Summer Scholars tackle big data to explore galaxy evolution
Associate Professor of Physics and Astrophysics Danilo Marchesini, Ph.D. candidate Kalina Nedkova,
and senior Erin Kado-Fong collaborate on their Summer Scholars project. (Alonso Nichols/Tufts University).
Managing the output of astronomical data presents a challenge to
researchers: currently, big data in the field of extragalactic
astronomy require human collection and interpretation. But in the
coming era of even bigger data, this sort of laborious process will
not be efficient enough.
"Using modern computing techniques to extract and analyze
astronomical data is becoming increasingly important as we enter the
era of 'big data' astronomy," says rising senior Erin Kado-Fong,
astrophysics major in the School of Arts and Sciences. This summer,
Kado-Fong and Kalina Nedkova, a doctoral candidate in physics in the
Graduate School of Arts and Sciences, are immersed in a project to
create a suite of tools that will allow for the unsupervised
extraction of photometric data for a large number of sources. Kado-Fong
and Marchesini enlisted Nedkova's assistance in collecting data that
would create a benchmark for the tools Fong is developing.
"Because light travels at a finite speed, light from galaxies that
are farther away takes longer to reach us, allowing us to directly
observe galaxies at different points of cosmic time," Kado-Fong
explains. "These data allow us to flesh out the evolutionary paths
of various types of galaxies.
Danilo Marchesini, an associate professor in the Department of
Physics and Astronomy, is Kado-Fong and Nedkova's project advisor as
part of Tufts Summer Scholars, a program that funds students who
delve into their academic passions with the support of a professor.
Nedkova received funding as a Tufts Graduate Summer Scholar, a new
scholarship to encourage open collaboration across various Tufts
schools and departments.
"In the past, many undergraduate Summer Scholars worked informally
with graduate students in their departments or labs, and we wanted
to offer further support in terms of finances and mentoring
resources," says Graduate School of Arts and Sciences Associate Dean
Sinaia Nathanson. "While in the lab, both students develop or deepen
ongoing relationships, and we believe that they are stronger
researchers and scholars because of this experience."
Using tools—some that Marchesini and his collaborators have
developed and optimized, and others that have been publicly
released—Kado-Fong is developing an automatic pipeline that applies
all of the tools to large datasets without the need of human
intervention. "Erin has also been improving these tools, as well as
making the whole machinery computationally fast and adapting it to
run on super-computers, such as the computing cluster at Tufts,"
says Marchesini. "This pipeline is applicable to large area surveys
to construct catalogs with exquisite quality as quickly as
Kado-Fong, Nedkova, and Professor Marchesini examine a mosaic of cosmic images in the lab. (Alonso Nichols/Tufts University).
While the current goal of the project is to apply these tools to
today's available astronomical surveys, the team's work could also
prepare for the arrival of the Large Synoptic Survey Telescope
(LSST), a survey telescope to be released in the next decade that
can process an unprecedented amount of astronomical data. With the
capacity to map the entire visible sky in just a few nights, "the
LSST will allow us to image millions of stars and galaxies using
several different filters," notes Kado-Fong.
The research grew out of Kado-Fong's previous collaboration with
Marchesini on an academic paper that explored images of galaxies in
the early universe. "His research focuses on galaxy evolution and
galaxies in the early universe, and my projects with him have been
centered around these topics," she explains. She wanted to combine
her previous work with a data-driven direction due to the inevitable
release of the LSST, which is predicted to process fifteen terabytes
of data every night.
Marchesini says Kado-Fong was a natural choice for a student
research partner. As an undergraduate, Kado-Fong was enrolled in his
graduate class, Stellar Structure and Evolution, and was conducting
research in galaxy formation and evolution under his supervision.
"When that project was coming to a conclusion, says Marchesini, "we
looked ahead. Erin had already worked with large data sets [studying
stars at Harvard last summer]. Under my supervision, Erin learned
how to reduce and analyze near-infrared spectra of distant
ultra-massive galaxies and how to determine their properties,
answering questions such as: How big are these galaxies? How old are
their stars? Are they still forming new stars? Do they host actively
growing super-massive black holes? Now, Erin is taking advantage of
the expertise and skills she developed during the two research
Nedkova is creating a mosaic of the images from data currently
available from the Dark Energy Survey (DES), a collaborative effort
to map millions of galaxies and find patterns in cosmic structure.
Although her area of study has focused on physics, Nedkova likes the
challenge of tackling big-data coding and computer science for this
project. "I convert DES data to usable, science-quality images,"
says. "I look at the coordinates of each picture and place it where
it belongs. It's useful because the images I produce will constitute
the perfect test bench for Erin to assess her new tools."
Nedkova's first research project at Tufts "continues to be a
valuable and rewarding experience," she says. "It has helped me
develop many of the skills needed for research in astronomy. Erin is
already familiar with a lot of astronomical tools, so collaborating
with her was a great way to start my research career."
While the three researchers say the project is in the developmental
stage, Kado-Fong notes that she's very proud of what they have
accomplished together to date. "I love setting up infrastructure
with a plan in mind," she says. "Code tends to get out of hand a
little bit at a time, so it's really rewarding to see that the
program I set up is robust enough to be flexible and adapt."
Looking ahead, Marchesini says the tools and infrastructure Kado-Fong
has developed will be applied to the construction of
multi-wavelength photometric catalogs from the NEWFIRM Medium-Band
Survey II, funded by NASA through their Astrophysics Data Archive
Program, and the project has received additional funding from the
Massachusetts Space Grant Consortium. Some tools are being applied
to ongoing and next generation surveys, such as the Hubble Frontier
Fields dataset, aiming at exploring the first billion years of
cosmic history and finding candidates of the first galaxies that
formed in the Universe, which will be among the objects the James
Webb Space Telescope will observe after its launch into space in
2018. "We expect the infrastructure to be fully developed and tested
by the end of 2016," says Marchesini, "and will continue to be used
for many years to come."