difference between Total energy and ENERGY| Total FORCE_EVAL ( QS ) energy ?

[Jun]

Hi all,

When doing electronic structure calculations in MD or geo_opt, it is
very often to see a small difference in energy between "Total
energy:      ****  " and " ENERGY| Total FORCE_EVAL ( QS ) energy
(a.u.):   **** " in output files. The difference is usually so small
that I thought it is just due to numerical error. (I am not saying the
difference due to applying some restraining potentials or in MIXED
calculations). However, I recently noticed that the difference can be
as large as 0.1 eV, which makes me wonder what is difference in
definitions of "Total energy " and "ENERGY| Total FORCE_EVAL ( QS )
energy". An example is as follows: (input file (input.inp) is
uploaded)

I have repeated the same ENERGY_FORCE calculation three times (there
may be some differences in the initial guess for wavefunction because
I used SCF_GUESS RESTART.), the energies I got are
1.  Total energy:
-3269.58265002363760
 ENERGY| Total FORCE_EVAL ( QS ) energy (a.u.):
-3269.578462626209330

2.   Total energy:
-3269.57862048247534
 ENERGY| Total FORCE_EVAL ( QS ) energy (a.u.):
-3269.578462728981776

3.   Total energy:
-3269.57793566477721
 ENERGY| Total FORCE_EVAL ( QS ) energy (a.u.):
-3269.578462722795848

In three cases, " ENERGY| Total FORCE_EVAL ( QS ) energy (a.u.)" are
almost the same, but  "Total energy: " are quite different. Especially
for case 1, the difference is up to 0.1 eV.

Any suggestion and comment are appreciated.

Cheers,
Jun