Hello all,
I was wondering is anyone has tried using the invariant descriptors generated either by symmetry functions or Chebyshev descriptors to understand similarity in structures.
I have a few structures that are very similar (with small displacements in the positions) and I am trying identity them and cluster them together using the descriptors.
Since the descriptor space is in high dimension, I cannot directly use L2-norm or cosine-similarity right? Any help/discussion will be useful.
Vaidish Sumaria