We’ve been doing some vibrationally resolved spectral analysis recently. What I have found, though, is that sometimes my spectral output changes dramatically (magnitude and shape), and I had a hard time figuring out why.
Recently, we put this to a test, and it looks like we’ve tracked it down to the length of the file name, or more likely the entire file path plus name.
What we did was take the same structure and same input files and just changed the file names. I was using long file names to capture the important features I was changing while exploring what level of theory we want to use.
What we saw was that with short file names/paths, we got reasonably looking spectra. At very long file names and paths, the spectra would warp, becoming smaller in magnitude and “stretched out” over a longer wavelength range. No errors anywhere that I am aware of, though we do not check the fchk files. If I really push the file name, I eventually get an error in the vibronic output file. At least this error doesn’t create a false spectrum.
Vibrational Resolved Spectra Simulation
spectra saved in xxx_spectra.dat
log of calculation in vibronic_spectra.log
Error: Unknown Spectrum Flag.
We don’t know for sure that the file name/path lengths is truly the root cause. It could be that there is something wrong with my supercomputer environment, but I doubt it.