# Energy and Enthalpy from conjugate gradient

Hi All,

I have an issue that seems to belong to accuracy of CG method. I am trying to determine energy of a system having ~ 30000 atoms with a dislocation dipole, relaxing using box-relax with zero applied pressure and CG with the following set up:

minimize 1.0e-20 0.0 10000 20000

It shows a difference in energy and enthalpy of the system. The difference is ~ 0.03 for the system. Then I do similar relaxation with applied uniaxial pressure e.g. 500 MPa, 1 GPa, 2 GPa etc. and I expect the difference between energy and enthalpy to be in the order of applied external pressure. However, I see no trend in the evolution of enthalpy vs. pressure.

On the other hand if do a MD simulation with press/berendsen and NVE at 0 K, it shows the expected trend and at zero external pressure, the energy is identical to enthalpy.

I would like to know if anyone has experienced a similar issue and what could be done to make CG useful for such simulations.

Regards,

Alankar

Hi All,

I have an issue that seems to belong to accuracy of CG method. I am trying
to determine energy of a system having ~ 30000 atoms with a dislocation
dipole, relaxing using box-relax with zero applied pressure and CG with the
following set up:

minimize 1.0e-20 0.0 10000 20000

It shows a difference in energy and enthalpy of the system. The difference
is ~ 0.03 for the system. Then I do similar relaxation with applied uniaxial
pressure e.g. 500 MPa, 1 GPa, 2 GPa etc. and I expect the difference between
energy and enthalpy to be in the order of applied external pressure.
However, I see no trend in the evolution of enthalpy vs. pressure.

On the other hand if do a MD simulation with press/berendsen and NVE at 0 K,
it shows the expected trend and at zero external pressure, the energy is
identical to enthalpy.

I would like to know if anyone has experienced a similar issue and what
could be done to make CG useful for such simulations.

just set your velocities to zero.

atomic velocities are completely disregarded
during minimization, but since the same routines
are called as during MD, e.g. to compute pressure,
their contributions are still included.

so using the following should give the expected behavior:

velocity all set 0.0 0.0 0.0

cheers,
axel.

Thanks.

Still not working.

I tried the following:

velocity all zero linear

and also

velocity all zero angular

Still shows descrepency in energy vs. enthalpy at 0 applied pressure.

Alankar

Thanks.

Still not working.

I tried the following:

velocity all zero linear

and also

velocity all zero angular

Still shows descrepency in energy vs. enthalpy at 0 applied pressure.

sorry, but if you don't want to follow

axel.

Sorry for the confusion.

When I tried

velocity all set 0.0 0.0 0.0

I got the following error:

ERROR: Use of velocity with undefined lattice

I figured out that that I was reading a data filed in my script and therefore had turned off the lattice command.

I turned the lattice command on; it ran well.

However, it could not help me in energy vs. enthalpy discrepancy.

Thanks,

Alankar

Sorry for the confusion.

When I tried

velocity all set 0.0 0.0 0.0

I got the following error:

ERROR: Use of velocity with undefined lattice

I figured out that that I was reading a data filed in my script and therefore had turned off the lattice command.

I turned the lattice command on; it ran well.

However, it could not help me in energy vs. enthalpy discrepancy.

i don’t believe that. how should this be possible?

if you have no kinetic energy and have a properly
relaxed system (pressure ~0), there is no other
contributing factor that would explain any difference.
have a look at the source code and you’ll see.

i can only assume that you must be doing
something wrong in your comparisons or
calculations that you are not really comparing
apples with apples somehow.

axel.

I got the following error:
ERROR: Use of velocity with undefined lattice

This is simply b/c the velocity command has a units box/lattice
option and the default is lattice. Use units box if you prefer.

But Axel is correct, velocities are ignored for minimization.

Steve