Overlaps in 2d hpmc simulation for mixture of ellipses and circles.

[sauravIISc]

Hi,

I am trying to simulate a mixture of circles and ellipses in a 2D box using hoomd.hpmc.integrate.Ellipsoid. In 2D, after a short preliminary run, I see overlaps between particles when I visualize them. However, mc.overlaps from the integrator gives me a value of 0. We know overlaps are prohibited for hard particles, and I see overlap when I visualize them. I suspect it could be due to some bugs in the integrator when simulations are performed in 2d because there were no overlaps when I tried the 3d box. I could also be making some mistakes in defining the parameters, so any help or clarification is appreciated. I have also attached a minimal script to see the overlaps.
minimal_script.zip

Thanks in advance,

[joaander]

You closed this discussion. Were you able to resolve the issue?

[sauravIISc]

Thank you for the reply. I solved the issue temporarily by defining a nonzero dimension (c) for the ellipse and circles in the z-direction, keeping everything else the same.

[joaander]

Yes, this is the workaround you should use. The fast numerical methods HOOMD-blue uses to detect overlaps assume that a, b and c are all non-zero. At the same time, your shapes will only rotate about the z axis in 2D boxes effectively giving you ellipses as desired.

I opened #1762 so that a future HOOMD-blue release will provide an error message when given Ellipsoid is given a 0 value.

[sauravIISc]

Thanks, it clarified the issue.