VASP with BoltzTraP

[goja25]

Hello all,
I wonder if it is somebody here who has been using BoltzTraP program together with VASP code. BoltzTraP (Boltzmann Transport Properties) is a program for calculating the semi-classic transport coefficients. I read that BoltzTraP program is already interfaced with AB-INIT and SIESTA programs, but I do not know how can I use it together with the VASP code. Is it possible to directly use the output files from VASP to the BoltzTraP program? I am new in this field and I would be grateful if someone gives me some tips on how to do this.

Thanks a lot.

[alirezaa66]

I'm sorry I had the same question but didn't understand how this paper is useful?

[kzhuo3]

I have been trying to do this too but I get stuck at some point. I will outline what I have done so far and maybe someone can comment on how to carry on.

Install the atomic simulation environment first and get all your python pathings in order. I am using Linux Mint 14 and found the ASE and python stuff all in the software center.

Next download and unzip boltztrap and find a "util" directory inside. Here you will find the interface code vasp2boltz.py and instructions vasp2boltz.txt. Follow the instructions of vasp2boltz.txt to make a python file. I just called it "d.py" to use on a Si primitive cell test case. Here are the contents:
======================================
#! /usr/bin/python
from ase import io
from ase.lattice.spacegroup import Spacegroup
import vasp2boltz

ao = io.read('POSCAR')

# Input spacegroup info for Si manually
sg = Spacegroup(227)
ao.info = {'spacegroup': sg}

# The remaining part takes care of writing the files required by BoltzTraP
bs = vasp2boltz.get_vasp_bandstructure()
vasp2boltz.write_bandstructure_boltztrap(bs)
vasp2boltz.write_structure_boltztrap(ao)
vasp2boltz.write_intrans_boltztrap()
======================================

Make a new folder somewhere (I call it Si) and put the following files in:
d.py
vasp2boltz.py
POSCAR
EIGENVAL
OUTCAR
Then in the terminal do: ./d.py

This will produce: energies.boltztrap, hte.intrans and hte.struct
I change their names to Si.energy, Si.intrans and Si.struct respectively.
The format of energies.boltztrap and hte.struct are detailed in section "1.3.1 File format description for the GENE file format" of the boltztrap user guide.
These are basically your VASP to boltztrap interface files. So far so good.


Then in the terminal do: "path to src"/x_trans BoltzTraP
and then HERE I GET STUCK because of some fortran error
================ BoltzTraP vs 1.2.2 =============
At line 70 of file generic_band.F90 (unit = 20, file = 'Si.struct')
Fortran runtime error: Bad integer for item 1 in list input
0.0u 0.0s 0:00.00 0.0% 0+0k 0+8io 0pf+0w

OK that is all I have understood so far. Any further advice anyone?

I think if I can get the last step to run correctly then I will get the required output.

[kzhuo3]

continuing my previous post ...

this error
================ BoltzTraP vs 1.2.2 =============
At line 70 of file generic_band.F90 (unit = 20, file = 'Si.struct')
Fortran runtime error: Bad integer for item 1 in list input
0.0u 0.0s 0:00.00 0.0% 0+0k 0+8io 0pf+0w

happens because the symmetry operations as produced by vasp2boltz.py in the .struct file are not integers. They come out as e.g.
-1.0 1.0 -1.0
1.0 0.0 0.0

I guess this is a bug in the vasp2boltz.py code which you can change if you know python. I found my own quick fix using some bash scripting to reformat the output hte.struct file.
sed 's/\.0//g' hte.struct >sometempfile; mv sometempfile Si.struct
This will remove all the .0 to make it integer like
-1 1 -1
1 0 0 etc

So now just rename stuff (hte.somestuff is the default output)
hte.intrans must be Si.intrans
hte.struct must be Si.struct (already done above)
energies.boltztrap must be Si.energy
BTW, the name of your work folder and files MUST be Si also or whatever system you are working on.

