In this second exercise we will find a transition path between a icosahedral-like minimum and a fcc-like minimum.
Download the 1.1 exercise into your $HOME folder and unzip it.
you@eulerX ~$ wget http://www.cp2k.org/_media/exercises:2017_ethz_mmm:exercise_1.2.zip you@eulerX ~$ unzip exercises:2017_ethz_mmm:exercise_1.2.zip
The goal is to check the dependence of the result on the initial guess. The relevant section is the following (neb_linear.inp):
&MOTION &BAND NPROC_REP 1 BAND_TYPE IT-NEB NUMBER_OF_REPLICA 20 &OPTIMIZE_BAND OPT_TYPE DIIS &DIIS MAX_STEPS 1000 N_DIIS 3 &END &END &REPLICA COORD_FILE_NAME a.xyz &END REPLICA &REPLICA COORD_FILE_NAME b.xyz &END REPLICA &END BAND &END MOTION
where the initial and final states are specified. There is a second version called (neb_better.inp). Check the differences… note that we input more configurations. This means that we start from a more advanced start with respect to a linear interpolation between a and b.
cp2k.popt -i neb_linear.inp -o neb_linear.out
. You will get the number NFRAMES (mark it).
rm path_linear.xyz ; for a in NEB_LINEAR-pos-Replica_nr_*xyz ; do m_xyzframes NFRAMES NFRAMES $a >> path_l.xyz ; done
(put the value of NFRAMES).
./myq4 path_linear.xyz > linear.q4.ene