For me, that is what's fascinating: the fact that these are enigmatic and mysterious entities that both exert gravity—black holes are the most extreme gravitating objects given their high density and compactness; and the compelling evidence for dark matter comes from the gravitational influence it exerts on the motions of nearby stars and gas. Both have consequences, in terms of the predictions of the general theory of relativity, and yet we simultaneously know a lot about them and nothing. That paradox excites me because there's room in that frontier to take creative, intellectual risks and come up with new ideas. That's where I have placed myself, and I'm working on trying to use the best datasets to see what give there is, what inconsistencies might pop-up between theory and the state of the art observational data.

For example, one intellectual issue that I'm interested in, as I said, is that there's a standard model that works incredibly well; however, there are a few gaps in this model. The question is whether these gaps portend new ideas and discoveries as we learn from history. For example, in the 1850s, when Uranus was discovered there was a wobble in the orbit of Uranus that deviated from the prediction of Newton’s laws of gravitation. The astronomer Urbain Le Verrier predicted that there must be another gravitating body nearby. Neptune was predicted and it was found, and all was well with Newtonian theory. The wobbling in the orbit of Uranus was not a case where Newton was proved wrong—it's just that we needed to refine the theory. We needed to add another body to the Solar System.

Similarly, as more accurate data for the orbit of Mercury become available, it was found that Mercury did not fit Newton's theory either. Urbain Le Verrier played the same game. He said, "Hey, maybe you have another planet between the Sun and Mercury. That could explain the wobble." He called it "Vulcan." People looked for it, claimed to find it but they didn't find it; it doesn't exist. What was needed in this instance was an entirely new theory. The theory of general relativity was needed to explain the orbit of Mercury, the precession of the perihelion of Mercury. Newton was upended in one case (Mercury), and Newton's ideas were just refined in the other (Uranus). A disagreement between data and our theoretical expectation can therefore be one or the other. I'm looking for these gaps, these mismatches between data and theoretical predictions, with the hope that it may point us to a brand-new discovery potentially of an entirely new theory. We are in need of a new explanation possibly for dark matter and a better understanding for black holes. That's what I am trying to do in my work. It's very exciting because one is challenging the current paradigm and looking for clues that would provide us ideas to come up with alternative explanations. We don't have one yet actually we don't have one that's compelling. Right now, I am definitely involved and immersed in trying to find these gaps.

We do live in a very peculiar universe. The inventory of our universe is most bizarre. Most of what we know is a tiny fraction of the universe, 4 to 5 percent is the matter that we know, the atoms that we know, the elements that are in the periodic table. The vast majority of stuff in the universe—both dark matter and dark energy, which dominate the universe—its fate and its contents are unknown. I'm being somewhat modest when I say that we're looking for a gap. We're actually waiting for some kind of revolution. But one never quite knows when revolutions start.

We are in the midst of figuring out what kinds of reformulations of these mysterious entities dark matter and dark energy might be plausible. For dark energy, at the moment we have a placeholder theory. We just have a way to frame what has been measured so far. There isn't a theory of when dark energy is generated, when it manifests in the universe, how it kicks in. There are many ideas that are floating around, but none are personally convincing to me or the cosmology community. At the moment, strangely, the formulation of dark energy that appears to work—the working model—is consistent with the infamous cosmological constant that Einstein craftily inserted into his equation way back when in order to hold the universe steady.

It turns out that there is a huge gap in understanding of the cosmological constant from the physics side. A physical explanation for the cosmological constant, for example, is 120 orders of magnitude discrepant from the measurement, from the supernovae data that the astronomers published about fifteen years ago. There's a yawning gap, and there's a lot of work on trying to come up with a model that folds in what we know about Big Bang cosmology. Now we have a lot of incontrovertible observational evidence, so any formulation that we come up with has to be consistent with what is observed now, and has to make further predictions as well that can be tested.

With the dark matter question, which I am a little more actively involved at this moment because there is a brand new dataset that is just incredible, much of the mapping of dark matter that I do is to try and understand how granular dark matter is, in terms of how it's spatially distributed in the universe. We know, for example, that dark matter is lightly smeared everywhere in the universe, but that there are regions where it is lumped and accumulated due to gravity. These regions that have copious amounts of dark matter reveal themselves to us because of the light bending that they cause, gravitational lensing of background galaxies that lie behind them.