Now we should have 6 files just to be clear
EIGENVAL, POSCAR, OUTCAR
Si.struct, Si.intrans, Si.energy

And finally you are good to go. Just do
"path to boltztrap src"/x_trans BoltzTraP

and it should at least produce some stuff. Watch out for the random capital letters!



Only problem now is to link the ASE spacegroup library with vasp2boltz.py. If you can't don't worry, just manually input the space group info into the d.py file I mentioned earlier. i.e. sg = Spacegroup("whatever group number").

[LABurton]

hello,

firstly thank you kzhuo3 for posting your progress. I'm also having trouble with this program. When i run the utility to generate the boltztrap files from the OUTCAR, POSCAR and EIGENVAL files, I receive the message "write_bandstructure_boltztrap: Fermi energy not known, nothing done".

At this point only the case.intrans and the case.struct files are generated but not the case.energy.

Is there a specific flag or something i need in the original VASP calculation? any help would be appreciated.

[kzhuo3]

Do you have this file: energies.boltztrap?
If you do, then rename it as the case.energy.

If not, the answer may be in the vasp2boltz.py script. If you read it, you should find lines such as
===========
# get the Fermi energy from a selfconsistent OUTCAR file (not from bs calculation!)
===========

============
if (E_Fermi_zero==True) and (bandstructure['E_Fermi']==None):
print 'write_bandstructure_boltztrap: Fermi energy not known, nothing done'
return False
============

Not good at reading python but it seems to me like vasp2boltz.py could not find the fermi energy in your OUTCAR file.

Try this in your terminal: grep fermi OUTCAR
That should tell you whether your OUTCAR contains a fermi energy or not. If it doesn't you could also try to alter the vasp2boltz.py file by manually assigning the fermi energy. I think (not sure) that you could type in a line "ef=1.23456" as "ef" should be the fermi energy variable in the script.

[kzhuo3]

And also, I'm trying to rewrite parts of the BoltzTraP codes in the src folder to make BoltzTraP work for 2D systems like graphene. If anyone knows which F90 files need to be altered and how to do so, pls post. Thanks.

[tuocvu]

Dear kzhuo3,
I also have the same interest as you, using BoltzTraP code for low-dim structure (nanowire, nanorod), i.e. specify space group manually as P1 (International Tabele # 1) Triclinic (# of Operator 1) 3D crystal because of vacuum wrap in vasp2boltz.
Have you succeeded with your 2D structure?
VNT

[kldas-05]

Hello all
firstly thank you kzhuo3 for posting this information.
I have a problem using the boltztrap. I used the vasp2boltz code, i change the name of the hte and energy files and when i use the boltztrap code, this error appears:

================ BoltzTraP vs 1.2.3 =============
STOP error in factorization.

Somebody knows how i can resolve this problem?

[kzhuo3]

kldas,

I don't know exactly but I'll share some insights anyway.

This error originates from the fite4.F90 subroutine. Comput. Phys. Commun. 2006, 175, 67-71, explains that this deals with the star function expansion for your BAND ENERGIES.

Internally, fite4.F90 uses a standard DGETRF subroutine (http://www.netlib.no/netlib/lapack/double/dgetrf.f) to do matrix LU decomposition. You get the error because DGETRF was not happy. In your .outputtrans, check the value for "int" and compare it with the possible "INFO" output arguments from the DGETRF manual in http://www.netlib.no/netlib/lapack/double/dgetrf.f to find out the problem with your matrix. My guess is there was a singularity somewhere i.e. inf or INFO > 0.

If you post your .intrans and .outputtrans and say a bit about your system, I'll see if I can help some more. Meanwhile, check that your band structure and .energy file is ok.

