Tag Archives: NewtonX

After I got the tests to run, I started working my way through the tutorial.  Specifically, I am trying part 4.  I managed to get to step 18 before I ran into problems.  In step 18, we are trying to generate the inital conditions for NewtonX to use.  To do this you run a line of code:

$NX/initcond.pl >initcond.log &

This line of code immediately spits out a message which is a number in brackets followed by another number.  The numbers change each time.  It then prints a message saying initcond.pl is dying now.  At this point, the console freezes and does not respond until you abort with ctrl-c.  It then spits out an exit code, “Exit 255.” (Which I have been unable to find more information on).

To try to figure out what was wrong, I examined the initcond.log file.  So far I have gotten two different messages which seem to alternate with no real reason.  The first occurs early in the run and says:

Cheking input files

Creating qvector...

It appears to die while trying to make the missing qvector.  The qvector it is referring to used to be a file in version 1.1 that contained the quantum vibrational number, but it isn’t referenced at all in 1.4.

The second error happens later in the run and says:

Starting run_g09_initcond.pl at Tue Nov 22 08:21:30 PST 2016

Finished run_g09_initcond.pl with ERROR at Tue Nov 22 08:21:30 PST 2016

This error seems to be referring to the same issue with NewtonX inserting an extra /g09/ which we were seeing earlier.  However, when I tried using the same set of variable definitions which worked in the tests, it did not change anything.  Also, I poked around in the file and found the same qvector error as earlier but instead of crashing after creating the qvector, it continued successfully.  So it seems like that error is one that it can sometimes get past.

So far I haven’t been able to figure out when one error occurs versus the other.  There doesn’t seem to be any particular logic to it.

-Kristine

First you need to install NewtonX.  To do this, you request a download link from the NewtonX site, follow the link, and download the binary file recommended (it will be called something like “NX-1.4.0-3-source+binary.tgz”).  Then copy the file to comet, and untar and unzip it by typing tar -zxf NX-1.4.0-3-source+binary.tgz.  This corresponds to section 9.1 in the documentation.  Do not go further and do 9.2; 9.2 is only for if you want to install the program manually.

After installing NewtonX, I spent a while trying to get the built in tests to run.  Eventually I came up with this procedure:

First we have to set up the variables so NewtonX can find Gaussian.  We do this through the following lines of code:

export NX=/home/kvorwerk/NX-1.4.0-2/bin
export g09root=/share/apps/compute/gaussian/09.E.01/g09
source $g09root/bsd/g09.profile
export g09root=/share/apps/compute/gaussian/09.E.01


The last line is necessary to avoid an error by which NewtonX adds an extra /g09/ into the path where it searches for gaussian.  Also note the line setting the NX variable would need to be modified to match the path where you installed NewtonX.

Next we make a directory to run the tests in.  This can be anywhere.  Then we select the tests to run by typing:

$NX/inp-testnx.pl

This shows a list of tests.  We want to run the gaussian09 one and the NewtonX ones; that is, 18, 19, 24, and 25.  To actually run the tests we type:

$NX/test-nx.pl >test.log

This creates a bunch of subdirectories and a file called test.log.  The test.log file contains basic output, and more output is in the sub directories.  When I ran the tests, the test.log file told me that:

18 finished, but with a message saying:

Files /home/kvorwerk/nx_test/TEST_NX/MD-SICH4-G09-TDDFT-NAD/RESULTS/dyn.out and   /home/kvorwerk/NX-1.4.0-2/bin/../test-nx/STANDARD-RESULTS/MD-SICH4-G0    9-TDDFT-NAD/RESULTS/dyn.out differ.  It seems to be important differences. Something may be wrong in the installation.

The files it refers to are easy to find.  However, I have been unable to find the file or script which compared them.

19 finished normally

24 finished with a message saying:

Files /home/kvorwerk/nx_test/TEST_NX/MD-ANALYTICAL-MODEL1/RESULTS/dyn.out and   /home/kvorwerk/NX-1.4.0-2/bin/../test-nx/STANDARD-RESULTS/MD-ANALYTIC    AL-MODEL1/RESULTS/dyn.out differ.  It seems to be only minor numerical differences. Probably installation is OK.

I do not think this error is worth worrying about.

25 finished normally.

The issue with test 18 may be due to us using a more recent version of gaussian (09.E) than the developers used to create the test.  Hopefully we will be able to find the comparisson they use and learn if this is indeed the case.

-Kristine