FROM GRAPHENE TO GALAXIES

Graduate Physics
at MIT

M A S S A C H U S E T T S I N S T I T U T E O F T E C H N O L O G Y
Janet Conrad Lead Professor and Originator of Course
Janet Conrad grew up in Wooster, Ohio. She became excited about particle physics
while at Swarthmore College. After graduating in 1985, she attended Oxford University
for her M.Sc. and Harvard for her Ph.D. She then spent 15 years with Columbia
University, as a postdoc and then faculty. She moved to MIT in 2008, where she
developed the 8.226 class. She studies neutrinos, a particle in the Standard
Model with distinctly non-standard behavior. She likes to search for and follow
up on experimental anomalies in neutrino physics, because she is sure nature
is trying to tell us something new – if we just listen closely enough.

Krishna Rajagopal Professor of Course

Krishna Rajagopal is a theoretical physicist who asks how the quarks that in ordinary
matter are found confined within protons and neutrons behave in extraordinary
conditions – like the trillions-of-degrees-hot quark soup that filled the universe
microseconds after the Big Bang, which can provide a testbed for understanding
how a complex world can emerge from simple underlying laws. Krishna grew up
in Toronto, graduated from Queen’s University in 1988, and completed his PhD
at Princeton in 1993. After stints at Harvard and Caltech, he joined the MIT
faculty in 1997. He is currently the Chair of the MIT Faculty.

Michelle Tomasik Physics Teaching Assistant

Michelle Tomasik is a Digital Learning Lab Fellow in the Physics Department at
MIT where she works on creating and evaluating digital content for EdX physics
classes and on-campus introductory physics. Michelle grew up on a farm in upstate
New York, studied chemistry and physics at Swarthmore College, and earned her
Ph.D. in physics at MIT studying interface effects on solar cell efficiency. She and
Janet have now taught two classes together: introductory electricity and magnetism,
and this class, 8.226. She has also taught introductory physics with Krishna.

Joshua Sokol Writing Teaching Assistant

Joshua Sokol is a freelance science journalist in Boston. He provided guidance
in writing for 8.226. His work has appeared in New Scientist, The Atlantic,
The Wall Street Journal, and elsewhere. Like Janet and Michelle, he attended
Swarthmore College, where he double majored in astronomy and English literature.
After undergrad, he worked for a few years as a data analyst for the Hubble Space
Telescope’s Advanced Camera for Surveys, and then pursued his master’s degree
in science writing from the MIT. Currently, he covers astronomy and other disciplines
of natural history.

2 DEPARTMENT OF PHYSICS | MASSACHUSETTS INSTITUTE OF TECHNOLOGY

Cover: Dark Matter Detective Lina Necib by David Sella. Inside: All photographs by David Sella.
Introduction
I think of the MIT Physics Department as being like one of the truly
great, magical cities of the world–like New York City, where I lived
for many years, or London, Paris, Jerusalem, Bombay, or Tokyo.
It is a vibrant, thriving and inspiring place to pursue science.
Because of its size, it is organized into four separate Divisions, each
its own borough. We have Experimental Nuclear & Particle Physics,
Theoretical Physics, Astrophysics, and our own alphabet city:
the Division of Atomic Physics, Biophysics and Condensed Matter.
Each borough seems to have its own distinct scientific style, and yet
when you look closely, you find they are united. Intellectual streets
connect particle physics to astrophysics through the neighborhood
of Dark Matter; astrophysics to atomic physics by way of the LIGO
District; and atomic physics to theoretical physics through the
(very hipster) Quantum Information Theory Quarter.
Because of its size and diversity, it is hard to sum up a great city in
just one picture. Instead, one needs a series of snapshots. That is what
this project is about. This booklet provides a snapshot of MIT Physics
in the spring of 2016, through a series of articles on the research of
graduate students across the department. Each story is written by
a different undergraduate from the 8.226 writing course, which I
taught with Krishna Rajagopal. This collection is in no way complete,
and it isn’t meant to be. But it captures the range of creative scientific
endeavors we pursue here and why it is a joy to be a part of the
MIT Department of Physics.

FROM GR APHENE TO GAL A XIES: GRADUATE PHYSICS AT MIT 3

 BY A MELIA SMITH

Alex
Tinguely
It’s official: our favorite spinning
space-rock is in trouble. The Earth
is rapidly becoming too hot to handle
because we can’t stop consuming energy
from fossil fuels. But, what if there were
a way to unleash oodles of energy from
extremely abundant fuel without producing
harmful byproducts? A phenomenon
so awesome that it lights the sun
and every other star in the sky?
This
does for them to exist apart. That said, it isn’t easy
to convince nuclei to get close enough to fuse,
because they are positively charged and repel each

process
other with colossal force.
It may be helpful to think of fusion like trying to

is called
get two people who hate each other to carpool:
As difficult as it may be, once you manage to
get them both in the same car, you save a lot of

nuclear
energy. Analogously, once you manage to get
the hydrogen nuclei to fuse into a single nucleus,
energy is saved by the combination. You can then

fusion, and
use that extra energy to do useful things, like
generate electricity.
Tinguely gets nuclei to smash into each other by

the good
getting them moving fast enough to plow through
the barrier of electrical force. For hydrogen atoms
in a gas, this means heating them to such a high

news is,
temperature that the electrons are ripped from
their nuclei, resulting in a plasma, a gas full of free
charged particles. As exotic as they may sound,

we can do plasmas are common. Here on Earth, they’re

responsible for the glow of fluorescent lights, the

it here
brightness of a flash of lightning, and the awe-
inspiring shimmer of the polar aurorae. They’re
even more mundane in outer space, where they

on Earth.
make up much of the matter between stars, and
the stars themselves. In fact, most natural fusion
reactions occur in plasmas at the hearts of stars.
This gives us an idea of how hard it is to make
The bad news, however, is that there are problems fusion happen on Earth: we need to create condi-
with our methods of harnessing it that make it tions comparable to the inside of a star. Great for
impossible to use as a viable power source. Solving dramatic flair, but also great for melting expensive
these problems could mean a bright future for scientific equipment.
our bleak energy landscape, which is what Alex Unsurprisingly, hot star-stuff can’t be allowed
Tinguely, a graduate student at the MIT Plasma to run amok on Earth: it would dissipate and
Science and Fusion Center, is working on. never produce any useful reactions. “You have
Tinguely is doing research on an impressive to somehow confine it,” explains Tinguely. “The
machine with an even more imposing name: Sun produces fusion energy, and it’s confined
the Alcator C-Mod Tokamak. It’s a device that, by gravity. If we wanted a reactor like that, we
against all odds, allows us to achieve nuclear would need something the size of the Sun in order
fusion reactions in the middle of Cambridge, MA. to have it confine itself, and we can’t have that
A fusion reaction involves taking two nuclei, on Earth, because Earth is like a million times
the cores of atoms, and smashing them together smaller than the Sun, right? But what we can do is
so they become a single nucleus. This releases confine it with magnetic fields.”
a lot of energy, because it takes less energy for That is exactly what the tokamak was designed
hydrogen nuclei to exist bound together than it to do.

6 DEPARTMENT OF PHYSICS | MASSACHUSETTS INSTITUTE OF TECHNOLOGY

A tokamak, such as the Alcator C-Mod, consists Using spectrometers inside the tokamak, Tinguely
of a donut-shaped vacuum vessel that contains is measuring the radiation emitted by the acceler-
the plasma, which is prevented from touching the ating runaway electrons in order to determine
walls by powerful magnetic fields. The plasma under what conditions they arise. If he can figure
interacts with the magnetic fields due to its abun- out the formation mechanisms, it might become
dance of charged particles. Jolts of current heat up possible to stop them from occurring or prevent
the plasma to the necessary temperatures. them from damaging the device.
So, to recap, we have a clever setup of magnet- For both plasma disruptions and runaway elec-
ically-confined plasma whizzing around at trons, figuring out how to prevent them or
temperatures hot enough to produce fusion reac- compensate for them is crucial for creating
tions. Why aren’t we using this to power the larger-scale fusion experiments, because it would
world? Unfortunately, there are problems that be wasteful to spend so much time and money
arise when you confine such an energetic mate- building a tokamak that’s going to destroy itself.
rial, and Tinguely’s goal is to solve them. Tinguely says, “if we can solve that, then we’re
One of Tinguely’s research objectives is to predict one step closer to having a fusion power plant
events called plasma disruptions in the tokamak. producing net energy that you can put on the grid
A plasma disruption is, in Tinguely’s words, “the for the world to use.”
death of a plasma,” in which it rapidly loses energy With regard to the awesome potential of fusion
and deposits itself on components of the machine. and the bright prospects of his current work, Alex
Disruptions often render the tokamak unable to Tinguely is optimistic. “I really think that it will
continue generating power without undergoing be the energy of the future,” he says. “But hope-
expensive repairs. Tinguely is looking at measure- fully, the near future, within our lifetimes.”
ments of parameters of the plasma, such as its
position, velocity, and trajectory in the vacuum Amelia Smith is pursuing a Bachelor of Science
vessel as well as the current. “Just by looking at degree in physics and linguistics. Her nebulous life
all the data we have, trying to create some sort of plans include getting a puppy and maybe doing
science or theater or something.
algorithm that, given another shot, would predict
another disruption,” he says.
If Tinguely is successful in creating an algorithm
to predict disruptions at Alcator C-Mod, it would
be a ray of hope for the future implementation of
tokamaks. When it comes to a plasma disruption,
“if we can predict it, then we can try to prevent
it or mitigate it,” he says. That would mean less
damage to the equipment and a greater possibility
of using it to generate usable power.
Another problem that arises in tokamaks is
the issue of runaway electrons. Electrons are
extremely light compared to the nuclei of atoms,
so they move much faster in the plasma. If the
electrons reach relativistic speeds, they can
escape confinement and harm the device. The
vessel and other components of the device can
experience “in a sense, very small explosions just
from being hit by a beam of really high energy
electrons,” says Tinguely.

