zadorlab / sella

A Python software package for saddle point optimization and minimization of atomic systems.
https://www.ecc-project.org/
Other
59 stars 19 forks source link

IRC failing to converge with IRCInnerLoopConvergenceFailure #20

Open Youbin-K opened 1 year ago

Youbin-K commented 1 year ago

Hi, I'm trying to run the IRC in metal slab systems, where it gives me the error of

sella.optimize.irc.IRCInnerLoopConvergenceFailure

Below is the out file:


Running relaxation using VaspInteractive


 Step     Time          Energy         fmax

IRC: 0 18:25:51 -166.366325 0.0239 Trying to close the VASP stream but encountered error: Traceback (most recent call last): File "/gpfs/scratch/ykim219/IRC_test/CCH3/relax_with_VASPInteractive.py", line 51, in run_vasp_interactive opt.run(fmax=0.025, steps=1000, direction='forward') File "/users/ykim219/.local/lib/python3.9/site-packages/sella/optimize/irc.py", line 101, in run for converged in self.irun(fmax, fmax_inner, steps, direction): sella.optimize.irc.IRCInnerLoopConvergenceFailure

During handling of the above exception, another exception occurred:

Traceback (most recent call last): File "/users/ykim219/.local/lib/python3.9/site-packages/vasp_interactive/vasp_interactive.py", line 1015, in _force_kill_process self.close() File "/users/ykim219/.local/lib/python3.9/site-packages/vasp_interactive/vasp_interactive.py", line 598, in close self._run(self.atoms, out=out, require_cell_stdin=False) vasp_interactive.utils.TimeoutException: Timed out closing VASP process.

Will now force closing the VASP process. The OUTCAR and vasprun.xml outputs may be incomplete Traceback (most recent call last): File "/gpfs/scratch/ykim219/IRC_test/CCH3/relax_with_VASPInteractive.py", line 57, in run_vasp_interactive() File "/gpfs/scratch/ykim219/IRC_test/CCH3/relax_with_VASPInteractive.py", line 51, in run_vasp_interactive opt.run(fmax=0.025, steps=1000, direction='forward') File "/users/ykim219/.local/lib/python3.9/site-packages/sella/optimize/irc.py", line 101, in run for converged in self.irun(fmax, fmax_inner, steps, direction): File "/gpfs/data/cfgoldsm/ykim219/Research/Source_files/ase/ase_3.22.1/ase/optimize/optimize.py", line 135, in irun self.step() File "/users/ykim219/.local/lib/python3.9/site-packages/sella/optimize/irc.py", line 138, in step raise IRCInnerLoopConvergenceFailure sella.optimize.irc.IRCInnerLoopConvergenceFailure

Would this possibly be due to the compatibility of the VASPinteractive and Sella? Or is this because of the algorithm of the IRC?

ehermes commented 1 year ago

This is probably caused by poor gradient precision in the inner loop. Can you try tightening your SCF convergence parameters? It's been a while since I've used VASP so I can't remember which parameters control its SCF convergence threshold...

Youbin-K commented 1 year ago

Hi, I tried tightening the SCF convergence parameter from 1e-5 to 1e-6 and 1e-7, which both also fails with the same errors. I've also tried to run with increased steps by dx=0.1 to dx=0.3, which also gives the same errors. Would it be worth trying to change the 'eta' value in the IRC and the 'steps' in the forward and reverse directions?