This phenomenon was predicted by Einstein, according to which when you have a distant galaxy and it is viewed through a screen of a massive lump of dark matter, light rays from these distant glowing galaxies get deflected and what you end up seeing is a distorted shape rather than the true shape of the background galaxy. Sometimes you have a huge lump of dark matter. Both dark matter and visible matter will bend light, but it's just that dark matter is the dominant matter component, and it's implicated in much of the light bending that we see. Sometimes there is so much dark matter concentrated that you split a single light beam into multiple beams, so you end up seeing multiple copies of the same, single background galaxy. In fact, you have only one true object that's emitting light, but you end up seeing multiple copies of it all distorted and misshapen.

The Hubble space telescope, with the exquisite resolution offered by its cameras, has allowed us to chart these multiple itsy-bitsy pieces which are actually images of the same individual object in detail. What I'm doing right now, for example, is looking at these extreme distortions, this strong gravitational lensing, and then doing a tomography of what all dark matter has to lie in the foreground to give us the distortions that we see. With a prior assumption for the shapes of normal undistorted shapes of galaxies, we can use the observed mangled shapes to back out the dark matter distribution. Galaxies are born with a range of shapes, and we know the distribution so we start with that, and then we look at the extreme distortions and try to figure out how much dark matter is causing these distortions.

With the Hubble space telescope, the resolution is so exquisite that we are able to chart very small lumps of dark matter. For example, with the latest data from this project called the Frontier Fields, gravitational lensed images produced by clusters of galaxies—these are very large repositories of dark matter in the universe—are detected. My group has just finished analysis, and the paper has just been submitted where we are able to chart lumps of dark matter that are as small as the dark matter content of a tiny dwarf galaxy in our neighborhood, about 1 billion times the mass of the sun. That's the total mass of the galaxy. We are able to resolve such small dark matter lumps in these very distant objects. These are very dim little blobs, fuzzy little galaxies, at huge cosmic distances that would otherwise be unavailable to us, in terms of directly looking at them and studying them. And here with the help of gravitational lensing we are mapping these tiny clumps.

What this mapping technique has allowed us to do is figure out how clumpy dark matter is on these very small scales. The reason this is an interesting exercise is that there's a concrete prediction from this theory of cold dark matter that you should have clumps of all sizes. In fact, you should have a never-ending set of smaller and smaller clumps in the universe. That's one of the key predictions. For example, if we see a cutoff in the clump size, that tells you that the nature of dark matter is something fundamentally different from that predicted by the cold dark matter theory. At the moment, cold dark matter looks pretty good, given the quality of the data. It looks like the theory works quite well, so we're still very much in search of the gap!

My work has focused on these objects called clusters of galaxies. These, as I said earlier, are the largest repositories of dark matter. They are 95 percent dark matter. Essentially, a cluster is a giant blob of dark matter that holds about 1000 galaxies that are swirling around but held in place by the gravity of the dark matter. The cluster acts as a very efficient lens in deflecting light, and the strength of the deflection, the distortion that we see, is proportional to the distance between us, this cluster, and the distant object. Since distances depend on the geometry of the universe, lensing allows you to measure dark energy as well, which is why I love these objects. They allow you to simultaneously map dark matter and get a handle on dark energy. This is not the usual way that astronomers have been measuring dark energy. These are a new class of objects that we are deploying for this task. Clusters are complicated objects in and of themselves, which is why they've not been so popular. However, the data has gotten so much better in the last five years that we understand the complexities and can model them extremely accurately.

Cluster environments are very violent places. There's a lot of galaxies transformation that is onging in these dense environments—galaxies smashing into each other, stars spilling over and dark matter flying about. These environments are very complicated, in terms of the physics and the transformation of galaxies that's occurring inside these objects, but we understand them rather well now, both with numerical simulations, as well as real data.