FROM GR APHENE TO GAL A XIES: GRADUATE PHYSICS AT MIT 7

 B Y B O R YA N A D O Y L E

Arolyn
Conwill
There are nearly 1030 bacterial cells on
Earth – that is one trillion bacteria for every
grain of sand on the planet! But despite their
omnipresence inside us and around us, as well
as their importance in soil and medicine, little is
understood about how different types of bacteria
engage in teamwork and warfare.
Arolyn
peting other bacteria. Interestingly, bacteria help
humans process nutrients within our intestines.
However, if those bacteria escape to another

Conwill,
location in the body, they can cause an infec-
tion. Antibiotics are designed to stop the bacte-
rial infection, by either killing the bacterial cells or

a physics disabling a key step in division.

At the same time, bacteria can develop the ability

graduate
to grow in the presence of antibiotics, a phenom-
enon called resistance. When resistant and non-
resistant bacteria live together, it can lead to what

student in
bacterial ecologists call the “cheater-cooper-
ator” setup between two bacteria. The cooper-
ator bacteria degrade the antibiotic around them

Jeff Gore’s
and are thus resistant to it. The so-called cheater
bacteria are not resistant to the antibiotic and
will die in its presence – but if both the resistant

laboratory
cooperator and non-resistant cheater bacteria are
grown together with the antibiotic, then even the
non-resistant bacteria can survive. The cooper-

at MIT, is
ator bacteria degrade the antibiotic, reducing the
concentration enough for the cheater bacteria to
live. The idea here, as Conwill said, is “if the coop-

working erators are deactivating some bad thing in the

environment, then maybe the cheaters don’t have
to.” This freeloading behavior from the cheater

to fill this bacteria allows them to divide faster and more effi-
ciently, because they forgo the work of degrading

gap in our
the antibiotics. Yet, without the cooperator
bacteria, the cheaters would not be alive at all.
Previous experiments have studied the cheater-

knowledge
cooperator situation, but Conwill tested a setup
in which this cooperation between the two types
of bacteria goes both ways. Specifically, bacteria
resistant to the antibiotic ampicillin survive in
using new technology to quickly and accurately the presence of ampicillin but not chloramphen-
count bacteria. “The work is really exciting because icol, another antibiotic. Chloramphenicol-resistant
there are a lot of new experiments that would not bacteria survive in the presence of chloramphen-
have been possible before,” says Conwill. icol but not ampicillin. But when grown together
with both ampicillin and chloramphenicol present,
Bacteria are small cells containing DNA, much
they both survive because they each degrade the
like their larger human cell counterparts. One
antibiotic that would inhibit the other’s growth.
bacterial cell can split into two identical copies
Understanding this bacterial teamwork is the
in a process called division. This allows a large
current focus of Conwill’s research.
number of bacterial cells to accumulate if there are
sufficient nutrients in the environment. A type of Thus far, she has been interested in how many
bacteria is successful if it divides quickly, outcom- bacterial cells are present over time for each type

10 DEPARTMENT OF PHYSICS | MASSACHUSETTS INSTITUTE OF TECHNOLOGY

of bacteria while they work together. To this end, the antibiotics and transfers the bacteria every
she performs laboratory experiments by growing hour. She tested the prediction by performing
both ampicillin-resistant and chloramphenicol- this experiment in the laboratory for twenty-four
resistant bacteria in a fraction of a milliliter of hours straight. Each replenishment and transfer
liquid containing nutrients and both antibiotics. step took her thirty minutes, leaving only thirty
Over time, the bacteria divide many times and minute breaks in between. According to Conwill,
degrade their respective antibiotics. After twenty- half of the planning for the marathon experiment
four hours, she replenishes the nutrients and both was finding enough “food that she was going to
antibiotics along with transferring one hundredth be able to eat really fast between time-points.”
of the bacteria to a new container, so that the Fortunately, she maintained enough energy over-
bacteria have more room to grow. She repeats night and discovered that frequently replenishing
these two steps of replenishing the antibiotics and the antibiotics so the concentration remains
transferring the bacteria each day for two weeks roughly constant does indeed stabilize the amount
straight. To her surprise, Conwill found that the of each bacteria.
relative abundance of the two bacterial popula- Besides day-to-day research combining exper-
tions increases and decreases by four orders of iments, models, and computer simulations,
magnitude every three days. Conwill is a leader in outreach and her commu-
Her explanation for these oscillations goes like nity. In April, she went to Washington D.C. with
this: when there is a large amount of ampicillin- MIT’s Science Policy Initiative Group to discuss
resistant bacteria in comparison to the amount of “science funding issues and encourage congress to
Chloramphenicol-resistant bacteria, then ampi- continue and/or increase their support.” Conwill
cillin will be rapidly degraded in comparison to is also an avid baker: “a couple years ago, I decided
chloramphenicol. Once most of the ampicillin has that I would learn how to make really fancy cakes.”
been degraded, the Chloramphenicol-resistant To improve her skill, she made each member of
bacteria will be able to rapidly grow. This leads Jeff Gore’s lab a birthday cake. One highlight was
to a small amount of ampicillin-resistant bacteria a quad-copter cake for a high school student in the
relative to the large amount of Chloramphenicol- lab over the summer.
resistant bacteria. At this point the same logic Conwill’s current results demonstrate and char-
applies with ampicillin and chloramphenicol acterize bacterial teamwork in the laboratory
reversed, resulting in a subsequent increase in setting. Going forward, Conwill hopes that
ampicillin-resistant bacteria. her work will be expanded to include an evolu-
For Conwill, though, the bacterial teamwork tionary component. Perhaps after growing the
experiment can be very time consuming, espe- ampicillin or Chloramphenicol-resistant bacteria
cially because she needs to vary an important together for a long time, researchers can observe
parameter: the frequency at which she replen- some bacteria evolve resistance to both ampicillin
ishes the antibiotics and transfers the bacteria. To and chloramphenicol, while other bacteria lose
speed up the work, she created a mathematical their resistance altogether. In conjunction with
model and used a computer to simulate the exper- Conwill’s current results, she believes these exper-
iment with different frequencies. These simula- iments would help scientists better understand the
tions allowed Conwill to understand the bacterial mysterious details of antibiotic resistance in our
teamwork setup without performing thousands food supply and bacterial infections.
of laboratory experiments, which she says “would
have been a total nightmare.” Boryana Doyle is a rising senior at MIT, majoring
in physics. After MIT, she hopes to attend medical
In her simulations, Conwill found that the school. Thus far, she has enjoyed advising freshman,
amount of each bacteria remains constant over sailing on Boston Harbor, and performing theoretical
time rather than oscillating if she replenishes biophysics and experimental biology research.

