[Pw_forum] hydrogen adsorption on graphene

Paolo Giannozzi giannozz at democritos.it
Thu Jul 31 17:55:26 CEST 2008


s305igun at mail.chem.itb.ac.id wrote:

> I need your help, I was calculated hydrogen adsorption on 
> graphene layer 

what a waste of CPU time: H2 physisorps on graphene, nothing
interesting will happen

and this is my trouble output: [...]

>      from cdiaghg : error #       296
>      info =/= 0

> Would you like to explain to me about info =/= 0?

this not-so-obvious error message was written "just in case",
because it should never occur, except for pathological systems.
Most occurrences of such error can be tracked to bad input data.
I am too lazy to look at your input to check whether it is good
or bad, but the graphene and graphene+H2 input, when read by
"xcrysden --pwi ....", give a bunch of errors like:
    WARNING: atom "3" has huge number of bonds !!!
and xcrysden fails, so it doesn't bode well.

This being said: sometimes the subspace diagonalization can
fail even for perfectly legitimate data. This might be a
consequence of a bug, but it is more likely a numerical
instability. It seems to happen more frequently with the
parallel algorithm, but soemtimes also in the serial
algorithm. A few cases are currently under investigation,
but it is exceedingly difficult and time-consuming to
understand what is going on when the errors is produced
erraticaly inside a lapack routine and only on large systems.

Paolo
-- 
Paolo Giannozzi, Democritos and University of Udine, Italy


More information about the Pw_forum mailing list