Geophysical Journal International, Volume 230, Issue 1, July 2022, Pages 29–49, https://doi.org/10.1093/gji/ggac042
Title:Machine learning based tool for building damage assessment.
Abstract: On April 6, 2009, a strong earthquake (6.1 Mw) struck the city of L’Aquila, which was severely damaged as well as many neighboring towns. After this event, a digital model of the region affected by the earthquake was built and a large amount of data was collected and made available. This allowed us to obtain a very detailed dataset that accurately describes a typical historic city in central Italy. Building on this work, we propose a study that employs machine learning (ML) tools to predict damage to buildings after the 2009 earthquake. The used dataset, in its original form, contains 21 features, in addition to the target variable which is the level of damage. We are able to differentiate between light, moderate and heavy damage with an accuracy of 59%, by using the Random Forest (RF) algorithm. The level of accuracy remains almost stable using only the 12 features selected by the Boruta algorithm. In both cases, the RF tool showed an excellent ability to distinguish between moderate-heavy and light damage: around the 3% of the buildings classified as seriously damaged were labeled by the algorithm as minor damage.
F. Di Michele, E. Stagnini, D. Pera, B. Rubino, R. Aloisio, A. Askan & P. Marcati Natural Hazards volume 116, pages3521–3546 (2023)
Poster Presenter: Ourania Giannaopoulou (GSSI)
Title: Modeling intertwining of growing shoots
Abstract: In recent years there has been a growing interest in developing biomimetic robotic devices using plants as prototypes. In particular, climbing plants capitalise on other structures to navigate through different territory and in some cases they intertwine two or more shoots forming braid-like structures through a phenomenon called intertwining. This strategy evolved by plants, is considered as a very effective way to improve their material properties and also their stability. In this work we present a mathematical model for intertwining that is based on morphoelasticity, a framework that allows us to take into account elastic effects on the plant's evolution by modeling the shoots as thin continuum Kirchhoff rods. The proposed approach can accommodate different mechanisms common among plant shoots: circumnutation, a circular periodic motion; gravitropism, the reaction to the perception of gravity; proprioception, a straightening mechanism; allotropism, sensing of other shoots; and finally point (such as a point light source) and directional tropism (such as sunlight). The model presented takes into account relevant aspects of plant physiology such as a variable bending stiffness along the shoot as well as the interplay between the different mechanisms during intertwining therefore making it amenable to bioinspired robotic applications.
Abstract: The quantum Navier-Stokes equations describe a compressible fluid flow subject to a density dependent viscosity and a dispersive stress tensor. The system can be seen as viscous correction of the Quantum Hydrodynamic system (QHD) arising e.g. as prototype model in the description of superfluidity and Bose-Einstein condensates. We consider the QNS system on the whole space with non-trivial farfield behaviour providing the suitable framework to study coherent structures and singular limits.First, we prove global existence of finite energy weak solutions. Second, we investigate the low Mach number limit for the proof of which an accurate dispersive analyis of acoustic oscillations in quantum fluids is needed (Bogoliubov dispersion relation and adapted Strichartz estimates).Based on joint works with P. Antonelli, P. Marcati (GSSI) and S. Spirito (Università dell’ Aquila).
Poster Presenter: Lucia Nasti (GSSI)
Title: Searcher-Shoot: a Reinforcement Learning Approach to understand climbing plant growth
Abstract: Plants' structure is the result of constant evolution towards the adaptation to the surrounding environment.From this perspective, our goal is to investigate the growth behavior of a peculiar plant structure, namely the searcher shoot, by providing a Reinforcement Learning (RL) environment, that we called Searcher-Shoot, which considers the mechanics due to the mass of the shoot and leaves. Assuming that the plants can maximize their length, avoiding a maximal stress threshold, we explore whether the mass distribution along the stem is efficient, formulating this hypothesis as a Markov Decision Process (MDP). By exploiting this strategy, we are able to mimic and thus understand the plant's behavior, finding that shoots decrease their diameters smoothly in order to distribute the mass efficiently. The strong agreement between our results and the experimental data allows us to remark on the strength of our approach in the analysis of biological systems traits.
