Calculations on solid LiH with large Gaussian basis sets

This page contains information on how to run Hartree-Fock (HF) calculations on solid LiH with CRYSTAL using the basis set proposed by Paier et al.[1]

To run HF calculations with such a large basis set in the required very severe computational conditions, a static dimension (fixed long time ago when smaller basis sets and looser computational conditions could be used) had to be removed.

This has been done in the demo version of the CRYSTAL06 code that can be downloaded from the web site.

In the demo version all functionalities of the full version are active, the only constraint being the number of atoms, restricted to 4 per primitive cell.
To obtain the evaluation copy you have to register first, and then follow instructions to download executables.
Installation instructions and scripts to compute wave function and properties are also supplied.

An example of input deck (corresponding to case A of Table 1 in ref. [2]) for solid LiH at the experimental lattice constant with the basis set proposed by Paier et al. follows:

LiH  Basis set Paier et al. PRB 2009 (Case A)
CRYSTAL
0 0 0
225
4.084
2
3  0.0  0.0  0.0
1  0.5  0.5  0.5
END
1 12
0 0 3 1.00 1.
0.30797000E+04 0.00023473
0.46152000E+03 0.00182450
0.10506000E+03 0.00959330
0 0 1 0.00 1.
0.27463675E+02 1.00000000
0 0 1 0.00 1.
0.68559258E+01 1.00000000
0 0 1 0.00 1.
0.17679972E+01 1.00000000
0 0 1 0.00 1.
0.51181842 1.00000000
0 0 1 0.00 1.
0.20167548 1.00000000
0 2 1 0.00 1.
0.21240865E+01 1.00000000
0 2 1 0.00 1.
0.10736812E+01 1.00000000
0 2 1 0.00 1.
0.56838662 1.00000000
0 3 1 0.00 1.
0.92833840 1.00000000
0 3 1 0.00 1.
0.49583000 1.00000000
0 4 1 0.00 1.
0.12073480E+01 1.00000000
3 19
0  0  3  2.  1.
0.70681000E+05 0.00000544
0.13594000E+05 0.00003328
0.31004000E+04 0.00019175
0  0  1  1.  1.
0.13360341E+04 1.00000000
0  0  1  0.  1.
0.44429982E+03 1.00000000
0  0  1  0.  1.
0.14779702E+03 1.00000000
0  0  1  0.  1.
0.49209451E+02 1.00000000
0  0  1  0.  1.
0.16428957E+02 1.00000000
0  0  1  0.  1.
0.55293994E+01 1.00000000
0  0  1  0.  1.
0.19052824E+01 1.00000000
0  0  1  0.  1.
0.70025874 1.00000000
0  0  1  0.  1.
0.29958682 1.00000000
0  0  1  0.  1.
0.16636288 1.00000000
0  2  2  0.  1.
0.28500000E+02 0.00036754
0.66400000E+01 0.00322359
0  2  1  0.  1.
0.15709110E+01 1.00000000
0  2  1  0.  1.
0.74875864 1.00000000
0  2  1  0.  1.
0.38614089 1.00000000
0  2  1  0.  1.
0.22620503 1.00000000
0  3  1  0.  1.
0.77920820 1.00000000
0  3  1  0.  1.
0.40789925 1.00000000
0  4  1  0.  1.
0.73706300 1.00000000
99 0
END
TOLINTEG
6 6 6 15 30
SETINF
2
41 28
43 20
SHRINK
12 12
FMIXING
50
END

Note that the basis set of Li reported in Table 1 of ref. [1] is misprinted. The exponent of one of the p shells has to be changed from 0.15709110 to 0.15709110E+01 [3].

All calculations reported in ref. [2] can be repeated by modifying the above input deck.
In particular, the following options have to be modified:

cases B-L: the keyword TOLINTEG for accuracy on one- and two-electron integrals (see column TI in Table 1 [2])
cases M-O: the SETINF option for multipolar expansion (inf(41)=Coulomb series, inf(43)=Exchange series) (see column ME in Table 1 [2])
cases P-R: the keyword SHRINK to modify the sampling of the irreducible Brillouin Zone (see column IS in Table 1 [2])

We refer to the CRYSTAL06 User's Manual for further details.

[1] J. Paier, C.V. Diaconu, G.E. Scuseria, M. Guidon, J. VandeVondele and J. Hutter, Phys. Rev. B 80, 174114 (2009)
[2] B. Civalleri, R. Orlando, C.M. Zicovich-Wilson, C. Roetti, V.R. Saunders, C. Pisani and R. Dovesi, Phys. Rev. B submitted
[3] J. VandeVondele, private communication