Jessie Muir on the mystery of dark energy

The universe is expanding faster and faster all the time, and Jessie Muir wants to understand why. Muir is a postdoctoral researcher at Perimeter Institute and a member of the international Dark Energy Survey (DES) collaboration. She co-leads the DES analysis team that seeks to understand the mysterious dark energy driving the universe's accelerating expansion by analyzing galaxy clustering and weak lensing. In this episode, she chats with Lauren and Colin about her work at the intersection of theory and experiment – and how drawing science-themed cartoons helps her grasp and share complex topics in captivating, accessible ways. View the episode transcript here.

We are looking for feedback about our audience’s preferences as we chart the path forward. Please take a few minutes and fill out our podcast listener survey.

Conversations at the Perimeter is co-hosted by Perimeter Teaching Faculty member Lauren Hayward and journalist-turned-science communicator Colin Hunter. In each episode, they chat with a guest scientist about their research, the challenges they encounter, and the drive that keeps them searching for answers.

The podcast is produced by the Perimeter Institute for Theoretical Physics, a not-for-profit, charitable organization supported by a unique public-private model, including the Governments of Ontario and Canada. Perimeter Institute acknowledges that it is situated on the traditional territory of the Anishinaabe, Haudenosaunee, and Neutral peoples. Perimeter’s educational outreach initiatives, including Conversations at the Perimeter, are made possible in part by the support of donors like you. Be part of the equation: perimeterinstitute.ca/donate

Episode Transcript

(00:00):
(bright music)
- Hello, and welcome back toConversations at the Perimeter.
Today, Lauren and I are really excited
to share this conversationthat we had with Jessie Muir
who's a postdoc hereat Perimeter Institute,
and she studies the mysteriousphenomena of dark energy,

(00:23):
which is believed todrive the acceleration
of our universe's expansion.
- I learned so many thingsfrom this conversation
that I didn't knowbefore about dark matter,
dark energy, gravitational lensing,
and I even learned a newscientific term that I really love
called galaxy clumpiness.
It was just fascinating tohear about how Jessie's work
really relies on an interplaybetween theory and experiment.

(00:46):
And she told us about her work
as part of the Dark EnergySurvey collaboration
and how her team works
to process and analyze mindboggling amounts of data.
- What I found also fascinating
was not only does she do allof this computational work,
but she actually went tothe telescope in Chile
on top of a mountain
where this Dark Energy Surveyis doing its observations,

(01:07):
so she got a real hands-on experience
of what it's like to be an astronomer.
- She also told us abouta series of cartoons
that she works on to helpcommunicate her science
and make it more accessible for everyone.
I know you're gonnaenjoy this conversation,
so let's step inside theperimeter with Jessie.
(bright music continues)
- Hello, Jessie, andthank you for being here

(01:28):
at Conversations at The Perimeter.
- Hey, thanks for having me.
- We're really excitedto chat with you today.
In particular, I'm excitedto learn about dark energy,
which is related to some work
that you're gonna tell us about,
and dark matter, all things dark,
because we haven't reallytalked to any experts
about what these things reallyare or what they aren't.

(01:48):
Can you start us off by telling us,
what do we know about darkenergy and dark matter?
Are they even related
aside from both having dark in the name?
- I think the main thing that relates them
is that they have dark in the name,
and they're labels that we give
to components of the matterand energy in the universe
that we are fairly sure are there

(02:09):
based on how they influence visible matter
that we can see and measureand detect and study,
but we fundamentallydon't know what they are.
But these are twodifferent unknown things.
We can get into this in more detail,
but sort of the simplisticdescription I give
of what makes them different
is dark matter seems tobe some type of particle,
but it clumps up underthe influence of gravity,

(02:29):
so it's not uniform in space.
It behaves in some wayslike ordinary matter
that we're familiar with.
It just doesn't seem tointeract through light
or through other forces or if it-
- Because ordinary matterdoes that as well, right?
It clumps in areas of high gravity?
- Yeah, yeah, so thething that gravity does
is it causes mass to wannafall towards other mass.
And it seems, as far as we can tell,
that both dark matter and ordinary matter

(02:51):
seem to feel gravity in the same way,
whereas dark energy seems likewe're not sure what it is,
but it seems to be more likesome property of space itself.
So dark matter clumps upunder the influence of gravity
and we can see how it influencesthe formation of galaxies
and how stars move ingalaxies among other things.
And dark energy, we learnabout and we've detected

(03:11):
based on its influence of thevery large-scale universe.
So large and small scales
kind of have somewhat different meanings
depending on what field you're in.
In cosmology, we tendto refer to small scales
as anything under about30 million light years.
- (laughs) Just tiny.- So it's, you know,
maybe a little bitdifferent than the scales
of, like, colleagues overdoing quantum stuff here at PI.

