[gmx-users] Steepest descent

[Andrew Horsfield]

Hi,

Another problem with my simple calculation.

I relaxed the isolated molecule without any difficulty. I then solvated 
it, and tried to rerelax it using steepest descent. The output is:

  Step =     0, Dmax = 1.00e-02 nm, Epot = -8.68044e+03 Fmax = 1.82046e+04
  Step =     1, Dmax = 1.00e-02 nm, Epot = -1.00319e+04 Fmax = 6.48675e+03
  Step =     2, Dmax = 1.20e-02 nm, Epot = -1.16666e+04 Fmax = 3.59934e+03
  Step =     3, Dmax = 1.44e-02 nm, Epot = -1.28017e+04 Fmax = 1.99101e+03
  Step =     4, Dmax = 1.73e-02 nm, Epot = -1.36208e+04 Fmax = 1.65291e+03
  Step =     5, Dmax = 2.07e-02 nm, Epot = -1.38893e+04 Fmax = 1.57456e+03
  Step =     8, Dmax = 6.22e-03 nm, Epot = -1.38896e+04 Fmax = 1.56714e+03
  Step =    21, Dmax = 1.82e-06 nm, Epot = -1.38896e+04 Fmax = 1.63983e+03
  Stepsize too small Converged to machine precision,
  but not to the requested precision (500)
  You might need to increase your constraint accuracy

I have tried turning off constraints, using double precision binaries,
conjugate gradient, PME and cut-off for the electrostatics, but the result
is always the same. What else can I try to fix this?

Cheers,

Andrew