The ABINIT Users Mailing List ( CLOSED )

["David M. Wood" <dmwood@mines.edu>]

Howdy, all!

I've been using 4.6.5 (sequential, built with g95) under Mac OS X  
(10.4.6) for some semiconductor phonon calculations.  Often things  
work perfectly, but equally often I am getting a file open error:  
Here's the tail end of a log file when the problem has occurred:

PROBLEM:

At SCF step    9       vres2   =  1.97E-10 < tolvrs=  1.00E-08  
=>converged.
-open ddk wf file :lro_DS2_1WF7
-open ddk wf file :lro_DS2_1WF8
-open ddk wf file :lro_DS2_1WF9
======================================================================== 
========

----iterations are completed or convergence reached----

outwf  : write wavefunction to file lro_DS3_1WF1
Fortran runtime error: Unattached unit

[In this case the file lro_DS3_1WF1 WAS written, at least partially,  
and files WF7, WF8, WF9 already existed]
This message comes from outwf.F90, line 257


When the problem does NOT occur, things proceed normally: at the same  
point in the job I see in the log

NO PROBLEM:

At SCF step    9       vres2   =  3.01E-10 < tolvrs=  1.00E-08  
=>converged.
-open ddk wf file :lro_DS2_1WF7
-open ddk wf file :lro_DS2_1WF8
-open ddk wf file :lro_DS2_1WF9
======================================================================== 
========

----iterations are completed or convergence reached----

outwf  : write wavefunction to file lro_DS3_1WF1

Thirteen components of 2nd-order total energy (hartree) are
1,2,3: 0th-order hamiltonian combined with 1st-order wavefunctions...


Strangely, identical runs with a different cutoff may proceed to  
successfully compute and write ALL WF files, in this case, in order:

WF7-WF8-WF9-WF1-WF4-WF10

and finish the run with no difficulties.

It is as if there is some strange race condition which sometimes  
causes the problem and sometimes not.
Has anyone else experienced such a problem? (I saw nothing about this  
in the forum archives.) Is it unique to g95 or more general?

Thanks for any insight you may have on this issue.
DMW

David M. Wood, Dept. of Physics, Colorado School of Mines, Golden, CO  
80401
Phone: (303) 273-3853; Fax: (303) 273-3919