Creating higher level single point input files (for MOLPRO) from an xyz file and a MOLPRO input template can be rather tedious. The hitch with this one is sometimes the default (designated by -a) geometry optimization fails, and I need to run one or more subsequent geometry optimizations (designated -a2, -a3, etc). The original method I had used the below bash line:

 for i in $( ls geom-method-tests/c6h7-int*rowb97xd-631*.xyz ) ; do outname="$( basename $i "-a.xyz" )-uccsdtf12-vtzf12-ad.inp" ; echo $outname; cp c6h7-template-uccsdtf12-vtzf12-ad.inp $outname ; tail -n13 $i >tmpfile ; sed -i -e '/geom={/r tmpfile' $outname ; done

One downside with the above code is when the xyz file ends with something other than -a.xyz basename does not properly parse the file name (the new input file ends up with -a2.xyz in the middle of it). Thankfully, replacing basename with parameter expansion (## and % in this case) solves this.

for i in $( ls geom-method-tests/c6h7-int*rohcth407hcth407-631*.xyz ) ; do filename=${i##*/} ; basename1=${filename%.xyz} ; suffix=${basename1##*-a} ; basename2=${basename1%-*} ; outname="${basename2}-uccsdtf12-vtzf12-ad${suffix}.inp" ; echo ${outname} ; cp c6h7-template-uccsdtf12-vtzf12-ad.inp $outname ; tail -n13 $i >tmpfile ; sed -i -e '/geom={/r tmpfile' $outname; echo "---------------" ; done

While the above is rather long, I left in all of the steps for clarity. Thus far it has worked like a champ!

X11SDV-12C-TLN2F stability issues

This post fits with the other X11SDV posts in 2H of 2019. Both of my X11SDV-12C-TLN2F boards had problems where they would run fine for hours or days at full load, but then randomly I would find them shut down.


Things I tried that did not work:

 - separating their electrical connections onto independent battery backups (previously they both shared a single 1500 VA UPS). This was about the same time I was replacing batteries in the UPS units and thought power blips might then lead to overload situations.

- swapping RAM between the units

- swearing a lot

- removing the AOC-SHG3-4M2P cards having the scratch SSD drives


Things I tried that did work:

 - Replacing the Fortron 250 W PSU (FSP250-60FAG) with a Fortron 300 W PSU (FSP300-57FCB)


Running these boards with 128 GB of RAM, boot SSD, and the PCI-E card + 2 SSDs drew 150-180 W. Hence, I figured the 250 W PSU would be fine, but one of the rails must not have been sufficient to supply the required load.