[gmx-users] negative eigenvalues occured and not "nearly zero"

[silvester.thu]

Hello gmx-users:

I'm recently using Normal Mode Analysis(NMA) of GROMACS package to calculate the foundamental frequency of graphite systems. But I got some results that are unexplainable, and I can not continue my work now. 

For all the examples below, I use the double precision GROMACS functions: grompp_d, mdrun_d, and g_nmeig_d. For the EM part, I used the conjugated-gradients algorism, and for the NMA part, I used "nm" integrator and included the full precision structure file: em.trr into the .tpr input file.

For the systems that I applied no restraints, the first six eigenvalues should be zero, or at least "nearly zero" if we considering the accuracy of numerical calculating. But the typical first twelve eigenvalues I got have the values below:
(after EM, the Fmax<5e-4)

    1        -0.37993
     2       -0.335177
     3       -0.206887
     4       -0.111813
     5      -0.0381566
     6      -0.0211911
     7       0.0189381
     8       0.0370517
     9       0.0385126
    10        0.093188
    11        0.132927
    12        0.169143

the first eigenvalue has larger absolute value than the 7th eigenvalue, which should not happen for an unrestrained system. WHY the first six eigenvalues are not zero or nearly zero? I guess there might by something wrong in my input files, but I can not figure out where it is. By the way, the topology files and coordinate files are surely correct.

Another problem is:
For the system that I applied restraints to, which removed the translation and rotation about the center of mass, the eigenvalues should be all positive. But I still found that negative values occured. 
     1       -0.135621
     2      -0.0745555
     3      -0.0390086
     4      -0.0108285
     5       0.0101438
     6       0.0553579
     7        0.146471
     8        0.162898
     9        0.344921
    10        0.396553
    11        0.486316
    12        0.544625
WHY? 

If anyone used the NMA of GROMACS before and got correct output, it will be very helpful if you can send me your parameters or input files or give me some advice. If you need any more information, please e-mail me.

I really need your help, thanks.

Yours Sincerely

Yue Shao

E-mail: silvester.thu at gmail.com




___________________________________________________________________________
___________________________________________________________________________

Shao Yue
Institue of Biomechanics and Biomedical Engineering
Department of Engineering Mechanics
Tsinghua University
P.R. China
____________________________________________________________________________
____________________________________________________________________________
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20080425/c8169b2d/attachment.html>