SDF expansion array returns all zeros for concave polygons at density=0.3

[aidexs]

Hello,

I'm researching the relationship between pressure and the scale distribution function for concave particles. I'm using simple polygons, but when trying to get sdf.expansion and sdf.compression with SDF, sdf.compression works normally while the sdf.expansion array is all zeros. When back-calculating with beta_P, it seems s(0-) should have a value. My number density is 0.3. What might be the cause?

My particle:

concave_vertices = [
(0,1),
(-0.5, -0.25),
(-1,-1),
(0, -0.5),
(1, -1),
(0.5, -0.25),
]

Here's the key code:

sdf_compute = hoomd.hpmc.compute.SDF(xmax=0.05, dx=5e-4)
simulation.operations.add(sdf_compute)
s0_compression_sum = 0
s0_expansion_sum = 0
num_valid_s0 = 0
for i in range(int(num_runs)):
simulation.run(100)
s0_compression = sdf_compute.sdf_compression
s0_expansion = sdf_compute.sdf_expansion
print(f's0_compression={s0_compression}')
print(f's0_expansion={s0_expansion}')
print(f's0_avg={(sdf_compute.betaP/0.3 -1)*4}')
s0_compression_sum += s0_compression
s0_expansion_sum += s0_expansion
num_valid_s0 += 1

Could the issue be with xmax=0.05 being too small? Or do concave particles need special settings for expansion calculations? Let me know if more info is needed.

[joaander]

To achieve samples on the expansion side, you need to have particles arranged so that they start to overlap when moved further apart. Star shaped polygons like yours don't tend to arrange themselves this way. I use a staple shape in the validation scripts to achieve a reasonable number of samples on the expansion side: https://github.com/glotzerlab/hoomd-validation/blob/trunk/hoomd_validation/simple_polygon.py.

[aidexs]

I see. I only considered the difference between concave particles and convex particles at the beginning, without taking into account that some concave particles will not overlap. Thank you for your help.