The ABINIT Users Mailing List ( CLOSED )

["H.L.Shih" <hlshih@gate.sinica.edu.tw>]

Dear Abinit Users:

I've found when running as parallel, using odd number of processors will 
result in convergence problem (at least for my interesting system, say 
spinel Fe3O4 and BN:V_B 32-atom supercell), however, testing with 
simple iron and nickle bulk wont have the same error. 

Table I. Fe convergence test (ecut=30 Ha)
============================
nprocs  scf     Etot
(#)     lopp    (Ha)
----------------------------
2       25      -25.3339
3       25      -25.3339
4       25      -25.3339
5       25      -25.3339
6       25      -25.3339
7       25      -25.3339
============================

Table II. spinel Fe3O4 (no. of procs vs. scf loop and etot w.r.t. each 
cases )
==============================
nprocs  scf     Etot
(#)     loop    (Ha)
------------------------------
2       16      -192.71086
3       100*    -190.04405
4       16      -192.71086
==============================
* unconverged after 100 iterations.
                                                                              
      
for the unconverged case, i found the following error in *LOG* file:
---------------------------------------------------------------
-P-0001  cgwf : WARNING -
-P-0001   New trial energy at line       4 =   4.970215E-01
-P-0001      is higher than former:  4.970215E-01
-P-0001
---------------------------------------------------------------


Any suggestion.

BR,
J
-- 
---------------------------------
Jason Shih
Academia Sinica Computer Center
No.128, Sec. 2, Academia Rd., 
Nangang District, Taipei City 115,Taiwan (R.O.C.)
Tel: +886-2-27899960
Fax: +886-2-27899949
---------------------------------