[QE-users] Testing Al PBE pseudo

[Zack Gainsforth]

Hi all,

I figured this out and since the question seems to have popped up at least
twice before in the QE forum without a clear answer I thought it may be
worth mentioning the solution.

The answer seems to be that the configts card needs to use the valence
states only.  Namely, in my case:

configts(1)='1s2 2s2 2p6 3s2 3p1'

fails, while

configts(1)='3s2 3p1'

is fine.  Looking at
http://www.democritos.it/pipermail/pw_forum/2012-April/023896.html and
https://lists.quantum-espresso.org/pipermail/users/2012-October/025269.html,
one can see that the same error is occurring and in those cases the
configts card either misses some states, or includes too many.

Hopefully this saves someone a little time down the road.

Cheers,

Zack


On Fri, Aug 21, 2020 at 11:56 AM Zack Gainsforth <zackg at berkeley.edu> wrote:

> Hello all,
>
> I’m having difficulty with testing a pseudo.
>
> I started with the Al.pbe-n-kjpaw_psl.1.0.0.in file from the PS Library.
> I’m using QE installed through synaptic, which is ld1.x version 6.4.1.  I
> also tried this with 6.6 that I downloaded from the website and compiled.
>
> In both cases, I can run the .in file to generate the pseudo without
> complaint, and I see no obvious issues in the output.  When I plot the log
> derivatives they seem to match up very nicely.  When I try to run a test
> configuration though, it fails.
>
> It looked like it was complaining about the wavefunctions having nodes, so
> I tried adding rho0=0.01 and out of desperation 0.1!  To no avail.
>
> My question is whether I am doing something else incorrect in how I’m
> setting up the test?
>
> Thanks,
>
> Zack
>
> Here is the .in file I’m using to run the test (the original file from PS
> Library is exactly the same except I’ve added a &test section and changed
> iswitch=2 instead of 3):
>
>  &input
>    title='Al',
>    zed=13.,
>    rel=1,
>    config='[Ne] 3s2 3p1 3d-1',
>    iswitch=2,
>    dft='PBE'
>  /
>  &inputp
>    lpaw=.true.,
>    pseudotype=3,
>    file_pseudopw='Al.pbe-n-kjpaw_psl.1.0.0.UPF',
>    author='ADC',
>    lloc=-1,
>    rcloc=1.9,
>    which_augfun='PSQ',
>    rmatch_augfun_nc=.true.,
>    nlcc=.true.,
>    new_core_ps=.true.,
>    rcore=1.8,
>    tm=.true.
>  /
> 6
> 3S  1  0  2.00  0.00  1.70  1.90  0.0
> 3S  1  0  0.00  6.00  1.70  1.90  0.0
> 3P  2  1  1.00  0.00  1.70  2.00  0.0
> 3P  2  1  0.00  3.50  1.70  2.00  0.0
> 3D  3  2  0.00  0.10  1.70  1.90  0.0
> 3D  3  2  0.00  0.30  1.70  1.90  0.0
> &test
>     file_pseudo='Al.pbe-n-kjpaw_psl.1.0.0.UPF'
>     nconf=1
>     configts(1)='1s2 2s2 2p6 3s2 3p1'
> /
>
> With both versions, the test runs the AE calculation OK, and then when it
> starts to run the pseudo:
>
> With 6.6 I see:
>
>        Warning: n=1, l=0 expected 0 nodes, found *
>        Setting wfc to zero for this iteration
>        (This warning will only be printed once per wavefunction)
>
>        Zero norm: self consistency problem; state:  1 (l=  0, j=0.0)
>
>        Warning: n=2, l=0 expected 1 nodes, found *
>        Setting wfc to zero for this iteration
>        (This warning will only be printed once per wavefunction)
>
>        Zero norm: self consistency problem; state:  2 (l=  0, j=0.0)
>
>        Warning: n=2, l=1 expected 0 nodes, found *
>        Setting wfc to zero for this iteration
>        (This warning will only be printed once per wavefunction)
>
>        Zero norm: self consistency problem; state:  3 (l=  1, j=0.0)
>
>        Warning: n=3, l=0 expected 2 nodes, found *
>        Setting wfc to zero for this iteration
>        (This warning will only be printed once per wavefunction)
>
>        Zero norm: self consistency problem; state:  4 (l=  0, j=0.0)
>
>        Warning: n=3, l=1 expected 1 nodes, found *
>        Setting wfc to zero for this iteration
>        (This warning will only be printed once per wavefunction)
>
>        Zero norm: self consistency problem; state:  5 (l=  1, j=0.0)
>        Zero norm: self consistency problem; state:  1 (l=  0, j=0.0)
>        Zero norm: self consistency problem; state:  2 (l=  0, j=0.0)
>        Zero norm: self consistency problem; state:  3 (l=  1, j=0.0)
>        Zero norm: self consistency problem; state:  4 (l=  0, j=0.0)
>        Zero norm: self consistency problem; state:  5 (l=  1, j=0.0)
>         [many lines removed]
>      Message from routine run_pseudo:
>      Warning: convergence not achieved
>        Zero norm: self consistency problem; state:  1 (l=  0, j=0.0)
>        Zero norm: self consistency problem; state:  2 (l=  0, j=0.0)
>        Zero norm: self consistency problem; state:  3 (l=  1, j=0.0)
>        Zero norm: self consistency problem; state:  4 (l=  0, j=0.0)
>        Zero norm: self consistency problem; state:  5 (l=  1, j=0.0)
>
>
> With 6.4,
>
>        Warning: n=2, l=0 expected 1 nodes, found 0
>        Setting wfc to zero for this iteration
>        (This warning will only be printed once per wavefunction)
>
>        Zero norm: self consistency problem; state:  2 (l=  0, j=0.0)
>
>        Warning: n=3, l=0 expected 2 nodes, found 0
>        Setting wfc to zero for this iteration
>        (This warning will only be printed once per wavefunction)
>
>        Zero norm: self consistency problem; state:  4 (l=  0, j=0.0)
>
>        Warning: n=3, l=1 expected 1 nodes, found 0
>        Setting wfc to zero for this iteration
>        (This warning will only be printed once per wavefunction)
>
>        Zero norm: self consistency problem; state:  5 (l=  1, j=0.0)
>
>
>  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>      Error in routine new_paw_hamiltonian (1):
>      negative rho
>
>  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20200825/dbbead3a/attachment.html>