ewald accuracy

LAMMPS LAMMPS Mailing List Mirror
1

Dear lammps users,
I want to know how exactly the ewald accuracy is calculated in the code. I have set the the accuracy in the input script as 1e-6, but in the code I see that its value is 0.00033, why is it?

Thanks
Arshia

2

Dear lammps users,
I want to know how exactly the ewald accuracy is calculated in the code. I
have set the the accuracy in the input script as 1e-6, but in the code I see
that its value is 0.00033, why is it?

because it is computed to be like that.

just read the code and the documentation.
it is all there, well documented and
referenced. what else do you need?

axel.

3

Paul can answer this, he just reworked
the accuracy criterion for Ewald to
make it consistent with PPPM.

Steve

4

Arshia,

A few points to note:

  1. We’ve recently changed the way LAMMPS computes ewald accuracy. See the March 2, 2012 change here: http://lammps.sandia.gov/bug.html

  2. You need to make sure you’re using the latest version of LAMMPS if you want to have those latest changes included. You didn’t say which version (date) you’re using.

  3. In the current version, LAMMPS reports two estimated force accuracies: absolute and relative. You can specify either one from the input deck (see “force” option here: http://lammps.sandia.gov/doc/kspace_modify.html ). You didn’t say which of those that 0.00033 represents.

  4. If you post your input, it would be easier for us to diagnose what the issue might be.

  5. The accuracy LAMMPS reports is only an estimated accuracy.

  6. When I specify 1e-6 relative accuracy in a simple input file, I see 1e-6 relative accuracy reported back. Not sure what the difference is with your case.

  7. If you also specify the gewald parameter, LAMMPS you probably won’t get the accuracy you specify.

  8. See: http://lammps.sandia.gov/doc/kspace_modify.html and http://lammps.sandia.gov/doc/kspace_style.html for more information.

Paul