ESPNN 0.1.4

Electronic Stopping Power Neural Network predictor

ESPNN - Electronic Stopping Power Neural Network

The ESPNN is a python-based deep neural network that allows the user to predict the electronic stopping power cross-section for any ion and target[^1] combination for a wide range of incident energies. The deep neural network was trained with many tens of thousands curated data points from the IAEA database. See more details of the ESPNN in this publication.

You can use the ESPNN package remotely or locally. Find below all the usage options available.

Run ESPNN online

The ESPNN package can be used remotely in the Google Colab platform[^2]. There, you'll find a jupyter notebook with a quick tutorial on how to use the ESPNN. You can also make a copy of the notebook to your own personal Drive and compute the stopping power of any projectile-target combination.

Install ESPNN

To use the ESPNN in your computer, first you'll need to install it. We recommend using a python virtual environment to this end (for example, see anaconda or virtualenv). If you are not familiar with virtual environments and would like to rapidly start using python, follow the anaconda indications according to your operating system:

• Install anaconda in Linux
• Install anaconda in Windows
• Install anaconda in macOS

Using pip

The simplest way to install the ESPNN is via pip. Indistinctively, Ubuntu, Windows and macOS users can install the package by typing in the terminal or the anaconda bash terminal:

pip install ESPNN

Using this repository

You can also install the ESPNN package by cloning or downloading this repository. To clone (make sure you have git installed) this repo, use the following commands in your terminal/anaconda bash terminal:

git clone https://github.com/ale-mendez/ESPNN.git
cd ESPNN
pip install ESPNN/

If you downloaded the zip, change your directory to your download folder and, in your terminal/anaconda bash terminal, type

pip install ESPNN-master.zip

Run ESPNN locally

Once you've installed the ESPNN package in your preferred environment, you can run it by using a jupyter notebook or directly from terminal.

Using a notebook

A basic tutorial of the ESPNN package usage is given in prediction.ipynb. The package requires the following parameters:

• projectile: Chemical formula for the projectile
• target: Chemical formula for the target

Optionally, you can provide these parameters:

• projectile_mass: Mass in amu for the projectile
• target_mass: Mass in amu for the target

import ESPNN
ESPNN.run_NN(projectile='He', projectile_mass=4.002602, target='Au', target_mass=196.966569)

The package automatically produces a matplotlib figure and a sample file named XY_prediction.dat, where X is the name of the projectile and Y is the name of the target system.

ls -a
.  ..  HHe_prediction.dat  prediction.ipynb 

Other optional arguments

The energy grid used for the ESPNN calculation can be customized with arguments

• emin: Minimum energy value in MeV/amu units (default: 0.001)
• emax: Maximum energy value in MeV/amu units (default: 10)
• npoints: Number of grid points (default: 1000)

Furthermore, the figure plotting and output-file directory-path can be modified via

• plot: Prediction plot (default: True)
• outdir: Path to output folder (default: "./")

ESPNN.run_NN(projectile='He', projectile_mass=4.002602, target='Au', target_mass=196.966569, emin=0.01, emax=1, npoints=50)

From terminal

The ESPNN package can also be used from terminal with a syntaxis analogous to the above given:

python -m ESPNN H Au -Ym 196.966569

Additional information about the optional arguments input can be obtained with the -h, --help flag:

python -m ESPNN -h

Funding Acknowledgements

The following institutions financially support this work: the Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET) by the PIP-11220200102421CO and the Agencia Nacional de Promoción Científica y Tecnológica (ANPCyT) of Argentina PICT-2020-SERIEA-01931. CCM also acknowledges the financial support of the IAEA.

[^1]: ESPNN first release considers only mono-atomic targets. [^2]: A Google account is required.