[gmx-users] Essential Dynamics Sampling (EDS) MD

[Ran Friedman]

Dear all,

I'm resending this message (please see below)  with some additional
info, hoping that someone can at least point me to what I should check.

I ran the simulation with the -debug flag and got some "nan" values in
mdrun.log:

dekin = nan, ekin = 175815  vcm = (     nan      nan      nan)
mv = (     nan      nan      nan)
PC: pres (3x3):
   PC: pres[    0]={         nan,          nan,          nan}
   PC: pres[    1]={         nan,          nan,          nan}
   PC: pres[    2]={         nan,          nan,          nan}
PC: ekin (3x3):
   PC: ekin[    0]={         nan,          nan,          nan}
   PC: ekin[    1]={         nan,          nan,          nan}
   PC: ekin[    2]={         nan,          nan,          nan}
PC: vir  (3x3):
   PC: vir [    0]={ 5.66646e+04,  1.12632e+02, -1.88577e+03}
   PC: vir [    1]={ 5.10852e+02,  5.71922e+04,  3.69013e+03}
   PC: vir [    2]={-1.82055e+03,  3.43891e+03,  6.20980e+04}
PC: box  (3x3):
   PC: box [    0]={ 9.73688e+00,  0.00000e+00,  0.00000e+00}
   PC: box [    1]={ 3.24470e+00,  9.11683e+00,  0.00000e+00}
   PC: box [    2]={-3.24470e+00,  4.55842e+00,  7.97596e+00}
fshift after SR (27x3):
   fshift after SR[    0]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
   fshift after SR[    1]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
   fshift after SR[    2]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
   fshift after SR[    3]={         nan,          nan,          nan}
   fshift after SR[    4]={         nan,          nan,          nan}

What am I missing?

Ran Friedman wrote:
> Dear all,
>
> I'm trying to run MD with ED sampling. The target structure is a protein
> taken from one simulation, the .tpr file is taken from a different one.
> I'm using a subset of the CA atoms to calculate the eigenvectors. The
> problem is that the MD run crashes after set-up.
>
> Depending on the machine, it either crashes without any error message or
> complains about "Too many iterations in routine JACOBI".
>
> Any suggestions?
>
> Thanks,
> Ran.
>
>