Search

Artificial intelligence is paving the way for a new era of algorithms focusing directly on the information contained in the data, autonomously extracting relevant features for a given application. While the initial paradigm was to have these applications run by a server hosted processor, recent advances in microelectronics provide hardware accelerators with an efficient ratio between computation and energy consumption, enabling the implementation of artificial intelligence algorithms ‘at the edge’. In this way only the meaningful and useful data are transmitted to the end-user, minimising the required data bandwidth, and reducing the latency with respect to the cloud computing model. In recent years, European Space Agency is promoting the development of disruptive innovative technologies on-board Earth Observation missions. In this field, the most advanced experiment to date is the Φ-sat-1, which has demonstrated the potential of Artificial Intelligence as a reliable and accurate tool for cloud detection on-board a hyperspectral imaging mission. The activities involved included demonstrating the robustness of the Intel Movidius Myriad 2 hardware accelerator against ionising radiation, developing a Cloudscout segmentation neural network, run on Myriad 2, to identify, classify, and eventually discard on-board the cloudy images, and assessing of the innovative Hyperscout-2 hyperspectral sensor. This mission represents the first official attempt to successfully run an AI Deep Convolutional Neural Network (CNN) directly inferencing on a dedicated accelerator on-board a satellite, opening the way for a new era of discovery and commercial applications driven by the deployment of on-board AI.

The true color composite is a widely used Earth observation product for displaying satellite imagery. As it is often used in communication with non-expert audiences in various media, it is desirable that such a product approximates the color perceived by the human eye. Additionally, as the Sentinel-2 mission with its high resolution multispectral imagery and short revisit times is delivering unprecedented amounts of data, any algorithm for computing the composite should be efficient. In this paper we define the natural color product, propose two efficient approaches for computing it, analyze the results, and implement the products on a satellite imagery service for interactive use. Our algorithms work on a per-pixel basis and hence parallelize naturally. The presented approaches are general and not limited to Sentinel-2 data.

We introduce a theoretical framework for the calculation of uncertainties affecting observables produced by Monte Carlo particle transport, which derive from uncertainties in physical parameters input into simulation. The theoretical developments are complemented by a heuristic application, which illustrates the method of calculation in a streamlined simulation environment., 12 pages, 9 figures

We present advances in the development of methods to predict the effect that uncertainties in physical data needed for generic Monte Carlo simulations induce on the observables resulting from the simulation. Under wide conditions the PDF for the input (uncertain) physical data determine univocally the PDF for the output of the simulation: this general result has clear and powerful applications both in the case of a single (or few) uncertain input data and in the case of many unknown variables, provided they are totally correlated or uncorrelated. Although this is a typical example of forward propagation problem, our results can be applicable also in the backward case, for instance when one is trying to assess conditions for a robust design.

Radioactive decays are of concern in a wide variety of applications using Monte-Carlo simulations. In order to properly estimate the quality of such simulations, knowledge of the accuracy of the decay simulation is required. We present a validation of the original Geant4 Radioactive Decay Module, which uses a per-decay sampling approach, and of an extended package for Geant4-based simulation of radioactive decays, which, in addition to being able to use a refactored per-decay sampling, is capable of using a statistical sampling approach. The validation is based on measurements of calibration isotope sources using a high purity Germanium (HPGe) detector; no calibration of the simulation is performed. For the considered validation experiment equivalent simulation accuracy can be achieved with per-decay and statistical sampling.

The simulation of radioactive decays is a common task in Monte-Carlo systems such as Geant4. Usually, a system either uses an approach focusing on the simulations of every individual decay or an approach which simulates a large number of decays with a focus on correct overall statistics. The radioactive decay package presented in this work permits, for the first time, the use of both methods within the same simulation framework - Geant4. The accuracy of the statistical approach in our new package, RDM-extended, and that of the existing Geant4 per-decay implementation (original RDM), which has also been refactored, are verified against the ENSDF database. The new verified package is beneficial for a wide range of experimental scenarios, as it enables researchers to choose the most appropriate approach for their Geant4-based application.

We present results from a recently launched project to study computational issues related to the quantification of non statistical uncertainties in numerical (Monte Carlo) simulations: they derive from different areas of the process of simulation[1], like e.g. epistemic uncertainties[2], experimental errors in physical data, error propagation from the employed numerical algorithms, etc., This paper addresses the development of methods to predict the effects of a set of correlated, partially correlated or uncorrelated physical uncertainties on the observables produced in a Monte Carlo simulation. It also provides some insight on the computational effort needed and on the possible software solutions to be implemented in the kernel of Monte Carlo codes to facilitate the quantification of uncertainty in experimental use cases.

Ongoing investigations for the improvement of Geant4 accuracy and computational performance resulting by refactoring and reengineering parts of the code are discussed. Issues in refactoring that are specific to the domain of physics simulation are identified and their impact is elucidated. Preliminary quantitative results are reported., To be published in the Proc. CHEP (Computing in High Energy Physics) 2012

The Statistical Toolkit is an open source system specialized in the statistical comparison of distributions. It addresses requirements common to different experimental domains, such as simulation validation (e.g. comparison of experimental and simulated distributions), regression testing in the course of the software development process, and detector performance monitoring. Various sets of statistical tests have been added to the existing collection to deal with the one sample problem (i.e. the comparison of a data distribution to a function, including tests for normality, categorical analysis and the estimate of randomness). Improved algorithms and software design contribute to the robustness of the results. A simple user layer dealing with primitive data types facilitates the use of the toolkit both in standalone analyses and in large scale experiments., To be published in the Proc. of CHEP (Computing in High Energy Physics) 2012

Several models for the simulation of photon elastic scattering are quantitatively evaluated with respect to a large collection of experimental data retrieved from the literature. They include models based on the form factor approximation, on S-matrix calculations and on analytical parameterizations; they exploit publicly available data libraries and tabulations of theoretical calculations. Some of these models are currently implemented in general purpose Monte Carlo systems; some have been implemented and evaluated for the first time in this paper for possible use in Monte Carlo particle transport. The analysis mainly concerns the energy range between 5 keV and a few MeV. The validation process identifies the newly implemented model based on second order S-matrix calculations as the one best reproducing experimental measurements. The validation results show that, along with Rayleigh scattering, additional processes, not yet implemented in Geant4 nor in other major Monte Carlo systems, should be taken into account to realistically describe photon elastic scattering with matter above 1 MeV. Evaluations of the computational performance of the various simulation algorithms are reported along with the analysis of their physics capabilities.

ISICS, originally a C language program for calculating K-, L- and M-shell ionization and X-ray production cross sections from ECPSSR and PWBA theory, has been reengineered into a C++ language class, named ISICSoo. The new software design enables the use of ISICS functionality in other software systems. The code, originally developed for Microsoft Windows operating systems, has been ported to Linux and Mac OS platforms to facilitate its use in a wider scientific environment. The reengineered software also includes some fixes to the original implementation, which ensure more robust computational results and a review of some physics parameters used in the computation. The paper describes the software design and the modifications to the implementation with respect to the previous version; it also documents the test process and provides some indications about the software performance., Preprint submitted to Computer Physics Communications

The Coster–Kronig enhancement factor calculation for M3 shell x-ray production cross sections was found to be incorrect in both ISICSoo class (Batic et al. (2012) [1]) and isics program (Cipolla (2013) [2]). The affected functions of ISICSoo class have been corrected. The resulting X-ray production cross sections are modified by less than 15%, while ionization cross sections are unchanged.