Skip to main content

Artificial Intelligence Data Analysis


Minutes of the 1st Advisory Board meeting

Minutes of the 1st Advisory Board meeting

Report on code and simulations selected for inclusion in AIDAdb

The report, in collaboration with Cineca WP10, will be written at the end of Task 2. It will focus first on the “low-level data”, i.e. the numerical simulations of interest and available to the project, and on their storage on the AIDdb. The second part of the report, in collaboration with WP4, will focus on the “high-level data”, i.e. the data obtained by exploiting the simulation data.

Report on the identification of numerical simulations to analyze and on the collection of data by virtual spacecraft

Once WP7 finalized the selection of numerical simulation data for inclusion in AIDAdb (deliverable D7.2), we will produce a report on the identification of numerical data suitable to be analyzed through virtual spacecraft techniques. Existing numerical runs will be selected to cover different regions of interests and different specific physical processes, e. g. particle heating, reconnection exhaust, magnetopause crossing, etc. Through these numerical simulations, we will provide virtual spacecraft measurements, which will be employed to support the design and training phase of novel techniques of artificial intelligence for data analysis and interpretation (AIDApp), aiming at identifying regions of scientific interest, based on time series of fields and particle velocity distributions. If needed, we will also plan and design new numerical simulations, in collaboration with AIDApp Data Assimilation and Analysis Tools (WP4 and WP6), able to simulate realistic conditions of the heliosphere environment.

Report on Analysis and selection of the software framework

This report will show the performances of different Neural Networks using different AI frameworks. We will perform a selection of benchmark cases, including multiple types of NN and multiple topologies. This report present the results from the analysis of multiple AI frameworks. Possible software includes: Keras, Torch, TensorFlow, Theano, etc. The goal of this report is to select a single AI framework to be used in the AIDApp package.

Report on Python tool to download data from open-access archives

The report will present the analysis of different available Phyton tool (spacepy,sunpy,astropy, PlasmaPy) and of other tools (irfu-matlab, SPEDAS) in order to decide by M6 which Phyton tool will be used as tool to download in situ, remote and ground plasma data from open-access archives (Tool 1 of AIDApp). Namely, if AIDA will use existing available Phyton tools, adapt available MATLAB or IDL tools to Phyton or a combination of both approaches.

Report on application of DA techniques to OpenGGCM

The report at the end of Task 1 will detail initial activities performed with the aim of developing as part of the AIDA Python package tools to assimilate observational data into code and to couple it with existing codes. We will use in particular the example of c s\introducing data assimilation techniques in the global magnetospheric code OpenGGCM. The initial model sensitivity study will provide information on the temporal evolution and spatial structure of the ensemble variance, on the sensitivity of the model to the variation of the model inputs, and on the domain of influence of observations. This last analysis in particular, which can be performed before implementing observation assimilation by analysing the ensemble correlation between state variables at different grid points, will provide guidelines on the variables and locations which should be privileged for assimilation during the rest of the work. The report will also describe the infrastructure developed for the assimilation of observations into OpenGGCM.

Exploitation plan version 1

First version of the exploitation plan of the outcomes after the end of the project.

Deliverable progress report

Report on the progress of all the different deliverables of the project

Report on the statistical methods included in AIDApp

We will produce an open source code for the statistical analysis of variables, dedicated to in situ space missions. Unifying several analysis methods, we will provide a complete suite of data analysis tools for heliophysics. This major deliverable consists in building an integrated software library that accumulates analysis methods able to reveal the structure and topology of the fluctuating fields, such as the velocity, the density and the magnetic field of the interplanetary plasma. The library will be versatile enough to ingest data from various missions, thanks to the interaction with the other WPs. The toolbox will apply a full package of analysis methods, achieving the following three major products: low-order statistics, high-order statistics, and graphics. The low-order statistics will consist of a Python software aimed to the characterization of the large scale profiles (averages, averaged-profiles and variances). The high-order statistics deliverable will be dedicated to the power spectra and structure function analysis. Finally, an open source graphical user interface, as MATPLOTLIB and gnuplot, will be produced for AIDApp.

Report on energetic particle analysis

The deliverable will consist of a toolbox that performs a detailed analysis of particle data, from several space missions. This module of the AIDApp will deal, automatically, with very large particle dataset. Plasma data are inherently 4-dimensional (one time coordinate, and 3 dimensions in the velocity space). These data possibly contains the most unexplored treasure of information, important for the physics of the heliosphere and, in general, for astrophysical processes. In this regard, the AIDApp, because of its versatility, will represent the first publicly available performing software for energetic particles analysis. This algorithm will help to identify possible correlations between fluctuating fields, structures, and acceleration of energetic particles.

Report on the Mathematical analysis of NNs and recommendations for developers

An extension on the benchmarks used for D2.1 is used for the analysis of multiple types of NNs that will be added to the AIDApp package. The benchmarks will be based on the spacecraft data and the exact applications required by the Consortium: time series, event catalogs and images. This deliverable will show a detailed analysis of the internal operations of the NNs. This in-depth analysis allows to select the best NN topology to a particular space data application. Practical recommendations are presented for each of the possible space application in the project, and for future users of the AIDApp package.

