|
|
pymbar调用MBAR计算PMF的输出结果如下:
Binning data...
Evaluating reduced potential energies...
Running MBAR...
K (total states) = 17, total samples = 142714
N_k =
[11314 9990 11112 8861 7883 4661 7781 8741 8105 8059 7063 8092
8317 7566 8936 8067 8166]
There are 17 states with samples.
Initializing free energies to zero.
Initial dimensionless free energies with method zeros
f_k =
[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]
Determining dimensionless free energies by Newton-Raphson / self-consistent iteration.
self consistent iteration gradient norm is 9.0031e+07, Newton-Raphson gradient norm is 6.3272e+08
Choosing self-consistent iteration on iteration 0
self consistent iteration gradient norm is 5.7465e+07, Newton-Raphson gradient norm is 1.8677e+07
Choosing self-consistent iteration for lower gradient on iteration 1
self consistent iteration gradient norm is 3.9352e+07, Newton-Raphson gradient norm is 2.2391e+06
Newton-Raphson used on iteration 2
self consistent iteration gradient norm is 3.8027e+05, Newton-Raphson gradient norm is 787.14
Newton-Raphson used on iteration 3
self consistent iteration gradient norm is 101.9, Newton-Raphson gradient norm is 4.4636e-05
Newton-Raphson used on iteration 4
self consistent iteration gradient norm is 6.5189e-06, Newton-Raphson gradient norm is 1.6842e-19
Newton-Raphson used on iteration 5
self consistent iteration gradient norm is 2.3465e-20, Newton-Raphson gradient norm is 1.9608e-22
Newton-Raphson used on iteration 6
Converged to tolerance of 1.257696e-14 in 7 iterations.
Of 7 iterations, 5 were Newton-Raphson iterations and 2 were self-consistent iterations
Final dimensionless free energies
f_k =
[ 0. -6.35575236 -10.35715215 -12.5221046 -13.38232867
-13.51146418 -13.41177059 -13.78373329 -14.5068132 -15.38831936
-16.27500615 -17.20686221 -18.04267401 -18.68934648 -19.12060088
-19.19076083 -19.03618277]
MBAR initialization complete.
PMF (in units of kT)
bin f df
1.81 24.835 0.512
1.83 24.562 0.461
1.84 23.407 0.290
…………(下略)…………
请问老师,哪部分是reweight的结果呢?是红字部分吗?但这个结果可以解决主楼的问题吗?
(只有红色部分看上去像是reweight结果,但刚刚检查了一下,其实这是17个窗口的帧数除以相应窗口的correlation time得到的“有效帧数”,但有几个窗口的数值对不上)
刚才追查了mbar.py中的相应函数代码,发现完全看不懂,太难了。。。
|
|