about Coulombic energy computation with coul/long+pppm vs coul/cut

first off, please always reply to the list, too.

hi,Steve and Axel

In my simulation, E_cut=3383.7 and E_long=2387.4. And I test
cutoff=350,400,450, E_cut is the same 3383.7.

More details here.

For coul/cut test,I use the following setting:

boundary p p p
kspace_style pppm 1.0e-8
pair_style coul/long #same system with a vacuum space about 50A.
pair_coeff * *
variable E_cut equal ecoul

For coul/long+pppm, the setting is

boundary s s s
kspace_style none
pair_style coul/cut 400 # the box size is 60x60x320, and
(60^2+60^2+320^2)=331
pair_coeff * *
variable E_long equal ecoul+elong

In my simulation, E_cut=3383.7 and E_long=2387.4. And I test
cutoff=350,400,450, E_cut is the same 3383.7.

you are comparing apples and oranges here.
one input is with periodicity, the other without.
surely you get different results. you *have* to.

PEPCAC!

axel.

Axel,

I do some new simulations ( simulation input files and results are attached ). The molecule size is smaller than (60x60x320), and I set the box size to be (600x600x1000). both of pppm+coul/long and coul/cut
are with periodicity.
Then I found:

  1. Changing the cutoff for coul/long (10,20,30,40) doesn’t effect the result.
  2. pppm/coul/long and coul/cut have the same force.
  3. coul/cut with or without periodicity have the same result.
  4. The Coulombic energy computed by pppm+coul/long still is smaller than that by coul/cut.
    With large a vacuum space, ecoul+elong become a litter larger,but still is smaller than the result of ecoul/cut. E_cut=3383.7 vs E_long=2561.7

Cun

test.tar.gz (76.8 KB)