Clusters have now become new cosmological tools, very effective astrophysical laboratories. Now that we understand them, they serve as unique laboratories. Before that, of course, they were enigmatic objects; we didn't understand them. Now we understand them quite well. I like them because they are complex, and the complexity requires of us a more detailed physical understanding. With the data, we are now forced to go beyond a simple mathematical model. You have these various observational views of clusters, clusters viewed through many different wavelengths, and this has now led us to have a composite understanding of these cosmic objects. The dark matter in a cluster is inert and is just sitting there doing nothing, but you have all these galaxies that you see glowing in the optical, giving us optical light.

We have to square everything that we see and come up with a synthetic model. I'm interested always in making comprehensive models, not just an explanation that works for one tiny aspect of the phenomenon. I'm interested in a more complete understanding. Clusters are quite enticing as cosmic objects.

Structure formation in the late universe—what I work on—is somewhat decoupled from the details of the very early universe and the pre-Big Bang conditions. It's contingent on our universe having had an inflationary stage, because without all of that you can't explain the structure formation that we see today. The pre-Big Bang theories are not as directly relevant. That's a big problem with those theories anyway. You cannot test them empirically in the ways in which we have become accustomed to in science. That's why it's hard to discriminate amongst these various string theory models, all of which are compelling in some way but they can't be independently tested. They all need to patch onto the Big Bang and inflation and structure formation that follows afterwards. You cannot independently test the validity of those theories at the present time, but the course of future science cannot be predicted.

For the work I do, knowledge of the initial conditions that started off our universe do not have a direct bearing, all of those details can be a somewhat uncertain, and I can still do the kinds of testing and probing of dark matter that I do. In many ways, the late universe, if you will, which is 400,000 years after the Big Bang pretty much, is somewhat decoupled. The theories that describe structure formation have to be consistent. The initial conditions for this theory are provided by inflationary models like Alan Guth's models, but the multiverse, per se, is not directly relevant. However, in order to make sense of why the universe that we have is as it is, why structure formation is the way it is, the idea of a multiverse is a very attractive explanation. I, for one, believe in the multiverse, and I use the word believe very intentionally here.

It is definitely the golden age in cosmology because of this unique confluence of ideas and instruments. We live in a very peculiar universe—one that is dominated by dark matter and dark energy—the true nature of both of these remains elusive. Dark matter does not emit radiation in any wavelength and its presence is inferred by its gravitational influence on the motions of stars and gas in its vicinity. Dark Energy, discovered in 1998, meanwhile is believed to be powering the accelerated expansion of the universe. Despite not knowing what the dark matter particle is or what dark energy really is, we still have a very successful theory of how galaxies form and evolve in a universe with these mysterious and invisible dominant components. Technology has made possible the testing of our cosmological theories at a level that was unprecedented before. All of these experiments have delivered very exciting results, even if they're null results. For example, the LHC, with the discovery of the Higgs, has given us a lot more comfort in the standard model. The Planck and WMAP satellites probing the leftover hiss from the Big Bang—the cosmic microwave background radiation—have shown us that our theoretical understanding of how the early fluctuations in the universe grew and formed the late universe that we see is pretty secure. Our current theory, despite the embarrassing gap of not knowing the true nature of dark matter or dark energy, has been tested to a pretty high degree of precision.                      

It's also consequential that the dark matter direct detection experiments have not found anything. That's interesting too, because that's telling us that all these experiments are reaching the limits of their sensitivity, what they were planned for, and they're still not finding anything. This suggests paradoxically that while the overall theory might be consistent with observational data, something is still fundamentally off and possibly awry in our understanding. The challenge in the next decade is to figure out which old pieces don't fit. Is there a pattern that emerges that would tell us, is it a fundamentally new theory of gravity that's needed, or is it a complete rethink of some aspects of particle physics that are needed? Those are the big open questions.

I've been trying to test aspects and predictions of this standard cosmological model. As I said, the standard model seems to be working very well. At the moment, I am part of the establishment that is still finding strong support for the model. There's nothing amiss that has emerged yet with the current paradigm, but what has been crazily suggested is an alternative to dark matter. A theory that entirely dispels the need for any unseen matter. There's a theory called MOND, which is Modified Newtonian Dynamics. It's a theory that was proposed by Jacob Bekenstein and Mordehai Milgrom, two Israeli astrophysicists. The idea here is that in the regime when accelerations are very small you have to amend Newton's laws. Remember, I was talking about how, when you find mismatches, you could interpret it as a current theory just needing a little refinement, or a fundamentally new theory. This model suggests that all you need is a slight tweak to Newton's laws and that would be sufficient to explain away dark matter; there's no need for dark matter in MOND.                                 

