optimized EOS coordinates and final CVOLOPT

job314

Alessandro, I am having a blast finding the volume of the crystals using EOS which I can then use for supercell etc. I was wondering about the technically correct sequence here:

-I run EOS, it prints
SORTING VOLUMES/ENERGIES

 VOLUME (A^3)    ENERGY (a.u.)

 443.177886   -2.716943460865E+03
 451.049595   -2.716949077158E+03
 458.957263   -2.716952819603E+03
 466.951330   -2.716954739628E+03
 471.547619   -2.716955035989E+03
 475.033963   -2.716954889255E+03
 483.208420   -2.716953349891E+03
 491.475047   -2.716950220038E+03
 499.836229   -2.716945606438E+03

+++++++ FITTING USING ALL POINTS +++++++

USERS OF THIS EOS MODULE SHOULD KINDLY CITE THE FOLLOWING REFERENCE:

A. Erba, A. Mahmoud, D. Belmonte and R. Dovesi, J. Chem. Phys., 140, 124703 (2014)

EQUATION OF STATE VOL(A^3) E(AU) BM(GPa) BM PRIME

MURNAGHAN 1944 471.7053 -2716.95501994 53.34 3.41
BIRCH-MURNAGHAN 1947 471.7038 -2716.95502102 53.40 3.42
POIRIER-TARANTOLA 1998 471.7006 -2716.95502119 53.41 3.43
VINET 1987 471.7019 -2716.95502111 53.40 3.43

POLYNOMIAL FITTING VOL(A^3) E(AU) BM(GPa)

THIRD ORDER POLYNOMIAL 471.6937 -2716.95502707 53.73
FOURTH ORDER POLYNOMIAL 471.6893 -2716.95503729 54.28
FIFTH ORDER POLYNOMIAL 471.6651 -2716.95503712 54.28

Now, is there a trick to make software plots lattice parameters for that minimum volume? I could not find any

Since I could not find any, I take the two lattice parameters around that lowest energy volume and extrapolate lattice parameters to it. My second question then is - should I run one more CVOLOPT using that EOS lowest volume extrapolated lattice parameters to arrive to something that is not only lowest volume, but also optimal lattice parameters for that volume OR whatever comes out from EOS is already optimal lattice parameters for that minimal volume?

thank you

job314

Correction, I tried my last suggestion and, I do not understand why, volume changes when running CVOLOPT. The name implies that volume will be constant but it is not, it changes, which defeats the purpose of me wanting to keep volume fixed and finding the optimal lattice parameters for that volume

INFORMATION **** CVOLOPT **** CONSTANT VOLUME OPTIMIZATION

aerba

Hi,

Keeping in mind that geometry optimizations (unconstrained, volume-constrained, pressure-constrained) are all numerical processes with final results that thus may depend on numerical aspects of the implemented algorithms:

If you are interested in the structure of the true minimum of the PES, the best way to get it is directly from an unconstrained optimization (either from an OPTGEOM calculation or from the PREOPTGEOM within the EOS). By definition, numerically, this is the geometry corresponding to the lowest energy. In both cases, the optimized structure (volume, lattice parameters, atomic positions) is printed in full.
If, after an EOS calculation, you are interested in getting the structure for a specific volume V (not the minimum), then the strategy you described is exactly what I would do (and I've often done): I make sure to input a structure corresponding to the desired volume (typically by manually adjusting the lattice parameters) and I run a further CVOLOPT calculation.

Note: During a CVOLOPT calculation the volume of the system is expected to remain almost constant (within a certain numerical error). So small variations may be observed. If instead you experience large variations, there may be a problem.

job314

Hi Alessandro, naturally I am interested in PES minimum obtained from EOS. WHat would be the procedure here? I know the volume from EOS calculation but I do not know what are the optimal lattice parameters corresponding to that volume. So I do not know what parameters to take from the lowest energy EOS volume, can you illustrate that in the example?

I always do PREOPTGEOM with EOS, but I can't find the parameters printed out for that lowest volume

EOS
RANGE
0.94 1.06 8
PREOPTGEOM
MAXCYCLE
500
END

job314

In other words, it always prints coordinates for the optimized minima at each point in EOS (e.g. points corresponding 0.94, 0.96, 0.98 etc EOS optimizations). I do not believe (or I can't find at least) that it then prints coordinates and lattice parameters for that global minimum that it calculates between those EOS points that is exact EOS minimum

aerba

Hi,

I think that my point is that if you want to find the minimum of the PES you really do not need to run an EOS calculation. You can simply run an OPTGEOM calculation. EOS is useful if you need to take the pressure into explicit account and/or compute the bulk modulus.

If you have already run an EOS calculation with the PREOPTGEOM option, the structure corresponding to the minimum is the one obtained at the end of the pre-optimization, thus before all the various 0.94, 0.96, etc. compressions/expansions are explored. So in this case, all you need to do is look in your output file for the first occurrence of the "OPT END" string. You will find the optimized structure printed in its vicinity.

Hope this helps,

job314

HI Alessandro, I need to think about it. OPTGEOM will perform something at lattice parameters that are 1.0 with respect to EOS. Those are not optimal. I am looking for lowest global energy structure with optimal volume. I will revert to you with an example

job314

In other words, I am looking for something similar to FULLOPTG but I want to do it via systematic scanning of lattice parameters, thus EOS

job314

I suppose I am trying to do something like this with EOS, then use that minimum set of lattice parameters (minimum energy corresponding volume) and use it in all of my Raman calculations

https://www.vasp.at/wiki/index.php/Fcc_Si

aerba

Maybe I understand where the confusion may come from. Till CRYSTAL09, the OPTGEOM option corresponded to a geometry optimization of just the atomic coordinates within a fixed cell. Since CRYSTAL14 (and thus in versions 14, 17 and 23) the OPTGEOM option corresponds to a full geometry optimization (i.e. OPTGEOM is now equivalent to FULLOPTG).

So, OPTGEOM provides the minimum energy structure.

I would need to better understand what you mean by systematic scanning of lattice parameters to comment on that point.

Let me just reiterate on the numerical nature of both the OPTGEOM and EOS options: the minimum structure from both approaches should be the same. If they differ, it is just due to numerical reasons and, personally, I would tend to trust OPTGEOM better.

job314

Yes, we finally approach the same idea. Yes, FULOPTG - at least in other software packages FULLOPTG equivalent is rarely done, cell shape optimization is done systematically similarly to EOS. See example above what is done - a series of scans where volume is optimized and different deviation from 1.0 lattice parameters. VASP does that with ISFI=4.

But do you expect FULLOPTG result in lattice parameters that closely correspond those of the minimum in fitted EOS?

aerba

Absolutely yes, the optimized structure you get from OPTGEOM (i.e. a full optimization of both atomic coordinates and cell shape/volume, in CRYSTAL) is very very very close the minimum from fitted EOS, and if they differ I would blame EOS over OPTGEOM (meaning that maybe more points in the EOS scan would be required). Additionally, OPTGEOM is much much faster than EOS as only 1 versus N optimization processes are actually performed.

So, to get the fully relaxed structure, I strongly recommend to use OPTGEOM in CRYSTAL. This is what I usually do before running frequency and Raman calculations.

job314

OK, so to me that makes EOS completely redundant... I just used the final structure of the first optimization cycle before it started EOS and calculating Raman now

aerba

For the purpose of finding the minimum energy structure to then do Raman calculations, it is.

EOS gives you much more than that of course: the p(V) or, equivalently, V(p) relation (i.e. structure as a function of pressure), the bulk modulus K(p), and allows to compute the enthalpy H(p).