FROM GR APHENE TO GAL A XIES: GRADUATE PHYSICS AT MIT 11

BY E L E N A D E L A PA Z

John
Hardin
We’ve learned a lot about atoms in the past
150 years. The atom has gone from merely a
hypothesis to a staple of elementary physics,
from the unknown to a symbol of science in
pop culture. Atoms, we are taught, make up
matter as we know it. But what are the
components of atoms?
The nucleus
Thankfully, with the computation power devel-
oped in the last 50 years, physicists can simu-
late the behaviors of quarks and the particles they

of an atom
make up, through a process called Lattice QCD
(LQCD). In LQCD, by choosing discrete points
in space and time, and defining the strength of

consists of
the interaction they want to model, scientists can
solve numerically for solutions that specify the
behavior and decay of quarks. Testing these solu-

protons and
tions is the work of MIT graduate student John
Hardin. Along with other members of the MIT
Hadronic Physics group, he works on the phys-

neutrons,
ical hardware that will detect the products of
predicted quark interactions.

which in
But there is a catch– some of the mesons, a type
of subatomic particle predicted by LQCD, have
not been observed experimentally. The typical

turn are
components of a meson are a single quark and a
single antiquark, both of which contribute spin,
charge, angular momentum, and other charac-

made up
teristics to the particle as a whole. There are strict
rules about how those contributions relate to each
other. However, some of the mesons predicted

of quarks.
by LQCD have a momentum, spin, and charge
that could not result from a quark-antiquark pair.
The additional component would come from the
excitation of gluons. These mesons, which have
significant gluonic contributions, are known as
In the same way that protons and electrons have exotic mesons.
electric charge in accordance with the electro- If LQCD is a perfect representation of QCD, the
magnetic force, quarks have “color” charge, which nonexistence of these mesons would mean huge
is carried by gluons, the particles that bind quarks problems for the dominant theory in the field for
together. The theory that describes this interac- the last 30 years – or imply that the methodology
tion is called quantum chromodynamics, or QCD. in performing the numerical calculation uses too
The primary goal of any theory is to successfully many approximations to be useful. “A deviation
explain observable phenomena, but the observa- from the spectra implies that something about
tion aspect is difficult for QCD. LQCD is wrong,” says Hardin. If those mesons do
That’s because a fundamental point of QCD is appear, it would reaffirm the validity of QCD, and
the idea that you cannot separate two quarks. If predictions based on LQCD would be limited only
you do, the energy in the gluon field will result by computational power. So which is it?
in the spontaneous creation of another quark/ In Virginia’s Thomas Jefferson National
anti-quark pair. We are able to study protons and Accelerator Facility, known as JLAB, Hardin is
neutrons by examining their interactions individ- trying to find out using an apparatus designed to
ually, but we cannot simply do this with quarks. solve this problem. JLAB is home to four halls –
As a result, creating experiments that test the within each, specialized spectrometers are dedi-
validity of QCD requires a careful approach. cated to investigating collisions with photons

14 DEPARTMENT OF PHYSICS | MASSACHUSETTS INSTITUTE OF TECHNOLOGY

and a stationary target, usually protons. In Hall between JLab and MIT. He spends 4-6 weeks
D, Hardin and his colleagues work on GlueX, per year on shifts, making sure data collection
which looks at the products of collisions between is running smoothly. As Hardin and his group
light and protons. These products often include prepare to build and install their upgraded PID
mesons, which then decay further into other apparatus, this amount of time will only increase.
particles. By looking at the decay patterns, As of now, installation of the upgraded PID is
researchers can infer the characteristics of these scheduled for May 2017.
mesons, and, most importantly, whether or not With improvements in simulations and the
they consisted of one quark and one antiquark – construction of the upgraded particle identi-
or whether they are exotic. The ultimate goal of fier on his hands, Hardin is quite busy. But the
GlueX is to find all of the mesons predicted by most interesting data collection won’t occur until
LQCD, both exotic and already-observed. the upgraded PID is brought to GlueX. Exotic
In order for GlueX to achieve this goal, however, mesons have long been hypothesized, but this
it needs an upgrade. The current way the detector will be a huge opportunity to observe them. No
identifies particles is a mixture of drift cham- matter what GlueX finds, it will open the door
bers, which provides particle tracks, a time-of- to new questions. “If it exactly reproduces the
flight calculator, and calorimeters, which measure spectra, that would be… interesting,” says Hardin.
particle energy. However, at present the detector “Then we would get more computers, to do more
is not adequate to see all mesons predicted by LQCD.” But perhaps, a failure to precisely repro-
LQCD. Two of the longest-lived mesons that duce the mesons predicted would have just as
will be detected by GlueX are kaons and pions. much impact. It would raise questions as to how
However, GlueX’s current particle identification LQCD simulations were performed, and which
system cannot distinguish between them, which assumptions they made, and raise questions about
means the distribution of meson products can’t be QCD itself. Either way, when data collection
understood fully. In addition, seeing a kaon gives starts, we are sure to learn more about how the
important information about the quark content of strong force works.
the meson – in particular, it is characteristic of a
“strange” quark. “If you want to get the full meson Elena De La Paz will graduate with a Bachelor of
spectra,” says Hardin, “you need to be able to iden- Science degree in physics with a minor in mechanical
tify strange content.” Upgrading the particle engineering in June. After graduating, she plans to
work in San Francisco. For the past 4 years she has
identification detector, or PID, to be statistically learned more than she could have ever imagined,
significant within a higher energy range will be and feels truly fortunate to have received excellent
extremely important for reproducing the mesons education as a scientist and engineer.
predicted by LQCD.
Building an upgraded PID has long been a
priority. Hardin joined the group building the
detector in 2012, when development was still in its
infancy. The upgraded detector will use a special
quark prism, called a DIRC bar. After hitting the
target, the photons will travel through the DIRC,
and emerge with an angle dependent on their
velocity, producing a pattern like a fingerprint for
the kaon or pion that made it.
The current PID is being calibrated, and the appa-
ratus is being prepared for the installation of the
upgraded PID, so Hardin commutes regularly

FROM GR APHENE TO GAL A XIES: GRADUATE PHYSICS AT MIT 15

 BY A N A GLIDDEN

Lina
Necib
For all the glittering stars we
see populating the sky, another
unidentifiable material is thirty
times as prevalent. Understanding
this material, mysteriously named
“dark matter,” is a unifying problem
in physics that pulls together
scientists trying to describe
nature at drastically different
scales– from galactic interactions
to fundamental particles.
MIT
But although the mystery of dark matter still
haunts us today, it took a long time for physi-
cists to appreciate the magnitude of the problem.

graduate
Fritz Zwicky first stumbled onto the problem
of “missing mass” in 1933 when he analyzed the
movement of a group of galaxies. Zwicky found

student
that more gravitating mass was needed than
could be explained by the stars he saw. He wrote,
“It is not certain how these startling results

Lina Necib
must ultimately be interpreted.” Yet, the discon-
nect between visible mass and unseen mass was
dismissed for several decades.

works hard The problem of missing mass was not understood

until astronomer Vera Rubin came on the scene in

to solve this
the 70’s. She determined that there was a uniform
halo of non-luminous material around galaxies,
extending past their visible edge. Astonishingly,

dilemma.
this mysterious halo composed the majority of
the galaxy’s mass, pulling the stars into a complex
choreography. “Think of it as a puppet show,”

With
Necib says. Dark matter acts like hidden strings,
rotating galaxies and pulling clusters together.
Finally, the hunt was on to discover what made

complex
up dark matter. “There were different theories,”
Necib explains. “Either it is a particle that we don’t
know, or it is modified gravity, which means we

computer
don’t understand the laws of gravity and that is
why things are acting up.”
The best test in deciphering which of these theo-

models, she ries held up came in 2006. Within the Bullet

Cluster, two groups of galaxies collided and passed
through each other. It was clear from X-ray obser-

sifts through vations that gas clouds from the two galaxies
were colliding and heating up. Yet, using gravita-
tional lensing to map all the gravitating matter, it
became apparent that the majority of the groups’
mounds of astrophysical data, looking for the mass passed through without colliding – some-
smallest suggestion about the identity of dark thing that normal gas could never do. Only an
matter and how it interacts with known particles unknown particle could explain this: dark matter
like photons and electrons. So far, dark matter had to be a substance, not just modifications to our
has remained elusive, refusing to reveal its true current understanding of gravity. But, the ques-
nature, but offering hints along the way. “This tion remained: what is this particle? Even now, we
problem in physics has grown over the past 20 still don’t understand it.
years, especially in the past decade,” she says, and ***
the payoff for solving it could be huge. “It might Dark matter captivates Necib. From her early
complete our picture of particle physics.” childhood in Tunisia, Necib was enthralled by

