Indeed Nick, my fourth block was defined as an integer.

 Fixed, and working fine now.


 Thanks!

-Salvador


________________________________
From: [email protected] <[email protected]> on behalf of Nick 
Papior <[email protected]>
Sent: Thursday, September 17, 2015 2:39 PM
To: [email protected]
Subject: Re: [SIESTA-L] Error (parse.f?) on siesta-trunk-462

parse.F90 is a routine used to "parse" fdf input, hence you have an error in 
your fdf input file.

I would suspect your kgrid_monkhorst_pack grid to be ill-defined.

I guess you have
%block kgrid_monkhorst_pack
 <int> <int> <int> 0
 <int> <int> <int> 0
 <int> <int> <int> 0
%endblock
which in reality should be
%block kgrid_monkhorst_pack
 <int> <int> <int> 0.
 <int> <int> <int> 0.
 <int> <int> <int> 0.
%endblock

*NOTE* the additional dot after 0.

2015-09-17 19:02 GMT+00:00 Salvador Barraza-Lopez 
<[email protected]<mailto:[email protected]>>:

 Dear all,

 I want to use the exchange-correlation potential from Berland and Per 
Hyldgaard (PRB 89, 035412 (2014)) to work with layered materials. This has been 
implemented on the trunk-462 version of the code.


 There is a reproducible error in my calculations related to the parsing module 
related to generating filename strings. The error is:


*************************************************************
ERROR

PARSE module: reals: Not enough reals in line

     File: parse.F90
     Line:   596
*************************************************************
Stopping Program
*************************************************************


The error has been reproduced on three different architectures, and with the 
following mpi/compiler/linear algebra libraries:

intel 14.0.3

mkl 14.0.3

mvapich2 2.1


and

gcc 4.5.0

gotoblas2

openmpi 1.4.2


 I've compiled all in debug mode and the error persist. I am wondering if there 
a fix available.


 Thank you!

-Salvador



--
Kind regards Nick

Responder a