Hi Giacomo,

Sorry for the late reply. I can confirm that 999 works in this scenario even using parallel execution settings. I tried to pass the HPC the command to run the simulation on a single CPU core but it kept complaining that it is not set up to do so, so I ran it on multiple CPUs and it worked. However, it is printing the eigenvectors for K points 48 and 80. Is there a way to print the eigenvectors at the Gamma point only? Additionally, when I try:

NEWK
4 4
1 2
66 -505
67 999
END

It will not print the eigenvectors (just the eigenvalues), even though the initial crystal run for calculating the ground-state via DFT was ran with SHRINK 4 4. There is no error at the end of the output file, it just ends. I'm not sure why, to be honest, but maybe it's got something to do with the number of CPUs this is running on? I might try to run it with less than 80 CPU cores (which is the standard number I use for Crystal simulations) and see what happens. In the meantime if you have any insight I would greatly appreciate it,

Thanks in advance for your help,

Chris

Hi Giacomo,

Thanks for the input. It seems that this would only print the eigenvalues. The eigenvectors won't be printed and I get this error at the end of the output file:

FERMI ENERGY AND DENSITY MATRIX CALCULATION ON COMPUTED EIGENVECTORS
DENSITY MATRIX AT SCF CYCLE ( 3+1)
SPIN LOCKING: NO ENERGY GAP COMPUTED
CORE DENSITY MATRIX CALCULATION
TTTTTTTTTTTTTTTTTTTTTTTTTTTTTT NEWK TELAPSE 16.38 TCPU 14.78
ERROR **** properties **** END OF DATA IN INPUT DECK

My input is:

NEWK
6 6
1 2
66 -505
67 -505

..ran via crystal_properties binaries. Could it be because 67 works only by printing at the end of the SCF cycles?

Chris

Hi Eleonora,

Thanks for your reply. So the calculation can simply be ran as:

NEWK
6 6
1 66 67

? Or do I have to run two separate calculations, one with "1 66" and one with "1 67" on the second line?

Chris

Hello,

Is there a way to print the eigenvectors and eigenvalues of a system after a simulation has finished running?

Chris

Thanks Giacomo, much appreciated,

Chris

Hi Giacomo,

Gotcha! Thanks for the thorough explanation. Using the Chromium example from the website (https://tutorials.crystalsolutions.eu/tutorials/soc/Cr3_so_colu_fmix.d12) I think it is straightforward to do the conversion to CRYSTAL.

One final question, if I may. Should I add:

TWOCOMPON
SOC
ENDtwo

at the beginning or at the end of the SCF block, or it does not matter?

Chris

Hi Giacomo,

Thanks for your quick reply!

I see. So for the elements I'm interested in (Nickel and Cobalt) the "pob-DZVP" and "pob-TZVP" should be all I need for an SOC calculation, right? Since they are double- and triple-zeta valence + polarization enabled.

Chris

Hi,

I am curious if Crystal accepts something like the cc-pVDZ-DK Gaussian basis set for the calculation of SOC effects in transition metals? From what I gather, it is an all electron basis set (https://www.basissetexchange.org/basis/cc-pvdz-dk/format/crystal/?version=0&elements=28). Would this require the keyword INPSOC in the basis set block of the input file?

Chris

Hi Giacomo,

That's a very useful answer, thank you!

Chris

Hello all. I have a question about CRYSTAL23 and how it works when it comes to calculating eigenvalues for certain k-points. Let's look at the example of SiO2 cell with 12 ions:

*** K POINTS COORDINATES (OBLIQUE COORDINATES IN UNITS OF IS = 6)
1-R( 0 0 0) 2-C( 1 0 0) 3-C( 2 0 0) 4-R( 3 0 0)
5-C( 1 1 0) 6-C( 2 1 0) 7-C( 2 2 0) 8-C( 0 0 1)
9-C( 1 0 1) 10-C( 2 0 1) 11-C( 3 0 1) 12-C( 1 1 1)
13-C( 2 1 1) 14-C( 2 2 1) 15-C( 0 0 2) 16-C( 1 0 2)
17-C( 2 0 2) 18-C( 3 0 2) 19-C( 1 1 2) 20-C( 2 1 2)
21-C( 2 2 2) 22-R( 0 0 3) 23-C( 1 0 3) 24-C( 2 0 3)
25-R( 3 0 3) 26-C( 1 1 3) 27-C( 2 1 3) 28-C( 2 2 3)

So here it shows 28 k-points for the 6x6x6 Monkhorst-Pack grid, and if I understand correctly, the reason why it's not showing all 216 points is because the program recognizes the crystal symmetry so it knows that the energy at (1 0 0) is the same as at (0 1 0), (0 0 1), (-1 0 0), etc?

If that is the case, then to build a high-symmetry path along Gamma-X-K-Gamma, for example... we could just take the points 4-R( 3 0 0) and 7-C( 2 2 0) and divide them by IS, which in this case is 6 to get the points (0.5, 0, 0) and (1/3, 1/3, 0). Would that be correct? Because that would make the extraction of eigenvalues quite simple for this path.

Thanks in advance,

Chris