18 DEPARTMENT OF PHYSICS | MASSACHUSETTS INSTITUTE OF TECHNOLOGY

physics. It was not just a path to understanding Earth as the WIMP interacts with other parti-
how the universe operated – it was a way to gain cles. From these calculations, she can set bounds
respect in a male-dominated society. Her drive to on the WIMP’s characteristics. Necib alternates
be a physicist has never waned. After completing between testing her theories with experimental
her senior thesis on dark matter as an undergrad- data and using experimental results to make new
uate at Boston University, Necib arrived at MIT theories. Remarkably, dark matter signals could
ready to delve deeper into the subject. “When I already be lurking in the data, unknown to phys-
came to MIT my advisor asked me, ‘Do you want icists. “Theorists have the job to actually suggest
to work on jets, dark matter, or supersymmetry?’” new experimental ways to find dark matter,” says
Necib says. “And I replied ‘Dark matter it is, Necib. “At this point you have to be more and
because that seems awesome.’ So that’s what I’ve more creative.”
been doing over the past four years.” MIT shapes how Necib searches for dark matter.
To unravel its identity, Necib uses the standard “It’s a good place to explore your full poten-
model for dark matter – called the weakly inter- tial. You are pushed to very, very high standards
acting massive particle, or WIMP – as a foun- and very high limits. But still you have a lot of
dation. The WIMP needs to explain a decade’s people around to help you,” she says. As a grad-
worth of odd data. Though Necib is a particle uate student, Necib faces setbacks on a daily basis.
physicist, her data often comes from astrophys- Sometimes her code is buggy; other times her
ical observations. The vast scales, quantities, and computer has issues. When she makes a plot to
exotic phenomena of the sky offer her a tremen- compare her most recent model with the data, she
dous search area for dark matter. waits, staring at the screen. The plot finally pops
While dark matter was first apparent as a missing up, and it looks wrong. But, a few iterations later,
mass, other anomalies across the sky may also her understanding grows and the plot is beautiful.
be dark matter signals. As WIMPs can release Necib’s day-to-day struggle to understand the
gamma rays when two of them bump into each data parallels the larger problem physicists face in
other and annihilate, any unexplained gamma unraveling the nature of dark matter. The life of
ray emission could come from dark matter. Using a dark matter detective is tedious and often frus-
gamma ray telescopes, Necib says, “There should trating. Each scrap of understanding takes an
be no place we don’t look.” Yet, observations enormous amount of time and patience. But, this
of excess gamma rays alone are not enough to is just the type of task that Necib thrives on – tech-
confirm the presence of dark matter. nical and hard.
Any detection of dark matter has to be verified in
a second way. For instance, physicists are hoping Ana Glidden graduated in June from MIT with a
Bachelor of Science degree in physics. She plans on
to directly observe dark matter with detectors pursuing research in astrophysics after enjoying
made of heavy nuclei like xenon or germanium her experience with the Undergraduate Research
that recoil or loose an electron when they interact Opportunities Program. In her free time, Ana
with a WIMP. Physicists could also make dark enjoys backpacking and canoeing.
matter by smashing protons together at the Large
Hadron Collider. The goal is to observe WIMPs
with the same characteristics in several exper-
iments. With so many different ways to look for
dark matter and so many different signals to parse,
Necib says, “You have to be very open-minded in
physics and basically let the data speak to you.”
Necib’s role is to calculate the rate at which dark
matter signals may appear in experiments on

FROM GR APHENE TO GAL A XIES: GRADUATE PHYSICS AT MIT 19

BY WILLIA M A . RUT TER

Nancy
Aggarwal
A silent explosion rocks the very fabric
of space-time. The perpetrators of the
chaos – two massive black holes – have
ended a journey of epic scale. Caught
in each other’s gravitational grasp some
point long past, the singularities have
spiraled towards each other for an
unknowable time, circling like boxers
before the first punch.
Together,
Aggarwal. “I think I took it as a challenge.”
But she observed a profound gender bias in her

the black
college cohort. “People know how to solder things,
people have worked with Legos,” she says. “I wasn’t
really doing that when I was a kid, and I saw it

holes weigh
was the same for other women around me.” So
Aggarwal began mentoring women in science and
technology, saying, “Hey – you can do this. This is

in at over
not something that has to be only men.” She decided
experimental physics’s hands-on culture would
allow her to further blaze the path. She remembers

sixty times
thinking “I want to change the face of this – I want
to be that woman who’s interested in robotics and
who’s interested in making things with her hands.”

the mass
Committed to experimental physics, Aggarwal
moved to MIT’s Kavli Institute post graduation.
Since then she has worked with Professor Nergis

of the sun
Mavalvala, studying and designing quantum-
optical systems to further improve the sensitivity of
the LIGO observatories.
When Aggarwal arrived, LIGO’s search had
and the ultimate collision and combination turns proceeded for eight years without seeing a gravi-
three suns’ worth of that mass straight into tational wave. Scientists agreed that LIGO’s noise
energy.This energy becomes ripples in reality, floor, the point at which internal fluctuations in
gravitational waves, radiating outward as final, the detector shroud useful signals, was too high.
fleeting evidence of this cosmic clash. Important parts of this noise floor come from the
These ripples made waves of a different sort 1.3 quantum mechanics of the lasers LIGO uses to
billion years later: they became the first gravi- detect gravitational waves. Aggarwal’s work focuses
tational waves detected by humans. The Laser on lowering these particular noise sources with
Interferometer Gravitational Observatory, LIGO, quantum states of light called “squeezed states.”
caught the vibrations flitting through the Earth Squeezed states are a consequence of the quantum
in late 2015. A true feat of experimentation, the mechanical uncertainty inherent in measuring
discovery astounded and invigorated the world- light. Light has two principle measurable qual-
wide physics community. ities: power, which is essentially brightness,
It was, however, simply the next chapter in a and phase, which describes what point the light
larger story for Nancy Aggarwal, a Ph.D. candi- currently inhabits in its electromagnetic cycle.
date in physics at the Massachusetts Institute of Light’s inherent uncertainty can be shared equally
Technology. Hailing from India, Aggarwal grad- between power and phase, as is the case for laser
uated from the Indian Institute of Technology light. In squeezed states, however, careful manip-
Bombay in 2011. At IIT, Aggarwal’s high school ulation has “squeezed” uncertainty from one
love for math and physics translated predictably quality into the other. It is this exchange of uncer-
well into university – “I was that kid who didn’t tainty between power and phase that will prove
like anything else!” – and she immediately began invaluable for lowering LIGO’s noise floor.
pursuing a degree in physics. Although not much of One source of quantum noise in LIGO is shot
a tinkerer prior to college, she nevertheless decided noise – the uncertainty in measuring flows of
to explore experimental physics. “All of my class- distinct objects solely due to their distinct nature.
mates, everyone was doing theoretical physics,” says The laser LIGO uses to detect gravitational waves

22 DEPARTMENT OF PHYSICS | MASSACHUSETTS INSTITUTE OF TECHNOLOGY

is a flow of distinct photons, and the resultant shot analogous to the way a guitar string vibrates at
noise creates uncertainty in the laser’s phase. Laser higher frequencies when you move your finger
interferometers like LIGO work by measuring up the fretboard. Any uncertainty in laser power
phase precisely, so this is a problem. becomes uncertainty in cavity length, and there-
fore uncertainty in resonance frequency. This
LIGO could decrease shot noise by increasing its
frequency uncertainty becomes phase uncer-
laser power. The more powerful a laser is, the more
tainty as the light leaves the resonance cavity.
photons it makes. With more photons, each indi-
Overall, the system takes some of the laser’s power
vidual photon is less important to the big picture,
uncertainty and turns it into phase uncertainty –
washing out shot noise and decreasing phase
a squeezed state is born.
uncertainty. This solution, however, is not perfect.
Increasing laser power comes hand in hand with Aggarwal’s system is noteworthy for two
increasing the power’s uncertainty. Greater uncer- reasons. First, the lensless design means it can
tainty in power leads to greater uncertainty in the create squeezed states at any laser wavelength
bouncing force the laser’s photons exert on mirrors LIGO could want. Second, Aggarwal’s squeezer
in LIGO. This uncertainty in force leads to an design lends itself to a convenient, self-contained
uncertainty in the mirrors’ movement, as if they package. “We have squeezers, but currently they
take a room full of space and three Ph.D. students
are being buffeted by a gale of photons. This mirror
to make,” says Aggarwal. By contrast, Aggarwal’s
movement ultimately raises the noise floor, making
squeezer is quite compact – no larger than a piece
it harder to see gravitational waves.
of carry-on luggage. “I could imagine,” Aggarwal
It turns out that this phase versus power uncer- continues, “putting it in a box and just giving it to
tainty tradeoff is fundamental to lasers. “There isn’t LIGO, saying ‘Hey LIGO, here’s this box, and that’s
really any getting away with it,” says Aggarwal, your squeezer.’” These words hint at a deeper
because it comes down to the same quantum implication of Aggarwal’s work: ease of scaling.
uncertainty that causes squeezed states. But this As LIGO, propelled by its success, begins to estab-
means that squeezed states can offer a solution: if lish further observatories around the world,
inserted correctly, squeezed states can exchange the convenience and flexibility of Aggarwal’s
power uncertainty with phase uncertainty for squeezer will allow for expedient implementation
the whole LIGO system. By controlling the type of squeezers in new systems.
of squeezing, scientists can therefore control the Still, the work is ultimately very personal for
uncertainty exchange, even tailoring it for listening Aggarwal. “I designed and built the entire exper-
at different gravitational wave frequencies. iment in an empty room,” she says, “from
Here, Aggarwal hopes to contribute conven- designing shelves to designing very small and
tional methods using advanced lenses to generate precision measurement components.” The system
squeezed states. This only works at certain laser is her creation – not only designed with her mind,
wavelengths, however, constraining the lasers but also built with her hands. In an intellectual
LIGO can use. Further, conventional squeezers whirlwind of quantum physics and gravitational
require meticulously aligned, room-sized optical waves, Aggarwal still challenges that bias against
setups to operate. Aggarwal wishes to create hands-on, practical work she discerned in college.
squeezed states without these lenses. Now a mentor of a different sort, Aggarwal leads
Aggarwal begins by creating resonance cavi- by example, squashing both scientific and societal
ties out of two tiny mirrors, suspending one of barriers with each squeezed state.
the mirrors on a thin, spring-like piece of metal.
William Rutter graduated with a Bachelor of Science
When a laser shines into the resonance cavity, the
in physics in June 2016. He is now pursuing a career
suspended mirror recoils from photons hitting in design and communication with the hope of
it, changing the length of the cavity and its corre- spending life listening to and telling stories of
sponding resonance frequency. This effect is all stripes.