Title:Historical earthquakes in the L’Aquila Area: 1461, 1703, 1762
Abstract: The aim of this work is to propose a physics-based simulation of three historical earthquakes that hit the surrounding area of the city of L’Aquila (Abruzzo central Italy) in 1461 , 1703 and 1762, with magnitudes 6.4 Mw , 6.7Mw and 6.0 Mw, respectively. Two events (1461 and 1762) are placed, by the available literature, on the fault structure named San Pio delle Camere , whereas the third events had epicenter in Montereale (AQ) area. For all the events the physical parameters characterizing the earthquakes such as fault plane, epicenter, and magnitude are considered to be fixed. Starting from them three stochastic rupture scenarios are generated from each earthquake using three different slip distributions for the SanPio delle Camere 1461 and 1762 events. In a similar way four stochastic rupture scenarios are generated using four different slip distributions for the Montereale 1703 event. The generated scenarios were evaluated in relation to the possibility to reproduce the macroseismic intensity field available from the historical catalogs. The simulated values of peak velocity are used to derive the value of the macrosiesmic intensity obtained by a suitable empirical relationship specifically derived for Italy. For the numerical simulations we used a three-dimensional soil model used and validated in a previous study related to the 2009 L’Aquila earthquake. For the SanPio delle Camere area events the considered slip distributions are able to reproduce quite well the macroseismic effect of the 1461 earthquake. While none of the three scenarios developed satisfactorily reproduce the 1762 earthquake. For the Montereale event from our study, although restricted to only four simulated scenarios, we demonstrate that it is possible to obtain a satisfactory reconstruction of the ground shaking, employing a stochastic source constrained on a limited amount of ex-ante information. Indeed all the considered slip distributions are able to reproduce quite well the macroseismic effect of the 1703 earthquake. Among the four selected scenarios, the one named as SC3 and that corresponds to a maximum slip in the central part of the fault structure, better approximates the intensity of the damage in comparison with the available data from historical catalogs. However this is not sufficient to definitively constrain the slip distribution in a deterministic way, but it provides information about the most likely slip. To investigate the possible correlations between the MI and slip distribution it is necessary to analyze many recent earthquakes for which both slip distributions and damage patterns are available. This is the subject of a new study currently underway.
Accepted for publication at International Conference on Computational Science and its Applications 2023.
2) Insights on the 1703 L’Aquila earthquake using 3D physics-based numerical simulations.
Univaq-GSSI-INGV-Polimi. In preparation.
Poster Presenter: Giacomo Vecchiato (GSSI)
Title: An optimal control approach to the problem of the longest self-supporting structure
Abstract: The characterisation of the self-supporting slender structure with the furthest length is of interest both from a mechanical and biological point of view. Indeed, from a mechanical perspective, this classical problem was developed and studied with different methods, for example, similarity solutions and stable manifolds among others. However, none of them led to a complete analytical solution. On the other hand, plant structures such as tree branches or searcher shoots in climbing plants can be considered elastic cantilevered beams. In this paper, we formulate the problem as a non-convex optimisation problem with mixed state constraints. The problem is solved by analyzing the corresponding relaxation. With this method, it is possible to obtain an analytical characterization of the cross-section.
Poster Presenter: Delyan Zhelyazov (Universidad Nacional Autonóma de México)
Title: Existence and stability of traveling waves in quantum hydrodynamics with viscosity
Abstract: We investigate existence and stability of shock profiles for 1D compressible Euler equation with dispersion and linear viscosity terms in the context of quantum hydrodynamics. Existence of possibly non-monotone shock profiles is proved without any restriction on the shock amplitude. The spectral problem of the linearized system around the traveling wave is also considered. We derive a sufficient condition for the stability of the essential spectrum, and using an Evans function technique we obtain bound for the absolute value of possible unstable eigenvalues. Using numerical computations of the Evans function we provide numerical evidence for the stability of the point spectrum of an oscillatory profile. Moreover, for a subset of parameters leading to monotone profiles the stability of the point spectrum is proved analytically.
Poster Presenter: Hao Zheng (Chinese Academy of Science)
Title:On 1D collisional QHD system and the relaxation-time limit
Abstract: In our recent work, we consider the 1-dimensional collisional Quantum Hydrodynamics (QHD) system and its relaxation-time limit. The well-posedness of the Cauchy problem of QHD system is established for solutions with strictly positive density in the space of finite energy and finite generalised chemical potential (GCP), which is developed in the authors’ previous work [Antonelli, Marcati, Zheng, CMP 2021] for the non-collisional QHD system. The estimates provided by the total energy and the GCP functional allow us to consider the dissipation and the relaxation-time limit of the system. Last, we show an explicit convergence rate by using the method of relative entropy.