Question about split_dos

[flyhigh]

Hello,

Recently, I downloaded the scripts "vtstscripts.tgz".
I calculated d-band center of Pt surface with this, and I got the results below.
Only Emax is different in the case (1) and the case (2).
I do not understand the meaning of the last message,"Can't take sqrt of -415.985 at /bin/vtstscripts/dosanalyze.pl line 292."
Why did it make problems to increase Emax?
Logically, it does not seem to have problems.
In other words, how can I decide the range of Emin and Emax?
Average Energy (band center) is affected by the range of Emin and Emax too much.

Any advice would be welcome.
Thanks.


case (1)
$ dosanalyze.pl e=-7,0 d

Integrate from -7eV to 0eV.
WARNING: For DOS0, ALL bands are analyzed
WARNING: band center has shifted by more than 10% compared to full integrated range

Total States: 69.43606769
Average Energy (band center): -2.25169355987311
Standard Deviation: 2.48325612734491

case (2)
$ dosanalyze.pl e=-7,1 d

Integrate from -7eV to 1eV.
WARNING: For DOS0, ALL bands are analyzed
Can't take sqrt of -415.985 at /bin/vtstscripts/dosanalyze.pl line 292.

[graeme]

If you can post the directory with the data, I'll try to reproduce the problem and find a solution.

[flyhigh]

[quote="graeme"]If you can post the directory with the data, I'll try to reproduce the problem and find a solution.[/quote]

I checked the script "dosanalyze.pl ", and I found the error message is related to the variance.


290 $eneavg = $dstsum/$dossum;
291 $enevar = ($dssqsum*$dossum - $dstsum*$dstsum)/($dossum*$dossum);
292 $enestd = sqrt($enevar);

In case I used ISPIN=2 in INCAR file, can this problem be caused by this?

[graeme]

If I can see the files that you are analyzing, I should be able to debug this.