What is ASE?

Overview

Teaching: 10 min
Exercises: 0 min

Questions

• How can ASE help me in my research?

• In what way is the ASE code structured?

Objectives

• List the key features of ASE

• Identify and describe the core ASE classes

• Describe how ASE fits within the larger atomistic modelling software ecosystem

• Understand the difference between class, instance and method in object oriented programming

ASE is a Python library for atomistic modelling

• The atomic simulation environment (ASE) is Python library for atomistic modelling; it allows you to set up, run and analyse atomistic simulations using Python scripts.
• Systems are described by a set of atomic positions, and calculations are derived from these.
• There are two core classes in ASE: Atoms and Calculator.

NumPy

ASE makes strong use of NumPy arrays. NumPy arrays have a host of useful features such as array slicing, and ensure that there is efficient performance even for thousands or millions of atoms. If you are unfamiliar with NumPy arrays there is a brief overview on the ASE website.

ASE uses several design patterns from object oriented programming (OOP).

• You do not need to be familiar with the intricacies of OOP to start coding with ASE, however some basic terminology is useful.
  • An object is an combination of data alongside operations for manipulating and returning information on that data.
  • A class is a category of objects - for example, Composer().
  • An instance is an object that belongs to a class - for example, the instance beethoven belongs to the class Composer.
  • A method is a fuction that belongs to an object - for example, the class Composer contains the function count_symphonies().
• To use ASE you will need to create an instance of a pre-defined class.
• To take an example most will be familiar with, lets assume you have a class called Composer().
• We can create an instance of that class called beethoven.

beethoven = Composer(birth_year=1770)

• This then allows us to access the methods and data associated with beethoven and the Composer class more generally:

print(beethoven.count_symphonies())
print(beethoven.birth_year)

9
1770

The Atoms class is used to represent molecules and materials

• When working with ASE, molecules and materials are represented by an Atoms object.
• Atoms represents (at minimum) a collection of atoms of any chemical species and associated positions

The Calculator class calculates basic properties of an Atoms object

• A Calculator object can be attached to an Atoms object.
• The Calculator takes the atomic numbers and positions from Atoms and calculates basic properties such as energy or forces.
• Calculator objects can themselves be split into three types:
  • in-built calculators that run the simulation within the same Python interpreter process. - file-based calculators that run the simulation as a sub-process, with communication mediated through input and output files.
  • calculators that run the simulation as a sub-process, with communication via pipes. Each approach has its own advantages and limitations which will be outlined later in the course.

Further information

For further information about code structure, implementation and features we recommend reading this topical review from Larsen et al.

ASE plays nicely with a variety of atomistic modelling tools

• The modular design of ASE allows it to play nicely with (“interoperate”) with many different atomistic simulation codes.
• This includes codes implementing density functional theory, (semi-)empirical methods, tight-binding models and classical interatomic potentials.
• For a full list of supported codes please see the ASE documentation.

Course structure

The structure of this course is designed to reflect the structure of ASE. First we introduce the Atoms object (episodes 2-4) and Calculators object (episodes 5-7). We then outline how to simulate more complex and computationally demanding systems by parallelising over multiple compute cores (episode 8). Finally we apply our understanding to three common tasks: molecular dynamics, geometry optimisation and electronic structure calculations (episodes 9-11). As the course progresses we move from smaller code snippets introducing the core ASE objects, to extended pieces of code for more complex tasks. Throughout the course you will see boxes like this. They will provide Python hints, signposting to other resources, discussion prompts or exercises for you to complete.

Key Points

• ASE is a Python library for atomistic modelling

• ASE uses several design patterns from object oriented programming (OOP)

• The Atoms class is used to represent molecules and materials

• The Calculator class calculates basic properties of an Atoms object

• ASE plays nicely with a variety of atomistic modelling tools