Some of my work confronted MOND theory with clusters, these complicated objects. It turns out that galaxies are also repositories of dark matter, and you know that there's a lot of dark matter in them when you map the speeds of stars around a galaxy to figure out how the stars are held together. Unlike the solar system where, because the sun is the dominant gravitational object, Mercury is moving the fastest. As you go further out in the solar system, the speeds of the outer planets are slower as they are further away from the sun. In a galaxy, the situation is the opposite, strangely. It turns out that the speeds of stars from the center out in a galaxy, actually rise and they flatten out, suggesting that there is gravitationally important material smeared everywhere in the galaxy. There's a notion of a dark matter halo. We believe that all galaxies are held together by the gravity of dark matter. In fact, there’s a lot of dark matter in galaxies.

This Modified Newtonian Dynamics theory suggested by Milgrom and Bekenstein, exquisitely explains these rotation curves (speeds of stars) of individual galaxies; however, it fails miserably to explain the properties of clusters. That's why I like clusters—they are a powerful test-bed for theories. They tried modifying MOND, making it a little more sophisticated to explain light bending without dark matter, but it turns out that MOND theory does not survive the test of gravitational lensing by clusters of galaxies.

That’s the one and only competing theory at present, I wouldn't call it a total alternative because it didn't have the entire framework of structure formation like the standard model, current cold dark matter model does. MOND theory is in a bit of hot water, so that's not a viable alternative. It was a very interesting approach, and there are still people who are invested in that theory and who are trying to refine it to make it work, but not with much success at the moment.

What's incredible about this golden age of cosmology is the time between the proposal of a new radical idea and when it's tested has shrunk. You could come up with a new idea. I have one in the running right now, which is a new idea of how the first black holes formed. The standard idea is that they are the end states of massive stars. You form the first stars in the universe, they burn up all their hydrogen, they leave these little seed black holes, and then these grow very rapidly in the early universe and generate all the massive black holes that we see in the centers of most galaxies, including the ones that glow as quasars, which can be ~1 billion times the mass of the sun. We're seeing quasars out very large cosmic distances, so clearly quasars form early and often!

Now we are detecting quasars when the universe was about 1 billion years old, when the first structures formed from the initial fluctuations that we see in the microwave background when the universe was 400 to 500,000 years old. We're starting to see the seeds of the formation of the first structures. But we're seeing these quasars about 1-2 billion years after the Big Bang. These black holes powering quasars can be about 10 billion times the mass of the sun—these black holes—that are actively feeding, that we see as quasars, these bright beacons. The question is, how can you form them from the tiny seeds, remnant of the first stars, within 2 billion years to these huge masses?                                 

We came up with a solution about ten years ago, suggesting that there's an easy solution. Instead of making tiny seeds, if the physics permitted you to make very massive black holes, to assemble 10,000, a million times the mass of the sun initial black holes from the get-go, that resolves this timing/feeding problem. We proposed this idea of Direct Collapse Black Holes that are very massive and form very early in the universe. You basically bypass the formation of a star. Instead of forming a star, you somehow directly form a black hole. To do that, you need certain physical conditions. It turns out that in the early universe those conditions are available. You need large pockets of gas that swirl very rapidly to the center and form a huge dense concentration in the center of a proto-galaxy.

We've made predictions for these Direct Collapse Black Hole seed models. Of course, there are consequences for making black holes, or a population of them, this way rather than through stars. The key predictions of this scenario are that you should see very bright quasars much earlier into the universe, because you form very massive seeds to start with. This is a prediction that is going to be tested by the James Webb Space Telescope that we expect to be launched and going up in 2018.