(03:32):
- So if that's small, what is large?
- Generally, we work in alittle bit of different units
in cosmology, but like30 million light years
is kind of the benchmark foronce you're looking above that,
the universe isn't necessarily uniform
but in a statisticalsense, it becomes uniform.
So I guess you can maybe picture
looking at, like, a zoomed-inor zoomed-out picture

(03:53):
of, like, a lawn of grass.
If you're looking on small scales
that are sort of comparable to the size
of, like, little clumps of grass,
you might be concerned
with, like, oh, how isthis blade of grass growing
and how is it interactingwith its neighbors?
And so that would belike individual galaxies
forming and observing.
And when you zoom out, youknow, you can still see
that, you know, the groundisn't completely uniform,
there's still blades of grass there,
but you can sort of get a sense

(04:14):
of, like, the global properties
of, like, this grass tendsto grow in little clumps
or is it more spread out
or, you know, do we thinkit was grown there wildly
or using sod, or I don'tknow, maybe this is getting
a little bit obtuse-- No, I actually like that.
You know, it made methink of a golf course
where it's all grass,
but you look from above
and there's different characteristics,
different ways it grows.
And you mentioned darkenergy in comparison

(04:35):
sort of being an element ofspace-time, is that right?
It's something intrinsic to it?
- For this, maybe it's kind of useful
to, like, tell a little bit of the story
of how we learned about dark energy.
Up until the '90s,
we knew that there'smatter in the universe.
We've known there's dark matter,
sort of first hints showedup in, like, the '30s,
and then Vera Rubin made some measurements
of the motion of stars and galaxies

(04:57):
in, I believe, the '60s, maybe '70s.
So we've kind of known about dark matter.
We've had a good understandingof how gravity works
since Einstein published histheory of general relativity.
And given those things,
we know that mass attractsmass through gravity,
we know there's matter in the universe,
and so your expectation
is even if everythingis sort of thrown out
by the Big Bang in the early universe,

(05:19):
what you'd expect gravity to be doing
is that all that matteris being thrown out,
the universe is expanding,
gravity should be actingsort of as a friction.
It should be slowing that down.
Given a universe that containsmatter and that has gravity,
you expect to see
that the expansion of theuniverse is decelerating.
And what we found, or whatseveral teams of scientists
and since many haveconfirmed in the late '90s,

(05:40):
was that the universe'sexpansion is not slowing down,
it's actually accelerating.
And so previous to that,people were kind of looking
at, like, all right,we can measure the rate
at which it's decelerating
to learn about how much matter there is
and some stuff about the geometry
of the large-scale universe.
And this finding that theuniverse is accelerating,

(06:01):
like, it's like if you threwa baseball up in the air
and instead of coming back down,
it, like, zips off insome other direction.
So there's gotta be someother sorts of energy there,
and the simplest descriptionthat we can come up with
that dark energy could be
that would give us the sortof observable properties
that we're seeing
is that if empty space just hadsome intrinsic energy to it.
So sometimes people willcall this vacuum energy,

(06:23):
sometimes people will callit the cosmological constant,
and so what that means isit's some energy density
associated with empty spacethat's both constant in space,
so same everywhere in the universe,
and constant in time, sothe same energy density
throughout the history of the universe.
And it seems to havebeen causing acceleration
of the expansion of the universe
just in the relatively recent past.

(06:43):
Here, recent being on cosmologist scales
of the last couple billion years.
And so the picture you can have there
is the universe is expandingand it has some matter density,
but as the universe expands,
the same number of particlesare around roughly,
and that matter gets diluted.
So as the universe progressesthrough its history,
the matter density will drop,
and at a certain point,
the average density ofmatter in the universe

(07:05):
drops below that vacuum energy,that cosmological constant,
and that's when the universestarts accelerating.
So these different componentshave different influences
on the behavior of thespace-time in the universe,
and this is something
we can get out of Einstein'sgeneral relativity,
we can relate the behavior ofspace-time to the stuff in it.
And so when the relative contribution
to the total energydensity of the universe

(07:28):
switches from being matter-dominated
to dark energy-dominated
or cosmological constant-dominated
depending on which model you wanna use,
the expansion startsgetting faster and faster.
So we don't know what dark energy is,
but we can sort of place constraints
and say, is it a constant?
Does it have some time evolution?
Is it something that maybeinteracts with matter?
And given one of these assumptions,
you can go through anddo your calculations

(07:50):
for how that should affectthe expansion history,
how it'll affect how thematter is clumping up
to form galaxies and things,
and we can kind of testand constrain those.
That's a lot of the motivation
behind what I and a lotof other cosmologists do.
- It seems like a lot of thework that you specifically do
is trying to look atthe role of statistics
in understanding some of these properties,

(08:10):
so can you tell us in generalhow statistics comes in?
- So it comes in in acouple different ways.
One is, you know, ifwe're trying to describe
the large-scale universe,
you know, we look out in the universe
and we see millions andmillions of galaxies,
like the experiment I work on,
which I think we'll touch on later.
Like, we're working with a data set
with a couple hundredmillion galaxies imaged,
and that's only, like,one part of the sky,

(08:32):
and it's not looking out
as far as, like, futuretelescopes will be able to look.
We want to be able to test our theories
or to constrain the question
of whether dark energy's density,
like, varies in time or not,
which is sort of one of thestraightforward questions
you can ask about that model.
You wanna find things aboutthose measurements we're making
that you can actuallypredict with your theory.

