[gmx-users] Steepest descent vs Conjugate gradient energies

[Andrew Horsfield]

Hi,

I get different energies from using steepest descent and conjugate
gradient optimizers. (I a using gromacs 3.0.5, with PME for
electrostatics). The problem appears to be that the long ranged coulomb
term is not included in the steepest descent calculation:

Steepest descent
----------------
   Energies (kJ/mol)
           Bond          Angle    Proper Dih.  Improper Dih.          LJ-14
    2.59408e+02    9.03184e+02    6.64435e+02    2.15660e+02    1.03799e+03
     Coulomb-14        LJ (SR)   Coulomb (SR)   Coulomb (LR)      Potential
    1.64964e+04   -7.15170e+03   -5.94253e+03    0.00000e+00    6.48285e+03
    Kinetic En.   Total Energy    Temperature Pressure (bar)
    0.00000e+00    0.00000e+00    0.00000e+00    0.00000e+00

Conjugate gradient
------------------
   Energies (kJ/mol)
           Bond          Angle    Proper Dih.  Improper Dih.          LJ-14
    2.59488e+02    9.02182e+02    6.64151e+02    2.15576e+02    1.03894e+03
     Coulomb-14        LJ (SR)   Coulomb (SR)   Coulomb (LR)      Potential
    1.64972e+04   -7.15239e+03   -5.94202e+03   -2.35473e+04   -1.70643e+04
    Kinetic En.   Total Energy    Temperature Pressure (bar)
    0.00000e+00    0.00000e+00    0.00000e+00    0.00000e+00

Is there a reason for this?

Cheers,

Andrew