
1.000000  -474.392029  -103.202522  -465.290649  -336.001678
2.000000  -471.920166  -122.774239  -466.924316  -356.056152
3.000000  -473.729126  -126.067245  -467.559814  -338.434265
4.000000  -470.662170  -121.280731  -465.025696  -371.711914
5.000000  -475.543701  -107.152908  -467.782623  -384.397583
6.000000  -472.631958  -117.776604  -466.166626  -358.762207
7.000000  -475.125946  -117.924507  -470.068909  -366.489777
8.000000  -472.335083  -97.736244  -462.545410  -358.059052
9.000000  -473.706116  -100.887009  -471.232727  -319.063080
10.000000  -477.310669  -91.141815  -470.217285  -335.902954
11.000000  -474.757172  -92.951813  -471.183716  -327.236603
12.000000  -476.350311  -84.839539  -469.967926  -344.779968
13.000000  -480.117004  -86.455193  -471.249084  -306.114471
14.000000  -471.353333  -86.339157  -471.060791  -324.502075
15.000000  -478.203552  -71.673683  -474.668976  -307.050629
16.000000  -479.165161  -85.107513  -477.765442  -314.934784
17.000000  -478.340668  -75.278351  -476.254822  -327.682922
18.000000  -473.911987  -83.797638  -478.664795  -309.315247
19.000000  -476.786499  -91.984390  -479.992493  -322.357452
20.000000  -478.927826  -100.777863  -482.816803  -320.063354
21.000000  -480.460876  -96.634804  -480.216309  -327.306519
22.000000  -474.064606  -105.588707  -486.441895  -365.361267
23.000000  -477.911804  -97.156754  -484.829285  -341.619141
24.000000  -478.648193  -99.559036  -479.883484  -340.641876
25.000000  -472.125824  -104.962997  -487.581238  -333.393951
26.000000  -485.988861  -85.459961  -484.417480  -307.918304
27.000000  -484.931335  -108.146156  -485.709625  -339.144897
28.000000  -478.451111  -104.143417  -486.556183  -334.229218
29.000000  -475.137848  -90.729675  -487.771271  -325.844971
30.000000  -483.564087  -85.601746  -492.515167  -303.332855
31.000000  -482.912079  -94.166672  -490.017212  -333.517761
32.000000  -478.855347  -90.254486  -496.321381  -306.370789
33.000000  -484.706238  -104.702026  -491.746674  -318.092560
34.000000  -486.294495  -94.498283  -485.954468  -325.412781
35.000000  -487.990967  -81.701225  -482.665833  -364.669556
36.000000  -484.663788  -101.429947  -489.174683  -344.040710
37.000000  -490.228119  -89.068947  -491.011230  -349.456055
38.000000  -481.703430  -101.721565  -487.405792  -366.052551
39.000000  -482.488403  -88.403488  -488.902466  -343.280945
40.000000  -484.626770  -102.168518  -492.261627  -362.391235
41.000000  -484.468872  -104.910881  -491.425903  -353.970398
42.000000  -483.734802  -96.607147  -489.960022  -335.697205
43.000000  -481.612854  -84.228065  -488.876587  -365.214081
44.000000  -479.892548  -85.868683  -488.594116  -355.748047
45.000000  -487.144745  -72.582977  -488.639221  -365.704620
46.000000  -482.595276  -88.166504  -492.805939  -362.576508
47.000000  -488.102997  -87.397392  -480.761169  -353.823486
48.000000  -487.386230  -93.904411  -484.718872  -368.828369
49.000000  -484.514435  -102.363945  -488.070496  -346.731201
50.000000  -483.913025  -102.513077  -482.503052  -369.271790

作者
Author: snljty2 时间: 2024-12-3 15:38
这个参数控制随机数种子，这个不一样模拟结果当然不一样。即使输入文件严格一致，因为并行计算的顺序问题长时间模拟后结果也不会严格一致。参考卢老师博文《数值误差对计算化学结果重现性的影响》。

作者
Author: raku 时间: 2024-12-3 16:25
种子不一样结果肯定不一样，修改为一样后再试试。通常就没问题了

作者
Author: xxx127 时间: 2024-12-3 17:18

snljty2 发表于 2024-12-3 15:38
这个参数控制随机数种子，这个不一样模拟结果当然不一样。即使输入文件严格一致，因为并行计算的顺序问题长 ...