(08:52):
And with our theory of the universe,
we're not able to say, "Ithink I'm gonna see a galaxy
at this location in space orthis coordinate on the sky."
What we can say is we have somepicture or some description
of how a universe thatstarted out very uniform,
so the density being basicallyalmost the same everywhere
but with tiny density fluctuations,

(09:14):
and then over time,given our understanding
of, like, what types ofmatter are contributing
to those fluctuationsand how gravity works,
how they grow over time.
So what you get is not a description
of, okay, I expect tosee a galaxy in spot A
and a galaxy in spot B,
but you can say I expectthat the sort of size
and fluctuations in density,
so, like, how do you compare

(09:34):
sort of the highestdensities in the universe
to the lowest densities,
and you can make predictions about that.
And you can also, in the same way
that, you know, our torturedlawn of grass analogy,
like, you might be able to tie,
like, how you put theseeds down on the ground
to how, like, clustered the grass is.
Are you seeing grass ina bunch of little tufts,
or is it pretty spread out uniformly?

(09:56):
We can make predictions
for, like, do we expect to see galaxies
distributed at random,
or do we expect to seethem clumped together?
And we can make predictions
for basically the probabilityof finding galaxies
separated by a givendistance in the universe
compared to an average distribution.
So we're describing statistical properties
of the distribution ofmatter in the universe.

(10:17):
And then statisticscomes in in another way
as, like, all right,given these measurements
of statistical properties in the universe,
how can we use that
to tell us about the physics of our model?
We have these measurements of,like, how close or far away
we expect to see galaxies to one another.
We can predict that with our model,
but we know our model hassome assumptions in it
and we need to be ableto do these calculations,

(10:37):
we need to make some assumptions.
But a lot of my day
and a lot of the work I dowith my close colleagues
is making sure that, all right,
we're trying to use these measurements
to say something very fundamental
about physics in the universe
of, like, does dark energyvary with time or not?
And we wanna make sure
that we don't mistake some complication
in, like, how supernova
blow gas out of galaxies or something.

(10:59):
Like, one of our bigchallenges in cosmology
is trying to make sure uncertainties
about the detailed calculations
of that smaller-scale astrophysics,so just galaxy scales,
doesn't influence the inferences
that we're making from the larger scales,
or wanna get as muchinformation out as possible
without biasing ourselves
and tricking ourselves into thinking
we discovered something about dark energy

(11:20):
when, really, we're not understanding
our modeling predictions.
So we do a ton of tests,we use a ton of simulations
to really make sure thatwe do that rigorously,
and then translating these comparisons
of model predictions data
into information aboutparameters of a model
or which model is better than another one
is the whole sort of subfieldof study in cosmology itself.

(11:42):
- Yeah, I would assume
this must be a really challenging problem
when you have so much data.
And I'm just curious, like,when you have all this data,
how do you go aboutapproaching the problem
of when you need tolook at the observations
you already have
versus when you need togo and collect more data
in a new way?
- There's a lot of value
in looking at what data we have on hand

(12:05):
and looking for new ways toextract information out of it.
So a lot of the measurements we make
of these statisticalproperties of galaxies
are looking at, like, the distances
between pairs of galaxies,
and you can go to sort of, wesay higher order statistics,
so that's, you know, statisticsbased on pairs of galaxies.
You can look at triplets of galaxies
and see, like, what kind oftriangles you expect to see
of different sizes and length scales.

(12:27):
And there's a whole field of research
which these calculationstend to be a bit harder
and the measurementstend to be a bit harder
of understanding, like,what kinds of physics,
either new or what we know about,
can you get more information from,
like, taking these maps we already have
and, like, pushing them harder
to get more and moreinformation out of that.
But then going and gathering more data
'cause the more galaxies youmake these measurements for,

(12:48):
the smaller the error barson those measurements are,
so, like, when you make a comparison
of your model prediction to the data,
if your data are more precise,like, they're measured well,
having more galaxies is good for that.
You can know that if you seea little bit of a deviation
between your prediction and the data,
you can be more confident that it's real
and not some, like, statisticalfluctuation or noise.

(13:09):
And I think most, if not all, cosmologists
are kind of engaged a bitin both of these things.
We're consistently planning,
like, working on the current generation
of experiments gathering data,
and sort of looking to thenext generation of experiments
which we'll be turning on.
And also, there's sort of alot of complementarity there.
So the experiment that Iwork on is a galaxy survey

(13:30):
called the Dark Energy Survey,
which is a survey that's mappedthe distribution of matter
in a patch of the sky
measuring a couplehundred million galaxies,
and we have the biggestdata set of its type,
so it's the most statisticallypowerful galaxy survey
of its type, which we canmaybe touch on it in a bit.
And so the constraints we can get
from studying that map of theuniverse is really exciting

(13:53):
and, you know, sort of pushing the bounds
of what we can do in cosmology.
It's also crucial as sort of a workshop
for developing techniques we'll need
when we go to the nextgeneration experiment,
which we'll get evenmore precise constraints,
and, you know, I mentioned wehave to spend a lot of time
accounting for, like, are theapproximations we're using
for our calculations accurate enough?

(14:13):
And as your measurements get more precise,
that answer can veryeasily turn from yes to no,
and so we have to, like,push the bounds on that
every time our data get more precise.
- You mentioned the Dark Energy Survey,
the experiment that you're working on.
Can you tell us sort of thegoals and motivations of that
and how it actually works?
Is this a telescope outin space or on a mountain,

(14:35):
or is it something else entirely?
- I guess maybe as a,like, basic definition,
a galaxy survey is someexperiment usually run by,
I think always run bya large collaboration
which you try to systematically,
like, observe a patch of the sky
and make a really uniform map
of the distribution of galaxies.

