Question about dosanalyze.pl

[pnita]

Dear all,

I have been using the dosanalyze.pl for a while without any unusual output. Anyway, I just re-downloaded a VTSTTools from the website a couple days ago and tried to use the dosanalyze.pl again. I was trying to evaluate the d-band center of atom no.1 in my system, so I typed the following command:

login3% dosanalyze.pl d 1

And here is the result I obtained:

No energy range specified, integration will cover the whole range.
Total DOS is zero.

I did try with other atoms as well as all atoms, but I always received the message "Total DOS is zero."

I'd really appreciate any advice on this matter.

Thank you very much.
Panithita

[graeme]

Hm, it would be interesting to know what has changed. Can you upload your files so we can see the problem?

[pnita]

I've attached the DOS1 and the dosanalyze.pl script that I used here. I hope it is okay to attach as a txt file. I was trying to upload the original output file but there was a problem with the extension.

Thank you very much for your help, and please let me know if you need other information.

[graeme]

The DOS1 has a format that the script doesn't understand -- there are more columns than expected. Can you also updated the DOSCAR, or the entire calculation? Are there, perhaps, f-electrons in your calculation?

Well, something is clearly wrong with the script that should be fixed. Any help would be appreciated.

[pnita]

Sorry for late response. I'm attaching the original DOSCAR, OUTCAR and the split_dos script here.

My system doesn't involve with f-electrons. It is a clean surface of vanadium. So each column in the DOS1 file refers to:

Energy s py pz px dxy dyz dz2 dxz dx2y2

Please let me know if you need further information.
Thank you very much for your help.

[lgxyz]

So, is there any progress for this problem. I have got the same question: Total DOS is zero.Thank you.

[graeme]

Our script does not properly work with LORBIT=11, in which the bands are lm decomposed. This addition really needs to be added, but in the mean time, it should work if you remove the LORBIT tag.

[lgxyz]

hi, I want to calculate the d band center of the whole system, including all the atoms. What should I need type?
I type that "dosanalyze.pl d 1" and it give the only value of no. 1 atom's d band center. I want the d band center of the whole system. Waht should I do?
Thanks very much!

[graeme]

The input format looks a little clunky, but you can add as many atom numbers as you want, and the script will add them up for you. We really should have options to include atom ranges and all atoms.

[pradeep]

Dear Users and developers,
I am writing this email to get some help in connection with the WRONG results obtained upon compilation of the dosanalyze.pl code. I followed the following steps to get numerical estimate of the d-band centre of a molecular cluster, say, for instance, M9, where M is any metal in the first/second row.

I did the following to install the VTSTOOLs in a workstation:

First step: I used the command,

$ svn co http://theory.cm.utexas.edu/svn/vtsttools

,to download the code.
Then. I defined a path in the .bash_profile as follows:
export PATH=/home/user/Henkelman/vtsttools/vtstscripts:$PATH

Second step
I have created a folder, wherein I kept all the three files, POSCAR, OUTCAR, and DOSCAR. This is the working dir for the present case.

Third step
I issued the following command:

DOS-send]$ split_dos
(THIS HAS GENERATED THE FOLLOWING INFORMATION)

The OUTCAR exists, we use it to get the Fermi level,
the RWIGS tag and the number of spins.
Fermi level: -3.0573
Spin polarized calculation
LORBIT = 11
WARNING: not completely test for vasp.4.*
Use at your own risk. Please check results for consistency.
7 3006
0 7 3006
1 3008 6007
2 6009 9008
3 9010 12009
4 12011 15010
5 15012 18011
6 18013 21012
7 21014 24013
8 24015 27014
9 27016 30015
​
Fourth step

I used the command ls to see if the recommended files DOS0, DOS1 ... etc were generated?

[pradeep@london00 DOS-send]$ ls
DOS0 DOS1 DOS2 DOS3 DOS4 DOS5 DOS6 DOS7 DOS8 DOS9 DOSCAR OUTCAR POSCAR tmp.dat tmp.dat2

(Yes, there were such files generated as expected)

Fifth step

I then used the following commands separately to check whether the program called dosanalyze.pl works properly?

$ dosanalyze.pl w=2.6 a 2-5

(This did not give anything. The program stands by without printing anything on the screen)

$ dosanalyze.pl e=-9.5,1.2 a 9

(This one gave the following)...


integrate from -9.5eV to 1.2eV.
Total DOS is zero.

Clearly, it gives me a feeling that the program did not compile successfully.

Please note that I had an email communication with a member of the group, and according to him, the input files I sent him could successfully run in his/her computer. I used three different work stations with different linux-based architectures, but all ended up with the same error. I am thus wondering to know what is the major and serious mistake at my end.

I need your kind and prompt advice in resolving the aforesaid problem at your earliest.

Thank you very much for reading my post and replying ..
Best regards
PRV

[graeme]

I seem to remember some incompatibility with our dosanalze script and the LORBIT 11 tag. I'll look into this today.

Actually, to help me out, can you try rerunning your vasp calculation using LORBIT=10 rather than 11. This will print out the band-decomposed DOS but without the phase factors. See if the dosanalyze script works with that output.

At the same time, I'll see if I can update the script to sum the phase components so that it works with LORBIT 11.

[graeme]

Please update your vtstscripts with svn and see if my updated dosanalyze.pl script works for you. It is untested, so please use with caution and check your results. If you see any problems, I'll fix them.

[pradeep]

Thank you very much for the modifications to the program. It looks like the program is now running, at least it produces some numericals which are yet to be compared with the experimental/theoretical ones feasible in the literature (that is, to check the accuracy). I am posting here some of my test results for the Ag9 cluster, for instance.

