Atomistic simulations are increasingly widespread in modern condensed matter physics, computational chemistry, and materials science. A wide range of approaches from quantum chemistry to machine learning interatomic potentials (MLIPs) are used, with correspondingly wide-ranging software implementations. Because of this diversity, it can be challenging to track the origin and consistency of calculations and utilize the in-house tools of different research groups. Without shared collaborative tools, multiple variants of high-level and code-independent algorithms, such as geometry optimisation or transition state search, are implemented by multiple groups. This inefficiency results in both needless duplication of efforts, code and an increase in potential bugs or non-maintained code. 

The Atomic Simulation Environment (ASE) is a community-driven Python package that provides standardised tools for representing and manipulating atomic structures, running calculations, and derived higher-level algorithms. It interfaces with around 100 file formats and 30 simulation codes, acting as an essential "glue" for work spanning multiple packages. Originally designed and still widely used for running electronic structure calculations and manipulating atomic structures, ASE is increasingly used for more complex atomistic simulation workflows and as a lingua Franca for fitting of machine learning models such as MLIPs, as well as for their evaluation. The existence of ASE frees the developers of new packages to focus on novel aspects and has become a critical piece of research infrastructure for an increasingly broad community of users and developers.  A 2017 paper describing ASE has attracted over 500 citations every year for the past 5 years [1], and 530 packages have listed ASE as their dependency on the Python Package Index, demonstrating the broad adoption of ASE. 

The 2025 CECAM workshop: “The atomic simulation environment ecosystem: Present and perspectives” addressed the increasing challenge of maintaining ASE due to its rapid growth in recent years. The main proposals from the workshop were a new plugin system,  a decoupling of the structure representation (the Atoms object) from the calculation itself (the Calculator object), which are currently strongly intertwined and frequently cause bugs, and a restructuring of the data obtained from calculators. These proposed changes, while conceptually small, will change the fundamental aspects of how every single user and developer operates and interfaces with ASE. The smoothness of the transition from ASEv3 to ASEv4, its take-up and success will strongly depend on stress-testing the modifications to the core ASE functionality in a wide range of community-led efforts, as well as the engagement of the developers of the derived packages with the plugin infrastructure.

Building on last year’s progress, we decided to form a group of early career researchers to work on a follow-up CECAM workshop to focus on ensuring robust integration of the upcoming changes in ASE into the wide range of community-developed packages. To ensure the sustainability and usability of ASE, we will engage the developers of electronic structure codes frequently used through ASE, as well as contributors and developers whose packages rely on ASE and were not present at the 2025 workshop, all of whom are represented among our proposed invited speakers. The workshop will serve both to identify and resolve the remaining cases where recent ASE changes do not fully suit the variety of ways it is used in practice, and as a hackathon to expand the examples, documentation, and testing coverage around the new ASE core & extended ecosystem redesign. This collaborative effort is timely and essential to ensure that the modifications planned for the next year remain compatible with real-world scientific applications and to finalise the necessary work for a stable new major version release of ASE.