Extreme Value Theory
Classical Extreme value laws have been recently found for orbits of dynamical systems. The theory has been devised for a special class of observables which allow for a link between the concept of extremes and the lack of Poincaré recurrences around a chosen point of the phase space. My contributions in this fields of research are directed towards the understanding of the relation between the recurrences and the extreme events [9], the nite time behavior of asymptotic laws [1,2,4,5] and the generalizations of theoretical aspects to physically meaningful systems [6,8,10,11,13,18].
Critical Phenomena in Complex Systems: Climate, Finance and Epidemiology
The understanding of the mechanism regulating the transitions between different attracting states in complex systems is a general problem in statistical mechanics. Systems which feature critical phenomena range from spin glasses up to finance [20], the climate systems and epidemiology [60]. I have been involved in developing rigorous statistical methods for detecting the transition thresholds in datasets [3,12,17] and in the modeling of systems at bifurcation points via the so called ARMA (Auto Regressive Moving Average) processes technique [15].
Turbulent and Geophysical flows
Providing a statistical description of turbulence, by combining theoretical findings with high quality experimental datasets, is helping in understanding several features of turbulent flows as the dissipation anomaly or the existence of singularities in the Navier Stokes equations [29,43]. I am actually contributing to this research field by developing statistical techniques based on the Extreme Value Theory and the ARMA process analysis which allows for quantify the distance between observations and theoretical models in a rich model-parameter space [14,53].
[73] Katerina Giamalaki, C Beaulieu, SA Henson, AP Martin, D Faranda, H Kassem. Net heat flux responses to extreme atmospheric sea level pressure patterns in the North Pacific. Submitted to Journal of Geophysical Research - Oceans, 2020.
PhD, Earth Sciences and Applied Mathematics
Hamburg University, Hamburg, Germany Mention: 1.0/1.0 magna cum laude January 2013
Thesis: Extreme Value Theory for geophysical flows. [.pdf]
Supervisor: Prof. Valerio Lucarini.
Master of Science, Physics
Bologna University, Bologna, Italy Mention: 110/110 cum laude July 2010
Thesis: Dynamical indicators of Stability in Non-Linear maps of low dimensionality. [.pdf]
Supervisor: Prof. Giorgio Turchetti.
Bachelor Degree, Atmospheric Physics and Meteorology
Bologna University, Bologna, Italy Mention: 110/110 cum laude July 2008
Thesis: Characterization of stable and radioactive atmospheric tracers at the GAW M.te Cimone Station [.pdf]
Supervisor: Prof. Laura Tositti.
PROFESSIONAL EXPERIENCE
London Mathetmatical Laboratory - September 2017 - Now
External Fellow
Climate Dynamics, Dynamical Systems, Turbulence.
CNRS - October 2015 - Now
Researcher (Chargé de Recherce)
Extreme weather events, Singularities in Navier Stokes equations, Financial Crises.
LSCE - CEA October 2014 - September
Research Fellow (CDD). Grant director: Dr. Pascal Yiou.
-Reasearch activities: Thermodynamics and Mechanisms of geophysical
flow. Study of the weather analogues with techniques of dynamical systems.
CNRS URA 2464 / SPEC - CEA April 2013 - September 2014
-2014 Distinguished citizen of the year of the town S.Agata Militello (Italy). Prize assigned under the patronage of the Ministry of Home affairs.
-2014 Winner of the Outstanding Student Poster (OSP) Awards at the European Geophysical Union general assembly - section Nonlinear Processes in Geosciences. Vienna, April 2014
-2013 Selected for participating in the 1st Heidelberg Laureate Forum: a one-week event combining scientific, social and outreach activities featuring the winners of the awards in Mathematics and Computer Science, the Abel Prize, the Fields Medal (including the Nevanlinna Prize), and the ACM Turing Award, September .
-2012 Winner of a sponsorship within the HPC-Europa2: Pan-European Research Infrastructure for High Performance Computing which covered a two months visit to LadHyX (Ecole Polytechnique, Palaiseau, France) to collaborate with the Dr. P. Manneville and an amount of 50.000 CPU hours to perform simulations on fluid dynamic systems on CINES-GENCI facilities located in Montpellier, France.
INSTITUTIONAL RESPONSABILITIES
-Elected member of the LSCE laboratory council.
-EGU Nonlinear Processes in Geophysics, Scientific Officer.
-Organiser of the bimestrial inter-laboratory seminars “Climate and Statistical Mechanics”, between the laboratories SPEC & LSCE at CEA Saclay, France.
INTERESTS
Cooking, classical music, building-bricks design.
NORTH ATLANTIC ATMOSPHERIC CIRCULATION INTERACTIVE ANALYSIS
Download the daily values of dimension and persistence from 1948 to present*
The figure displays some analyses performed using the instantaneous dimension d - the higher d, the more unpredictable is the atmospheric circulation - and the persistence θ - the lower θ the more stable is the atmospheric circulation - of the sea level pressure field (in hPa) extracted from the NCEP Database. Please cite [30,38] for research use. a) Sea-level pressure fields over the North-Atlantic showing the domain of the analysis for the selected day. b) Distribution of the best 2% analogues per decades. c) Distribution of the best 2% analogues per month of the year. d) Scatter plots of dimension d and inverse persistence θ for all the data (ligth gray), the data of the month considered (dark gray) and the day of the analysis (red star). e) Weather regimes, computed as the days beyond the 0.15 quantiles of the d and θ distributions.
* Last available date depends on the NCEP and GFS updates. Thanks to M Carmen Alvarez Castro for providing the R code, and Yoann Robin for the Python Package.
COVID 19 DAILY FIT
DISCLAIMER: the code used for the forecast is adapted from the fitvirus program by Copyright (c) 2020, milan batista. These predictions are a research product and should not be used in any way as an actual forecast of the epidemics. This forecast does not replace or contradict any estimates reported by official sources from the French governement, that will be far more accurate since based on additional information. Please refer to the official French governement website to know how to behave during the Covid-19 epidemics. I decline any responsability for the use of this analysis without specifying the disclaimer.
The figure displays the fit of the covid19 epidemics data from France (downloadable here) with a logistic model. The generic expression of this model is y=c/(1+exp(-(t-b)/a)), where the variable t is the time and a refers to the infection speed, b is the day with the maximum infections occurred and c is the total number of recorded infected people at the end of the infection. Solid lines display the best prediction based on the actual total number of cases in France, dotted lines the prediction in case data are affected by a 30% error in the last data. Black display the cumulative total number of infected people. Red displays the number of new infections per day. Dotted blue is an estimate of the epidemic peaks. Data updated daily, in the evening. Forecasts for the previous days are reported in this tweet. The model shows extremely high sensitivity to the last data point and enormous fluctuations of the total infected patients. Estimates should be handled with extreme care to forecast the long term behavior of the epidemics
01.69.08.85.41
PHoCEA DRF - Tous droits réservés - Mentions légales - Protection des données personnelles - Dernière mise à jour : 29-05-2020