(14:56):
So instead of, like, pointing a telescope
at an individual galaxyor a group of galaxies
and taking detailed pictures,
we're trying to just map the sky
so we can make thesestatistical measurements.
The Dark Energy Survey
is what is known as an imaging survey,
which means on our telescope,
we basically have a giant digital camera,
and we can, like, take pictures of the sky
as opposed to, like, measuringthe colors very precisely.

(15:17):
That giant digital camera iscalled the Dark Energy Camera,
which we're very creative
with names clearly.- That's a good name for it.
- And it is on a four-meter telescope
called the Blanco Telescopein Cerro Tololo in Chile.
So it's on a top of a mountain.
You put telescopes on tops of mountains
because there's water in the atmosphere
and, like, turbulence in the atmosphere
can make images of space look blurry,

(15:38):
and so you wanna go towhere there's not much water
in the atmosphere andthere's not much atmosphere,
so, generally,observatories are in deserts
and on tops of mountains.
- You've said this is areally big collaboration.
Can you give us a sense of how big
and how the different teams
in this collaboration are organized?
- Dark Energy Survey has, Ithink, about 400 people in it.

(15:59):
It's been going for over a decade
so I think the camera was installed
on the telescope in 2011.
So this camera was builtspecifically for this survey.
It's specialized to bemore sensitive to red light
than your average chip thatwould be in a digital camera.
The CCD chips,
or the little chip that wouldbe in your digital camera,
for the telescope is likethree feet across so it's big.

(16:20):
So this collaboration worked on things
from planning the surveyto building the camera
to installing it to running the shifts,
so we did something like760 nights of observing
between, I think, 2013 and 2019.
And then there's a whole team of people
that go from sort of raw imagesfrom the big digital camera
and turn that into catalogsof where do we see galaxies,

(16:42):
what are their colors,what are their shapes?
These teams all overlap andpeople move between them,
but then there's going from those catalogs
to making these statistical measurements.
And then where I kind oflive within the collaboration
at the sort of end of that
is trying to go from thosestatistical measurements
to inferences about the physics.
So I've been talking specifically
about measurements of galaxy clustering.

(17:04):
The image we have also letsus map the distribution
of structure in the universe
using how the shapes of distant galaxies
get a little bit distortedby gravitational lensing
when their light passesthrough clumps of matter
along the line of sight.
- And then the light isactually bent a little bit
by the gravity of what is passing by?
- Like a beam of lightwill get a bit deflected

(17:25):
by a gravitational potential.
And, you know, if we're looking out
over millions or billions oflight years in the universe,
there's sort of structuresin the universe,
these structures, I mean, like, galaxies
and groups of galaxies
and they kind of end up being aligned
in this kind of filamentary structure.
So light from more distant galaxies
is going through the large-scale structure
between us and them and getting deflected.

(17:46):
So we can both look at thefact that galaxies tend to live
in high-density regions of the universe
and that those high-density regions
also cause the most deflection
and therefore distortion tobackground galaxy shapes.
Those are both tools we have
to map the distributionof matter in the universe.
There are other teamsin the collaboration.
There's a team thatfocuses on galaxy clusters,

(18:08):
so, like, large groups of galaxies.
There's a team that looks for supernova
and uses those measurements
to learn about theexpansion of the universe.
But this data set is really rich
and lets you do a lot ofthings not just in cosmology,
and I'm sure I'm leavingout something in cosmology,
but the fact that we have760ish nights of observation
over the course of six years,

(18:29):
imaging each patch of the sky
I think something like 50 times,
so like 10 times in each of 5 colors.
It also is really goodto see things moving.
So there's a whole group,
which I'm very impressedby but I am not part of,
but finding, like,things like dwarf planets
or comets in the solar system.
- Wow, all from the sameessential piece of equipment

(18:49):
and experiment?- Exactly.
- Maybe this is a silly question,
but why so much observation?
And how much of the sky areyou actually looking at?
- The survey area covers aboutone-eighth of the total sky,
so it's kind of looking outthe south pole of our galaxy.
So it turns out if you're trying
to look at distant galaxies,
the Milky Way is kind of a hindrance
'cause it's hard to see stuff behind it
when you're looking throughthe disc of our galaxy.

(19:11):
- So are you lookingperpendicular to the disc?
- Yeah, sort of looking down,
and there's some other patches
added onto the survey footprint
to increase overlap withother kinds of measurements.
So there are other experiments
that map the large-scale universe
using light from the veryearly universe that was emitted
in the first couple hundredthousand years of the universe
when atoms first formed.
- Is this the cosmic microwave background?

