Hello Ewerton Ferreira da Silva Souza ,

There are many causes to Quantum Espresso problems. Once you have an known structure, you must find a proper way to make an input. Could you share your input, or show what your structure is? Generally, I create a new input of the structure, and kind of "ignore" the past errors of the code. I have never seen this one you are showing, that's why I cant tell exactly what happened. But I can try to make it work.

Best regards,

Ricardo Tadeu

Heey Ewerton Ferreira da Silva Souza

For a quick check, double-check the compatibility of the pseudopotentials (format and cutoffs used). Try a different diagonalization scheme, such as 'cg' if you're using 'david,' or vice versa.

Best,

Gabriel Vinicius

Ricardo Tadeu Maia

My input for vc-relax calculation, i think this my problem is in my pseudopotential, but i'm not sure. I tried:

1- use ibrav=0 and lattice parameters of lattice;

2- cell_dofree = shape / all/ xy / z / volume... for the cell move any way...

I never see this error in any way before...

&SYSTEM

a = 8.54435e+00

c = 5.69623e+00

degauss = 1.00000e-02

ecutwfc = 110

ecutrho = 1320

ibrav = 6

nat = 37

ntyp = 2

occupations = "smearing"

smearing = "gaussian"

nspin = 2

starting_magnetization(1) = -0.7

/

&electrons

electron_maxstep = 300

mixing_beta = 0.4

/

&ions

/

&cell

i have tried many ways in this section...

/

ATOMIC_SPECIES

Fe 55.84500 Fe.pbe-spn-rrkjus_psl.1.0.0.UPF

C 12.011 C.pbe-n-rrkjus_psl.1.0.0.UPF

ATOMIC_POSITIONS {angstrom}

Fe 0.000000 0.000000 0.000000

Fe 1.420020 1.420020 1.420020

Fe 0.000000 0.000000 2.840040

Fe 1.420020 1.420020 4.260060

Fe 0.000000 2.840040 0.000000

Fe 1.420020 4.260060 1.420020

Fe 0.000000 2.840040 2.840040

Fe 1.420020 4.260060 4.260060

Fe 0.000000 5.680080 0.000000

Fe 1.420020 7.100100 1.420020

Fe 0.000000 5.680080 2.840040

Fe 1.420020 7.100100 4.260060

Fe 2.840040 0.000000 0.000000

Fe 4.260060 1.420020 1.420020

Fe 2.840040 0.000000 2.840040

Fe 4.260060 1.420020 4.260060

Fe 2.840040 2.840040 0.000000

Fe 4.260060 4.260060 1.420020

Fe 2.840040 2.840040 2.840040

Fe 4.260060 4.260060 4.260060

Fe 2.840040 5.680080 0.000000

Fe 4.260060 7.100100 1.420020

Fe 2.840040 5.680080 2.840040

Fe 4.260060 7.100100 4.260060

Fe 5.680080 0.000000 0.000000

Fe 7.100100 1.420020 1.420020

Fe 5.680080 0.000000 2.840040

Fe 7.100100 1.420020 4.260060

Fe 5.680080 2.840040 0.000000

Fe 7.100100 4.260060 1.420020

Fe 5.680080 2.840040 2.840040

Fe 7.100100 4.260060 4.260060

Fe 5.680080 5.680080 0.000000

Fe 7.100100 7.100100 1.420020

Fe 5.680080 5.680080 2.840040

Fe 7.100100 7.100100 4.260060

C 2.750671 7.082327 2.888623

K_POINTS {automatic}

6 6 6 0 0 0

Gabriel Vinicius I didn't understand, can you give more details?

Heey Ewerton Ferreira da Silva Souza

at first glance, your pseudopotentials look fine. I'd just adjust ecutrho to be about four times ecutwfc, as the ratio used is very large and may lead to matrix errors. I usually set ibrav=0 and define the lattice vectors manually within the CELL_PARAMETERS card, rather than using a specific symmetry like ibrav=6. In the electrons card, add diagonalization = "cg" (if it does not work, try "david").

Try and see how it goes,

Best,

Gabriel Vinicius

I just tried the input you provided. I think it's possible to remake it, because the way I see it, you are not exploring the symmetry of the system as much as you could. So, is this system a Iron BCC crystal doped with only 1 carbon? If this is the case, you can try to create an simple input for pure Iron BCC, do the vc-relax, and it will run normally. Then, for the doped with 1 carbon, you substitute the central Iron atom of the lattice, once that this way QE will recognize the symmetry of the system, and this speeds the calculation a lot. Do you agree with this suggestion?

Ricardo Tadeu Maia In my case, the carbon isn't for doping.

It's an impurity in the Unit Cell like in literature.

Thus, I have what maintain the iron and introduce the carbon in the structure.

Using carbon like doping I already achieved, but is not my main goal.

Gabriel Vinicius i will try your solution.

Gabriel Vinicius I tried, but I was unsuccessful... keep moving...

Hello Ewerton Ferreira da Silva Souza

I modified the input and I came to this new one. You showed a 3x3x3 Iron BCC structure, with only one Carbon atom in the interstitial site. So, I kept this proportion and modified the unit cell with VESTA, positioning the interstitial site to be in the center of the system, because in this way, QE will recognize 16 symmetry operations, and make calculation a lot faster. The point is that this calculation is very demanding because there are 54 atoms of Iron and 1 Carbon. If you could make a closer proportion, it would be faster, because a 2x2x2 cell would have less than 20 Iron atoms. Other point, you used:

ecutwfc = 110

ecutrho = 1320

Are you using a cluster? A calculation with these cutoffs for this system is quite consuming. Anyway, I made a 2x2x2 cell with carbon impurity, and you see if this fits for you. Let me know if you have any progress.

Ricardo Tadeu Maia but why did you use ibrav=0 (that is discouraged by the QE developers) for a simple cubic lattice? Setting ibrav=1 and removing the cell_parameters region provides a leaner and possibly faster input file.

Ewerton Ferreira da Silva Souza, your cutoff for the wavefunction is likely too high. You are using ultrasoft pseudopotentials which usually need a cutoff in the region of 40-50 Ry (the cut-off is in Rydberg) and a cutoff on the kinetic energy 10-12 time this value (as you set in your input).

The subroutine that generates the error is one belonging to the scalapack libraries. It is possible that this signals a problem with either the size of your matrices (determined by the cutoff) o the parallelization of your machine.

Regards,

Roberto