[gmx-users] g_analyze -ee

[dina dusti]

Dear Gromacs Specialists,

I have one problem about g_analyze -ee, May I ask you to help me, Please?
When I do this program as "g_analyze -f .....xvg -av -ee", it is as followed:

Read 4 sets of 83001 points, dt = 6

                                      std. dev.    relative deviation of
                       standard       ---------   cumulants from those of
set      average       deviation      sqrt(n-1)   a Gaussian distribition
                                                      cum. 3   cum. 4
SS1   1.811675e+00   1.585145e-02   5.502120e-05       0.416    0.334
SS2   1.476938e+00   5.370597e-02   1.864162e-04       0.011   -0.009
SS3   1.482869e+00   5.208019e-02   1.807730e-04      -0.010   -0.034
SS4   1.475204e+00   5.259068e-02   1.825449e-04       0.012   -0.089


Back Off! I just backed up average.xvg to ./#average.xvg.2#
Set   1:  err.est. 0.000400002  a 0.402275  tau1 14.2321  tau2 255.692
Set   2:  err.est. 0.002757  a 0.870111  tau1 592.695  tau2 1081.59
Warning: tau2 is longer than the length of the data (498000)
         the statistics might be bad
invalid fit:  e.e. 0.476411  a 0.999161  tau1 572.985  tau2 2.48396e+10
Will fix tau2 at the total time: 498000
Set   3:  err.est. 0.00341541  a 0.998996  tau1 571.479  tau2 498000
Set   4:  err.est. 0.00360836  a 0.984581  tau1 578.453  tau2 39087.7



Then, I studied this article: J. Chem. Phys. 116, 209 (2002)
In it has been written " There are restrictions on the range of the fitting parameters: "a" should be between 0 and 1 and "tau1" and "tau2" should be larger than 0. When the longest correlation time, "tau2" , is longer than the averaging interval and "(1-a) tau2" is not negligible compared to "a tau1" , there is not enough statistics to estimate the error."

I thought that my problem is related to frequency to write coordinates and the other outputs to output trajectory file, then I changed it from 200 to 2000 in md.mdp file, but when I did g_analyze, again I took warning as following:

Read 4 sets of 8001 points, dt = 60

                                      std. dev.    relative deviation of
                       standard       ---------   cumulants from those of
set      average       deviation      sqrt(n-1)   a Gaussian distribition
                                                      cum. 3   cum. 4
SS1   1.811686e+00   1.577702e-02   1.763924e-04       0.422    0.376
SS2   1.476945e+00   5.345846e-02   5.976838e-04       0.004   -0.014
SS3   1.482819e+00   5.253597e-02   5.873700e-04       0.002   -0.034
SS4   1.475254e+00   5.289252e-02   5.913564e-04       0.021   -0.084


Back Off! I just backed up average.xvg to ./#average.xvg.1#

Back Off! I just backed up errorgyrate1.xvg to ./#errorgyrate1.xvg.1#
Set   1:  err.est. 0.000415045  a 0.741181  tau1 93.389  tau2 374.375
Set   2:  err.est. 0.00283072  a 0.964747  tau1 629.293  tau2 1869.63
a fitted parameter is negative
invalid fit:  e.e. 0.0025708  a 1.01686  tau1 576.241  tau2 663.907
Will fix tau2 at the total time: 480000
Set   3:  err.est. 0.00469726  a 0.997164  tau1 558.935  tau2 480000
Set   4:  err.est. 0.00371208  a 0.969329  tau1 551.965  tau2 21102.3

I don't know that where is my mistake and what thing should I change for fix of this warning!

Please help me.
Best Regards
Dina
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20120224/e1896a42/attachment.html>