"Error: Invalid gaussian directory" when use_gausoptonly is flagged as true

[jpetersen48]

Hello,

I recently attempted to use POLTYPE2 on the zwitterion of proline. When I submitted the job with the suggested flags for zwitterions:

optpcm=True toroptpcm=True torsppcm=True use_gausoptonly=True

I was given the error:

ERROR: Invalid Gaussian directory:

After testing the individual flags, I found that this error was only returned when use_gausoptonly=True was active. If it is relevant, the error that ends the job when that flag is inactive appears to be this, within the "-opt.log" file

PsiException: Could not converge geometry optimization in 5 iterations.

[misterbrandonwalker]

Hello,

Would it be possible to see a link of files (in some box link etc), for troubleshooting?

Thanks

On Mon, Sep 14, 2020 at 10:39 AM Jacob Petersen notifications@github.com wrote:

Hello,

I recently attempted to use POLTYPE2 on the zwitterion of proline. When I submitted the job with the suggested flags for zwitterions:

optpcm=True toroptpcm=True torsppcm=True use_gausoptonly=True

I was given the error:

ERROR: Invalid Gaussian directory:

After testing the individual flags, I found that this error was only returned when use_gausoptonly=True was active. If it is relevant, the error that ends the job when that flag is inactive appears to be this, within the "-opt.log" file

PsiException: Could not converge geometry optimization in 5 iterations.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/pren/poltype/issues/50, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKNB26NUTJYEBHI5OKZYIELSFY2M3ANCNFSM4RLYEEYQ .

[misterbrandonwalker]

Also, could you check to see if your gaussian executable works (test on anything)? Then also open any .com file and locate the referenced folder location, then check to see if this folder exists in your file system? If it does not, there should be a place in poltype.log file where it says, mkdir /folder.. etc .

On Mon, Sep 14, 2020 at 10:59 AM Brandon Walker notifications@github.com wrote:

Hello,

Would it be possible to see a link of files (in some box link etc), for troubleshooting?

Thanks

On Mon, Sep 14, 2020 at 10:39 AM Jacob Petersen notifications@github.com wrote:

Hello,

I recently attempted to use POLTYPE2 on the zwitterion of proline. When I submitted the job with the suggested flags for zwitterions:

optpcm=True toroptpcm=True torsppcm=True use_gausoptonly=True

I was given the error:

ERROR: Invalid Gaussian directory:

After testing the individual flags, I found that this error was only returned when use_gausoptonly=True was active. If it is relevant, the error that ends the job when that flag is inactive appears to be this, within the "-opt.log" file

PsiException: Could not converge geometry optimization in 5 iterations.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/pren/poltype/issues/50, or unsubscribe < https://github.com/notifications/unsubscribe-auth/AKNB26NUTJYEBHI5OKZYIELSFY2M3ANCNFSM4RLYEEYQ

.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/pren/poltype/issues/50#issuecomment-692152340, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKNB26IALKVDTAHEO5OIKF3SFY4WPANCNFSM4RLYEEYQ .

[jpetersen48]

Hello,

Here's a link to the entire directory that I was using poltype within.

https://notredame.box.com/s/avwnsdn6rbxyqrku0yccig66pi6fldf7

I have also tested gaussian again, I am able to perform calculations with gaussian. Please let me know if there is anything else I can do to assist.

[misterbrandonwalker]

Hey,

I have attached a screen shot of what I see when I attempt to access the link. I have logged in via my institution instead of University of Notre Dame.

Also, in the poltype job folder, there should be a *opt.com file. In that file there should be a folder location on the top of the file. I just want to make sure that this folder location exists on your cluster.

• Brandon

On Thu, Sep 17, 2020 at 11:14 AM Jacob Petersen notifications@github.com wrote:

Hello,

Here's a link to the entire directory that I was using poltype within.

https://notredame.box.com/s/avwnsdn6rbxyqrku0yccig66pi6fldf7

I have also tested gaussian again, I am able to perform calculations with gaussian. Please let me know if there is anything else I can do to assist.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/pren/poltype/issues/50#issuecomment-694339557, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKNB26M6CXO2R5GORP2DCITSGIYVRANCNFSM4RLYEEYQ .

[misterbrandonwalker]

It would seem I cannot attach a screenshot from email response. Basically it says the shared link or file is unavailable to you.

