Unable to perform dynamical matrix run for transition state

[nz24]

We are using vasp.5.3.3 with the vTST 3.1 code and tools. When running the dynamical matrix routine with minima found in the standard way with VASP, all works well. However, when we invoke ICHAIN=1 for a transition state gotten via NEB using force based optimizers, our run encounters a NaN for rms(c) after the prescribed number of non-self consistent steps (NELMDL=-10) and then seems to be stuck. If we include all the optimization parameters in the INCAR for a dynamical matrix run e.g. EDIFFG=-0.1, EDIFF=1e-05, IBRION=3 in addition to ICHAIN=1 and POTIM=0.0 and NSW=244, then the run returns the optimized TS in a single step but does not shift the x, y, z coordinates based on DISPLACECAR to get the dynamical matrix information. We have also tried just commenting out IBRION=3 so as not to confuse VASP on our wanting to do a dynamical matrix run rather than an optimization but that gets us back to NaN on for rms(c). Below we include the INCAR file we are using. Does someone have a sense of how our INCAR file needs to change to do the dynamical matrix routine on a TS calculated with force based optimizers? When compiling VASP 5.3.3 with vTST 3.1. tools we copied the *.F files into the vasp source directory and made the changes noted at http://theory.cm.utexas.edu/vtsttools/installation.html# and compiled. We did also just notice that the vtstcode directory also has files: bdr_changes, vasp-5.3.2-main.patch and vasp-5.3.2-mpmd.patch. We didn’t make those changes as we were using vasp-5.3.3 and the bdr_changes were not mentioned in the web site installation and would probably only affect bader volume calculations. If we should be making additional changes before compiling, please let us know. Any help would be most appreciated.

System=YBaZrO3
NPAR=4
#electronic
NELECT=640
IALGO=48
ISMEAR = 0
NELMDL=-10
LREAL= Auto
LWAVE=.FALSE.
LCHARG=.FALSE.
PREC=accurate
#nuclei
#criteria used during optimization (http://theory.cm.utexas.edu/vtsttools/dymmat.html suggests that these should still be included)
EDIFFG = -0.1
EDIFF = 1e-05
IBRION=3 !to disable VASP optimizers along with the POTIM=0.0 line below
#Dynamical Matrix
ICHAIN=1
POTIM = 0.0 ! with a time step of zero
NSW = 244 ! 3*numberAtoms/numberOfImages+1

[graeme]

Try setting EDIFFG=-1e-10 (or 0) and make sure that you have a DISPLACECAR with your diplacements listed simply as:

dx1 dy1 dz1
dx2 dy2 dz2
...