ABINIT Discussion Forums

Dear all

In the process of lattice optimization,there are some comments
1.--- !COMMENT
src_file: m_xgScalapack.F90
src_line: 251
message: |
xgScalapack in auto mode

2.getcut : COMMENT -
Note that boxcut > 2.2 ; recall that boxcut=Gcut(box)/Gcut(sphere) = 2
is sufficient for exact treatment of convolution.
Such a large boxcut is a waste : you could raise ecut
e.g. ecut= 40.897903 Hartrees makes boxcut=2

I have finished the ecut convergence,34Ha is enough.why the comment appeard?

In the process l get the ddb.file,
1.--- !WARNING
src_file: m_drivexc.F90
src_line: 1076
message: |
Density went too small (lower than xc_denpos) at 28058 points
and was set to xc_denpos = 1.00E-14. Lowest was -0.32E+02.
Likely due to too low boxcut or too low ecut for pseudopotential core charge.

2.--- !ScfConvergenceWarning
message: |
nstep 60 was not enough SCF cycles to converge.
I finished lattice optimization,Why is it hard to converge?

3--- !WARNING
src_file: m_crystal_io.F90
src_line: 192
message: |
Alchemical crystals are not fully supported by the netcdf format
Important parameters (e.g. znucl, symbols) are not written with the correct value

4.WARNING : still subject to testing - especially symmetries.

5. Alchemical pseudopotential with nlcc!

6.--- !WARNING
src_file: m_crystal_io.F90
src_line: 192
message: |
Alchemical crystals are not fully supported by the netcdf format
Important parameters (e.g. znucl, symbols) are not written with the correct value
7.--- !WARNING
src_file: m_hdr.F90
src_line: 4570
message: |
Restart of self-consistent calculation need translated wavefunctions.
How can i remove the warning?

In my process to get elastic contants
1.--- !WARNIN
Unstable eigenvalue detected in force constant matrix at Gamma point
The system under calculation is physically unstable.

2.--- !WARNING
src_file: m_ddb_elast.F90
src_line: 360
message: |

Unstable eigenvalue detected in force constant matrix at Gamma point.
The system under calculation is physically unstable.

3. ddb_piezo : WARNING -
Acoustic sum rule violation met : the eigenvalues of accoustic mode
are too large at Gamma point
Increase cutoff energy or k-points sampling.
The three eigenvalues are: -2.261762E-05 6.285754E-05 -2.261762E-05

Why is it there?How to eliminate them?

Looking forward to your repiy.
jlwindy

Hello,

these are all warnings, not fatal and not errors. Most of them are quite explicit actually.
Also note that for the most part these have nothing to do with elastic constants and post processing in general.

NB: you have told us nothing about your system. Please read the nettiquette and provide details.

jlwindy wrote:Dear all

In the process of lattice optimization,there are some comments
1.--- !COMMENT
src_file: m_xgScalapack.F90
src_line: 251
message: |
xgScalapack in auto mode

you may have compiled with scalapack...

jlwindy wrote:2.getcut : COMMENT -
Note that boxcut > 2.2 ; recall that boxcut=Gcut(box)/Gcut(sphere) = 2
is sufficient for exact treatment of convolution.
Such a large boxcut is a waste : you could raise ecut
e.g. ecut= 40.897903 Hartrees makes boxcut=2

I have finished the ecut convergence,34Ha is enough.why the comment appeard?

you probably used dilatmx. The easiest is to leave it as is, otherwise abinit is telling you you can remove/reduce dilatmx for the final structure, increase ecut, reconverge the forces, and proceed with a slightly better basis set.

jlwindy wrote:In the process l get the ddb.file,
1.--- !WARNING
src_file: m_drivexc.F90
src_line: 1076
message: |
Density went too small (lower than xc_denpos) at 28058 points
and was set to xc_denpos = 1.00E-14. Lowest was -0.32E+02.
Likely due to too low boxcut or too low ecut for pseudopotential core charge.

this is a warning, not a problem. You probably have vacuum. Due to Fourier transforms through reciprocal space, you can have a few points with negative density, which are removed.

jlwindy wrote:2.--- !ScfConvergenceWarning
message: |
nstep 60 was not enough SCF cycles to converge.
I finished lattice optimization,Why is it hard to converge?

unrelated to optimization. You may have vacuum or magnetism. As noted above, we have no way of knowing. 1) increase nstep and see if it converges at all. 2) increase nband with some empty bands 3) set diemac between 4 and 10 if you have vacuum.

jlwindy wrote:3--- !WARNING
src_file: m_crystal_io.F90
src_line: 192
message: |
Alchemical crystals are not fully supported by the netcdf format
Important parameters (e.g. znucl, symbols) are not written with the correct value

this is pretty explicit - what do you not understand? The netcdf refers to different output files

jlwindy wrote:4.WARNING : still subject to testing - especially symmetries.

without context I have no idea where this shows up.

jlwindy wrote:5. Alchemical pseudopotential with nlcc!

as 3 above: you are doing alchemy with non linear core corrections, and the cores are not mixed in a consistent way. This is a bit complex...

jlwindy wrote:6.--- !WARNING

you duplicate 3.

jlwindy wrote:7.--- !WARNING
src_file: m_hdr.F90
src_line: 4570
message: |
Restart of self-consistent calculation need translated wavefunctions.
How can i remove the warning?

you may have changed plane wave basis set, or k grid

jlwindy wrote:In my process to get elastic contants
1.--- !WARNIN
Unstable eigenvalue detected in force constant matrix at Gamma point
The system under calculation is physically unstable.

You calculate the bare elastic constants (rigid ion) and it tries to combine with the Gamma point phonons to see the internal relaxation which would happen. In your case you have an unstable mode, so it can not invert the dynamical matrix. You need to converge the initial structure correctly, and if there are still symmetry breaking phonons you have to get to a properly dynamically stable phase to calculate the elastic constants - they are not defined for an unstable system.

jlwindy wrote:2.--- !WARNING
Unstable eigenvalue detected in force constant matrix at Gamma point.
The system under calculation is physically unstable.

again you duplicate 1.

jlwindy wrote:3. ddb_piezo : WARNING -
Acoustic sum rule violation met : the eigenvalues of accoustic mode
are too large at Gamma point
Increase cutoff energy or k-points sampling.
The three eigenvalues are: -2.261762E-05 6.285754E-05 -2.261762E-05

Why is it there?How to eliminate them?

see above. If you find an unstable phonon mode, displace the atoms along the eigenvector and relax again to find a more stabilized structure.
Sometimes, if you are unlucky, the phase can have unstable modes even if it is basically relaxed, and it can be tough to get everything numerially clean.

Dear Verstraete
Thank you very much.Your reply has helped me a lot.I don't understand the phonon mode.l well learn it.

Best wishes.
jlwindy