Page 1 of 1

initialization in yambo 4.0.2

Posted: Thu Mar 17, 2016 2:04 pm
by caowd
Dear all

There are lots of Kpoints in my recent calculations and I find that if I use the yambo in 4.0.2 version, there could be an error like

"NetCDF: One or more variable sizes violate format constraints"

After I checked the files in the SAVE/ directory, I have found that the file ndb.kindx is too large.

However, if I switch to yambo 4.0.1 version, the error disappears. And for the ndb.kindx in the SAVE/ directory, there are two fragments "ndb.kindx_fragment_#".

I think there may be two possible reasons.
1. the netcdf which yambo in the newer version uses is different from that used previously
2. the yambo in the newer version needs more options like "yambo -xxx" to obtain the fragments of ndb.kindx

Re: initialization in yambo 4.0.2

Posted: Thu Mar 17, 2016 2:25 pm
by Daniele Varsano
Dear Wen-Dong Cao,
this is because netcdf admits a fixed number of variables. May be the number of variables admitted can be changed when compiling netcdf libraries.
Anyway, as you said, you can overcome the problem by fragmenting the databases. This is now done by adding the io verbosity when building up the input file:

Code: Select all

yambo -i -V io
and setting:

Code: Select all

DBsFRAGpm= "+QINDX"            # [IO] Space-separated list of +DB to be FRAG and -DB NOT to be FRAG. DB=(DIP,X,W,HF,COLLS,K,BS,QINDX,
Please note the "+" sign.

Best,
Daniele

Re: initialization in yambo 4.0.2

Posted: Fri Mar 18, 2016 7:24 am
by caowd
Dear Daniele

I have changed the input file as you suggested and it indeed works now. :D

Thank you very much.


Best Wishes.