(19:32):
- Exactly, yeah.- Okay.
- And so there's a lotof information gained
by analyzing those data sets together,
and so that's a whole team
that's using the overlap where the DES map
overlaps with the cosmicmicrowave background map
from something calledthe South Pole Telescope.
- Even though there'sbillions of years duration
between what's pictured in those maps,

(19:55):
do you compare one to the other
to show how things evolveand change over time?
- There's that element,
so you can analyze thecosmic microwave maps
and see what inferences thatwould give you about cosmology,
and then say, given our model,
what do we expect to seein the late universe?
If the maps are actuallyon the same patch of sky,
you get something additional.
Whereas, like, we kind of knowthe statistical properties

(20:16):
of the CMB, cosmicmicrowave background map,
and that light is also traveling
through the samestructures as the galaxies.
So the same structures that aredistorting the galaxy shapes
with, we call it, weakgravitational lensing
'cause it's, like, tiny distortions,
and that same distortionaffects the CMB light,
so you can use a cross correlation
or, like, look at therelationship between distortions

(20:37):
in the cosmic microwavebackground light and the galaxies
to be extra sure that the distortion
you're seeing in thegalaxies is from lensing
and not through some otherproperties of galaxies.
So it's kind of anadditional piece of data
you can throw at it
to really make sure ourmaps are more certain.
- I wanna go back to someterms you've said a few times,
which are galaxy clustersand galaxy clumps,

(20:58):
because when I was readingabout this Dark Energy Survey,
I found this really interestingthat galaxy clumpiness
is something that people actuallysay in a lot of this work.
Can you tell us why these are useful terms
to look into and define?
- Saying clumpiness, and as yousay, a lot of people use it,
is when we're describingstructure in the universe,
you know, we've got thisstory of the universe

(21:21):
of, like, once upon a time,
the universe was denserand much more uniform,
and over time, those smallfluctuations in density grow
to form structures,
and the properties of those structures
and how fast they grow dependon the physics of gravity,
it depends on how much matter you have.
If you turn up the amount of dark energy
and the universe expands faster,
that kind of acts againstthe pull of gravity,

(21:42):
so, like, the rate thatstructure forms in the universe
depends on the properties of dark energy
because it influences the expansion.
And so I guess I'm usingclumpiness or clumping
as like a shorthand
for the statisticalmeasurements we can make
for how matter isdistributed in the universe.
You know, sort of a keypiece of information
is just, like, how big arethe density fluctuations.

(22:03):
And by that, I don't mean likeif I hold up a ruler to them,
how far apart are they?
I mean, like, how much density
deviates from the average density
and how that varies whenyou look at it in space,
you can kind of make astatistical measurement,
which is, like, a statistical term
would be you'd measure thevariance of the density.
That variance will be small ifthe universe is very uniform
where the density is closeto average everywhere,

(22:25):
but if you have a big clump in one spot
and a void in another spot
and there's an extreme difference,
then this variance of thedensity will be higher
and sort of the universeis less uniform or clumpy.
- There's numerous teamsthat are part of the DES,
the Dark Energy Survey.
Can you go a little bit more in depth
about what you specifically aretrying to do with this work?

(22:47):
- The working group withinDES that I'm part of
is called Theory and Combined Probes,
which I help work onputting the pieces together
that we need to use to be ableto make the model predictions
that we compare to data,
and then, you know, doing that comparison
and doing the fits andmaking all the plots
and trying to make the plots pretty
and all these kind of things.
Like I was mentioning,when we have the two maps
from, say, the CMB and weaklensing in the galaxies,

(23:10):
having those twomeasurements of the universe
that you can put together,use them together,
it's greater than the sum of the parts
'cause you can get extra information
by combining these measurements.
- Are they consideredprobes, those different maps?
- We use probe just to refer
to, like, different kinds of measurements.
And I've been mainly workingon, the last couple years,
combined analysis of galaxy clustering,

(23:30):
so, like, do galaxies tend tobe close together or far apart
and how are they distributed,and the weak lensing,
so the distortions to thedistant galaxy shapes.
You know, I was talking aboutthose paired measurements
where you look at the distancesbetween pairs of galaxies.
You can do an analogous thing
by looking at how aligned are the shapes
that we see of distant galaxies
as a function of how farapart they're on the sky.
So if you have much more clumpy matter

(23:52):
along the line of sight,
you'll get more of this weak lensing,
and that'll cause theshapes of distant galaxies
to look more aligned.
Whereas if the universe is fairly uniform,
you won't have much lensing
and the shapes will lookpretty randomized on the sky.
So those are sort of two different
of these kind of measurements we can make
using pairs of things,
and then there's a thirdone where you can say,
all right, I've got thesepositions of galaxies
that are in the clumps of matterthat are doing the lensing

(24:15):
and then the shapes ofgalaxies behind them,
and so putting those things together
gives you some extra information.
We've got three kindsof measurements we make
from two kinds of maps,
and all of that togetheris combined probes.
- And I know you've saidthat in the analysis you do,
bias is something youhave to be careful about
in different forms,
and we had a question about this

(24:36):
that was sent in from Estefania,who's a student in Texas.
- I've noticed your emphasis
on the refinement of position cosmology.
How has your research
sought to alleviatepotential sources of bias
in cosmological analysis?
- I think that's a question
that I spend most ofmy time worrying about,
so it's a good question.
There are a lot of waysthat we approach this,

(24:58):
and so there's not one panacea.
It's a lot of trying to thinkof all the possible ways
that bias could enter our analyses
and trying to test for themand make analysis choices
to help protect us against them.
So one of the key things that we do
is we try to make as manychoices about our analyses,
like what length scales are we gonna use

(25:18):
in comparing our model to measurements
is, like, a very key one.
We try to make a lot of those choices
based on simulated data.
So the sort of simplestway we approach that
is, you know, we've got our machinery
to do a model prediction
for the observables we're gonna measure,
so we pick an input set ofcosmological parameters,
an input model, we makeour model prediction,
and then we treat that modelprediction as if it's data

(25:40):
and analyze it using our planned analysis.
And the reason why this is nice to do
is 'cause you know what the truth is,
you know what cosmologythat you computed it with.
And so you can make sure,
like, if that were the data you measured
and you were to go analyze it
using your parameter fitting methods
and what length scales you'recomparing model to data on,
you get out what you put in.

