My Story

Originally from Argentina, I got my MS in Astronomy from La Plata National University in 2005 and my PhD in Physics from University of Buenos Aires in 2010. During my graduate studies I focused on gravitational field theories and various extensions of General Relativity. Afterwards, I continued my work on theoretical physics and cosmology at Brandeis University. I joined CfA in 2013 to work computational stellar astrophysics. Currently, I am part of the Institute for Applied Computational Science at Harvard University, where I use machine learning algorithms to conduct data driven astrophysical research.   

I am interested in stellar evolution, activity, and rotation, as well as the impact that stars have on their orbiting planets: star-planet interaction. I use magnetohydrodynamic simulations and deep learning to study problems like the spin-down evolution of young, active stars, exoplanetary environments and their habitability, and the orbital evolution of cataclysmic variables. More recently I have been focusing on uncertainty quantification in deep neural networks. In particular, developing probabilistic models and bayesian neural networks to retrieve fundamental stellar properties from observable quantities.

 Studying Stars

    Learning Planets 

CECILIA GARRAFFO

Astrophysicist at the Institute for Applied Computational Science, Harvard University

 

Research

Stellar activity and rotation are two sides of the same coin. Rotation fuels activity by mixing the magnetized plasma in the convective envelope of the star. In turn, the resulting magnetic fields determine the stellar winds through which the star spins down. This self-regulating mechanism results in a very powerful tool, called Gyrochronology, that allows us to convert rotation periods, that we can measure to great precision, into stellar ages, that are very difficult to determine.

The magnetic activity of a star and its evolution is also crucial to the space weather of its close-in orbiting exoplanets, and understanding stellar winds can help us understand the evolution of close binary systems.

Stellar Activity and Rotation

Open cluster observations have shown a bimodal behavior in rotation periods of young, active stars difficult to explain with current models of angular momentum loss. I model stellar winds to understand how the rate at which a star spins down evolves throughout its life. 

Exoplanetary Environments

Stellar winds are a potential threat to close-in exoplanets' atmospheres. I model the coroane of planet hosting stars to asses their "habitability". I have studied the space environment of exoplanets orbiting M-Dwarfs and found that they are likely to suffer from strong atmospheric stripping and evaporation. 

Cataclysmic Variables

The orbital evolution of Cataclysmic Variables (CV), for periods longer than the period gap of ~3.2 hrs, is governed by their angular momentum loss through magnetized winds. I am interested in how the magnetic field alignment and complexity impacts their evolution, potentially explaining the existence of the period gap.

Neural Networks

for Stellar Evolution

Neural Networks are known to outperform traditional methods for highly non linear problems. Stellar evolutionary tracks are an are complex and difficult to reverse in order to retrieve fundamental properties from observables like. I have built a fully connected neural network that fully characterizes a star directly from its observed photometry. In addition, I have built a Bayesian Neural Network to determine uncertainties in the model and predictions. I use this to improve the model where it is needed.  

Skills & Specialties

Machine learning, artificial neural networks, probabilistic programming, bayesian inference, MHD modeling, large data, parallel computing, stellar astrophysics, black hole physics, and exoplanetary environments.

Python, Pytorch, Pyro, IDL, Fortran, Jupyter Notebooks, Git, Colab.

 

Highlights

A Complete Spin-Down Model that reproduces OC observations:

...and also explains the Cataclysmic Variables Period Gap! 

Watch my keynote presentation on Stellar Activity and Rotation at the Einstein Fellowship Symposium: 

Proxima b and TRAPPIST-1 

On the habitability of our temperate terrestrial, rocky neighbors: 

Our work suggests that such close-in planets can be exposed to strong dynamic pressure from the stellar wind, 2000 times stronger than the solar wind pressure at Earth or higher, and to fast variations of this pressure over timescales of days, creating conditions for potentially strong atmospheric stripping, heating and possible evaporation.

Simulation of the magnetic environment around the planet TRAPPIST-1 f

   In addition, we find that close enough planets do not have magnetospheres protecting them from the intense stellar winds and magnetic pressures. Instead, stellar magnetic field lines connect with the planetary field over most of the planets surface, allowing energetic particles to constantly precipitate on to the atmosphere.

Garraffo et al. 2016(arXiv:1609.09076), Physics TodayGarraffo et al. 2017 (arXiv:1706.04617), selected as a AAS NOVA highlight in 2017 and Press release by CfA, 2017.

Watch my ITC talk on this "Proxima b and TRAPPIST-1: Check the Space Weather Before Packing" 

 

I enjoy the fresh perspectives that arise from working with students.

At IACS I am involved in teaching two classes on Data Science (CS109/209A and B) 

While at CfA I have mentored several Masters students on projects ranging from Cataclysmic Variables mass transfer to exoplanets and the history of the Sun

I am a mentor at the YouthAstroNet program and for El Universo a Tus Manos Program for Undergraduate Science Majors at CfA, as well as an advisor for the Talented and Gifted STEM program for Latinas in the Boston Public Schools.

In addition, I am a teaching fellow and an invited Lecturer at Harvard, and I have been a teaching assistant for 6 years in Argentina.

The next generation

 

Short CV

Education

PhD in Physics, Mar 2010, University of Buenos Aires

 

M.S. in Astronomy, Dec 2005, National University of La Plata

Experience

Research Associate, IACS Harvard University,  2018 to present.

Research Associate, Harvard-Smithsonian CfA,  2018 to present.

Postdoctoral Fellow, Harvard-Smithsonian CfA,  2013 - 2018.

Research Associate at Brandeis University, 2010 - 2013. 

Doctoral Fellow of the National Scientific and Technological Research Council  Argentina, 2006 - 2010. 

Teaching experience of 6 years, mentoring for 3 years.

Metrics

35 publications in scientific, peer reviewed journals (13 as a first author)

967 citations (as of Aug 2020)

 

Impact

International Invited Speaker at

XXIII Festival de Astronomia 

Villa de Leyva Colombia

 

Music

I sing and play the bass guitar with Esteparios, a group of friends who get together to play latin music.

Find me here

33 Oxford Street

Cambridge, MA

USA 02138

cgarraffo@seas.harvard.edu

617-998-1468

  • White Twitter Icon
  • White Facebook Icon