How to solve "Eigenvector with band 1 has zero norm"?  [SOLVED]

Total energy, geometry optimization, DFT+U, spin....

Moderator: bguster

Locked
FBNIC
Posts: 3
Joined: Thu May 03, 2018 4:01 pm

How to solve "Eigenvector with band 1 has zero norm"?

Post by FBNIC » Thu May 03, 2018 4:10 pm

I am a beginner on Abinit Calculations. When I try to run the scripts as those in tutorial course, I found the following error messages, and the program terminated. Could you give any hints?

--- !ERROR
src_file: fxphas.F90
src_line: 188
mpi_rank: 0
message: |
The eigenvector with band 1 has zero norm.
This usually happens when the number of bands (nband) is comparable to the number of planewaves (mpw)


My Input file from website is in the following:
#################################################################
# Input file for the positron tutorial #
# Positron lifetime calculation within PAW #
# Si, 2 atoms in the box #
#################################################################
# Datasets definition
ndtset 2
positron1 0 ! Dataset 1 is a simple electronic GS calculation
positron2 1 ! Dataset 2 is a positronic GS calculation
getden2 1 ! in presence of the previous electronic density
kptopt2 0 ! Use only k=gamma point
ixcpositron2 1 ! We are using the Boronski and Nieminen parametrization
# Common input parameters
! Unit cell
acell 3*5.43 angstrom
rprim 0.0 0.5 0.5
0.5 0.0 0.5
0.5 0.5 0.0

natom 2
ntypat 1
typat 2*1
znucl 14
xred 0.0 0.0 0.0
0.25 0.25 0.25

! K-points and occupations
kptopt 1
ngkpt 4 4 4
nshiftk 1
shiftk 0.0 0.0 0.0
occopt 1
nband 6

posocc2 1 ! Occupation number for the positron (should be set <1 for bulk calculation with a small cell).
! Here the zero positron density limit is used, so results do not depend on posocc.

! Convergence parameters
ecut 8. pawecutdg 15.
iscf 17
nstep 50 tolvrs 1.d-8

! Miscelaneous
prtwf 0 prteig 0 ! To save disk space
optforces 0 optstress 0 ! Not relevant here

---------------------------------------------------------------------------------------------------------------------
Last edited by FBNIC on Thu May 03, 2018 4:21 pm, edited 1 time in total.

FBNIC
Posts: 3
Joined: Thu May 03, 2018 4:01 pm

Re: How to solve "Eigenvector with band 1 has zero norm"?

Post by FBNIC » Thu May 03, 2018 4:14 pm

I have checked:
1) Band number is less than mpw.
2) Increasing ecut is not effective.

In addition, I can run other scripts as tpositron2.in, tpositron3.in ... very well.
I do not know why the tpositron1.in file cannot run.

ebousquet
Posts: 469
Joined: Tue Apr 19, 2011 11:13 am
Location: University of Liege, Belgium

Re: How to solve "Eigenvector with band 1 has zero norm"?  [SOLVED]

Post by ebousquet » Thu May 03, 2018 4:46 pm

dear FBNIC ,
I've submitted the input you give in your post with the Pseudodojo pseudo of Si (http://www.pseudo-dojo.org/) and I don't get any error message, it runs nicely. So I don't really know what could be the problem in your run, could you make a run with the Pseudodojo pseudo to check?
Best wishes,
Eric

FBNIC
Posts: 3
Joined: Thu May 03, 2018 4:01 pm

Re: How to solve "Eigenvector with band 1 has zero norm"?

Post by FBNIC » Thu May 03, 2018 6:42 pm

Hey, ebousquet, thank you for your post in advance. Now, the script can work well using the potential, which download from the website you provided.

However, the potential provided in psps_for_tests seems not to work well, which meet the problem as my first post.

In addition, although the potentials from your provide website can work well, the results are completely different from the running results in the tutarial course.
Last edited by FBNIC on Fri May 04, 2018 12:45 am, edited 1 time in total.

ebousquet
Posts: 469
Joined: Tue Apr 19, 2011 11:13 am
Location: University of Liege, Belgium

Re: How to solve "Eigenvector with band 1 has zero norm"?

Post by ebousquet » Thu May 03, 2018 9:29 pm

Yep, in the future the automatic test will use the Pseudodojo pseudos.
The fact that the result is very different is because the ecut and k-points used in the tests/tuto are very low to have them running quickly. You have to test the convergence of the physical properties you want to analyze w.r.t. ecut and k-points anyway.
All the best,
Eric

Locked