(26:01):
- So you're essentially creatinga simulation for yourselves
to make sure that what you get out
corresponds to what you've created,
even though that's not the actual data
that you're working with.
- Exactly.
- You're making sure thatyou can trust the data
when you get it?
- Exactly, and then we can sortof take that a step further
and say, all right, we knowthat our model prediction
has some approximations andwe had to make some choices

(26:22):
over, you know, which software to use
and what settings to use.
Generally, the more accurate you wanna do
or the more detailedphysics you wanna put in,
the slower your calculation is.
And, like, in practice, we can'tdo the really slow versions
for every single comparisonto model to data,
or, you know, there might be some physics
we just know that wedon't know how to model.
So I was talking earlier about the effects

(26:42):
of, like, galaxies andsupernova pushing gas out.
On, like, cosmological small scales,
that's very uncertain modeling
and sort of figuring out feedback,
we call it baryonic feedback,
so supernova gas, stars,dust, galaxy messiness
can have a feedback effecton the large-scale structure
that we don't know how to model.
Characterizing that islike cutting-edge cosmology

(27:03):
that people are debatingand figuring out actively.
- I like that what most people, I think,
consider the real stuff of the world,
you know, stars and matterand animals and trees,
you're, like, eh, it's messiness,
that's getting in the way.- Exactly.
So I was gonna say, like,one thing that we can do
with these simulated analysis
is we can go get what'ssort of a large-ish
but plausible amount ofthis, like, baryonic,

(27:24):
this supernova feedback stuffthat could influence our data
that we know we're not modelingso we can't model it well,
and we can look at if that was real,
so we'd throw out a lot ofour small-scale data points
to, like, make sure we'renot sensitive to that.
So we use these simulationswhere the simulation is done
with a more complicated modelthan what we're fitting with,
and we can make sure
that we're not gonna falsely detect

(27:46):
that dark energy is varying with time
when it's just thatgalaxies are hard to model.
So that's one form of bias.
Like, we're trying to find the true value
or a range of values wherethe true value may live
for our cosmological model,
and we wanna make sure those estimates
have the true number in our error bars.
One way that we talkabout bias in cosmology

(28:06):
is, like, some effect thatyou're not modeling correctly
pushes your inferredparameter values around enough
that you might try to measure,
like, a parameter describedas dark energy time dependence
and it might move awayfrom what the true value is
because you haven't accountedfor something in your model.
We also try to accountfor and protect against
something that we callunconscious experimenter bias.

(28:28):
As scientists, we try as hard as we can
to make all the decisionthat goes into this analysis,
what points to measure,
what choices to make forour model as objectively
and in response to thesesimulated analyses as possible,
but, ultimately, you know,science is done by people
and people are subject
to all kinds of pressures and assumptions
and we might be interested inseeing how our measurements
are relating to previous measurements

(28:49):
or, like, there are specialvalues in the parameter space,
like detecting if darkenergy varies in time,
it's a very different resultthan if it's constant in time.
And so you wanna makesure, if at all possible,
that, even subconsciously,
our decisions on how to dothe analysis aren't influenced
by whether the resultsagree with our expectation.

(29:09):
And so we use a, we call ita blind analysis framework.
Exactly what that means dependsa lot on the experiment,
but, like, the main thing inprinciple is you make sure
that you're not lookingat your main results
until you've frozen in allthe decisions to get there,
and you hope that nothingunexpected shows up
after you, like, reveal the results.

(29:29):
In practice, things arenot always that tidy,
but generally, part of thisis if something does change
or you find something afterwards,
we really try to be rigorousabout, like, documenting it
and being clear of, like, whatdecisions were made before
versus after unblinding.
So it's kind of a similar motivation
to if you hear about in,like, medical fields,
like double-blind trials
where you test a newmedication against a placebo.

(29:50):
Like, in those experiments,
neither the patient nor the doctor
knows which is the real pilland which is the placebo.
And you do that because youdon't want sort of expectations
of whether somebody'sgonna feel better or worse
to, like, influence your interpretation
of some very complicated phenomena then.
- I guess I just assumethat that kind of blinding
was done in medicine
and the more, I don'tknow, human-scale sciences,

(30:12):
and that when you'redealing with the universe
at these enormous scales and galaxies,
my assumption was that, youknow, that's objective data
and it's observables andyou don't need to do that,
but clearly this is somethingyou need to be aware of.
- Even though, you know,we sort of guideline
and try to be as transparent as possible
about how choices are made,
there are choices that need to be made.
So, like, for example,we use these simulations

(30:33):
including all these messy galaxy physics,
and we wanna make sure thatour cosmology inferences
about dark energy aren't biased by that.
But, like, how do you quantify that?
What amount of bias islittle versus enough?
And, like, you have to set a threshold
and decide exactly what numbers
you're gonna look at to assess that,

(30:53):
and, you know, there's sort of things
that are better choices thanothers in sort of a broad sense
but when you get down to the specifics,
you wanna motivate things,
but there's a certainamount of arbitrariness
that does come into it,
and so we wanna make sure
that, yeah, if we're making that choice,
it's not informed in any wayby, like, what the science
coming out the end of the pipeline is.

