Dear All,
Please, can you help me to know how to restart my calculation from a previous run , in case of performing geometry optimization using PAW ?
Thank you in advance,
Ahmed,
How to restart the code from a previous run ?
Moderator: bguster
Re: How to restart the code from a previous run ?
Take the lattice parameters and the atomic positions obtained by Abinit at the end of the
previous structural relaxation run and change the input file accordingly.
Restart the run with the new input file and use irdwfk or getwfk to initialize the wavefunctions from the WFK file.
see http://www.abinit.org/documentation/hel ... tml#irdwfk
previous structural relaxation run and change the input file accordingly.
Restart the run with the new input file and use irdwfk or getwfk to initialize the wavefunctions from the WFK file.
see http://www.abinit.org/documentation/hel ... tml#irdwfk
Re: How to restart the code from a previous run ?
Dear Abinit's users
I need to restart my calculation from a previous run in case of performing geometry optimization using PAW.
In my previous calculation, I used ionmov=3 and optcell=1.
After a number of iterations, I do not still obtain convergence, but the calculation stopped abruptly. The iteration 19 is ended but the energy does not still converge and the iteration 20 began, but is not finished.
1- If I have to restart the calculation taking the lattice parameters and the atomic positions obtained by Abinit at the end of the previous structural relaxation run and change the input file accordingly. I must use the output acell and xred obtained in iteration 19 or the input acell and xred given in the beginning of iteration 20?
2-Then I to use DEN files because I have no WFK files? (prtwfk = 0).
Thank you in advance.
Nassim
I need to restart my calculation from a previous run in case of performing geometry optimization using PAW.
In my previous calculation, I used ionmov=3 and optcell=1.
After a number of iterations, I do not still obtain convergence, but the calculation stopped abruptly. The iteration 19 is ended but the energy does not still converge and the iteration 20 began, but is not finished.
1- If I have to restart the calculation taking the lattice parameters and the atomic positions obtained by Abinit at the end of the previous structural relaxation run and change the input file accordingly. I must use the output acell and xred obtained in iteration 19 or the input acell and xred given in the beginning of iteration 20?
2-Then I to use DEN files because I have no WFK files? (prtwfk = 0).
Thank you in advance.
Nassim
Re: How to restart the code from a previous run ?
Hi,
Yes you have to take by hand the lattice parameters, lattice vectors and atomic position from the last ionic step and put them into the input file.
You can restart from a _DEN _TIMX_DEN file if you want to speed up a little bit the first step of the new run but it is not mandatory.
Note that you can do the restart automatically if you use restartxf -2 in the case abinit is compile with netcdf ( presence of the _HIST file). If so, abinit reads the _HIST file to restart the optimization.
Cheers
Jordan
Yes you have to take by hand the lattice parameters, lattice vectors and atomic position from the last ionic step and put them into the input file.
You can restart from a _DEN _TIMX_DEN file if you want to speed up a little bit the first step of the new run but it is not mandatory.
Note that you can do the restart automatically if you use restartxf -2 in the case abinit is compile with netcdf ( presence of the _HIST file). If so, abinit reads the _HIST file to restart the optimization.
Cheers
Jordan