FROM GR APHENE TO GAL A XIES: GRADUATE PHYSICS AT MIT 23

 BY Z A NDER MOSS

Spencer
Tomarken
Consider, for a moment, the
millenia of collective endeavor,
the dedication of countless careers
to the understanding of the nature
of interactions between two isolated
particles. From the rigid body
collisions of Sir Isaac Newton to
the field interactions of Feynman
and Schwinger, the struggle of
fundamental physics has been
to grasp at a deeper understanding
of the simplest possible processes.
In light of
entity can be visualized as a free electron which
has been “dressed” by the electromagnetic fields
of the matter through which it propagates. As it

this history,
travels, the pull of these fields will alter the inter-
actions and mechanics of the particle, causing it to
appear heavier or lighter than the bare electron.
These quasiparticles are the focus of Tomarken’s
it is a daunting truth that the world around us
experimental work in graphene.
consists of condensed matter: systems of unthink-
ably large numbers of particles acting in concert. Graphene is a single sheet of carbon atoms linked
How can we ever hope to understand this reality together in a hexagonal “honeycomb lattice”.
if our two-body models are still incomplete? Elegant theoretical models exploit the symmetry
and repetition of the lattice to produce simple
Spencer Tomarken, a graduate student at MIT,
models of non-interacting electrons. The reality
is working towards such an understanding by
of these systems, however, is a complex inter-
probing pure samples of carefully arranged
play of electrons as they skitter across the surface.
matter with cutting edge experimental tech-
Electron interactions in graphene are signifi-
niques. He dedicates great time and effort to
cantly augmented by the lattice “dressing” effects.
preparing the perfect sample and performing the
Electrons under the influence of these abnor-
perfect experiment. His samples are idealizations
mally large interactions, called “strongly corre-
of more common objects in terms of their purity,
lated electrons”, are notoriously difficult to model.
structure, and the conditions of their testing.
Tomarken summarizes the state of the theory
Nevertheless, if Tomarken can reach through the
as “contentious” and submits that it is up to the
complexity and isolate a new behavior, then he has
experimentalists to explore the subtleties of these
learned something fundamental about the micro-
interactions. “We’re the ones, on the ground level,
scopic life of matter.
answering questions,” he says. “I have this many
The key to studying these systems, says electrons, … how do they organize themselves,
Tomarken, is emergence. Emergence is the and how does energy minimize itself? Often it’s
appearance of collective, coordinated phenomena counter-intuitive.” They must find ways to listen
in many-particle systems. In Tomarken’s words: for interactions among the overwhelming buzz of
“If you can understand one particle, it feels like disorder. This is Tomarken’s art.
you can understand N particles, but, in fact, you
His research group, headed by professor Ray
can’t. You would never be led to something like
Ashoori, has developed a technique of unprece-
superconductivity by understanding the nature of
dented sensitivity for measuring the energy struc-
a handful of particles and then generalizing.” He
ture of thin condensed matter systems. Since
says: “The whole point is that you get something
matter is quantum mechanical in nature, its
which is greater than the sum of its parts.”
constituent particles may only occupy a discrete
There is a gulf between fundamental and emer- set of energy levels. The configuration of a
gent phenomena that reductionist methods quantum system will have a telling signature in
cannot bridge. Theorists must build models that the pattern with which its component particles fill
can predict emergent phenomena using clever these energy states. Due to the crystalline struc-
analogies and incisive approximations, rather ture of materials like graphene, these energy levels
than ground-up constructions from small- separate into distinctive bands. Through great
scale interactions. Perhaps the most important experimental effort, these bands can be mapped
of these theoretical constructs is the quasipar- out in detail.
ticle. Although they appear in all kinds of phys-
Tomarken is working on the application of a
ical systems, the quasiparticle of greatest interest
novel experimental method, known as tunneling
to Tomarken is the “electron quasiparticle”. This

26 DEPARTMENT OF PHYSICS | MASSACHUSETTS INSTITUTE OF TECHNOLOGY

spectroscopy, to the exploration of graphene. he is really searching for surprising new physics.
He begins by fabricating single layers of pure Graphene may exhibit unexpected emergent
graphene, and then sandwiches them between phenomena due to the strength of electron interac-
two layers of insulating material (one thick and tion. The challenges of theoretical prediction have
one thin). The outside surfaces of these insu- left much of this territory uncharted, so the form of
lators contact conductors. Tomarken applies these phenomena, if they exist, is still a mystery.
an electric potential across the contacts, which Tomarken admits that the research can be frus-
causes electrons to “tunnel” through the insula- trating at times, fraught as it is with false starts,
tors and fill the graphene layer. It is a fundamental faulty samples, and failed experiments. “To make
quality of nature that no two electrons can play one device, it could take two to three weeks,
exactly the same role (“occupy the same state”) in and the odds of that being a perfect fabrication
a physical system. As a result, when electrons fill are maybe one in ten,” he says. “So you could
the graphene, they will be forced to settle into spend six to nine months making samples that
progressively higher energy levels. The top filled just fail. There are so many things that have to
level is called the Fermi level. coalesce and work perfectly to do this measure-
Once the electrons have piled up to the Fermi level ment.” Despite these frustrations, Tomarken
in the graphene, Tomarken applies a rapid voltage says the thrill of leading this exploration keeps
pulse, saturating the conductor on one side of the him going. “I touch everything and know every-
graphene (the side with the thin insulator) with thing about it. I’m the master of every detail. The
electrons. These electrons will tunnel through the upside of this is that you are in control… you
thin insulator layer into the graphene, and then nurse [the experiment] all the way through. It’s a
slowly trickle down (in energy) to the Fermi level. very cool feeling to bring one of these projects to
Crucially, the rate at which electrons with a given fruition. It’s your baby.”
energy tunnel into the graphene depends on the Tomarken sees an exciting future for this work
number of empty states available at the energy in in graphene systems. Using techniques less
question. If the tunneling energy is near a band of sensitive than his, experimentalists are already
states, the current will be large. If not, electrons exploring the dynamics of overlapping sheets,
will only trickle through. By varying the size which produce complex quasiparticle behaviors.
of his voltage pulse and measuring the current To add to the excitement, electron quasiparticles
at each step, Tomarken traces out the density in graphene behave as if they are effectively mass-
of energy states, illuminating distortions in the less. This strange behavior is responsible for the
band structure. These distortions are the signa- strength of electron correlation, and it may cause
tures of electron-electron interactions within the distortions in the response of these quasiparti-
graphene, and can be identified by comparison cles to external magnetic fields . Looking forward,
to the predictions of a non-interacting model. If Tomarken says, “We hope to explore physics that
Tomarken injects electrons slowly, and at energies no one else has been able to access.” Armed with
far away from the Fermi level, the data will closely a powerful tool for probing complex systems like
match such a model. The closer the incoming graphene, he says, “You trust that at high magnetic
electrons are in energy to the Fermi level of the fields and low temperatures, you’re going to see
graphene, however, the stronger the interactions. something interesting”.
By narrowing this energy gap, Tomarken isolates
the physics of interacting electrons. Zander Moss is a rising senior at MIT, majoring
Asked what he hopes to find, Tomarken replies that in physics and mathematics. He has had a great
there are familiar effects, like magnetic interac- deal of time studying the physics of neutrinos with
Professor Conrad and her research group, and he
tions between pairs of electrons, for which he is on hopes to continue to explore nature wherever he
the lookout. In the bigger picture, Tomarken says, ends up next.