(31:14):
It's part of the structureof the whole analysis
within our collaboration
and in, you know, many cosmology analyses.
So, recently finished a big analysis,
and sort of one of thedramatic stages at the end
is you write up everything youdid and all the tests you do
and have some collaboratorswho are experts
but not directly involved inthe project look that over
and say, "All right, I thinkyou've checked everything
you needed to check.

(31:35):
You have our okay to revealyour results or unblind them."
And so it always feelslike a bit of an event,
kind of a nerve-wracking event
when you, like, look at theresults for the first time.
So in that sense, it's definitely active.
But, yeah, helping developthe sort of technical method
for hiding the results fromourselves was my first project
in the Dark Energy Surveyas a graduate student.

(31:55):
There's varying degrees
of technical manipulations you can do,
'cause the trick is you wannahide the results for yourself,
but you wanna give yourselfenough access to the data
that you can test for all thethings you need to test for.
And that ends up beinga pretty tricky question
sort of on one extreme end
of, like, not doing verymuch technically for this
is just you all agree as a collaboration,

(32:16):
like, we're not gonna lookat plots of these parameters
or something like that,
which, like, does work for your purposes,
but also, when you havea big collaboration
and, like, it can benice to have something
a little bit harder toaccidentally peek at.
The method that I workedwith some collaborators
to develop and test and implement
actually transforms thesestatistical quantities
that we measure
from these three kinds ofstatistical measurements,

(32:38):
and we figured out a waythat you can transform them
that, like, still keep themall consistent with one another
so they look like they camefrom some valid universe,
but it looks like theycame from a different set
of cosmology parameters.
So we have these, like,transformed statistic measurements.
Most of the other collaborations
that are sort doing similar analyses,
they have some mechanism
for this kind of transformationof data on some level.

(33:00):
And I know in one of the other
sort of weak lensing surveys out there,
they have a much more technical,
like, encryption double keysort of way of doing this.
It's the technical aspect ofhow can we transform the data
and make sure we preserve theaccess we need to preserve,
and then there's also, like,how does your collaboration
work as a group,
and, you know, how do you decidewhen to reveal the results,
and what do you do ifsomething unexpected comes up?

(33:23):
And, you know, this maybealso ties into other ways
that bias comes up in conversation
of, like, personaldynamics in collaborations
and getting large groupsof people to work together.
And so it's a challengewithin any collaboration
and also, like, looking forward
to next-generation galaxy surveys,
which are gonna be even bigger,
of, like, how do you make sure
everyone has enough information

(33:43):
to understand what tests are done?
How can you make sureeveryone's voice gets heard
when you're having these conversations?
Often, when people arekind of stressed out
and pushing for results,
it's an organizational challenge as well.
And I think one additional benefit
of these sort of blindanalysis frameworks,
in addition to, youknow, helping make sure
that you have the most robust
and accurate science as possible,

(34:04):
is it's kind of a littlebit of a sociological break.
It's like if you all need to decide
that you've checked all thethings you need to check
to look at the results,
I think it functions very well
as sort of a pause for a collaboration
to say, like, we've beensprinting towards the end,
let's take some time,
take a week or two.- Take a breather. (laughs)
- In the same way asdeveloping, like, modeling
and data analysis techniques,
we're sort of a laboratoryfor future analyses,

(34:26):
these sort of blindinganalysis and strategies
for how to make decisionsand how to organize people
I think is another thingthat we learn a lot from
and see what works andwhat could work better.
And that's very tied in with the science
of how these large collaboration works.
And these large collaborations are hard,
we gather enough dataand do the work we need
to, like, figure out what the universe
can tell us about dark energy,

(34:47):
so it's really crucial thatpeople who are interested
can contribute and feel like their work
is valued and important.- It seems that
a lot of your work alsopretty fundamentally relies
on understanding this interplay
between experiment and theory,
so I'm wondering if you can tell us
a little bit more about that
and how experiments canhelp us improve theory
and theory can help usimprove experiments.

