Download one version of python 3 environment, available here. This program was tested on python 3.5.2
Help for installation on unix systems
tar xf Python-3.5.2.tar.xz #extract tarball
cd Python-3.5.2
./configure --prefix=$HOME ##configure python
make #compile
make test
make install
echo "export PYTHONPATH="${PYTHONPATH}:$HOME/local/lib/python3.5/site-packages"" >> $HOME/.bashrc #define pythonpath NOTE: local can be .local instead
. $HOME/.bashrc
We recommend easy_istall and pip, if you do not have root access.
easy_install-3.5 --prefix=$HOME/local cclib-1.4 #should be 1.4 *
easy_install-3.5 --prefix=$HOME/local pip
pip3.5 install --user numpy #make sure to install numpy first
pip3.5 install --user scipy
Unfortunately cclib-1.5 does not work properly with our code, so version 1.4 is recommended. If you'd still like to use 1.5 you'll need to modify the source code of cclib. The simplest (and least elegant) way to do this is to find and modify the following file:
/home/user/lib/python3.5/site-packages/cclib/parser/gaussianparser.py
where lines 677 and 678 have to be commented out with a '#' symbol at the beginning of each line. The code should look something like this:
if len(parts) > 4:
energy = parts[2].split('=')[1]
if energy == "":
energy = self.float(parts[3])
else:
energy = self.float(energy)
# Next 2 lines commented by User
# if len(self.scftargets[0]) == 3: # Only add the energy if it's a target criteria
# newlist.append(energy)
scfvalues.append(newlist)Download this repository and note where you decide to store it. For proper execution we strongly suggest to create a symbolic link to the main file:
ln -s /wherever/this/program/is/main.py $HOME/bin/GSTA-hc
Make sure to use the absolute path, otherwise the link might be broken.
After reloading the environment, the program should be ready to use.
. $HOME/.bashrc && . $HOME/.bash_profile
The utility will run interactively simply by executing it:
GSTA-hc
For more advance usage, flags are available:
GSTA-hc -f [freq.out file] -c [config file] -t [type]
Config file
The parameters given in a config file are optional. The program may request interaction for necessary information.
example: parameter = value
| parameter | var type | value |
|---|---|---|
| command | (str) | command to run Gaussian 09 |
| options | (str) | all options you wish to use with command |
| temp | (float) | temperature where data should be calculated |
| NTraj | (int) | number of independet trajectories |
| NCalc | (int) | number of Gaussian calculations allowed to be run at one time |
| rotation | (bool) | include or exclude rotation |
| end | (int) | value |
Note, that even though the number of independent trajectories can be one, for the linear regression and determination of heat capacity it must be at least two.
value can be:
- create velocity generating inputs.
- run velocity generating inputs.
- create BOMD inputs.
- run BOMD inputs.
- process BOMD data. Not available automatically.
After the simulations all terminated, the data is ready to be processed to obtain heat capacities. Unfortunately the automatization of the process is not yet stable, so we recommend to you the postprocess function:
GSTA-hc -t pp -f .Calcfile*
where .Calcfile* is the binary file used as a checkpoint during the run. This results in an output *cv.out.
The directory structure of one run is important. Do not move or remove files from the associated directories. Once Gaussian calculations have been submitted, a process monitors them in the background. This can be followed by the watchlog files in the directories. This daemon recalls the main program when a calculation is finished. This can be done manually if needed:
GSTA-hc -f [file to process] -t [type] -c [config]
Configuration file is only needed if the initial run was set to an end already achieved (see above). Type can be MOVE(only if rotation was excluded) VELGEN and MD. The simulations' length is set to 1 ps, the timestep is 0.1 fs at the moment. We wish to optimize these parameters to actual cases and implement the algorithm.
Due to differences in output syntax simulation with Gaussian16 is not stable. At the moment we suggest to use Gaussian09 instead.