WP6: Cross domain initiative and Space Economy

Alessia Tricomi (Università di Catania & INFN)

Antonio Stamerra (INAF)

1

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

WP6 – Proposal goals

• Optimization and adaptation of widely used software packages on the national Centre infrastructure, like Geant4 or FLUKA or generic high-performance techniques for data access/analysis; statistical and AI-based tools; data-interpretations tools.

​

• In the context of the Space Economy Italian Strategy, develop and deploy techniques to access, analyze and process the data from the Mirror Copernicus program, creating the conditions to enable radically innovative services. In particular, enable thorough and continuous observation programs for global and local processes, allowing external partners to operate a large variety of services, including the planning for emergencies, risks and resources.

​

​

2

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

WP6 Organization

• Mailing lists:
• Contact person (one for each Institution): cn1-spoke2-wp6-contacts@lists.infn.it
• All WP6 participants: cn1-spoke2-wp6-all@lists.infn.it

​

• In the table Institution who signed ICSC agreement declaring WP6 participation
• Some missing
• Not all involved in the flagships (see later)
• Some invloved only in IG
• Be-weekly meetings
• Every two weeks we alternate Tuesday at 3pm and Wednesday at 5 pm
• First 2024 meeting on January 10th at 5pm
• https://agenda.infn.it/category/1782/

​

3

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

WP6 Enrolled newcomers

• INFN
• TD 24Months From 1^st May 2023 INFN-Firenze Alessandro Bombini (WP2/WP6)
• TD 24Months From 1^st Oct 2023 INFN-Catania Giuseppe Piparo (WP6/WP2)
• INAF
• PhD From 1^st Nov 2022 INAF-BO Alex Ciabattoni (WP6/WP3)
• TD 24 Months From 1^st Nov 2023 INAF-Rm Michele Mastropietro (WP3/WP4/WP6)
• TD 12 Months From 1^st Jan 2024 INAF-Rm Marco Faltelli (WP4/WP6-IG Leonardo)
• UNICT
• RTDa From 1^st March 2023 Gialex Anestasi (WP6/WP3)
• RTDa From 1^st July 2023 Giuseppe Gallo (WP6/WP3)
• UNIFE
• PhD From 1^st Jan 2023 Alice Magnoni (WP6)

​

​

4

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

WP 2.6 Approach

• Planning and identification: landscape recognition for best solutions for the realization of heterogeneous and portable code (e.g. software frameworks, compilers, programming models, ...), for the integration of services into a data-lake infrastructure; cross domain software and services will be identified if appropriate. Moreover, solutions for handling user support, user fora, and training opportunities will be identified;
• A realization phase, in which the services and the support systems are put into place, at least in alpha/beta phase. These include the testbeds to be used for benchmarking of scientific and industrial solutions, the user support system, the training opportunities.
• A validation phase, in which experience on the supported services and codes are reported, to be used as a touch base before the end of the project.
• A wrap-up phase, in which results are reported for executed activities, and are disseminated via white papers for future and external use cases.

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

WP6 – Landscape

• Simulation packages in Fundamental research
1. A detailed simulation of the physics processes under study (p-p collision at LHC, CR spectrum, GW…)
2. A detailed simulation of “secondary” physics processes (interaction of CR with atmosphere, decay of unstable particles,…)
3. A detailed simulation of the interaction of the primary and secondary particles with the target medium (instrumented detector in standard experiments, biological body in dose related studies,…)
4. A detailed simulation of the readout electronics, where relevant, to mimic signals from real experimental setups

🡪 GEANT4 Flagship 6.2.2

6

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

WP6 – Landscape

• Data management and High-Rate Analysis for non-core Spoke 2 domains
• Large need of computing resources
• Large scale N-to-M (sites and datasets) data management
• WLCG / CERN tools developed for LHC applications ”exported” to many fields
• medical research
• genomic surveysheritage
• Space Economy
• BigData, High rate, secure access and storage
• High-rate analysis flagship 6.2.1 (application also to cultural heritage scenario)
• Link with interactive analysis WP2 flagship