[kldas-05]

Hello kzhuo3,
My sistem has a hexagonal structure
The .intrans file is:
===================================
GENE
0 0 0 0.0
0.72737172 0.0005 0.3 9.0
CALC
4
BOLTZ
.20
300. 50.
-1.
HISTO

==========================================

and the .outputtrans obtained is:

==========================================
dos normalization: 2.0000000000000000
About to call read_input
=================== INPUT VARIABLES ======================
Bandstyle: GENE
Set fermi: 0
Debug: 0
FermiE: 0.7274. step size: 0.0005 Ecut: 0.3000. # val. e: 9.000
Run type: CALC
Fourier expansion factor: 4
Calc type: BOLTZ
Fermi level range: (Ry) 0.2000
Max temperature: (K) 300.0 Temp step: (K) 50.0
Range around Ef where bands are given individual output (Ry) -1.0000
HISTO used for calculation of DOS
0 doping levels will be used
=========================================================
Band style: GENE
1 0 0 1
0 1 0
0 0 1

0 -1 0 2
1 -1 0
0 0 1

-1 1 0 3
-1 0 0
0 0 1

-1 0 0 4
0 -1 0
0 0 1

0 1 0 5
-1 1 0
0 0 1

1 -1 0 6
1 0 0
0 0 1

0 1 0 7
1 0 0
0 0 -1

1 -1 0 8
0 -1 0
0 0 -1

-1 0 0 9
-1 1 0
0 0 -1

0 -1 0 10
-1 0 0
0 0 -1

-1 1 0 11
0 1 0
0 0 -1

1 0 0 12
1 -1 0
0 0 -1

Number of kpts in IBZ: 4500
Input file read successfully
NON-CENTROSYMMETRIC. ADDING i
About to call subroutine bandana
============== OUTPUT from BANDANA ================
Egap: 0.000000 Energy range: 0.208699 - 1.500383. Bands range: 7 - 14
VBM: 0.727372 CBM: 0.727372 Efermi: 0.727372

============== End BANDANA ====================
Subroutine bandana executed successfully
Approx number of kpts in BZ : 108000
About to enter gen_lattpoints
======= OUTPUT FROM gen_lattpoints ======================
KXMAX,KYMAX,KZMAX 53 53 43
GMAX 260.49255978544358
19554 LATTICE POINTS GENERATED

SIZE INCLUDING STAR MEMBERS = 431875
USED TIME: 2.5940001010894775
=============== END gen_lattpoints ======================
Allocating engre...
engre allocated
About to enter fite4...
fite4 matrix setup: 604.066
inf = 3000
====================================00

You can see that the inf value is >0 but i don't know what is the error source.

[kzhuo3]

I'm almost certain your error is due to your band energies.


Let's focus on this part which analyses your band structure. I have a few questions and comments for you.
============== OUTPUT from BANDANA ================
Egap: 0.000000 Energy range: 0.208699 - 1.500383. Bands range: 7 - 14
VBM: 0.727372 CBM: 0.727372 Efermi: 0.727372


It says here that you have no band gap since your VBM, CBM and Efermi are all the same. Are you working on a 0 gap system like graphene? What is the chemical name of your material and how many valence electrons does it have?

Your Efermi is set to 0.727372 Ry (=9.896403814 eV). Are you sure it is so high?

In .intrans, you are using 9 valence electrons (sure?), then VBM should be at band 4 or 5 which is outside your band range from 7 - 14. Ideally, you want your bands to range from 1 to a few bands above the CBM. To get the larger band range, you can increase X in your .intrans line: 0.72737172 0.0005 X 9.0.

CALC
4 # I usually use about 20 or more for this value. I sometimes get errors with small values like 4.
BOLTZ

For testing purposes, you can use 300. 300. instead of 300. 50. to speed things up. This just uses a single temperature 300K.

[kldas-05]

The system in which I'm working is the AlB2, it's a 0 gap system with 9 valence electrons, I'm sure.
I moved the Fermi level to 0Ry, i used 20 and more in lpfac, the X value in the intrans line:0.72737172 0.0005 X 9.0 was increased up to 1 but the error continues to appear (U=3000) in the single temperature 300K

[kldas-05]

*inf=3000

[bioggg]

I am encountering the same problem but don't know how to solve it too. I think inf=3000 means that you have a singularity and I guess you knew it already. Please someone helps us out. Thanks a lot.