I am calculating potential energy /atom of the different chunks of spherical particle of radius 2 nm. So I create chunks by using following command:

compute 3 all chunk/atom bin/sphere 0 0 0 0.0 20.0 4 discard no units box

So 4 chunks are created. Then I use the following command to calculate pe/atom

compute myPE all pe/atom
fix 1 all ave/chunk 1 20000 20000 3 c_myPE file chunk.profile

The output I get is :

Chunk-averaged data for fix 1 and group all

Timestep Number-of-chunks Total-count

Chunk Coord1 Ncount c_myPE

20000 4 3978
1 2.5 53.7734 -5.9446
2 7.5 443.979 -6.51198
3 12.5 1123.25 -6.28759
4 17.5 2356.99 -6.03041

So I find that pe/atom decreases from Chunk4 to Chunk 2 (outermost shell to inner) which is expected. But I find that Chunk1 (innermost) has highest pe/atom, even higher than the chunk having surface atoms. How is this possible? what can be the possible cause?

You can debug this yourself. Create a dump

file with the per-atom values and atom coords you are averaging

with fix ave/chunk. Verify that the atoms in

the chunk you are asking about sum to the value

reported for the chunk.