I also ended up with empty vib_normal_modes and vibspectrum files even when using TM 6.3.1
I could reproduce that these where cases where I did one numforce run (which finished fine, files not empty) then deleted the numforce sub-directory and did an other numforce run.
This run also ended OK but the vib_normal_modes and vibspectrum files in the numforce sub-directory are empty and as they are copied back to the main directory also there.
Reasons for doing two numforce runs (aoforce and numforce afterwards also doesn't work) would be for example if one did an NumForce run where one has forgotten to specify -central and now wants to run NumForce -central afterwards (
I anyway wonder why Numforce -central with weight derivatives is not the default... as it can have huge effect on the results and is strongly recommended in the manual for good reasons) or to do to an other jobex run in between these two NumForce runs (which should use the full hessian...)
I think the reason for this problem is that NumForce script performes
$showdg "vibrational normal modes" > vib_normal_modes
[...]
$showdg "vibrational spectrum" > vibspectrum
and if the vibrational normal modes and spectrum results are already outsoucced in these files (due to the first NumForce/Aoforce run these locations are set) they can't be read by this comands as these files become overwritten before the data can be read
If I delete the
$vibrational spectrum file=vibspectrum
$vibrational normal modes file=vib_normal_modesout of the controle file before I start the second NumForce run everything works fine also for the second run.
I don't know if there are any side effects for other cases if these lines where deleted at the beginning of NumForce (they are added again later anyway). An other solution would be piping the result of sdg to some tmp file and moving it to the desired filename afterwards
$showdg "vibrational normal modes" > vib_normal_modes.tmp
mv vib_normal_modes.tmp vib_normal_modes
[...]
$showdg "vibrational spectrum" > vibspectrum
mv vibspectrum.tmp vibspectrum
In any case, this could be easily be fixed for the next version of TM (I don't have version 6.4 at my hand so I couldn't test if was already fixed) as this problem was also reported by cediev87 twice in last autum.