| Re: [science-iwg] Introducing myself |
|
Dear all,
my name is Giovanni Pizzi and I am a researcher at EPFL, Lausanne, Switzerland, in the group of prof. Nicola Marzari.
I jump in because I’m one of the developers of AiiDA (yes, the website is
www.aiida.net; and yes, it’s written in python).
I was supposed to write to this list a few days ago to introduce myself and AiiDA, but then I got delayed, and in the meantime Mario was faster than me ;-)
So, let’s remediate now:
AiiDA is a project jointly developed by EPFL and by the Bosch Research and Technology Center in Cambridge (MA, USA).
It is at the core of a Swiss 12-year project, MARVEL, involving 33 principal investigators across 11 Swiss Institutions (http://nccr-marvel.ch/en/project) and of the Horizon2020 European
projects on Exascale computing in Material Science, MaX (http://www.max-center.eu).
We are directly involved in both projects, as well as the CSCS supercomputing center where Mario is working (and indeed we are directly collaborating with Mario, who was visiting us yesterday).
The aim of AiiDA is to provide an infrastructure for managing large computations on supercomputers as well as pre-post-processing steps, automatically storing all the data, simulations, their relationship and the full provenance in a database,
ensuring reproducibility of simulations and allowing for efficient querying.
At the same time, we aim at providing a flexible interface for users, easy to use from python without knowing how a database works: the user just works on his simulations through AiiDA, and everything gets transparently recorded.
Moreover, we provide a workflow engine tailored for researchers - this means that workflow steps are python functions, rather than static blocks, because in most cases the logic of the workflow cannot is very dynamic and the researchers need to
‘play’ with it during the research. However, under the hoods, AiiDA keeps track of the full stack of calling/caller functions for provenance.
The infrastructure is mainly used now by physicists/material scientists working with atomistic simulations, but the code is plugin-based so it can be interfaced with any code with a command-line interface and input/output files that can be processed
using python.
I hope this short summary is already useful, you can find more information on the website, on the online docs
http://aiida-core.readthedocs.org/en/stable/, or on the publication 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) http://www.sciencedirect.com/science/article/pii/S0927025615005820
(or here if you cannot access the PDF: http://arxiv.org/abs/1504.01163)
But if you need more details, just ask!
Best,
Giovanni
|