band plot for GW0

Message

giacomo giorgi · #1 Post by **giacomo giorgi** » Thu Nov 19, 2009 9:49 am

Dear All,
does anybody know if there is a tool for plotting bands from GW0 calculations?

Best,
G

gbal · #2 Post by **gbal** » Fri Nov 20, 2009 2:12 pm

Hi, Do you get the band Structure for the GW0 calculations? or you are having problems to calculate it?

panda · #3 Post by **panda** » Sat Nov 21, 2009 11:32 pm

I am under the impression that, like everything else, you write a parser by hand and plot it using a plotting library. I have written a parser and plotting tool for the EIGENVAL and PROCAR file, but not for GW0 calculations. I have done these calculations, and am going to write a parser for the OUTCAR to get the optical constants, but pretty much all the tools are DIY, unfortunately. I would think that the format of the parser for the PROCAR or EIGENVAL file would be the same in both GW0 or regular calculations, or am I missing something? IE you could plot the band structure from your GW0 output in the same manner that you plotted your non-GW0 band structures; please correct me if I am wrong.

gbal · #4 Post by **gbal** » Wed Nov 25, 2009 4:26 pm

Hi, For plotting i also use "hand-made plotting tools" starting from the EIGENVAL file. But to calculate it, I use the entry file KPOINTS setting the line mode and the direction that i want, than i have a good band structure. But for GW0, the line mode in the KPOINTS always crashes the calculations. And I'm here trying to find this answer about GW bands

! If Anyone have this answer, please tell me.

panda · #5 Post by **panda** » Wed Nov 25, 2009 7:05 pm

so for the GWO calculations, which I have done but do not claim to understand fully, there is an EIGENVAL and PROCAR file if you so choose to print one. The kpoints that I have always taken are parsed directly from that file, so as long as the band structure from the EIGENVAL or PROCAR file is OK to take as the band structure for the GWO calculations then I can show you how I parse the EIGNEVAL file. You just take each kpoint from the header of the file and store it in a hash table. Then for each kpoint parse the associated energy and store that in a second hash table, then plot. I do not use a second entry file for KPOINTS, I think this is your problem. All of the information is self contained in the EIGENVAL or PROCAR file, assuming these files we can use for GW0 calculations. I will have a look this weekend at the differences between the PROCAR and EIGENVAL file for a GWO and a non GW0 calculation and report any differences next week, this has been bugging me too for some time.

gbal · #6 Post by **gbal** » Sat Nov 28, 2009 2:19 pm

Hej, I got your idea and i can also do that, but then i got a plot with just seven points along a line. This is not really good even for a picture, imagine for extract some data. And is not possible to use a bigger K-mesh(this already takes too much time). PROCAR have the same K-mesh as in the EIGENVAL, so, no difference at all for me. If You have any hints, it will very good!

panda · #7 Post by **panda** » Thu Dec 03, 2009 7:46 pm

Can you do each high symmetry zone as a different run to increase the number of k-points, and then "glue" the plots together so to speak?

gbal · #8 Post by **gbal** » Fri Dec 04, 2009 1:58 pm

It is not a bad idea, but it will not increase the points in enough number to make a calculation using the band. Using different calculations i can almost double the number of points, but then i also have some variation of the results. In the end, nothing like the real calculation over a line. Having more ideas, please post it!

panda · #9 Post by **panda** » Mon Dec 07, 2009 5:45 pm

What system are you studying? Have you tried optimizing the geometry first to a minimum and then running a single point calculation for the band structure part? Can you post your INCAR file and or some indication of the variation in the results so we can help you? Thanks