[Ape-users] Palladium pseudopotential

[Micael Oliveira]

Hi!

  It seems there are a couple of problems with your input file. First, 
it should be "Kr" instead of "Xe". Second, you are putting all your 4d 
electrons into the same spin-channel. In fact, about this later point, 
the code should complain about it. I will try to change this.

  Also, if you are using the GGA, you should change the mixing to linear 
mixing. The default is the Broyden mixing and it seems there is a 
problem with it when using the GGA that we where not able to solve yet. 
So, just adding "TypeOfMixing = linear" to your input file should do the 
trick.

  Finally, there is no much point in doing spin-polarized calculations 
if your atom is spin-unpolarized. This is not very important, as both 
calculations should yield the same results, but it will save you some time.

  Now, about the way to generate a good pseudo-potential for palladium, 
I will just give you a couple of general advices. First, Palladium is 
already quite heavy, so you should do scalar-relativistic calculations. 
Second, APE still does not have all the tools to test the 
transferability, but the calculation of the logarithmic derivatives has 
already been implemented, as well as the test for ghost-states.

  One final point. In the last months we fixed many bugs and implemented 
some new features, so maybe you should try the development version (the 
download instructions are on the web page).

  If you encounter more problems just let us know. Also, any comments or 
suggestions are welcome.

  Cheers,

Micael Oliveira


Zhou, Chenggang wrote:
> Dear All,
> 
> Very nice to join the APE group.  Previously, I was using ATOM utility 
> to generate pseudopotentials for SIESTA. Yesterday I found APE...it 
> looks very great than ATOM..but I am a little confused on the usage.
> 
> I had installed APE on my computer and the test of Cu runs pretty good.
> 
> Now, my question is, how to generate a GOOD pseudopotential for 
> Palladium with excellent transferability?
> 
> I modified the input file from the sample of Cu, something like that:
> 
> ========================================================
> 
> ##################################################
> Title = "Palladium"
> CalculationMode = ae
> Verbose = 40
> UnitsInput = 1
> UnitsOutput = 2
> 
> ##################################################
> # Hamiltonian
> ##################################################
> WaveEquation = schrodinger
> SpinMode = polarized
> 
> XFunctional = gga_x_pbe
> CFunctional = gga_c_pbe
> 
> ##################################################
> # Specie
> ##################################################
> NuclearCharge = 46
> 
> %Orbitals
> "Xe"
>  5 | 0 | 0 | 0
>  5 | 1 | 0 | 0
>  4 | 2 | 10 | 0
>  4 | 3 | 0 | 0
> %
> 
> ...
> 
> =========================================================
> 
> I didn't change the rest lines.
> 
> When I run $ape < Pd.ae <http://Pd.ae>
> 
> the following errors appear:
> 
> =======================================================
> ape < Pd.ae <http://Pd.ae>
>          APE - Atomic Pseudopotentials Engine - Version 0.7.2
> 
>               Program started on 2007/05/06 at 20:23:34
> 
> Calculation Type:
>   Atomic Calculation
> 
> Setting units
>   Input units system:  Atomic Units
>   Output units system: Atomic Rydberg Units
> 
> Initializing Wave-Equations Integrator
>   Stepping function: Embedded 8th order Runge-Kutta Prince-Dormand
>                      method with 9th order error estimate
>   ODE Integrator tolerance:  1.000E-07
>   ODE Integrator maximum number of steps:   5000
>   ODE Integrator minimum step size:  1.000E-15
> 
> Reading info about the atom
> forrtl: severe (174): SIGSEGV, segmentation fault occurred
> Image              PC                Routine            Line        Source
> ape                0000000000406427  Unknown               Unknown  Unknown
> ape                0000000000409568  Unknown               Unknown  Unknown
> ape                0000000000418352  Unknown               Unknown  Unknown
> ape                000000000041B2DC  Unknown               Unknown  Unknown
> ape                000000000041D9AE  Unknown               Unknown  Unknown
> ape                0000000000422B1A  Unknown               Unknown  Unknown
> ape                00000000004872AF  Unknown               Unknown  Unknown
> ape                00000000004A2415  Unknown               Unknown  Unknown
> ape                00000000004A2E5F  Unknown               Unknown  Unknown
> ape                0000000000405CF6  Unknown               Unknown  Unknown
> libc.so.6          0000002A95EBB197  Unknown               Unknown  Unknown
> ape                0000000000405C2A  Unknown               Unknown  Unknown
> 
> ============================================
> 
> What is wrong with that? 
> 
> And, how can I judge the quality of the generated pseudopotential? i.e. 
> the transferibility...
> 
> I knew we can use the bulk cohesive energy as a benchmark to evaluate 
> the quality. I tried many times using the ATOM tool to generate Pd pps, 
> but....I never get a result comparable with experimental value. HELP!! 
> Could anyone kindly tell me how can I generate a GGA-PBE pseudopotential 
> of Pd?
> 
> Any suggestions and solutions will be highly appreciated.
> 
> Thanks,
> 
> Chenggang Zhou
> 
> 
> 
> 
> 
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Ape-users mailing list
> Ape-users at www.tddft.org
> http://www.tddft.org/mailman/listinfo/ape-users