FROM GR APHENE TO GAL A XIES: GRADUATE PHYSICS AT MIT 27

 BY JUSTICE M ASON

Ryan
Lynch
In our universe, many mysterious
and beautiful objects are invisible to
the naked eye – but we have created
ways to see them. Telescopes, science’s
most traditional method, use portions of
the electromagnetic spectrum to study
celestial bodies.
For example,
Lynch is a working on statistical methods for fast
signal detection and processing for the MIT LIGO
project. Developing methods to accurately and

most amateur
quickly detect signals is a key step in gravitational-
wave astronomy because it will allow us to “hear”
the violent and beautiful outcries of the universe,

astronomers
making Lynch’s research very powerful for
furthering insight and knowledge of the universe.

use visible
In Lynch’s line of work, he has to find ways to
separate the important sounds of the universe
from the background noise. Imagine you are

light to see
given a list of characteristics describing the
voices of 5 different people. Furthermore, you
are standing on a busy street – many cars and

objects like
beeping horns – and are asked to distinguish a
person’s voice from the background noise. Given
the noise of the cars, people etc., imagine how

stars or
difficult it would be to accurately distinguish a
single voice. What if a friend joined you? There
would be a better chance that you both would

the moon.
hear similar sounds and agree on who spoke.
However, in actuality, you and your friend would
not hear exactly the same background noise, so
trying to agree on what was said would be more
difficult, but – as Lynch’s research shows – not
Scientists are also developing techniques to use impossible. This analogy is essentially the LIGO
other types of waves scan the sky – giving us a project and Lynch’s research problem.
better look into the unknown. Interpreting noisy signals is the core of Lynch’s
Gravitational waves are ripples in the fabric of the research with LIGO: fast identification of a
universe caused by high-energy objects and inter- signal’s source so as to alert astronomers to point
actions – like two black holes merging together! their telescopes at the right place – to catch a
These waves stretch and squish the length of the glimpse of the excitement! “Think of LIGO as
objects through which they pass by an incredibly two really big antennas,” Lynch says. Similar to
tiny amount – a fraction of the size of an atom. our earlier voice analogy, data received by LIGO’s
LIGO, the Laser Interferometer Gravitational- instruments must be shown to be a real signal,
Wave Observatory, is the leading institution and a source for those signals must be identified.
for detection and analysis of gravitational-wave In order to determine whether the received data
signals. “In maybe 40 years, 50 years, we could represents a signal, it must pass a series of tests.
see everything through gravitational waves,” Initially, because of the large quantity of data over
notes Ryan Lynch, a current graduate student in which to search, Lynch trims down the search-
the physics department at MIT. Lynch started space by selecting high-activity areas with high-
working on the LIGO project as an undergrad- amplitude data. Secondly, “it has to be received
uate researcher, eager about the possibility of by both detectors [at the appropriate time] and
discovery. “We were turning on the upgraded the signal has to look the same,” Lynch explains.
instruments … There was a possibility that I was Going back to the busy street, Lynch’s description
going to be one of the ones that detected things.” of LIGO’s signal detection process is like waiting

30 DEPARTMENT OF PHYSICS | MASSACHUSETTS INSTITUTE OF TECHNOLOGY

for moments when there is a loud outburst, then computer to develop models to predict whether
confirming that both you and your friend heard data it receives is a signal or noise by creating a
it, and that it sounded the same. Although this threshold range for each of these characteristics.
process determines whether the instrument Ultimately, we want be able feed data to a machine
received a signal, there is still the mystery of the and be given a quick confirmation of gravitational
origin of the signal. In his research, Lynch deter- wave detection as well as its source and source’s
mines the origin of the gravitational wave given location! Tools like those developed by Lynch will
the information that we know about the signal – help to answer many of the puzzling questions of
and sometimes the lack of information we know. the universe, specifically the most confounding of
There are many methods for analyzing signals for them all – the beginning.
which we have analytical solutions and models to If we can automate the processes of recognizing a
determine its source, but Lynch’s research deals gravitational-wave event, it will allow us to hear
exclusively unknown signals. In order to get a these signals much faster and more efficiently.
better understanding of Lynch’s conditions, take Lynch’s research has already made great impact
our busy street analogy and imagine that rather on the field; some of his code and models were
than hearing a voice you hear a generic sound. used in the first ever detection of gravitational
Suppose you hear a noise that is a combination of waves. Lynch’s research is helping to open up
a soprano and bass note from two singers. Your conversation between the universe and man that
task would be to listen to the note and distin- will allow us to learn even more about it. Lynch
guish the source of the noise – the singers in this explained that the majority of the purpose of his
example – using your knowledge of vocal ranges research is to study the sources of the gravita-
and the singer’s voices. Ultimately, you and your tional waves of the early universe. Given the vast-
friend want to decide which portion of the sound ness of the universe, these relics of the past have
comes from each tone. This is done by comparing been travelling for millions of years to tell us the
the received sound to each tone of the vocal secrets of the past and give physicists an idea of
range, quantifying their similarity to each, and what the “climate” of the early universe was like.
adding them up to get an overall correlation of the Lynch’s techniques for determining signal source
signal for each range. Similarly, Lynch’s research and location will tell us what created these grav-
compares the signal received by astronomical itational waves and where, constructing a map of
events to very generic templates: overlaying the the universe’s past, which may give us a look into
original, received signal to the signal generated its future. The prospects for astronomy and astro-
by each template to estimate their differences, and physics are strong now and only getting stronger
comparing each of the differences to get the best with this amazing research by Ryan Lynch and
fitting templates from the set of generic templates. the MIT LIGO team.
A larger portion of Lynch’s job is to write code
that conduct all of this complex analysis, which Justice Mason graduated with a Bachelor of Science
then makes identifying and locating gravitational- degree in aero/astro and physics from MIT in
wave sources much easier and faster. June. Since graduating, he works at NASA Jet
Propulsion Laboratory. He has enjoyed his time at
Lynch’s research is focused on training computers MIT, dancing with DanceTroupe and Ridonkulous,
to listen and interpret the sounds of the universe participating in the Undergraduate Research
Opportunities Program, and developing his
more effectively, through machine learning.
love of engineering and science.
Machine learning is taking characteristics of the
signals that help us define the data as a signal or as
noise. We then take the set of data we receive and
label it as a signal or noise. From here, we give this
training data to the computer, which allows the

FROM GR APHENE TO GAL A XIES: GRADUATE PHYSICS AT MIT 31

 B Y D U Y- A N H D O A N

Greg
Dooley
It has long been known that baryonic
matter, which make up stars, planets,
and living organisms, contributes
only a small fraction to the total mass
of the universe. The rest is dark matter,
a mysterious substance which has
never been directly observed, but
whose existence has been confirmed
from its gravitational interactions with
ordinary matter.
According
This model is simple yet effective in explaining
many properties of the universe, such as the exis-

to measure-
tence and structure of the cosmological micro-
wave background, the distributions of galaxies,
the chemical abundances of elements, as well

ments of
as the accelerating expansion of the universe.
However, there are problems yet to be resolved by
the ΛCDM model.

these inter- One such problem is the cuspy halo problem,

which is the discrepancy between simulations

actions, the
and observations of density distributions of
dark matter halos (a component of a galaxy that
envelops the galactic disc and extends well beyond

amount of
the edge of the visible of galaxy). In simulations,
the dark matter density profiles increase steeply

dark matter
at small radii (i.e. cuspy), whereas observations
of rotation curves (the plot of orbital velocity as
a function of distance from the center) of dwarf

in the uni-
galaxies suggest that the profile is much flatter
near the center. Another problem is the “too big
to fail” problem, which again shows disagree-

verse is
ment between simulations and observations. Cold
dark matter simulations predict circular veloci-

five times
ties of subhalos that are too high to be consistent
with any known Milky Way satellites. Scientists
propose different resolutions to help alleviate

as much as
these problems, introducing baryonic phenomena
such as supernova outflow, bursty star forma-
tion, or tidal disruption. Dooley, however, takes

ordinary
a different approach: he modifies the prop-
erties of dark matter to allow for non-trivial

matter,
self-interaction.
In the ΛCDM model, the dark matter is hypoth-
esized to be slow-moving compared to the speed
of light (thus, cold), and interact almost exclusively
yet there is much to be learn regarding its prop- through the gravitational force. Therefore, cold
erties. Greg Dooley, a graduate student at MIT, dark matter is collisionless – the interaction cross-
is running dark matter cosmological simula- section is so low that dark matter collisions, either
tions to study how dark matter models affect with itself or with baryonic matter, are virtually
the formation and evolution of clusters of non-existent. However, Dooley and his research
galaxies in the universe. group observe that, by prescribing a new type
Currently the most widely adopted cosmolog- of dark matter with strong self-interaction (self-
ical model is the Lambda-Cold Dark Matter, interacting dark matter), the cuspy halo problem
called the ΛCDM model, which postulates that and the “too big to fail” problem could be allevi-
the universe contains a cosmological constant Λ, ated. Self-interacting dark matter will have a non-
associated with dark energy, and cold dark matter. trivial probability of colliding with one another