Please note that I have followed the same procedure I mentioned in my post above, and got the following:

[pradeep@test]$ ../vtstscripts/dosanalyze.pl e=-10.555,2.26393294 a 1 (FOR ATOM 1)
Integrate from -10.555eV to 2.26393294eV.

Total States: 2301.40717960001
Average Energy (band center): -3.74219881603182
Standard Deviation: 1.24834620614851

[pradeep@test]$ ../vtstscripts/dosanalyze.pl e=-10.555,2.26393294 a 2 (FOR ATOM 2)
Integrate from -10.555eV to 2.26393294eV.

Total States: 2320.24772175
Average Energy (band center): -4.3038338526987
Standard Deviation: 1.32514446101451

[pradeep@test]$ ../vtstscripts/dosanalyze.pl e=-10.555,2.26393294 a 3
Integrate from -10.555eV to 2.26393294eV.

Total States: 2316.17725043
Average Energy (band center): -4.19806688495184
Standard Deviation: 1.30319453121946

[pradeep@test]$../vtstscripts/dosanalyze.pl e=-10.555,2.26393294 a 4
Integrate from -10.555eV to 2.26393294eV.

Total States: 2327.04565043
Average Energy (band center): -4.39146209424043
Standard Deviation: 1.508142222022

[pradeep@test]$ ../vtstscripts/dosanalyze.pl e=-10.555,2.26393294 a 5
Integrate from -10.555eV to 2.26393294eV.

Total States: 2324.20309432
Average Energy (band center): -4.34476151949066
Standard Deviation: 1.49390982799125

[pradeep@test]$ ../vtstscripts/dosanalyze.pl e=-10.555,2.26393294 a 6
Integrate from -10.555eV to 2.26393294eV.

Total States: 2339.38384843999
Average Energy (band center): -4.88307114527221
Standard Deviation: 1.52771097859715

[pradeep@test]$ ../vtstscripts/dosanalyze.pl e=-10.555,2.26393294 a 7
Integrate from -10.555eV to 2.26393294eV.

Total States: 2339.52426241
Average Energy (band center): -4.88542526533005
Standard Deviation: 1.52960207286349

[pradeep@test]$ ../vtstscripts/dosanalyze.pl e=-10.555,2.26393294 a 8
Integrate from -10.555eV to 2.26393294eV.

Total States: 2343.42582775
Average Energy (band center): -4.66556653774542
Standard Deviation: 1.45342340990055

[pradeep@test]$ ../vtstscripts/dosanalyze.pl e=-10.555,2.26393294 a 9
Integrate from -10.555eV to 2.26393294eV.

Total States: 2333.07178624999
Average Energy (band center): -4.41643980586526
Standard Deviation: 1.38055346208996


My next concern is: whether the numbers from 1 through 2 to 9 in the above example are atom1, atom2, atom3 .... and so on, Or, they represent to 1 atom, 2 atoms, 3 atoms, and so on?? (To my understanding they correspond to the former??)

Moreover, it has advised that: If no orbital flag is specified, the script analyzes the d-band. If no atom is selected, it analyzes all of them. If no w= tag specified, the center is calculated between 2.5 half widths at half height

To test this, I tried some and got the followings:

[pradeep@test]$ ../vtstscripts/dosanalyze.pl e=-10.555,2.26393294 a
Integrate from -10.555eV to 2.26393294eV.
Energy 0 in file DOS9 does not match first file.


If no orbital flag is specified, the script analyzes the d-band
[pradeep@ test]$ ../vtstscripts/dosanalyze.pl e=-10.555,2.26393294
Integrate from -10.555eV to 2.26393294eV.
Energy 0 in file DOS9 does not match first file.

[pradeep@ test]$ ../vtstscripts/dosanalyze.pl e=-10.555,2.26393294 1
Integrate from -10.555eV to 2.26393294eV.
When no band is selected, will analyze d band

Total States: 2129.2788202
Average Energy (band center): -3.9511130545712
Standard Deviation: 0.779050223672628


[pradeep@test]$ ../vtstscripts/dosanalyze.pl a 1
No energy range specified, integration will cover the whole range.

Total States: 2365.05374609001
Average Energy (band center): -3.56637053034908
Standard Deviation: 1.62469983852434

(THIS VALUE FOR THE d-band center is largely different from the ones shown above. Are they physically meaningful? I trust in this case the program covers the whole energy range as it displayed just above)



Finally, the following two options did not work correctly:

[pradeep@ test]$ ../vtstscripts/dosanalyze.pl w=2.6
Energy 0 in file DOS9 does not match first file.
[pradeep@test]$ ../vtstscripts/dosanalyze.pl w=2.6 a
Energy 0 in file DOS9 does not match first file.

[pradeep]

Some additional test results ...

[pradeep@test]$ ../vtstscripts/dosanalyze.pl e=-10.555,2.26393294 d 1-9
Integrate from -10.555eV to 2.26393294eV.
Energy 0 in file DOS9 does not match first file.

[pradeep@test]$ ../vtstscripts/dosanalyze.pl e=-10.555,2.26393294 d
Integrate from -10.555eV to 2.26393294eV.
Energy 0 in file DOS9 does not match first file.

(This probably means that if no atom is selected, the program does not analyze all of them??)

[graeme]

Again, thanks for the testing. I've updated the script and it seems to work for the integration of all atoms.

When you specify an atoms such as "a 2" it will analyze just that atom.

Finally, the most important issue for comparing/analyzing d-band levels is to keep a consistent energy range. The fact that it changes with the integration width is less important.