Hi all,

yes, my concern is the total electrostatic energy difference between this two packages, the real space cutoff are 12 A, and long range accuracy 1e-12, both in GULP and lammps, and i have tried to increase the cutoff and accuracy, the energy is still there. GULP doc say the real part energy has a erfc function, and the eta is a dumping function (as in the picture in previous email), but in lammps doc for (coul/long), no erfc function act on the coulomb formula.that is best reason i have now. any ideas?

Best Regards

Jiasen Guo

Hi all,

yes, my concern is the total electrostatic energy difference between this

two packages, the real space cutoff are 12 A, and long range accuracy 1e-12,

both in GULP and lammps, and i have tried to increase the cutoff and

accuracy, the energy is still there. GULP doc say the real part energy has a

erfc function, and the eta is a dumping function (as in the picture in

previous email), but in lammps doc for (coul/long), no erfc function act on

the coulomb formula.that is best reason i have now. any ideas?

your conclusion is not correct. there is *of course* an erfc()

function applied in coul/long. otherwise the whole thing would not

work.

you cannot see a call to erfc() in the source, since it is

approximated by a polynomial formula or tabulation.

...and it turns out that in the documentation file itself the

expression for coul/long is simply not represented.

axel.

any ideas?

Have you turned off tabulation using "pair_modify table 0"? Next would be to use an exact representation of erfc instead of the polynomial expansion. See previous mail list posts on this topic.

Stan