The first research grant shall be developed in close collaboration with Planetek Italia (12 months out of 18 in total), and is related to the following topic:
Machine Learning for Space Weather
The proposed research project is concerned with the study of "Space Weather Phenomena" and the development of knowledge about the mechanisms and effects of solar-derived perturbative phenomena developing in circumterrestrial space and impacting the ionized atmosphere (ionosphere). In the project emphasis is given to the study and modeling of the dynamics of the ionospheric plasma and the electron density irregularities in it on a global scale, in order to improve the capability of long-term (24-48 hours in advance) nowcasting and forecasting of the ionospheric response to Space Weather events over the Mediterranean area. The modeling approach is developed through innovative "machine learning" techniques, recently introduced (Cesaroni et al 2020), the results of which point to this as a strategy to extend the time horizon of ionospheric forecasting, a fundamental requirement for increasing knowledge of Space Weather phenomena in near-Earth space. In addition, the growing demand for semi-empirical approaches for real-time mitigation of errors introduced by the ionosphere on positioning and navigation systems makes the proposed topic a significant contribution in the area of "services and research for society" in relation to the strategic objective "Development of a National Space Weather Service" in the context of developing countermeasures to contain the negative effect that the irregular and disturbed ionosphere can have on technological systems in use in modern society such as, for example, navigation and positioning satellite systems (GNSS, GLobal Navigation Satellite Systems), trans-horizon HF radio communications, and L-band satellite communication systems. Such systems are of interest to a variety of end users who can be identified as users of the service in which the developed products may be embedded. Examples of users may include: precision agriculture operators, operators in the field of mapping, aviation, and radio communications operators for emergency management in civil defense.
Cesaroni, C., Spogli, L., Aragon-Angel, A., Fiocca, M., Dear, V., De Franceschi, G., & Romano, V. (2020). Neural network based model for global Total Electron Content forecasting. Journal of Space Weather and Space Climate, 10, 11.
The second research grant shall be developed in close collaboration with GE Avio (12 months out of 18 in total), and is related too the following topic:
Operative Framework For HPC (Off-HPC)
High-performance computing (HPC) clouds are becoming a complement or, in some cases, an alternative to on-premise clusters for running scientific-technical, engineering, and business analytics service applications. Most research efforts in the area of cloud HPC aim to analyze and understand the cost-benefit of migrating computationally intensive applications from on-premise environments to public cloud platforms. Industry trends show that on-premise/cloud hybrid environments are the natural path to get the best out of on-premise and cloud resources. Workloads that are stable from the point of view of required computing resources and sensitive from the point of view of the need to protect processed information can be performed on on-premise resources, while peak computational loads can take advantage of remote computing resources available in the cloud typically under a "pay-as-you-go" consumption mode. The main difficulties in using cloud solutions to run HPC applications stem from their characteristics and properties compared to traditional cloud services to handle, for example, standard enterprise applications, Web applications, data storage or backup, or business intelligence. HPC applications tend to require more computing power than application services typically delivered in cloud environments. These processing requirements arise not only from the characteristics of the CPUs (Central Processing Units), but also from the amount of memory and network speed to support their proper execution. In addition, such applications may have a particular and different execution mechanism than dedicated cloud application services that instead run 24/7. HPC applications tend to run in batch mode. Users execute a series of computational jobs, consisting of instances of the application with different inputs, and wait until results are generated to decide whether new computational tasks need to be submitted and executed. Therefore, moving HPC applications to cloud platforms requires not only a focus on resource allocation in the infrastructure in use and its optimization, but also on how users interact with this new environment. Research in the area of cloud HPC can be classified into three broad categories: (i) feasibility studies on adopting the cloud to replace or complement on-premise computing clusters to run HPC applications; (ii) performance optimization of cloud resources for running HPC applications; and (iii) services to simplify the use of cloud HPC, particularly for users who are not specialized in data and information processing and processing technologies. This research project intends to focus on study activities within the first category, in which, more specifically, there are four main aspects that should be considered: (i) metrics used to assess how feasible the use of HPC cloud is; (ii) resources used in computational experiments; (iii) computational infrastructure; and (iv) software, which includes both well-known HPC benchmarks and computational tools, algorithms, or methodologies related to specific business application cases. Currently, the company uses HPC applications running mostly on on-premise systems but faces issues related to the need for greater computational resources that can be met through flexible and scalable architectures provided by cloud technologies. The need is to build clear technology and governance references for cloud or hybrid infrastructures. The research project will therefore aim to carefully analyze the state of the art of hybrid HPC solutions, define criteria for benchmarking different solutions, develop an operational framework that includes the operational and economic management aspects of a hybrid HPC solution, and finally implement one or more industrial pilots.
DEADLINE: June 24, 2022 ALL INCLUSIVE GROSS AMOUNT (for 18 months): 29050,50 euro (i.e., 19367 euro annual gross amount)
NOTE: Foreign candidates are strongly encouraged to contact me by email if they need help/support in order to prepare their application: I will be glad to assist.
Here you can download an unofficial English translation of the call: RIPARTI-call
*******************************************************************************************
Prof. Massimo Cafaro, Ph.D. Associate Professor of Parallel Algorithms and Data Mining/Machine Learning Head of HPC Lab https://hpc-lab.unisalento.it Director of Master in Applied Data Science
Department of Engineering for Innovation University of Salento, Lecce, Italy Via per Monteroni 73100 Lecce, Italy
Voice/Fax +39 0832 297371
Web http://sara.unisalento.it/~cafaro Web https://www.unisalento.it/people/massimo.cafaro
E-mail massimo.cafaro@unisalento.it E-mail cafaro@ieee.org E-mail cafaro@acm.org
INGV National Institute of Geophysics and Volcanology Via di Vigna Murata 605 Roma
CMCC Foundation Euro-Mediterranean Center on Climate Change Via Augusto Imperatore, 16 - 73100 Lecce massimo.cafaro@cmcc.it
*******************************************************************************************