34 DEPARTMENT OF PHYSICS | MASSACHUSETTS INSTITUTE OF TECHNOLOGY

and scattering elastically. The scattering effect lations, which take months or even years on a
will be particularly strong in the central region, supercomputer. In the future, Dooley is hoping
where the density is high, resulting in significant to add more baryonic interactions, such as hydro-
tidal dispersal and stripping of stellar masses, dynamics, as well as stellar formation, galaxy
thus effectively reducing the central density of formation, supernovae, and active galactic nuclei
halos, resolving the aforementioned problems. feedback mechanisms, to the simulations. The
Simulations by Dooley’s research group show inclusion of baryonic physics would produce
that the density profile of self-interacting dark more precise predictions as well as observational
matter halo flattens out near the center (i.e. cusp- tests checking the validity of self-interacting dark
less), strikingly similar to the actual observed matter. Nevertheless, Dooley comments, “The
profile of galaxies. As the density profile is addition of baryonic physics would be very costly
reduced in the inner regions, the circular in terms of computational power... If a typical
velocity profile of subhalos also drops and agrees galaxy-scale simulation takes 2 months to run
with observations as well. on a supercomputer, the equivalent time for self-
There are several ways to observationally test the interacting dark matter simulation would be 3
hypothesis of self-interacting dark matter. Since months. If you include hydrodynamics on top of
the model predicts that ultra-faint dwarf galaxies that, the runtime would increase fourfold to 12
will not have a core, astronomers could look at months.” Dooley is investigating and applying
them and evaluate the claim. However, since analytical methods that would reduce the simu-
ultra-faint dwarf galaxies are very dim – roughly lation runtime, as well as trying to secure more
100,000 times dimmer than a typical galaxy computer nodes to run his simulations. “As it
like the Milky Way – they are very difficult to is a parallel code, the more computer nodes we
observe with sufficient details. Thus, the current have, the faster it is,” he said. At the same time, he
data is sparse and inconclusive, but future prog- expresses his hopes that advances in the field of
ress in telescope technology is expected to solve computing would result in faster runtime of high-
the problem. A different testing method is to resolution detailed simulations.
predict the number of satellite dwarf galaxies Since dark matter interacts very weakly with
orbiting around a major one, and compare the electromagnetic radiation, it cannot be observed
figures with observations. directly. Thus, computational simulations carried
The idea of self-interacting dark matter was out by Dooley and his research group are essen-
proposed in the early 2000s, and the initial simu- tial to the advancement of the field. They produce
lations of self-interacting dark matter were predictions of effects by dark matter that could
carried out by Professor Vogelsberger at MIT in be tested by actual astrophysical observations,
2012. Recently, Dooley has been working with either validating or disproving certain theoretical
Professor Vogelsberger and Professor Frebel at cosmological models, bettering our knowledge of
MIT, Professor Peter at Ohio State University, dark matter. Studies of dark matter properties is a
and Dr. Jesus Zavala from the Dark Cosmology field of unexplored horizons, and future progress
Center in Copenhagen, to run more simulations in both computing and telescope technology will
with increased details and resolution, as well as contribute to the joint effort by many physicists,
updated cosmological parameters measured by the theorists and experimentalists alike, to eventually
WMAP and the Planck missions. Since the scale discover the mysterious properties of the elusive
of cosmological simulations is very large, typi- yet prevalent dark matter.
cally most simulations are dark matter-only, and
Duy-Anh Doan graduated with a Bachelor of Science
ordinary matter is added accordingly, or ‘tagged’,
degree in physics and earth, atmospheric and plane-
toward the end. This approach helps to reduce tary science at MIT. After graduation, he will attend
the computational power required for the simu- Penn State to pursue a Ph.D. in astrophysics.

FROM GR APHENE TO GAL A XIES: GRADUATE PHYSICS AT MIT 35

BY A H M E T M US A B E YO G LU

Fahad
Mahmood
Superconductivity at room temperature
has long been an unattainable goal for
scientists and engineers working in
the field of condensed matter physics.
Although Heike Kamerlingh Onnes
first discovered the effect of super-
conductivity in 1911, the critical
temperature required to observe this
effect was only 4.2K, which made
its utilization impractical for many
potential applications.
Since then,
John Bardeen, Leon Cooper, and Bob Schrieffer
(“BCS”) in 1972.

scientists
Free electrons in a material, which conduct elec-
tricity through their motion, collide with the
neighboring atoms as they drift due to the

have worked
applied electric field and this is why most mate-
rials show resistance to electrical current.
However, according to the BCS theory, at fairly

diligently to
low temperatures free electrons in certain crys-
tals can pair up and form “Cooper pairs” as the
result of electron-crystal interactions, and these

obtain materials special pairs do not collide or interact with the

neighboring atoms as the regular electrons

that would
do. Furthermore these Cooper pairs will form
a collective group, known as a Bose-Einstein
condensate, and they will move as a group in

exhibit super-
unity, without being affected by the neighboring
atoms, or in other words without any resistance.
We can think of the electrons in a resistive mate-

conductivity rial as single individuals dancing on the floor of

a nightclub. As these singles try to move around,

at higher
every now and then they will collide or interact
with someone dancing on the floor and this will
slow down their movement in the desired direc-

temperatures.
tion. On the other hand, we can imagine the
group of Cooper pairs in a superconductor as a
group of couples dancing in a ballroom. Although
the ballroom will be packed with dancing couples,
It is still a mystery whether superconductivity these couples will follow a specific choreog-
at room temperature is actually achievable, raphy to move around without colliding with
but according to MIT graduate student Fahad other couples. The group of couples will move in
Mahmood shining ultrafast lasers on thin-film unity in a seamless way without any collisions as
superconductive materials could answer this long long as they follow the choreography. According
unsolved mystery. to Schrieffer, one of the contributors of the BCS
theory, the effect of low temperature supercon-
In 1986 it was discovered that cuprates, which ductivity in certain crystals can be thought of as a
are crystals made of copper and oxygen, exhibit special choreography of electrons when they form
superconductivity at temperatures as high as 90K Cooper pairs and dance in unity. [1]
and this new family of materials were named
high-temperature superconductors. Although “However, the effect of high temperature super-
this discovery facilitated many practical applica- conductivity observed in cuprates can not be
tions that were impossible with regular super- explained with the BCS theory,” Mahmood says.
conductors, the working principle of these new “Although we still don’t know how high temper-
materials could not be explained with the BCS ature superconductivity exactly works, we think
theory, which first explained the working prin- that phonons, magnons, or some other quasipar-
ciple of conventional low temperature supercon- ticle is probably responsible for it.”
ductors and garnered a Nobel Prize in Physics for Mahmood is working with ultrafast lasers and a

