Installation instructions

Electronic structure packages

The program employs two levels of electronic structure: semiempirical (or Low-Level LL) and ab initio/DFT (or High-Level HL). The LL calculations can be carried out with MOPAC or Entos Qcore, and the HL calculations with Gaussian 09 (G09).

While MOPAC comes with AutoMeKin, if you want to employ xtb method, Entos Qcore must be installed. Additionally, for the HL calculations G09 is the only choice right now.

To install Entos Qcore, we recommend following these steps:

1. Install miniconda3

2. Add the following line to $HOME/.condarc (create the file if not present):

auto_activate_base: false

which avoids activation of base environment.

3. Install qcore version 0.8.14 in a conda environment (qcore-0.8.14-env):

conda create -n qcore-0.8.14-env -c entos -c conda-forge qcore==0.8.14 'tbb<2021'

4. Activate the newly created environment:

conda activate qcore-0.8.14-env

5. After installation, each user will be asked to read the Software License Agreement to generate a unique token:

qcore --academic-license

Auto installer

This is the easiest way to use AutoMeKin. The auto installer script installs singularity in your computer and downloads the last release container image from sylabs (https://cloud.sylabs.io/library/emartineznunez/default/automekin) as $HOME/automekin_<tag>.sif. Note that this is done only the first time you use it unless a new image is available. Then, the script will detect singularity and the image (that must be located in your $HOME) and will only start an instance of the container.

1) Download AutoMeKin's auto installer script Automekin.sh (Last update May 26, 2021)

2) Make the script executable:

chmod +x Automekin.sh

3) Run the script:

./Automekin.sh

Note that depending on your Linux configuration, before running the autoinstaller you might need to change some parameters which will require admin or root privilege. If that is the case and once you changed the parameters with your admin or root accounts, no further admin or root privilege will be needed. Return to your user account and run the auto installer again.

4) Once the above steps are completed, singularity will be installed under ${TMPDIR-/tmp}/amk_installer-${USER}/software in bash shell script syntax and an instance of the container will be started using a sandbox image deployed under /tmp/selfextract.XXXXXX folder (where XXXXXX is a randomly generated character sequence). The container comes with all AutoMeKin's tools installed in $AMK plus vim, gnuplot and molden which can be run from the container. A bash shell session under $HOME will start under the deployed instance. Note that you can open new sessions and access AutoMeKin's output files from your Linux environment and use your own tools as well.

5) To exit the container just type:

exit

6) Once your calculations are done, remember to stop the instance:

./Automekin.sh stop

Important notes:

1. Make sure your auto installer is up to date (see above)
2. The autoinstaller also works on Ubuntu 20.04 LTS on Windows 10. To install Ubuntu 20.04 LTS on Windows 10, follow these instructions: https://docs.microsoft.com/en-us/windows/wsl/install-win10
3. AutoMeKin's third-party packages in the container are updated (see below the minimum required version numbers). Local installations of different versions of these Python packages might interfere in the execution of AutoMeKin

Singularity container

If singularity is already installed in your computer, you can obtain the container from sylabs. First check what the latest image (Tag) is by typing:

singularity search automekin

and replace <Tag> below by that number. Then, from your $HOME type:

singularity pull library://emartineznunez/default/automekin:<Tag>

You can start an instance of the container and run it using:

singularity instance start automekin_<Tag>.sif automekin
singularity run instance://automekin

which will allow you to run low-level scripts. You can stop the instance using:

singularity instance stop automekin

Note, however, that if you want to use g09 you must bind it to the container. To help you do so, we created this script Singularityg09.sh , which should be run with the complete path to the sif file as argument as in the example:

Singularityg09.sh $HOME/automekin_<Tag>.sif

Note that Singularityg09.sh will start a new instance of the container every time it is executed. To list the instances use:

singularity instance list

And stop them as indicated above.

Tarball

If you do not want to use singularity, you can install the code following the steps indicated in this section.

Download the tarball

Untar and unzip the file:

tar xvfz Amk-SOURCE-2021.tar.gz

Installing dependencies

Before installing amk for the first time, be aware that the following packages are needed:

- GNU Bash

- GNU bc

- environment-modules

- GNU Awk (gawk)

- GNU C Compiler (gcc)

- Gnuplot

- GNU Fortran Compiler (gfortran)

- GNU Parallel

- SQLite (version >= 3)

- Zenity

You can install the missing ones manually, or you can use the scripts located in amk-SOURCE-2021 and called install-required-packages-distro.sh (where distro=ubuntu-16.4lts, centos7 or sl7), which will do the work for you.The ubuntu-16.4lts script installs all dependencies, but for the RHEL derivatives (centos7 and sl7) you have to install parallel separately, and you have two choices:

a) install-gnu-parallel-from-source.sh. This script installs parallel latest version from source thanks to Ole Tange (the author). Also it can fallback to a user private installation into $HOME/bin if you have not administrator permisions to install it globally.

b) install-gnu-parallel-from-epel.sh. Enables the EPEL repository and installs parallel from it.

The program employs python3 and the following python3 libraries are needed (which can be easily installed with pip):

- ASE (version >= 3.21.1)

- Matplotlib (version >= 3.3.4)

- NetworkX (version >= 2.5)

- NumPy (version >= 1.19.5)

- SciPy (version >= 1.5.4)

These packages might also be useful to analyze the results:

- molden

- sqlitebrowser

Installation

Once the above packages are installed:

go to the amk-SOURCE-2021 folder.

cd amk-SOURCE-2021

Now type:

./configure

This will install amk in $HOME/amk-2021 by default. If you want to install it in a different directory, type:

./configure --prefix=path_to_program

Finally, complete the installation:

make
make install
make clean

The last command (make clean) is only necessary if you want to remove from the src directory the object files and executables created in the compilation process.

For convenience, and once “Environment Modules” has been installed, you should add the following line to your .bashrc file:

module use path_to_program/modules

where path_to_program is the path where you installed amk (e.g., $HOME/amk-2021).

Running the tests

To run the tests follow these steps:

1) Load the module (unless you use singularity):

module load amk/2021

2) Run the following script to run all tests:

run_test.sh

Note that each test takes from a few seconds to several minutes. The results of each test will be gathered in a different directory.

If you want to run a subset of tests use the following:

run_test.sh --tests=FA, FAthermo

which will run FA and FAthermo tests only. These are the tests available in this version: assoc, assoc_qcore rdiels_bias, diels_bias, FA_biasH2, FA_biasH2O, FA_bxde, FA_singletraj, FA, FAthermo, FA_programopt, vdW, FA_ck, FA_qcore, FA_bxde_qcore and ttors

Older versions

Download the tarball for the 2020 version (revision 993)

And follow the instructions given above (tarball).

Download tutorial2020

Return to AutoMeKin