Page 1 of 1

how can I write WAVECAR and charge density during the iterations?

Posted: Tue May 17, 2005 2:37 pm
by zj
Dear all,

By default, vasp only writes the wavefunctions and charge density when it finishes SCF calculation.

However, sometimes, for a very long job, one maybe want to save WAVECAR, CHG and CHGCAR files during its iteration. If the job stops accidentally, one can continue from the old charge density and can save much time.

Does vasp can do it when setting some keywords in INCAR file?

Thank you.

Jian

how can I write WAVECAR and charge density during the iterations?

Posted: Mon May 23, 2005 9:17 am
by admin
1) WAVECAR and CHGCAR are written at the end of the run, i.e.after NSW steps or if the geometry is converged within the given convergence criteria (EDIFFG) . You can re-start from the last WAVECAR (using the corresponding CONTCAR, and setting ISTART = 1)

2) stops can be enforced during runtime by providing the file
STOPCAR ,
which contains one line, either
LSTOP = .True. (stop at the next ionic step)
or
LABORT = .True. (stop at the next electronic step)
please mind that for the latter case WAVECAR and CHGCAR might contain
non-converged results.
[ Edited Mon Jun 27 2005, 03:28PM ]

how can I write WAVECAR and charge density during the iterations?

Posted: Wed Jun 15, 2005 8:09 am
by zj
yes, when creating a STOPCAR file, it write the WAVECAR, however, the job will also be terminated then.

Does vasp can just write out the WAVECAR and CHGCAR without being terminated?

how can I write WAVECAR and charge density during the iterations?

Posted: Mon Jun 27, 2005 1:57 pm
by admin
No, if you want to be sure that a calculation finishes properly, writing WAVECAR and CHGCAR, reduce the number of ionic steps for the run (NSW).

how can I write WAVECAR and charge density during the iterations?

Posted: Tue Oct 23, 2007 8:27 am
by skn
Is it possible to continue SCF calculation by keeping WAVECAR and CHGCAR files for a lower value of NELM in electronic steps and then restarting?

how can I write WAVECAR and charge density during the iterations?

Posted: Wed Oct 24, 2007 9:53 am
by admin
no, not by default. both are written after NELM steps (or after electronic convergence is reached), there is no input parameter to change this behavior