AiiDA
Automated Interactive Infrastructure and Database for Computational Science

AiiDA is a flexible and scalable informatics' infrastructure to manage, preserve, and disseminate the simulations, data, and workflows of modern-day computational science. Able to store the full provenance of each object, and based on a tailored database built for efficient data mining of heterogeneous results, AiiDA gives the user the ability to interact seamlessly with any number of remote HPC resources and codes, thanks to its flexible plugin interface and workflow engine for the automation of complex sequences of simulations.

Journal ref: G. Pizzi, A. Cepellotti, R. Sabatini, N. Marzari, and B. Kozinsky, AiiDA: automated interactive infrastructure and database for computational science, Comp. Mat. Sci. 111, 218-230 (2016)

Open access link: arXiv:1504.0116

Posts

Report on the AiiDA tutorial held at CINECA, 30 May-1 June 2018

👤 🕔 June 4, 2018 Comments Off on Report on the AiiDA tutorial held at CINECA, 30 May-1 June 2018

Another AiiDA tutorial, sponsored by PRACE and the H2020 MaX Centre of Excellence, was organized in collaboration with the Italian CINECA supercomputer centre, who hosted the event.
The tutorial took place between May 29 and June 1 2018, at CINECA Bologna (Italy).
More than twenty participants, with a strong female representation, joined in this introductory tutorial to the AiiDA framework.

Group photo from the AiiDA tutorial in May 2018

On the first day, an introductory talk was delivered by Spyros Zoupanos about the general architecture and design of AiiDA, after which the participants started with the hands-on exercises to get to know with AiiDA’s interface.
A more detailed presentation on the workflow system was given by Martin Uhrin.

Introduction by the organisers of the AiiDA tutorial (May 2018)

On the second and third day, Giovanni Pizzi and Leonid Kahle gave an exposé on their research projects that actually implemented and used the AiiDA platform to discover new two-dimensional materials and solid state ionic conductors in a high-throughput way, respectively.
This gave a good idea to the tutorial participants of the potential of AiiDA in a broader sense.

Room with participants at the AiiDA tutorial

The opportunity was also used to introduce participants to the concept of open source collaboration – in particular, Sebastiaan Huber gave a crash course on the use of Git and GitHub to contribute to open source projects (branches, forks and pull requests).

During the tutorial, a poster session was also organized, where the participants had the chance to present their own work and get to know that of their colleagues, in an informal setting with drinks and fingerfood.

Poster session at the AiiDA tutorial

In between these extra events, the main focus lay on introducing the participants to AiiDA through a hands-on tutorial and exercises, with all the instructors available to provide support and explanation in person.
The participants were very committed during the tutorial and already provided comments to improve the tutorial during the event.

Instructors helping out participants at the tutorial

At the end of the tutorial, all participants were asked to give feedback by completing an evaluation questionnaire, which will be used to further improve the tutorial and AiiDA itself.
The feedback was overall very positive, with the tutorial graded with an average vote of 9.0 out of 10, and all votes between 8 and 10.
All participants stated that the well-prepared and helpful instructors helped the participants to get really well acquainted with AiiDA and that it may very well help them with their own work and research.
The scientific talks that gave examples of AiiDA in real-world scientific projects were also well received as they gave a more intuitive example of the power and potential of AiiDA.

You can find more photos of the event on our Facebook page.

We acknowledge support from PRACE and from there H2020 MaX Centre of Excellence.

 

Report on AiiDA tutorial: 24-25 January 2017 at ICTP in Trieste, Italy.

👤 🕔 January 27, 2017 Comments Off on Report on AiiDA tutorial: 24-25 January 2017 at ICTP in Trieste, Italy.

The end of January of 2017 saw the eighth hands-on tutorial for AiiDA within the “Advanced Workshop on High-Performance & High-Throughput Materials Simulations using Quantum ESPRESSO” organised at and by ICTP in Trieste. The tutorial was kicked off by an introductory lecture by Giovanni Pizzi (EPFL), explaining the philosophy and architecture behind AiiDA, and how the ADES model facilitates and simplifies performing and sharing calculations, under the umbrella of complete data provenance. After the scene was set, the approximately 75 participants went to their workstations to start with the first of the hands-on sessions. A completely configured virtual machine was provided for each individual participant, to which they only had to login and they were on their way. The first session served to familiarize the users with Verdi, a command line interface which acts as the user’s tool to interact with AiiDA. With the tools in hand to put AiiDA to good use, the participants learned by example how a Quantum Espresso calculation can be easily launched as AiiDA takes care of a lot of the repetitive work behind the scenes.