Minutes of the 2nd Advisory Board meeting

Minutes of the 2nd Advisory Board meeting

Data Management Plan version 1.0

The Data Management Plan (DMP) is a live document. The first version will be reported on month 6 and will continue to evolve through the project. It will include all the decisions taken by the consortium regarding: handling of data during and after the project, type of data collected and processed, standards applied, type of access, and plans for its distribution during and after the project.

Data Management Plan version 2.0

Second version of the DMP for the project AIDA.

Graphic image, dissemination and outreach plan

The first step of the dissemination is the definition of a logo and graphic image of the project with colour palette and font to be used for all the following actions, in order to give to all partners a guideline for graphics and start drawing the layout of the website. For this task is important that all partners will share materials as soon as will be available. All the available material will be organized and shown in the best way will be possible with images, video created ad-hoc, news and articles, flyer/brochure and poster for specific events.

Indicators of penetration in traditional and social media and correcting efforts

The dedicated web site both with the most popular social media like Twitter, Facebook, LinkedIn and Research Gate, will be the online presence to be maintained updated. We will also regularly monitor the online presence of AIDA in social media by means of indicators of penetration that we will step by step choose and upgrade and we will correct the efforts to improve the online presence.

Definitions of the software management standards and testing suit

In this document we will report the decisions taken for the management and automatic testing of the AIDApp and AIDAdb packages. We will perform a selection of the version control system, its physical location, and set up all the necessary access permissions. We will also document the selection of specific small benchmark cases, representative of the functionalities of the AIDApp and AIDAdb products, to perform automatic tests at a given frequency (nightly, weekly), ensuring a continuous follow up of the performances of the products.

Front-end Python software

A Python package will be created, able to digest data in any format of interest for the users of space missions archives. The package will offer an high-level interface to existing ML libraries and routines, with different levels of verbosity, in order to accommodate the needs of both expert and non-expert users. The open source code will be made publicly available on a git repository since the start of the project so to attract the interest and possibly the inputs of a large community of coders.

Publicly available database of virtual spacecraft measurements of fields and particle VDs

Virtual spacecraft, in single or multi-point configuration, will be launched through the output of previously selected numerical simulations and will provide synthetic measurements of electric and magnetic fields, particle velocity distributions and their moments (density, mean velocity temperature, heat flux ect.), mimicking real measurements of satellites (time series). In order to launch a virtual satellite across the output data of any kind of simulations, where particle distribution functions and electromagnetic fields are known on grid points, we will implement advanced interpolation techniques in space and time to design the trajectory of the satellite a posteriori (i. e. not during the code run phase) and to collect values of the observables in between adjacent grid points. Trajectories of virtual spacecraft flying through the output of global simulations will be implemented in such a way to reproduce the trajectories of real spacecraft flying into space. These synthetic measurements will be collected and organized in the AIDA database (AIDAdb) and made publicly available.

Searching for OpenAIRE data...


Dynamic Time Warping as a New Evaluation for Dst Forecast With Machine Learning

Author(s): Brecht Laperre, Jorge Amaya, Giovanni Lapenta
Published in: Frontiers in Astronomy and Space Sciences, Issue 7, 2020, ISSN 2296-987X
DOI: 10.3389/fspas.2020.00039

A multi-fluid model of the magnetopause

Author(s): Roberto Manuzzo, Francesco Califano, Gerard Belmont, Laurence Rezeau
Published in: Annales Geophysicae, Issue 38/2, 2020, Page(s) 275-286, ISSN 1432-0576
DOI: 10.5194/angeo-38-275-2020

Identifying Magnetic Reconnection in 2D Hybrid Vlasov Maxwell Simulations with Convolutional Neural Networks

Author(s): A. Hu, M. Sisti, F. Finelli, F. Califano, J. Dargent, M. Faganello, E. Camporeale, J. Teunissen
Published in: The Astrophysical Journal, Issue 900/1, 2020, Page(s) 86, ISSN 1538-4357
DOI: 10.3847/1538-4357/aba527

Electron-Only Reconnection in Plasma Turbulence

Author(s): Francesco Califano, Silvio Sergio Cerri, Matteo Faganello, Dimitri Laveder, Manuela Sisti, Matthew W. Kunz
Published in: Frontiers in Physics, Issue 8, 2020, Page(s) 12, ISSN 2296-424X
DOI: 10.3389/fphy.2020.00317

Local Regimes of Turbulence in 3D Magnetic Reconnection

Author(s): G. Lapenta, F. Pucci, M. V. Goldman, D. L. Newman
Published in: The Astrophysical Journal, Issue 888/2, 2020, Page(s) 104, ISSN 1538-4357
DOI: 10.3847/1538-4357/ab5a86

Detecting Reconnection Events in Kinetic Vlasov Hybrid Simulations Using Clustering Techniques

Author(s): Manuela Sisti, Francesco Finelli, Giorgio Pedrazzi, Matteo Faganello, Francesco Califano, Francesca Delli Ponti
Published in: The Astrophysical Journal, Issue 908/1, 2021, Page(s) 107, ISSN 0004-637X
DOI: 10.3847/1538-4357/abd24b

