howto:running_qe_computation
Differences
This shows you the differences between two versions of the page.
Both sides previous revisionPrevious revisionNext revision | Previous revisionNext revisionBoth sides next revision | ||
howto:running_qe_computation [2019/08/13 12:50] – mtaillefumier | howto:running_qe_computation [2019/09/02 08:10] – [Molecular dynamics] mtaillefumier | ||
---|---|---|---|
Line 35: | Line 35: | ||
/ | / | ||
&system | &system | ||
- | ibrav=0, celldm(1)=1, | + | ibrav=0, celldm(1)=1, |
- | ecutrho = 300, occupations = ' | + | ecutrho = 400, occupations = ' |
smearing = ' | smearing = ' | ||
nat=8 ntyp=2 | nat=8 ntyp=2 | ||
Line 77: | Line 77: | ||
| | ||
| | ||
- | | + | |
- | | + | |
| | ||
| | ||
Line 326: | Line 326: | ||
^ QE input ^ CP2K input ^ | ^ QE input ^ CP2K input ^ | ||
|< | |< | ||
- | ecutwfc=30 | + | ecutwfc=25 |
- | ecutrho = 300 | + | ecutrho = 400 |
</ | </ | ||
& | & | ||
Line 334: | Line 334: | ||
& | & | ||
| | ||
- | | + | |
- | | + | |
&END PARAMETERS | &END PARAMETERS | ||
&END PW_DFT | &END PW_DFT | ||
Line 348: | Line 348: | ||
& | & | ||
| | ||
- | | + | |
- | | + | |
| | ||
| | ||
Line 454: | Line 454: | ||
</ | </ | ||
- | ===== Cell relaxation | + | ===== Molecular dynamics |
- | + | ||
- | Quantum expresso supports many types of calculations that cp2k supports natively. For instance, it is possible to do cell relaxation in cp2k with the sirius backend, something that QE can do as well. The following QE input file | + | |
- | < | + | |
- | & | + | |
- | calculation=' | + | |
- | restart_mode=' | + | |
- | pseudo_dir = ' | + | |
- | outdir=' | + | |
- | prefix = ' | + | |
- | tstress = false, | + | |
- | tprnfor = false, | + | |
- | verbosity = ' | + | |
- | wf_collect = false | + | |
- | / | + | |
- | & | + | |
- | ibrav=0, celldm(1)=1, | + | |
- | occupations = ' | + | |
- | nat=8 ntyp=2 | + | |
- | / | + | |
- | & | + | |
- | conv_thr = 1.0d-11, | + | |
- | mixing_beta = 0.7, | + | |
- | electron_maxstep = 100 | + | |
- | / | + | |
- | &IONS | + | |
- | ion_dynamics=' | + | |
- | / | + | |
- | &CELL | + | |
- | cell_dynamics=' | + | |
- | / | + | |
- | ATOMIC_SPECIES | + | |
- | Ge 0.0 ge_lda_v1.uspp.F.UPF | + | |
- | Si 0.0 si_lda_v1.uspp.F.UPF | + | |
- | CELL_PARAMETERS | + | |
- | | + | |
- | 0.00000000 | + | |
- | 0.00000000 | + | |
- | ATOMIC_POSITIONS (crystal) | + | |
- | Ge 0.00000000 | + | |
- | Si 0.00000000 | + | |
- | Si 0.50000000 | + | |
- | Si 0.50000000 | + | |
- | Si 0.75000000 | + | |
- | Si 0.75000000 | + | |
- | Si 0.25000000 | + | |
- | Si 0.25000000 | + | |
- | K_POINTS (automatic) | + | |
- | 2 2 2 0 0 0 | + | |
- | </ | + | |
- | + | ||
- | describes the computation of the cell relaxation for Si7Ge for instance. The cp2k input file for the sirius backend will look like | + | |
+ | Quantum expresso supports many types of calculations that cp2k supports natively. For instance, it is possible to do molecular dynamics in cp2k with the sirius backend, something that QE can do as well. To do this, we can start from the previous cp2k input file for Si7Ge and add a small section for setting the parameters relevant for the molecular dynamics. The '' | ||
+ | |||
< | < | ||
& | & | ||
Line 519: | Line 470: | ||
| | ||
| | ||
- | | + | |
| | ||
- | | + | |
- | | + | |
| | ||
| | ||
Line 581: | Line 532: | ||
OPTIMIZER LBFGS | OPTIMIZER LBFGS | ||
& | & | ||
+ | & | ||
+ | ENSEMBLE NVE | ||
+ | TIMESTEP 0.1 | ||
+ | STEPS 125 | ||
+ | TEMPERATURE 300.0 | ||
+ | & | ||
&END MOTION | &END MOTION | ||
&GLOBAL | &GLOBAL | ||
PROJECT Si7Ge | PROJECT Si7Ge | ||
PRINT_LEVEL MEDIUM | PRINT_LEVEL MEDIUM | ||
- | RUN_TYPE | + | RUN_TYPE |
&END GLOBAL | &END GLOBAL | ||
</ | </ | ||
Line 593: | Line 550: | ||
</ | </ | ||
- | you should obtain a file '' | + | you should obtain a file '' |
+ | {{: | ||
+ | The total energy given by QE and CP2K/SIRIUS are shifted by a small offset. More generally, converting the input file from QE to CP2K is not a warranty to obtain the same results for the total energy. The reasons for this multiple : | ||
+ | * Indicating the same cutoff does not warranty the fft grid size. | ||
+ | * QE and SIRIUS treat the radial integrals interpolation differently | ||
+ | * The functional in QE do not use libxc by default, while CP2K/SIRIUS does. | ||
+ | * The minimization method is sensitive to the initial states. | ||
+ | For all these reasons, we should not try to compare results at the binary level. |
howto/running_qe_computation.txt · Last modified: 2021/12/08 10:18 by mtaillefumier