38 DEPARTMENT OF PHYSICS | MASSACHUSETTS INSTITUTE OF TECHNOLOGY

thin-film material called LSCO cuprate, which The observation that charge density waves and
is a type of high temperature superconductor, to groups of Cooper pairs coexist in a supercon-
solve the mysteries of high temperature super- ductor has interesting implications for high
conductivity. He explains that, by shining a very temperature superconductivity. Although it is still
powerful laser pulse on the LSCO cuprate, super- unknown whether these two phases of electrons
conductivity can be disrupted and some of the compete with each other at any time, researchers
Cooper pairs, as energetic photons interact with including Mahmood’s group are devising new
them, can be separated into two different elec- experiments to understand the nature of interac-
trons. This is in principle similar to any phase tions for these two phases. A recent experiment
change we see in our daily lives; if you apply an done by another research group at Max Planck
excess amount of energy to a cube of ice, either Institute in Germany showed that by shining
by heating or by shining a powerful laser pulse, ultrafast laser pulses on a superconducting
it will melt into water. Just as the more ordered cuprate, the effect of charge density waves could
crystal structure of ice turns into the less ordered be eliminated and they were able to get super-
liquid form, the ordered group of Cooper pairs conductivity at 250K for a very short duration
will become an unordered roup of electrons. of time. This is a very important step towards
Mahmood is excited about what happens next. attaining superconductivity at higher tempera-
“Once the applied laser pulse ends, the supercon- tures and naturally raises the question: if charge
ductor will slowly return to equilibrium, which density waves in a superconductor can be elimi-
means that separated electrons will recom- nated completely, can superconductivity at room
bine to form Cooper pairs again. How exactly temperature become a reality?
do these electrons bind each other and through Mahmood thinks that if we can solve the mystery
what mechanism do they recombine into Cooper of high temperature superconductivity, it will
pairs? These are the questions that can give us provide us with the required information to
important insights about the working principles engineer new materials that could exhibit super-
of high temperature superconductivity and the conductivity even at room temperature. High
good thing is they can be answered by our ultra- temperature superconductivity is one of the
fast optical techniques,” says Mahmood. most exciting research areas in condensed matter
Besides disturbing the Cooper pairs and physics, because if we can achieve superconduc-
observing the recombination process, these ultra- tivity at room temperature, that would be one of
fast optical techniques can be also utilized to orga- the greatest feats of the human history. Practical
nize the electrons of a superconducting material implications of such an achievement would range
in an interesting way called charge density waves. from ubiquitous quantum computing and super-
Charge density waves are sinusoidal distribution efficient data centers to lossless transmission of
of electrons, similar to the water waves on the electrical power.
surface of the sea. This is an interesting forma-
tion, because in a regular material electrons form References:
[1] Schrieffer, John Robert. “Superconductivity:
a uniform distribution, similar to the smooth
A Dance Analogy.” Superconductivity. AIP, 07 July 2007.
surface of a lake. Even more interesting is the fact Web. 13 Apr. 2016. <http://www.aip.org/history/
that charge density waves and superconductivity exhibits/mod/superconductivity/03.html>.
can coexist in the same material. That is to say,
when a snapshot of the material is taken at a point
Ahmet Musabeyoglu graduated from MIT
in time using ultrafast spectroscopy, electrons this June with a Bachelor of Science degree in
could be observed in the charge density wave physics and electrical engineering. During his
state, however in another snapshot taken shortly time at MIT, he enjoyed playing soccer, building
after electrons could be observed in the supercon- electronics circuits, and watching a lot of movies.
ductivity state.

FROM GR APHENE TO GAL A XIES: GRADUATE PHYSICS AT MIT 39

 BY CONNOR POPIK

Anand
Natarajan
Entanglement, the phenomenon where
characteristics of one particle depend
on another, has captivated physicists
since Schrödinger first described it in
1935. Part of this interest stems from
entanglement’s useful real world
applications such as decryption
and information transfer.
But before
On the surface, testing whether two particles are
entangled or not – what physicists call separable
– seems easy. For electrons with spin, a quantum

physicists use characteristic that is either up or down, measuring

the spin of two entangled electrons will always

entangled
yield one electron with spin up and one with spin
down. For separable electrons, some measure-
ments would yield both electrons with spin up

particles to
or both with spin down, proving that the elec-
trons were originally entangled. But while deter-
mining whether two electrons are entangled after

accomplish
measuring them is simple, predicting whether
more than two electrons are entangled without
measurement is incredibly hard.

any of these At MIT’s Center for Theoretical physics, graduate

student Anand Natarajan is exploring the quantum

tasks, they are

separability problem, called QUSEP, by searching
for faster and more accurate algorithms to deter-
mine whether particles are entangled or sepa-

focused on
rable. Originally interested in particle physics while
completing his undergraduate degree, Natarajan
took theoretical computer science classes before

solving a
being introduced to the fields of quantum informa-
tion and complexity. “Eventually I realized there’s
this field where you can think about complexity

more difficult
theory and physics,” he says. “That’s what led me to
be interested in quantum information.”

problem:
But unlike the fields of classical information and
computing, which today require testing and imple-
mentation on physical computers, quantum infor-

predicting
mation and computing is far more reminiscent of
theoretical physics in the early 20th century. When
tackling problems such as QUSEP, “It’s mostly pen

whether
and paper or on the blackboard,” says Natarajan.
At its most abstract level, QUSEP can be thought

particles are
of as the problem of separating the continuous
“region” of all possible quantum states into two
regions representing unentangled and entangled

entangled in
states. Imagine a large pasture of cows, your unen-
tangled states, with a cluster of sheep, your entan-
gled states, in the middle and the task of placing a

the first place.

fence around the sheep so as to completely sepa-
rate them from the cows. From a high vantage
point, you could imagine that it is quite easy to
place a fence that completely separates the sheep
from the cows, but the quantum world is not

42 DEPARTMENT OF PHYSICS | MASSACHUSETTS INSTITUTE OF TECHNOLOGY

so kind. Grass has grown far above the animals, While possible solutions to QUSEP are fasci-
requiring you to check each animal before the nating in their own right, Natarajan’s research
fence is placed. By analogy, an algorithmic solu- into QUSEP also brings physicists closer to
tion to QUSEP would efficiently check and sepa- understanding a complexity class called QMA,
rate the sheep from the cows, or the entangled or quantum Merlin Arthur. QMA is the set of
states from the unentangled states. problems for which a quantum computer can
only quickly check solutions rather than quickly
But there is a catch that makes Natarajan’s search
finding the solution itself. So far, complexity
for more efficient algorithms far more complex:
researchers believe that there is a difference
QUSEP was conjectured in 2003 to be a member
between QMA(1) and QMA(2), the classes where
of a class problems called NP-hard. While clas-
a quantum computer verifies possible solutions to
sical computers can quickly check the solutions
one or two problems respectively.
to NP-hard problems, computers cannot quickly
find the solutions themselves. This distinction Natarajan likes to think of the difference between
has far-reaching consequences for the problems QMA(1) and QMA(2) as interrogations of one or two
theorists can solve efficiently. Since the time it witnesses aided by a quantum computer that attempts
takes classical computers to solve problems like to determine whether they are lying or not. “If I want
QUSEP scales exponentially, even the fastest to interrogate one witness, they can come up with
computers may not find a solution in a human whatever lie they want to tell”, says Natarajan. “But if I
lifetime. While QUSEP’s membership in the have two witnesses who are not allowed to communi-
NP-hard class is still debated, complexity theo- cate, it’s much harder to coordinate their lies”.
rists doubt classical computers will ever be able Remarkably, if Natarajan’s group is able to find an
to solve QUSEP easily. efficient quantum algorithm to solve QUSEP, they
This revelation presents a challenge to researchers will also prove that there is no difference between
like Natarajan who are searching for efficient a quantum computer interrogating a single person
algorithms to solve QUSEP. While researchers or two people, an exciting discovery in the field
must accept the possibility no efficient classical of quantum information. While special cases of
algorithms, they can still try to optimize inef- this problem have been proven, researchers are
ficient algorithms to study QUSEP. Towards still searching for a general solution. “People have
this end, Natarajan has developed an algorithm gotten tantalizingly close to this,” says Natarajan.
around the sum-of-squares hierarchy along with At its core, Natarajan’s research is just as focused
fellow researchers Aram Harrow and Xiaodi Wu. on identifying the questions that we will never
The group’s algorithm reduces QUSEP to an opti- answer efficiently as it is on answering questions
mization problem, which is like finding the highest that we can answer. Problems like QUSEP and
peak in a mountain range. The key development complexity classes like QMA form distinct limits
of Natarajan’s group was the inclusion of mathe- on our theoretical understanding of physics. Even
matical constraints that allowed their algorithm to if a time efficient QUSEP algorithm is proved
find solutions faster than previous algorithms. “We impossible, that failure would yield signifi-
used this miraculous result from some people in the cant advances in physicists’ understanding of the
math literature,” says Natarajan. “When you add differences between QMA(1) and QMA(2). But for
these special constraints… this hierarchy converges the time being, “QMA is much less well-under-
at a finite level to the exact value.” In addition, stood,” says Natarajan, leaving the deceptively
Natarajan’s group examined the algorithm’s perfor- simple separability problem still unsolved.
mance when very close to the boundary between Connor Popik is a rising senior majoring in physics
the region of entangled states and the region and economics at MIT. After graduation he hopes to
of unentangled states and found it as precise as eventually pursue a career that links biophysics and
previous algorithms. healthcare economics at a biotech company.

FROM GR APHENE TO GAL A XIES: GRADUATE PHYSICS AT MIT 43

Every day, across forty-three orders of magnitude in distance scales,
scientists are making discoveries that affect society. To support sound
public policy about the issues which arise from this research, it is essential
for physicists to engage in the public debate. MIT 8.226 explores some of
the questions that our present scientific achievements raise, both for our own
community and for the public at large. By considering topics as diverse as
climate change and nuclear nonproliferation, this class makes the case
that physics is a necessary part of the national discourse.

These essays were written as an exercise for MIT Class 8.226: 43: Orders of Magnitude
March 2017

http://stellar.mit.edu/S/course/8/sp12/8.226
PROFESSOR J.M. CONRAD