Visualizing and Interpreting Unsupervised Solar Wind Classifications

Author(s): Jorge Amaya, Romain Dupuis, Maria Elena Innocenti, Giovanni Lapenta
Published in: Frontiers in Astronomy and Space Sciences, Issue 7, 2020, ISSN 2296-987X
DOI: 10.3389/fspas.2020.553207

The Challenge of Machine Learning in Space Weather: Nowcasting and Forecasting

Author(s): E. Camporeale
Published in: Space Weather, 2019, ISSN 1542-7390
DOI: 10.1029/2018sw002061

Energy conversion in turbulent weakly collisional plasmas: Eulerian hybrid Vlasov-Maxwell simulations

Author(s): O. Pezzi, Y. Yang, F. Valentini, S. Servidio, A. Chasapis, W. H. Matthaeus, P. Veltri
Published in: Physics of Plasmas, Issue 26/7, 2019, Page(s) 072301, ISSN 1070-664X
DOI: 10.1063/1.5100125

Statistical Analysis of Ions in Two-Dimensional Plasma Turbulence

Author(s): Francesco Pecora, Francesco Pucci, Giovanni Lapenta, David Burgess, Sergio Servidio
Published in: Solar Physics, Issue 294/9, 2019, ISSN 0038-0938
DOI: 10.1007/s11207-019-1507-6

Decomposition of plasma kinetic entropy into position and velocity space and the use of kinetic entropy in particle-in-cell simulations

Author(s): Haoming Liang, Paul A. Cassak, Sergio Servidio, Michael A. Shay, James F. Drake, Marc Swisdak, Matt R. Argall, John C. Dorelli, Earl E. Scime, William H. Matthaeus, Vadim Roytershteyn, Gian Luca Delzanno
Published in: Physics of Plasmas, Issue 26/8, 2019, Page(s) 082903, ISSN 1070-664X
DOI: 10.1063/1.5098888

Parametric Instability in Two-dimensional Alfvénic Turbulence

Author(s): Leonardo Primavera, Francesco Malara, Sergio Servidio, Giuseppina Nigro, Pierluigi Veltri
Published in: The Astrophysical Journal, Issue 880/2, 2019, Page(s) 156, ISSN 0004-637X
DOI: 10.3847/1538-4357/ab29f5

Current Sheets, Magnetic Islands, and Associated Particle Acceleration in the Solar Wind as Observed by Ulysses near the Ecliptic Plane

Author(s): Olga Malandraki, Olga Khabarova, Roberto Bruno, Gary P. Zank, Gang Li, Bernard Jackson, Mario M. Bisi, Antonella Greco, Oreste Pezzi, William Matthaeus, Alexandros Chasapis Giannakopoulos, Sergio Servidio, Helmi Malova, Roman Kislov, Frederic Effenberger, Jakobus le Roux, Yu Chen, Qiang Hu, N. Eugene Engelbrecht
Published in: The Astrophysical Journal, Issue 881/2, 2019, Page(s) 116, ISSN 0004-637X
DOI: 10.3847/1538-4357/ab289a

Single-spacecraft Identification of Flux Tubes and Current Sheets in the Solar Wind

Author(s): Francesco Pecora, Antonella Greco, Qiang Hu, Sergio Servidio, Alexandros G. Chasapis, William H. Matthaeus
Published in: The Astrophysical Journal, Issue 881/1, 2019, Page(s) L11, ISSN 2041-8205
DOI: 10.3847/2041-8213/ab32d9

Characterizing Magnetic Reconnection Regions Using Gaussian Mixture Models on Particle Velocity Distributions

Author(s): Romain Dupuis, Martin V. Goldman, David L. Newman, Jorge Amaya, Giovanni Lapenta
Published in: The Astrophysical Journal, Issue 889/1, 2020, Page(s) 22, ISSN 1538-4357
DOI: 10.3847/1538-4357/ab5524

ViDA: a Vlasov–DArwin solver for plasma physics at electron scales

Author(s): Oreste Pezzi, Giulia Cozzani, Francesco Califano, Francesco Valentini, Massimiliano Guarrasi, Enrico Camporeale, Gianfranco Brunetti, Alessandro Retinò, Pierluigi Veltri
Published in: Journal of Plasma Physics, Issue 85/5, 2019, Page(s) 1/25, ISSN 0022-3778
DOI: 10.1017/s0022377819000631

Interplay between Kelvin–Helmholtz and lower-hybrid drift instabilities

Author(s): Jérémy Dargent, Federico Lavorenti, Francesco Califano, Pierre Henri, Francesco Pucci, Silvio S. Cerri
Published in: Journal of Plasma Physics, Issue 85/6, 2019, Page(s) 1/19, ISSN 0022-3778
DOI: 10.1017/s0022377819000758

Simulation of Plasmaspheric Plume Impact on Dayside Magnetic Reconnection

Author(s): J. Dargent, N. Aunai, B. Lavraud, S. Toledo‐Redondo, F. Califano
Published in: Geophysical Research Letters, Issue 47/4, 2020, Page(s) 9, ISSN 0094-8276
DOI: 10.1029/2019gl086546