感谢！
但这篇博文里说种子数目不允许用户设定，那也就是说这种差别不可避免吗？
主要是我的两次模拟中有一次是错误的，我该如何避免这种错误呢

作者
Author: xxx127 时间: 2024-12-3 17:20

raku 发表于 2024-12-3 16:25
种子不一样结果肯定不一样，修改为一样后再试试。通常就没问题了

感谢！！但二楼发的博文里说种子数目不允许用户设定
我的两次模拟中有一次是错误的，如果种子数目是随机生成的，那错误不可避免吗

作者
Author: Lance先生 时间: 2024-12-3 21:28
gromacs是基于麦克斯韦-玻尔兹曼统计对体系的原子随机赋予初速度，所以参数一致往往几次模拟也很难得到完全相同的结果。

作者
Author: student0618 时间: 2024-12-3 21:34

xxx127 发表于 2024-12-3 17:20
感谢！！但二楼发的博文里说种子数目不允许用户设定
我的两次模拟中有一次是错误的，如果种子数目是随机 ...

什么是“错误的模拟”？报错崩溃了嗎？
不能得到想要的結果可以是因為隨機性、用作參考的有問題、或者真是參數有問題等。除非是报错，否则不会有一条轨迹“错误”、一条轨迹正确这种事情。

作者
Author: xxx127 时间: 2024-12-4 10:55

student0618 发表于 2024-12-3 21:34
什么是“错误的模拟”？报错崩溃了嗎？
不能得到想要的結果可以是因為隨機性、用作參考的有問題、或 ...

没有崩溃
但是一个液滴内(rightin)所有分子间的coul-SR和lLJ-SR每一时刻都为0

附我认为错误的模拟：
@ s0 legend "Coul-SR:leftin-leftin"
@ s1 legend "LJ-SR:leftin-leftin"
@ s2 legend "Coul-SR:rightin-rightin"
@ s3 legend "LJ-SR:rightin-rightin"
   0.000000  -478.628479  -118.852997 0.000000 0.000000
1.000000  -474.392029  -103.202522 0.000000 0.000000
2.000000  -471.920166  -122.774239 0.000000 0.000000
3.000000  -473.729126  -126.067245 0.000000 0.000000
4.000000  -470.662170  -121.280731 0.000000 0.000000
5.000000  -475.543701  -107.152908 0.000000 0.000000
6.000000  -472.631958  -117.776604 0.000000 0.000000
7.000000  -475.125946  -117.924507 0.000000 0.000000
8.000000  -472.335083  -97.736244 0.000000 0.000000
9.000000  -473.706116  -100.887009 0.000000 0.000000
10.000000  -477.310669  -91.141815 0.000000 0.000000
11.000000  -474.757172  -92.951813 0.000000 0.000000
12.000000  -476.350311  -84.839539 0.000000 0.000000
13.000000  -480.117004  -86.455193 0.000000 0.000000
14.000000  -471.353333  -86.339157 0.000000 0.000000
15.000000  -478.203552  -71.673683 0.000000 0.000000
16.000000  -479.165161  -85.107513 0.000000 0.000000
17.000000  -478.340668  -75.278351 0.000000 0.000000
18.000000  -473.911987  -83.797638 0.000000 0.000000
19.000000  -476.786499  -91.984390 0.000000 0.000000
20.000000  -478.927826  -100.777863 0.000000 0.000000
21.000000  -480.460876  -96.634804 0.000000 0.000000
22.000000  -474.064606  -105.588707 0.000000 0.000000
23.000000  -477.911804  -97.156754 0.000000 0.000000
24.000000  -478.648193  -99.559036 0.000000 0.000000
25.000000  -472.125824  -104.962997 0.000000 0.000000
26.000000  -485.988861  -85.459961 0.000000 0.000000
27.000000  -484.931335  -108.146156 0.000000 0.000000
28.000000  -478.451111  -104.143417 0.000000 0.000000
29.000000  -475.137848  -90.729675 0.000000 0.000000
30.000000  -483.564087  -85.601746 0.000000 0.000000
31.000000  -482.912079  -94.166672 0.000000 0.000000
32.000000  -478.855347  -90.254486 0.000000 0.000000
33.000000  -484.706238  -104.702026 0.000000 0.000000
34.000000  -486.294495  -94.498283 0.000000 0.000000
35.000000  -487.990967  -81.701225 0.000000 0.000000
36.000000  -484.663788  -101.429947 0.000000 0.000000
37.000000  -490.228119  -89.068947 0.000000 0.000000
38.000000  -481.703430  -101.721565 0.000000 0.000000
39.000000  -482.488403  -88.403488 0.000000 0.000000
40.000000  -484.626770  -102.168518 0.000000 0.000000
41.000000  -484.468872  -104.910881 0.000000 0.000000
42.000000  -483.734802  -96.607147 0.000000 0.000000
43.000000  -481.612854  -84.228065 0.000000 0.000000
44.000000  -479.892548  -85.868683 0.000000 0.000000
45.000000  -487.144745  -72.582977 0.000000 0.000000
46.000000  -482.595276  -88.166504 0.000000 0.000000
47.000000  -488.102997  -87.397392 0.000000 0.000000
48.000000  -487.386230  -93.904411 0.000000 0.000000
49.000000  -484.514435  -102.363945 0.000000 0.000000
50.000000  -483.913025  -102.513077 0.000000 0.000000

