The blip trial wave function

General discussion of the Cambridge quantum Monte Carlo code CASINO; how to install and setup; how to use it; what it does; applications.
Post Reply
Nguyen_Oanh
Posts: 17
Joined: Mon Sep 12, 2016 10:32 am

The blip trial wave function

Post by Nguyen_Oanh »

Dear user!
I am using CASINO program with the blip wave function of graphene monolayer and C2HF monolayer.
I create the blip wave function from Quantum espresso directly with the plan wave and pw2casino.dat file.
I tested for my wave function and it is nonlocal orbital.
1. I have use pseudopotential from QE, and convert from QE to CASINO. (QE: i use format : C.pbe-hgh.UPF)
But I have a problem :" Invalid value of l_local in c_pp.data". If I use the pseudopotential c_pp.data from CASINO library, it ok. But I know that, pseudopotential in the QE have to the same form in CASINO.
2 . And when I run test for Graphene, the out file were had a problem: " In a reduced-periodicity blip calculation, your atom coordinates need to be
in the centre of the region spanned by the blip grid, which is the
parallelepiped defined by the lattice vectors starting from the origin. The
problem occurs for atom 1 in your basis (direction of third lattice vector)." How to fix it?
3. I already read in the manual but I don't understand about the localized orbitals. I want to create a blip representation of the localized orbital directly from QE directly.
I can run in QE with the type word: ./pw.x -pw2casino -i input.file ( the form of nonlocal orbitals with no symmetry)
How I can create the blip wave function with the localized orbitals? Can you explain detail for me?
Yours sincerely!
Katharina Doblhoff
Posts: 84
Joined: Tue Jun 17, 2014 6:50 am

Re: The blip trial wave function

Post by Katharina Doblhoff »

Dear Nguyen,
ad 1.)
So what is your value for l_local? Open the file with the casino tabulatd (i.e. converted) PP and have a look. That may bring us closer to the problem. How did you convert it? If you have trouble converting: Have a look in the forum, there is an old post on exactly that issue somewhere (I remember answering it).
ad 2.)
Casino already gives you the answer: Most likely, you have an atom sitting at 0,0,0 or you have your slab not centered in the middle of the unitcell. For pwscf that is fine since it uses 3D with vacuum, but casino cannot handle this since it really uses 2D periodicity. Hence you have to put your slab into the middle of the cell.
ad 3.)
for localized orbitals, I recommend reading section 27 and especially 27.2.2 of the manual where a step-wise instruction is given (but I have never done it myself). But do you expect localization schemes to work well for your system?
Nguyen_Oanh
Posts: 17
Joined: Mon Sep 12, 2016 10:32 am

Re: The blip trial wave function

Post by Nguyen_Oanh »

Dear Katharina Doblhoff !
1. I have used calculate the pseudopotential to create the wave function in QE. After, I convert these pseudopotential from QE to CASINO in up2fcasino ultility in QE to have the from pseudopotential in CASINO (HF). I have the new form of pseudopotential :
" Converted Pseudopotential in REAL space for C
Atomic number and pseudo-charge
6 4.00
Energy units (rydberg/hartree/ev):
rydberg
Angular momentum of local component (0=s,1=p,2=d..)
-3 " ( i think that this is a problem in here).
2. So, can you explain clearly about how to put my slab into the middle of the cell. Will I change these element in input file with QE.
3. I convert directly from the plan wave to the blip wave in QE with pw2casino file. I already the manual at 17.2.2. When I see in out file my blip function is nonlocal orbitals. I mean that, how can I build the blip wave functions with the local orbitals directly in pw2casino.dat file?
( Do I need centes.dat file with pw2casino.dat file?)
I reaaly expect the localization schemes to work well for my system
Thank you!
Katharina Doblhoff
Posts: 84
Joined: Tue Jun 17, 2014 6:50 am

Re: The blip trial wave function

Post by Katharina Doblhoff »

Hi!
ad 1.)
-3 " ( i think that this is a problem in here).
- Obviously. Have a look what your original file says. If that is the only problem, you can also edit the -3 manually, but be careful whether the rest went well.
ad 2.)
If you have your slab lying orthogonally to z and the lenght of your unit-cell in z-direction is for example 20, then make the z-components of your atoms centered around z=10 (since you have a single layer, shift the z-value of all atoms to z=10).
Nguyen_Oanh
Posts: 17
Joined: Mon Sep 12, 2016 10:32 am

Re: The blip trial wave function

Post by Nguyen_Oanh »

Thank you very much!
i will test again
Neil Drummond
Posts: 113
Joined: Fri May 31, 2013 10:42 am
Location: Lancaster
Contact:

Re: The blip trial wave function

Post by Neil Drummond »

Dear Nguyen,

I think your question 3 is simply about re-representing your Bloch orbitals in a blip basis? If so, you don't need to provide centres.dat. (Note that using a localised basis set, e.g., blips, is a separate issue from transforming your Bloch orbitals to a set of localised orbitals. As Katharina says, you probably don't want to attempt to construct truncated localised orbitals for graphene. Localisation works best for molecular crystals and such like.)

Best wishes,

Neil.
Nguyen_Oanh
Posts: 17
Joined: Mon Sep 12, 2016 10:32 am

Re: The blip trial wave function

Post by Nguyen_Oanh »

Dear Neil Drummond!
I alway have problem with the wave function that I create with QE.
When I used QE to generate with the wave function (plane wave /blip wave).
The problem is that: ERROR : PWFDET_SETUP
An insufficient number of eigenstates have been supplied.

So, How can I fix it?
Neil Drummond
Posts: 113
Joined: Fri May 31, 2013 10:42 am
Location: Lancaster
Contact:

Re: The blip trial wave function

Post by Neil Drummond »

Dear Nguyen,

The error message means that the number of available orbitals is less than the the number of spin-up or spin-down electrons - which obviously isn't going to work.

Perhaps you should show your Quantum Espresso input files and the CASINO input file on the forum, and someone familiar with Quantum Espresso may be able to help.

Best wishes,

Neil.
Post Reply