LIGGGHTS generating "Complex" eigenvalues for superquadric orientation tensor

Submitted by coshivam on Fri, 03/16/2018 - 15:06

I am trying to simulate ellipsoidal particles using superquadrics atom_style. Therefore, I started testing with the input file that was released by LIGGGHTS in the example folder (It is attached). For visualization in Paraview, I have compiled it with the Tensorglyph plugins.
I have also attached the video of visualization. I am not able to understand, why after the collision the semi-axes values of both the ellipsoids are changing. For getting a closer look at the problem, I calculated the eigenvalues of both the ellipsoids and found that they are
1, a+ib, a-ib such that a^2 +b ^2 =1. I am not able to understand why LIGGGHTS generated these complex eigenvalues (because for ellipsoid the eigenvalues should be a positive number) and why the semi-axis of the ellipsoids are changing with time?
Are these two problems related?
I humbly appreciate your help.

Best Regards,
Shivam

AttachmentSize
Microsoft Office document icon LIGGGHTS input file4.44 KB
Binary Data Video1.2 MB