# Pair_style Table RSQ help & Pair_Write Issue

I have previously been having issues with the pair_style table and pair_write commands. It has been suggested that it is due to the large range of my table and rapidly varying r values. I have tried many things to no avail. A new question in my mind as I attempt to fix this issue on my own is how to put together an RSQ table? I ask this because I have searched far and wide and been unable to find a good example of a table using RSQ correctly. Additionally the doc page is a bit criptic in the area of RSQ only stating that it is the separation distance squared.

So when working with a range of about .001 to 10 Angstroms I imaging my first and last ‘r’ values are (.001)^2 and 10^2 but how do I fill the space between? Is it simply along the lines of (.002)^2, (.003)^2, … etc. until I reach something near 10?

Joe Morris

I have previously been having issues with the pair_style table and
pair_write commands. It has been suggested that it is due to the large
range of my table and rapidly varying r values. I have tried many things to
no avail. A new question in my mind as I attempt to fix this issue on my
own is how to put together an RSQ table? I ask this because I have searched
far and wide and been unable to find a good example of a table using RSQ
correctly. Additionally the doc page is a bit criptic in the area of RSQ
only stating that it is the separation distance squared.

So when working with a range of about .001 to 10 Angstroms I imaging my
first and last ‘r’ values are (.001)^2 and 10^2 but how do I fill the space
between? Is it simply along the lines of (.002)^2, (.003)^2, … etc. until I
reach something near 10?

the easiest way to get a good reference for such a table, is to use
the pair_write command, e.g. with lj/cut
i would recommend to use the melt example and replace the analytical
lj/cut with the tabulated version
(but you probably want to either extend the cutoff or shift the
potential to zero at the cutoff, to have a smooth behavior around the
cutoff).

then, if you find out good settings to match the analytical to the
tabulated variant, i would move on to more complex stuff.
i have used tabulated potentials for testing purposes quite a few
times and it can be made to work fine.

axel.