​

7

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

WP6 – Landscape

• Tools for Space Economy “Space Economy” is defined … as the full range of activities and the use of resources that create value and benefits to human beings in the course of exploring, researching, understanding, managing, and utilising space
• Data management and processing tools on the Datalake;
• Image analysis (satellites, drones, …) and their correlation with other data sources using AI techniques;
• Certification of satellite images
• Mainly (but not only) data from satellite used to infer properties at earth/soil level
• (Satellite) Imaging reconstruction flagship 6.2.3
• Strong link with IG (see later and dedicated talks)

​

​

8

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

WP 2.6 General Milestones

• M9-M15: investigations and identification of package(s) to be modernized; report and detailed plan; landscape analysis of solutions for high intensity analyses; choice of proof(s) of concept to be realized;
• M24: pilot implementation and first evaluation of performance for the selected package(s); pilot implementation of the high intensity solution(s);
• M25-M36: benchmarking and testing activities executed; final report including performance gain assessment; white paper produced to the larger scientific and industrial community; showcase of the analysis techniques and white paper for external users

​

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

FLAGSHIP 2.6.1: Extended Computer Vision at high rate

Link to Flagship document: https://docs.google.com/document/d/1gmL_Dxx7nvjPqebmQoIHtcotzbG5a2UB

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

Use case definition

The idea is to perform image segmentation on Spectral datacubes, instead of canonical RGB images;

The idea is to perform datacube segmentation through deep spectral clustering;

and do it fast, on multiple images in a distributed manner.

https://docs.google.com/document/d/1gmL_Dxx7nvjPqebmQoIHtcotzbG5a2UB

Use Case Expected Activities

1. M1-M12 (Sept-2023 - Aug 2024, synced with ICSC MS9): Survey of the State-of-the-Art; tracking of R&D technologies to be used; selection of datasets for use cases (at least one).
2. M13-M24 (Sept 2024 - Aug 2025, synced with ICSC MS10): Implementation of the selected technology(ies); test and validation on selected dataset(s). Proof-of-Concept deployment.

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

Example of Deep Clustering datacubes

- Astrophysics

https://docs.google.com/document/d/1gmL_Dxx7nvjPqebmQoIHtcotzbG5a2UB

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

FLAGSHIP 2.6.2: Enhancing Geant4 Monte Carlo Simulations through Machine Learning Integration

Link to Flagship document: https://docs.google.com/document/d/1gmL_Dxx7nvjPqebmQoIHtcotzbG5a2UB

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

Understanding Geant4

Geant4 is a software toolkit for the simulation of the passage of particles through matter, widely used in physics, astrophysics, and medical physics.

Core Features:

• Robust framework for modeling complex physical processes.
• Exceptional accuracy and precision in simulations.
• Wide range of applications including radiation therapy, space missions, and particle detectors.

​

Computational Demand: Geant4's detailed simulations require significant computational resources.

CPU Load: Due to its complex computations, simulations can be time-consuming, especially when high precision is required.

​

​

year

# of papers

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

Use case definition

Employ Neural Network models to emulate high-density voxel detector outcomes using low-density data (GEANT4DNA)

Accelerated simulation processes. Maintenance of a fine level of spatial resolution with significantly reduced computational demand.

https://docs.google.com/document/d/1TzBjzWeJ7R0RxylZGbJycOQfL6Qmu0qO/

Use Case Expected Activities

1. Phase 1 (M1-M6 | ICSC M13-M18 - MS7): Examine modern technologies and methodologies. Define a simple test case for utilizing ML to accelerate a typical MC Simulation, identifying the Geant4 segment for modification.

​

• Phase 2 (M7-M18 | ICSC M19-M30 - MS9): Implement chosen technologies. Develop and integrate machine learning models with MC simulations. Conduct testing and validation on selected datasets. Construct a Proof-of-Concept (PoC) system. Share codebase in a public repository for collaborative advancement.

​

• Phase 3 (M19-M24 | ICSC M31-M36): Optimize the PoC system. Undertake rigorous testing to ensure reliability and accuracy. Prepare for broader deployment. Explore advanced techniques like Deep Learning, and enhance existing use cases.Disseminate
• findings through conference presentations for peer review and feedback.

