[lammps-users] question on format of CH.airebo file

Albert,

The beta0-5 numbers in Lindsay are the values of the spline coefficients for the quintic spline function between 109.5 and 120 degrees.

The function has the form g(x) = beta0 + beta1x + beta2x^2 … beta5*x^5All you have to do in the CH.airebo file is make the appropriate substitution under the comment #g_c1 and g_c2.

The T0 is more difficult. You have to calculate the spline coefficients for a tricubic spline Tij(Nit,Njt,NijConj), then substitute these coefficients in to the CH.airebo where it says #T_ij. This of course is easier said than done. You need a code to calculate the coefficients for the tricubic spline from the node values which is nontrivial. Brenner’s original code has the subroutine to calculate the spline coefficients in it. With some minor modification, you can get it to output the coefficients. See table 5 (Brenner 2002) for an explanation of the values of the T tricubic spline. Also note that Stuart (2000) and Brenner (2002) have slightly different values for the node values of T.

Hope that helps.

Dave

If it could be of any help I had to refit all the bicubic and tricubic splines in the AIREBO potential file used fin LAMMPS and I have followed the paper in attachment to determine the interpolation coefficients with the same results of the professor Brenner’s code.

Beware that in the original paper (2002) describing the potential, there are few not corrected typos if compared to the MD FORTRAN code that you can find in http://www.mse.ncsu.edu/CompMatSci .

Regards,

Fabio Pavia

Brown University

Lekien-tricubic spline.pdf (273 KB)

Fabio,

Thanks, let us know when you make any progress on this issue. Actually, it is much better to let the code developer answer this question. He knows the detail better than anybody else.

Albert

You should communicate this to Steve Stuart's group
at Clemson. They have been doing careful tests of the AIREBO
in LAMMPS against their code. If you have found specific
issues you think need to be modified in the LAMMPS implementation
they would probably like to hear about them or they may have
some feedback for you.

His email is ss at g.clemson.edu.

Steve