附我认为正确的模拟：
@ s0 legend "Coul-SR:leftin-leftin"
@ s1 legend "LJ-SR:leftin-leftin"
@ s2 legend "Coul-SR:rightin-rightin"
@ s3 legend "LJ-SR:rightin-rightin"
0.000000  -478.628479  -118.852997  -467.359863  -342.902283
1.000000  -474.392029  -103.202522  -465.290649  -336.001678
2.000000  -471.920166  -122.774239  -466.924316  -356.056152
3.000000  -473.729126  -126.067245  -467.559814  -338.434265
4.000000  -470.662170  -121.280731  -465.025696  -371.711914
5.000000  -475.543701  -107.152908  -467.782623  -384.397583
6.000000  -472.631958  -117.776604  -466.166626  -358.762207
7.000000  -475.125946  -117.924507  -470.068909  -366.489777
8.000000  -472.335083  -97.736244  -462.545410  -358.059052
9.000000  -473.706116  -100.887009  -471.232727  -319.063080
10.000000  -477.310669  -91.141815  -470.217285  -335.902954
11.000000  -474.757172  -92.951813  -471.183716  -327.236603
12.000000  -476.350311  -84.839539  -469.967926  -344.779968
13.000000  -480.117004  -86.455193  -471.249084  -306.114471
14.000000  -471.353333  -86.339157  -471.060791  -324.502075
15.000000  -478.203552  -71.673683  -474.668976  -307.050629
16.000000  -479.165161  -85.107513  -477.765442  -314.934784
17.000000  -478.340668  -75.278351  -476.254822  -327.682922
18.000000  -473.911987  -83.797638  -478.664795  -309.315247
19.000000  -476.786499  -91.984390  -479.992493  -322.357452
20.000000  -478.927826  -100.777863  -482.816803  -320.063354
21.000000  -480.460876  -96.634804  -480.216309  -327.306519
22.000000  -474.064606  -105.588707  -486.441895  -365.361267
23.000000  -477.911804  -97.156754  -484.829285  -341.619141
24.000000  -478.648193  -99.559036  -479.883484  -340.641876
25.000000  -472.125824  -104.962997  -487.581238  -333.393951
26.000000  -485.988861  -85.459961  -484.417480  -307.918304
27.000000  -484.931335  -108.146156  -485.709625  -339.144897
28.000000  -478.451111  -104.143417  -486.556183  -334.229218
29.000000  -475.137848  -90.729675  -487.771271  -325.844971
30.000000  -483.564087  -85.601746  -492.515167  -303.332855
31.000000  -482.912079  -94.166672  -490.017212  -333.517761
32.000000  -478.855347  -90.254486  -496.321381  -306.370789
33.000000  -484.706238  -104.702026  -491.746674  -318.092560
34.000000  -486.294495  -94.498283  -485.954468  -325.412781
35.000000  -487.990967  -81.701225  -482.665833  -364.669556
36.000000  -484.663788  -101.429947  -489.174683  -344.040710
37.000000  -490.228119  -89.068947  -491.011230  -349.456055
38.000000  -481.703430  -101.721565  -487.405792  -366.052551
39.000000  -482.488403  -88.403488  -488.902466  -343.280945
40.000000  -484.626770  -102.168518  -492.261627  -362.391235
41.000000  -484.468872  -104.910881  -491.425903  -353.970398
42.000000  -483.734802  -96.607147  -489.960022  -335.697205
43.000000  -481.612854  -84.228065  -488.876587  -365.214081
44.000000  -479.892548  -85.868683  -488.594116  -355.748047
45.000000  -487.144745  -72.582977  -488.639221  -365.704620
46.000000  -482.595276  -88.166504  -492.805939  -362.576508
47.000000  -488.102997  -87.397392  -480.761169  -353.823486
48.000000  -487.386230  -93.904411  -484.718872  -368.828369
49.000000  -484.514435  -102.363945  -488.070496  -346.731201
50.000000  -483.913025  -102.513077  -482.503052  -369.271790

作者
Author: sobereva 时间: 2024-12-5 02:22

xxx127 发表于 2024-12-3 17:18
感谢！
但这篇博文里说种子数目不允许用户设定，那也就是说这种差别不可避免吗？
主要是我的两次模拟中 ...

我的博文根本没说GROMACS里随机数种子不允许自己设，ld-seed就是设这个的，默认为-1，代表赝随机数

欢迎光临计算化学公社 (http://bbs.keinsci.com/)