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

16

Voxelized Detectors: Utilized to model the spatial distribution of energy deposits in simulations, crucial for accuracy.

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

FLAGSHIP 2.6.3: Satellite Imaging

Link to Flagship document: https://docs.google.com/document/d/1gmL_Dxx7nvjPqebmQoIHtcotzbG5a2UB

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

Use case definition

Employ AI models to analyze (satellite) high-res images for ecosystems monitoring and risk management

Use of high-performance computing resources is pivotal, accelerating image analysis for early symptoms detection and enabling early warning systems

https://docs.google.com/document/d/1TzBjzWeJ7R0RxylZGbJycOQfL6Qmu0qO/

Use Case Expected Activities

• First period ( ICSC month 12-18): Survey of the State-of-the-Art; tracking of R&D technologies to be used; selection of datasets for use cases (at least one)

🡪D1: report on technologies to be used, selection of at least one test dataset.

​

• Second period m19-m22 (corresponding to MS8): first experimentation with data sources and algorithms, demonstration on the feasibility of choices

🡪D2: report on the experimentation and of technical choices; first code repository available

• Third period m23-m36 (corresponding to MS10): Implementation of the selected technology(ies); test and validation on selected dataset(s). Proof-of-Concept deployment.

🡪 D3: Report on the work carried out; release of the developed code on public repository.

• Intermediate report at MS9

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

Example of ongoing work (two dedicated talks later on…)

• Sentinel-2A is a satellite of the Copernicus constellation specifically designed for vegetation monitoring and natural disaster management.

​

• It acquires images in various spectral bands, allowing the calculation of many vegetation indices.

​

• First reproducibility tests were carried out using clustering techniques based on indices of vegetation susceptible to flavescence dorée symptoms.

​

​

19

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

Connection with IG

IG related to WP6 activities

• Hammon - UnipolSAI + SOGEI
• Risk assessment su immagini urbane (UNISALENTO, INAF)
• AGRI - Intesa Sanpaolo
• Analisi di immagini agricole per capire la redditività aziendale (UNICT, UNINA, UNIMIB, UNISAL, UNIBA)

​

​

​

​

• Connection with other IG/WPs
• IDL - Leonardo
• Data management e data certification su dati satellitari (INFN, INAF)

20

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

Connections with other WPs

• WP4 - Innovation grant "IDL" (coordinator Leonardo)
• Blockchain Software acquisition and deployment
• Francesco Visconti (Ric. INAF) and Marco Faltelli (TD INAF)
• Corsika Atmospheric simulator and Simtelarray interface for high-energy ground based experiments and: optimization and engineering
• F. Visconti and Michele Mastropietro (TD INAF)

• WP3
• subproject of the WP3 Flagship UC “Pipeline optimization for space and ground based experiments (PSGE)” - Valentina Fioretti (Ric. INAF)
• (see next slide)

21

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing

GEANT4 simulations in HPC environments, with the simulation of the NASA COSI Anti-Coincidence System (ACS) as a test case

​

• subproject of the WP3 Flagship UC “Pipeline optimization for space and ground

based experiments (PSGE)”

• new methodologies for multi-threading and multi-node computation in a pipeline for Geant4 multi-purpose simulations in HPC architectures while exploring new I/O interfaces (e.g. CAD geometries, databases).
• Bologna Geant4 Multi-Mission Simulator (BoGEMMS) as baseline simulation framework

​

​

• WPs: WP3, WP6
• Flagship UC: WP3.4 PIPELINE
• Members:
• V. Fioretti (INAF OAS Bologna) - coordinator
• A. Ciabattoni (ICSC PhD, UniBo & INAF OAS Bologna)
• S. Lotti (INAF IAPS)

​

​

22

See A. Ciabattoni lightning talk!

https://agenda.infn.it/event/38374/contributions/215725/

Lot of activities on going…

Stay tuned and take part to the meetings!

23

Missione 4 • Istruzione e Ricerca 

ICSC Italian Research Center on High-Performance Computing, Big Data and Quantum Computing