ABINIT Discussion Forums

This is maybe a stupid question, but anyway. I relaxed a structure, so the output is:


acell 5.6536178713E+00 5.6536178713E+00 3.5015770549E+01 Bohr
etotal -6.5707157781E+01
fcart 0.0000000000E+00 0.0000000000E+00 -3.5967159592E-06
0.0000000000E+00 0.0000000000E+00 4.8345461423E-06
0.0000000000E+00 0.0000000000E+00 8.6940498475E-07
0.0000000000E+00 0.0000000000E+00 -2.1072351678E-06
strten 1.9025208509E-08 1.9025200579E-08 1.4137761027E-07
0.0000000000E+00 0.0000000000E+00 0.0000000000E+00
xcart -8.1632957222E-21 -3.7079117888E-21 -4.3466537136E-02
2.8268089356E+00 -1.6320588999E+00 3.3543295591E+00
2.8268089356E+00 1.6320588999E+00 3.4355039260E+00
2.8268089356E+00 1.6320588999E+00 -1.9573247089E+00


Then I proceeded to calculation of Band structure. To the input file I just copied relaxed output above. I used all the general parameters as in relaxation.
But output file of the Band calculation gives me different energy, forces and strain, while all the material geometry stays the same (xcart, acell).


etotal -6.5707171472E+01
fcart 0.0000000000E+00 0.0000000000E+00 -3.0722155786E-04
0.0000000000E+00 0.0000000000E+00 2.6852582422E-04
0.0000000000E+00 0.0000000000E+00 2.6263546629E-04
0.0000000000E+00 0.0000000000E+00 -2.2393973266E-04
strten -1.8029036033E-07 -1.8029036424E-07 1.9155627909E-06
0.0000000000E+00 0.0000000000E+00 0.0000000000E+00


This structure clearly is not relaxed anymore and I don't understand why.
How to preserve relaxation results in further calculations?

Hello CHILL,

Looking at the information you provided I believe that your relaxation calculation went well. Your cartesian forces and strain tensor look well relaxed.

For your bandstructure calculation, I assume that you increased the density of your k-mesh from your relaxation calculation (or at least you should have). This increase in the k-mesh density is possibly responsible for the differences in your forces (which are still small) and a slight increase in your strain tensor (which is still great).

To tell if things are really changing in your calculation you can use the total energy which you provided. As the difference between your relaxation energy and band structure energy are very small this tells me that your calculation is well converged (changes in the total energy on the order of 1E-4 are normal.)

From what I see here, you are well on your way to getting the band structure of a nicely relaxed compound.

Congratulations!