It will be very exciting to be either proved right or wrong. That's the beauty of science. The fact that you can sit in your little office in New Haven and make a calculation that a community of people will take seriously enough to test it, and do so within ten, fifteen years is awesome. It's super exciting to be part of that process. What a privilege to be born at this time when that's actually possible. Of course, it would be nice to be right.

The question of whether you can be proven wrong multiple times and yet be a respected scientist is an excellent one. There have been many cases where you've had brilliant minds who have come up with powerful ideas, many of which have been proven to be wrong, and yet they are considered, still, the brightest astrophysicists.

The name that comes to mind automatically for me is Arthur Eddington, who formally opposed the idea of black holes. The fact of these enigmatic objects possessing a singularity—that peculiar nature of a black hole, was something that was very troublesome and unappealing to him. He opposed, famously, Chandrasekhar, who came up with the idea of stellar evolution leading you to these very compact objects, including black holes, as their end product. He contested that very fiercely. To his death Eddington didn't believe in black holes. And yet they have been shown to be right and real. Yet he was an incredible astrophysicist and it would be right to say, reading his papers and his works, he probably understood general relativity at a profound level that very few since him do.

The allure of a career in cosmology will always be there. As a young girl growing up in India, I started out being an amateur astronomer. I looked through a telescope saw Halley's Comet. The allure of the night sky will always be with us. There is never going to be a dearth of people who have a sense of curiosity and wonder about the universe and will follow through on it. The question is whether society will value that, and whether there will be a society, after all.

For me, what is a little ironic about people who want to establish colonies on the moon or on Mars is that we have this beautiful planet and we can still save it. We can take stewardship and do something about climate change. It's not yet time to give up on this planet and what we have here. Personally speaking, though, I don't like the world very much. I don't like the inequities, I don't like the strife, don't like wars. For me, working in cosmology is a form of escape. It is definitely a form of escape. Actually, it's a meditation and a form of escape.

I've been very fortunate. Much of my career has been not strategized, but I've ended up with some of the most fantastic mentors. It first started with this astronomer who was the Director of the Nehru Planetarium in New Delhi, Nirupama Raghavan. She's the one who gave me my first taste of research. Unfortunately, she passed away before I got tenure, so she didn't live to see that, although she saw me growing in the field. After that, I was very lucky to get a full scholarship to go to MIT. My parents are academics. I came from India as a young person to MIT and there I had the most amazing experiences learning, but also in terms of the people that I worked with, in terms of research.

My undergraduate thesis was with Alan Guth, who is an incredible human being, very thoughtful, both in his scientific work and as a person, and also very much someone who is driven by the love of what he does. That's what's been important and interesting for me in my career. I've somehow encountered these amazing minds, who are also incredible human beings, but for whom the love of what they do is palpable, and the passion is palpable.

For my PhD, I went to work, specifically, to Cambridge with Professor Martin Rees. It was just incredible, how much energy, time, and effort he gave to me, to talking about science. And not just about the work, but also about a life in science, and about not so much careerist matters, but about intellectual stuff, about ideas. This sparked an interest that's become more and more enduring for me: the interest in how new, radical ideas get accepted, how they get contested, what's the pushback, and what it takes to have an idea tested—how to stick it through—and how sometimes you may be the lone person supporting an idea or pushing it forward, and if you think it's a good idea, how to shepherd it until it's proven right or wrong. I was very fortunate and lucky, I realize, in the mentors that I've had.

I've also had a brief detour in the history and philosophy of science. There, I had some very exciting mentors as well. I interacted a lot with Evelyn Fox Keller, who thinks very deeply about science. That has also prompted me to be the kind of scientist I wanted to be: someone who thinks a lot about the implications of what they're doing, not just in terms of the immediate sense, the implications of a particular calculation, but the broader scheme of things, of a scientific idea and how it becomes knowledge, that whole process. Being much more thoughtful about what I do and the larger context, I learned from all these people.

Now I've been teaching at Yale for more than ten years, so I've had some fantastic students and colleagues who have helped me in refining and shaping how I think. In particular, one of the things that I am after is clarity. I want to understand things at a very profound, clear level. I'm aspiring to this, I'm still learning, I'm still working on getting there. A lot of my students, my colleagues that I engage with, that I argue with, have helped me tremendously and have been helpful in this quest.