Astrophysics, transients, supernovae, type Ia supernova, spectroscopy, radiative transfer, numerical simulations, statistical methods (e.g. Principal Component Analysis), machine learning.
Type Ia Supernovae are among the most luminous transients in our local Universe and hold a good degree of homogeneity. Their homogeneity allowed to use them as standard candles, to measure the distance of remote galaxies and study the expansion flow of the Universe. This led to the Nobel Prize awarded discovery of the Dark Energy, a mysterious repulsive fifth force that acts on the largest scales. Although Hydrogen and Helium are by far the most common elements in the Universe, Type Ia Supernovae are among the few astrophysical objects with a negligible amount of such elements. From an anthropocentric point of view, they are the most important mechanism for spreading intermediate and heavy mass elements that now constitute the Earth and our bodies.
Albeit being relatively common, quite homogeneous and significantly bright, the precise nature of Type Ia Supernovae is not clearly determined. In particular the nature of the progenitor of the explosion is heavily discussed and uncertain. They are widely accepted to be the thermonuclear explosion of an interacting white dwarf. But the mass of the white dwarf at the time of the explosion, the nature of the companion, and the explosion mechanism still remain uncertain.
Current Research: SN Ia, Radiation Transport.
I have experience in radiation transport by Monte Carlo methods. I studied the structure of the peculiar SN1991T (Sasdelli et al., 2014) using
the "abundance tomography" technique. Abundance
tomography is a method developed to study the stratification structure of supernovae fitting time
series of spectra.
The aim is understanding the reasons behind the spectroscopic diversity among SNe Ia with
small Δm15(B), that is bright SNe Ia.
Current Research: SN Ia, Statistical methods.
I have a good experience with datamining techniques. I developed a new
framework that creates a metric space to study large collections of SN Ia spectral series. I used
an improved version of the so called principal component analysis (PCA) to train such a metric
space.
With it I reduced the size of the large
parameter space of spectral series of SN Ia to few significant coefficients.
I used the derivative of
the flux over the wavelength to circumvent problems due to reddening and flux calibration. This simple idea was not used in Astronomy before and is the most
important idea in my work.
It is important to note that the derivative
operation does not remove any of the information present in the spectra (apart from a global
normalization).
The second part of the analysis is devoted to give physical meaning to the principal components.
I used a modern statistical tool called Partial Least Square regression (PLS). This tool allows to
predict a group of quantities called "responses" from a group of quantities called "predictors". It
is a standard technique in chemometrics and completely new in Astrophysics. The predictors are the components
from the PCA space (i.e. the spectral series) and the responses are light curve parameters and
standard spectral indicators as the intrinsic B − V color at maximum, the absolute Bmagnitude,
the Δm15(B). In practice, it means predicting these quantities using the
spectral series, without the need to estimate the reddening or the distance for the SN.
Work in progress: SN Ia, Statistical methods.
Studying methods to predict physical quantities using SN Ia spectra has opened a significant number
of promising applications. Using publicly available SN Ia spectra I have started to investigate some
of them:
 I am studying a new way to compare model predictions with observations in a systematic way using my metric space for type Ia supernova spectra. I am investigating realizations of the three major classes of explosion models which are presently discussed: delayeddetonation Chandrasekharmass explosions, subChandrasekhar mass detonations, and doubledegenerate mergers, and compare them with data.
 Disentangling color (curves) from reddening using the information of SN Ia spectra as a predictor
for the intrinsic color. This can be done in many different photometric bands and can increase our
knowledge of SN Ia reddening law(s).
 The PLSframework makes possible statistical studies of the relation between common properties
of SN Ia and their spectral characteristics. E.g. narrow circumstellar/interstellar
lines, host galaxy properties, evolution with redshift of the distribution of spectral properties of
SN Ia.
 It can allow to study systematically HVFs (High Velocity Features).
These are
components of the lines of some elements (e.g. CaII H&K) with a significantly higher velocity than
the photoshere and are a challenge to models. These feature, at early
epochs, are present in the large majority of SNe Ia.
 The techniques developed for SN Ia spectra are quite general and could be applied whenever
one has large databases of astronomical spectra.