[misterbrandonwalker]

Also, please attach your source for your envioronment

[jpetersen48]

Oh, sorry, I forgot to change permissions. Here's a new link with my .bashrc profile included. Let me know if you need anything else, or if this doesn't work again.

https://notredame.box.com/s/qime8z0e28ks1n38d68d6314o5tmfqgm

[misterbrandonwalker]

Okay so a couple of suggestions. 1) Just keep the poltype code directory seperate from things like GMDA or your poltype jobs. So like you can keep a seperate folder somewhere else "PoltypeJobs", then within this folder you can submit jobs according to the documentation.

2) What is basically happening is that you are telling poltype you want to use gaussian, but I dont see it in your.bashrc so it doesnt know where to find it. There is example bashrc on readme.install on the gh-pages branch. Try making a new bashrc like the example and then try again.

[jpetersen48]

For point 1, that makes sense. Point 2, I think I address that in the submission script pol.sh, but let me know if I'm incorrect:

module load tinker/8.8.1/intel/19.0
module load gaussian
module load python/3.7.3
module load conda

export PATH='/afs/crc.nd.edu/x86_64_linux/c/conda/miniconda2/4.5.4/bin:$PATH'
source activate poltype
export g09root=/afs/crc.nd.edu/x86_64_linux/g/gaussian/09/pgi/16.3/build_RH7/g09
export GAUSS_SCRDIR=/scratch365/jpeter20/poltype/
export GDMADIR=/afs/crc.nd.edu/group/corcelli/Poltype/gdma/bin/
export PATH=/scratch365/jpeter20/poltype/gdma/bin/:$PATH
export PSI_SCRATCH=/scratch365/jpeter20/poltype/
export PATH=/afs/crc.nd.edu/x86_64_linux/t/tinker/8.8.1/tinker/source/:$PATH

nohup python PoltypeModules/poltype.py

[misterbrandonwalker]

Oppps I didnt look inside your .sh. Hmmm here is one point of difference between our sources. source $g09root/g09/bsd/g09.profile Maybe try doing a "which g09"? Also could you copy that error and search for it via grep in PoltypeModules/poltype.py , there should be a reference to a self.gausdir and self.gausexe, write before the error statement (below), can you print these variables to see what poltype thinks they are? if self.which(os.path.join(self.gausdir, self.gausexe)) is None: print("ERROR: Invalid Gaussian directory: ", self.gausdir)

$g09root/g09/bsd/g09.profile (just showing top where it seems to be setting env for gaussian on our cluster) i#

These commands set up for use of Gaussian 09. They should be source'd

into each Gaussian 09 user's .login file, after setting the following

environment variable:

#

g09root -- Directory which contains the g09 main directory. Defaults

to users's home directory if not defined before entry.

#

Top directories for the program:

# gr=$HOME if [ "$g09root" ] then gr=$g09root fi export GAUSS_EXEDIR="$gr/g09/bsd:$gr/g09/local:$gr/g09/extras:$gr/g09" export GAUSS_LEXEDIR="$gr/g09/linda-exe" export GAUSS_ARCHDIR="$gr/g09/arch" export GV_DIR="$gr/gv" if [ -e "$GV_DIR/gview.app" ]; then alias gv='open $GV_DIR/gview.app' else alias gv="$GV_DIR/gview.csh" fi if [ "$PATH" ]; then export PATH="$PATH:$GAUSS_EXEDIR" else export PATH="$GAUSS_EXEDIR" fi export _DSM_BARRIER="SHM" if [ "$LD_LIBRARY64_PATH" ]; then export LD_LIBRARY64_PATH="$GAUSS_EXEDIR:$GV_DIR/lib:$LD_LIBRARY64_PATH" else

On Thu, Sep 17, 2020 at 1:28 PM Jacob Petersen notifications@github.com wrote:

For point 1, that makes sense. Point 2, I think I address that in the submission script pol.sh, but let me know if I'm incorrect:

export PATH='/afs/crc.nd.edu/x86_64_linux/c/conda/miniconda2/4.5.4/bin:$PATH' source activate poltype export g09root=/afs/crc.nd.edu/x86_64_linux/g/gaussian/09/pgi/16.3/build_RH7/g09 export GAUSS_SCRDIR=/scratch365/jpeter20/poltype_indole2cooh/ export GDMADIR=/afs/crc.nd.edu/group/corcelli/Poltype/gdma/bin/ export PATH=/scratch365/jpeter20/poltype_indole2cooh/gdma/bin/:$PATH export PSI_SCRATCH=/scratch365/jpeter20/poltype_indole2cooh/ export PATH=/afs/crc.nd.edu/x86_64_linux/t/tinker/8.8.1/tinker/source/:$PATH

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/pren/poltype/issues/50#issuecomment-694418169, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKNB26LKT7GYNLBMRYVO7HLSGJIMRANCNFSM4RLYEEYQ .

