next up previous contents
Next: Utility scripts Up: Job control (c-shell scripts) Previous: Job control for initialization

   
Job control for iteration (run_lapw or runsp_lapw)

In order to perform a complete SCF calculation, two types of scripts are provided on the distribution. For non-spinpolarized calculations use: run_lapw, for spin-polarized calculations use: runsp_lapw. Cases with/without inversion symmetry and with/without semicore or core states are handled automatically by these scripts. All activities of these scripts are logged in short in :log (appended) and in detail together with convergence information in case.dayfile (overwriting the old ``dayfile``). You can always get help on its usage by invoking these scripts with the -h flag.

run_lapw -h

PROGRAM:        /zeus/lapw/WIEN97/bin_7/run_lapw

PURPOSE:        running the nonmagnetic scf-cycle in WIEN
                to be called within the case-subdirectory
                has to be located in WIEN-executable directory

USAGE:          run_lapw [OPTIONS] [FLAGS]

OPTIONS:
-cc LIMIT ->    charge convercence LIMIT (0.0000)
-ec LIMIT ->    energy convercence LIMIT (0.0001 Ry)
-fc LIMIT ->    force  convercence LIMIT (0 mRy/a.u.)
-e PROGRAM ->   exit after PROGRAM ()
-i NUMBER ->    max. NUMBER (20) of iterations
-s PROGRAM ->   start with PROGRAM ()
-r NUMBER ->    restart after NUMBER (20) iterations (rm *.broyd*)
-nohns NUMBER ->do not use HNS for NUMBER iterations 

FLAGS:
-h/-H ->        help
-I    ->        with initialization of in2-files to "TOT" 
-p    ->        run k-points in parallel (needs .machine file [speed:name])
-so   ->        run SCF including spin-orbit coupling
-it   ->        use iterative diagonalization after first cycle
-it0  ->        use iterative diagonalization (also in first cycle)

CONTROL FILES:
.stop           stop after SCF cycle
.fulldiag       force full diagonalization

ENVIRONMENT VARIBLES:
SCRATCH         directory  where vectors and help files should go

Calling run_lapw (after init_lapw) from the subdirectory case will perform up to 20 iterations (or whatever you specify with switch -i) unless convergence has been reached earlier. You can choose from three convergence criteria, -ec (the total energy convergence is the default and is set to 0.0001 Ry), -fc (magnitude of force convergence) or -cc (charge convergence), but only one criterium can be specified. Be careful with these criteria, different systems will require quite different limits (e.g. fcc Li can be converged to $\mu$Ry, $\mbox{YBa}_2\mbox{Cu}_3\mbox{O}_7$ only to 0.1 mRy). With -e PROGRAM you can run only part of one scf cycle (e.g. run lapw0, lapw1 and lapw2), with -s PROGRAM you can start at an arbitrary point in the scf cycle (e.g. after a previous cycle has crashed and you want to continue after fixing the problem) and continue to self-consistency. Before mixer is invoked, case.clmsum is copied to case.clmsum_old, and the final ``important`` files of the scf calculation are case.clmsum and case.scf.

Invoking

run_lapw -I -i 30 -fc 0.5

will first set in case.in2 the TOT-switch (if FOR was set) to save cpu time, then run up to 30 scf cycles till the force criterion of 0.5 mRy/a.u. is met (for 3 consecutive iterations). Then the calculation of all terms of the forces is activated (setting FOR   in case.in2) for a final iteration.

Parallelisation is described in Sec. 5.4.

iterative diagonalization, which can save computer time in cases with ``few electrons'' and ``large matrices'', is described in Sec. 7.2. It needs the case.vector file from the previous scf-iteration and this file is copied to case.vector_old when the -it switch is set.

You can save computer time by performing the first scf-cycles without calculating the non-spherical matrix elements in lapw1. This option can be set for N iterations with the -nohns N switch.

For magnetic systems which are difficult to converge you can use the script runfsm_lapw (see section 4.5.3) for the execution of fixed-spin moment (FSM) calculations.

end:run_lapw


next up previous contents
Next: Utility scripts Up: Job control (c-shell scripts) Previous: Job control for initialization

2000-04-11