This thread is locked.Only browsing is available.
Top Page > Browsing
found minus eigenvalues of OLP at kloop=13 k1= 0 k
Date: 2005/04/17 11:53
Name: s. j. luo   <sjluo@phys.ntu.edu.tw>

Hi,
When I use openmx2.3 to do calculation. I find an information "found minus eigenvalues of OLP at kloop=13 k1= 0 k2= 0 k3= 0 and found minus eigenvalues of OLP kloop=13"in the file "*.std". What is it's meaning? and what should i do in the situation?

Thanks in advance!
sj luo
メンテ
Page: [1]

Re: found minus eigenvalues of OLP at kloop=13 k1= 0 k ( No.1 )
Date: 2005/04/19 23:17
Name: T.Ozaki

Hi,

The minus eigenvalue means that the basis set you specified is overcomplete
for your system. Since the localized atomic basis set used in OpenMX is not
a complete basis set, one may meet such a situation in some case.

Especially, this appears when a highly DENSE structure such as FCC and HCP
is calculated using MANY basis functions with a LONGER cutoff radius.
Once, the basis set becomes overcomplete in a bulk system, then the calculation
is invalid because of the division by near zero.

In another thread "Num. of grids overlapping with atom * = 0", you have shown
your input file. In this file, you seem to use quite longer cutoff radii for H,
C, and N, O elements. Maybe, the longer tail of basis functions is the reason
why you find the minus eigenvalue of overlap matrix. It would be better to use
a basis set with a shorter cutoff for your system.

Regards,

T.Ozaki
メンテ

Page: [1]