The virtual environment in which the participants were working came with a database filled with pre-computed calculations, which allowed the aptitude of AiiDA in working with big data sets to be displayed effortlessly. The QueryBuilder, the tool provided by AiiDA to efficiently query a user’s collection of calculations, was introduced and a step-by-step tutorial showed the participants how it could be used to efficiently analyze a large collection of calculations on perovskites. The final session detailed how the users could easily create such a big collection of calculations themselves, in a high-throughput manner, through the concept of Workflows.

Throughout the tutorial, five members of the AiiDA team were present to answer any questions of the participants and give introductory talks at the beginning of a new hands-on session. After the hands-on sessions, a good portion of the participants gave their feedback on the quality of the tutorial and the impact on their skills in using AiiDA. The results are very promising while at the same time they provide the AiiDA team with valuable information on how to improve both the tutorial and AiiDA itself.

The feedback on the quality of the tutorial itself, in terms of content and planning, were overall very positive

 

 

AiiDA tutorial Jan 2017 feedback 2

Participant feedback on the quality of the tutorial

The results of the feedback also show that the tutorial really helped improve the skill and knowledge of the participants regarding the use of AiiDA

 

AiiDA tutorial Jan 2017 feedback 1

The skill level of the participants before and after the tutorial

The AiiDA team thanks the organisers of the event for making this tutorial possible and all the participants for their enthusiasm. Until the next AiiDA event!

Report of the AiiDA tutorial of last year (June 2016) in Lausanne

👤 🕔 January 19, 2017 Comments Off on Report of the AiiDA tutorial of last year (June 2016) in Lausanne

Report from the AiiDA coding week – Dec 2016

👤 🕔 December 13, 2016 Comments Off on Report from the AiiDA coding week – Dec 2016

Last week (5-9 Dec 2016) we had a coding week for AiiDA, to which 15 people participated (roughly half of them core developers from the AiiDA, and the other half were developers who were interested in contributing to the platform).

The coding week has taken place in Leysin, in the Swiss Alps. A few discussion or work topics had been identified before the start by the AiiDA team, and have been addressed during the week. In particular, the following topics have been discussed and the result has been summarised in a draft document for later implementation:

  • New more efficient implementation of the AiiDA daemon
  • How third-parties can contribute to Materials Cloud sections
  • Redesign of the ORM classes (in particular Node) to improve the way multiple backends are supported

A few discussions have also led to new branches of development, that will be soon merged into the develop branch (and appear in a future release):

  • AiiDA plugin interface integrated with pip
  • Various improvements and bug fixes to the SQLAlchemy backend implementation

Finally, the remaining work topics have been taking up a large part of the coding sessions, and most of the code has already been merged into the develop branch:

  • Redesign of the AiiDA documentation, with a more clear structure, removal of duplicates, a new and more clear installation section.
  • Simplified procedure to install AiiDA.
  • New testing infrastructure to test in parallel any database backend (currently, both Django and SQLAlchemy)
  • Removal of the DbPath (transitive closure) table and triggers, and replacement with a dynamically created query (using either PostgreSQL recursive queries or PostgreSQL functions). The performance of these queries has been optimised.

Beside programming, most participants have taken place to an outdoor hiking activity on Wednesday morning to Berneuse, close to Tour d’Aï, that was a great team building opportunity. Participants have also enjoyed after-dinner sessions with board and card games.

The outcome of the feedback, summarised below, has been very positive, both in terms of participation, organisation, and improvement to the code. The results indicate also that similar events should be organised again in the future, with a frequency of about twice a year.

Some photos of the event can be found on our Facebook page.

— Giovanni Pizzi

Summary of the results of the feedback form

Results of the feedback form for the AiiDA Coding week (Dec 2016)