inverse participation ratio

[eminamitani]

construct module to calculate inverse participation ratio

[Sctdhy]

nmodes=len(eigenvector)
iprdat=np.zeros(nmodes)
for i in range(nmodes):
vec=eigenvector[:,i]
inner=vecvec
ipr=innerinner
print(ipr.sum())
iprdat[i]=ipr.sum()
In the original paper, it seems that we should first calculate the sum of squares for the three directions of an atom within a particular pattern before squaring it again.

[eminamitani]

I made get_IPR_rev module to evaluate the following equation directly. The eigenvector of vibrational mode is normalized, thus, the denomintor is 1. Thus, I evaluate IPR by using np.power(eigenvector_for_mode,4).

[eminamitani]

And, this method and the previous version return the same results, since both evaluate the sum of respective elements to the 4th power, 1x^4+1y^4+1z^4 ..... Nx^4+Ny^4+Nz^4

[Sctdhy]

Thank you,Prof. Minamitani. I've sorted this problem.