[misterbrandonwalker]

I also see this in your output log. Is the source activate poltype supposed to be like "conda activate poltype"? Like is poltype the name of your conda env?

/opt/crc/c/conda/miniconda2/4.5.4/etc/profile.d/conda.sh: line 15: dirname: command not found /opt/crc/c/conda/miniconda2/4.5.4/etc/profile.d/conda.sh: line 16: dirname: command not found

On Thu, Sep 17, 2020 at 1:44 PM Brandon Walker notifications@github.com wrote:

Oppps I didnt look inside your .sh. Hmmm here is one point of difference between our sources. source $g09root/g09/bsd/g09.profile Maybe try doing a "which g09"? Also could you copy that error and search for it via grep in PoltypeModules/poltype.py , there should be a reference to a self.gausdir and self.gausexe, write before the error statement (below), can you print these variables to see what poltype thinks they are? if self.which(os.path.join(self.gausdir, self.gausexe)) is None: print("ERROR: Invalid Gaussian directory: ", self.gausdir)

$g09root/g09/bsd/g09.profile (just showing top where it seems to be setting env for gaussian on our cluster) i#

These commands set up for use of Gaussian 09. They should be source'd

into each Gaussian 09 user's .login file, after setting the following

environment variable:

#

g09root -- Directory which contains the g09 main directory. Defaults

to users's home directory if not defined before entry.

#

Top directories for the program:

# gr=$HOME if [ "$g09root" ] then gr=$g09root fi export GAUSS_EXEDIR="$gr/g09/bsd:$gr/g09/local:$gr/g09/extras:$gr/g09" export GAUSS_LEXEDIR="$gr/g09/linda-exe" export GAUSS_ARCHDIR="$gr/g09/arch" export GV_DIR="$gr/gv" if [ -e "$GV_DIR/gview.app" ]; then alias gv='open $GV_DIR/gview.app' else alias gv="$GV_DIR/gview.csh" fi if [ "$PATH" ]; then export PATH="$PATH:$GAUSS_EXEDIR" else export PATH="$GAUSS_EXEDIR" fi export _DSM_BARRIER="SHM" if [ "$LD_LIBRARY64_PATH" ]; then export LD_LIBRARY64_PATH="$GAUSS_EXEDIR:$GV_DIR/lib:$LD_LIBRARY64_PATH" else

On Thu, Sep 17, 2020 at 1:28 PM Jacob Petersen notifications@github.com wrote:

For point 1, that makes sense. Point 2, I think I address that in the submission script pol.sh, but let me know if I'm incorrect:

export PATH='/afs/ crc.nd.edu/x86_64_linux/c/conda/miniconda2/4.5.4/bin:$PATH' source activate poltype export g09root=/afs/ crc.nd.edu/x86_64_linux/g/gaussian/09/pgi/16.3/build_RH7/g09 export GAUSS_SCRDIR=/scratch365/jpeter20/poltype_indole2cooh/ export GDMADIR=/afs/crc.nd.edu/group/corcelli/Poltype/gdma/bin/ export PATH=/scratch365/jpeter20/poltype_indole2cooh/gdma/bin/:$PATH export PSI_SCRATCH=/scratch365/jpeter20/poltype_indole2cooh/ export PATH=/afs/ crc.nd.edu/x86_64_linux/t/tinker/8.8.1/tinker/source/:$PATH

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/pren/poltype/issues/50#issuecomment-694418169, or unsubscribe < https://github.com/notifications/unsubscribe-auth/AKNB26LKT7GYNLBMRYVO7HLSGJIMRANCNFSM4RLYEEYQ

.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/pren/poltype/issues/50#issuecomment-694426727, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKNB26NY6E2NGFZRV5PRGVLSGJKHJANCNFSM4RLYEEYQ .

[mquevill]

[The following is specific to Notre Dame's CRC computing cluster.]

The default version of Gaussian is 16B01 (on the CRC), which uses the executable g16 instead of g09. [If you want to keep using g09, run module load gaussian/09E01, but it is deprecated on the CRC.]

Poltype should be able to find Gaussian for you once the module is loaded, so you shouldn't need to set the g16root.