(35:09):
- So I think cosmology as a field
is really defined by this interplay.
You can go back towards sortof early days of cosmology
where, you know, Einsteindeveloped general relativity
and had this assumption thatthe universe should be static.
And when you look at what the equations
tell you about the universe,
it tells you it's gonna beexpanding or contracting,

(35:29):
so we, you know, stuck aconstant in the equation,
and if you tune it to a specific value,
given the other propertiesof the universe,
you can get the universe
to not be expanding or contracting at all.
And then just a few years later,
Edwin Hubble measured the fact
that the universe was accelerating,
so they throw out thatterm, it's not needed,
you know, we're gonna expect to find
the universe that's decelerating.
And then, you know, you get to the '90s

(35:49):
when people go and measure that,
and you realize, oh, it'sactually accelerating,
which brings the constant back
but tells you it needs a different value.
And there's countlessstories within the field
where the data tells you youneed some aspect of the theory,
and then now, dark energy couldbe a cosmological constant,
and so far, sort of all the observations
we've made of the universeseem to prefer that

(36:10):
or there's not evidencefor some other property,
but we don't think that's the whole story.
And why don't we thinkit's the whole story
would be a reasonable question.
So, you know, this cosmological constant
would be some, like, vacuum energy,
and we can look to particlephysics colleagues down the hall
and they predict that thereshould be some vacuum energy.
It's difficult to predict,
but if you kind of make some estimates

(36:32):
based on our knowledge of particle physics
of what the value of thatenergy density should be,
you get a number that's,like, absurdly larger
than the number we measure.
So given, like, particlephysics energy scales,
the value of this energy density we find
is, like, very tiny but nonzero.
And so you want to knowwhy that's the case,
and so there's a lot ofwork being done by theorists
to think of different modelsthat could explain this.

(36:53):
Or you might ask, like, couldthe universe be accelerating
not because there's some extra substance
but because we need toextend general relativity
on large scales?
And then you can say like, all right,
but how would thatmanifest in the universe?
Those models are predictions
for, like, ways that youcould extend your description
of gravity beyond general relativity
while still respecting allthe very tight constraints

(37:16):
we have on gravity
from, like, measurementsof the solar system
and lab experiments,
sort of gives you a set ofeffects that you can go look for.
My team within the Dark Energy Survey
that I co-lead with another postdoc
who works at the JetPropulsion Laboratory for NASA
in particular focus
on taking these different proposed models

(37:36):
for, you know, maybe different ways
you could model dark energy
or modifications of your theory of gravity
and going and taking our galaxy cluster
and weak lensing data andtesting those extensions
to the sort of simplestdescription of the universe.
In a similar way to whenwe constrain properties
of the simplest model,
we can vary the input parameters

(37:58):
describing these kinds ofmodifications of gravity
or dark energy properties
and place sort of limits
on what those parametersare allowed to be.
Part of this big analysis we just finished
was testing a set of sixof these kinds of models,
and seems like the sort ofsimplest cosmological model
lives to fight another day,
given our data, what's the largest amount

(38:19):
of, like, time dependencethat dark energy can have
in some range.
- That connection betweentheory and experiment
is something that you very tangibly had
because you've not onlyworked on the theory side
but you actually wentto the telescope, right?
- One benefit of workingin a large collaboration
that's trying to do over700 nights of observing
over the course of six years

(38:40):
is they needed people todo shifts on the telescope.
Some observatories, I think,in next-generation survey,
they're doing a lot more,like, remote observing,
but it can be helpful tohave people in the room.
So I did two observing shifts for DES.
You fly into a little beach town
and then ride a van for threehours into the mountains,
and you stay in a astronomers' dorm
with, like, a little cafeteria

(39:01):
and go work on the telescope every night.
- After you told us aboutit at first, I looked it up.
I wanted to see what it looked like,
and it looks so much like what I pictured,
you know, this classicdome-shaped observatory,
but then there's these barren,
there's a few buildings aroundat the top of this mountain,
but then it's sort of barren.- Yeah, it's a desert.
- Yeah, what's it like togo to the top of a mountain

(39:22):
and live in an astronomers' dormitory?
It seems like such a unique experience.
- I think it's probably
one of the most, like, incredibleexperiences of my life,
and I feel very gratefulthat I got to do it,
especially because, you know,I usually work with data
that's in a very, like, processed form,
and so this is a very different way
of interacting with the experiment.
- That's data as it's pouring in

(39:43):
in real time from the universe, right?
- Yeah, so each exposurewith the Dark Energy Camera
is like 30-second exposures,
and you see, like, the raw image
of the different, like,chips that make up the CCD
that measures the image.
And so they pop up on thescreen as they come in,
and the thing that I find really striking
is just how messy they look.

(40:04):
So you see a lot of noise,
you see, like, streaks fromsatellites going through them.
One of the shifts I was on,
there was a bit of dust on one of them
so we spent a lot of time
trying to figure out if a little squiggle
was something we could dosomething about or not.
- So even a mountaintop
is not completely free of distortions
and issues to deal with.- Exactly.
And there is a lot of work
that goes into combining multiple images

(40:26):
to beat down the noise.
There's ways of correcting,
you know, so you can look atthe shapes of, like, stars,
which are, like, in principle,
from our point of view,like point objects,
and people look at howtheir shapes get distorted,
and there's a lot of complicated modeling
to correct for that kind of distortion.
And also, like, theoptics of the telescope
might be slightlydifferent towards the edge,
towards the center.

(40:46):
The science, the darkenergy constraints we do,
would not be possiblewith all that hard work
and technology developmentand analysis development
of my many colleagues.
So this is really a team effort
and is not something that's possible to do
without a big team of hardworking people,
and I think getting to go, youknow, sit in the control room
and sort of see the earlyiteration of the data
I think felt very valuableto me in that sense.

(41:08):
- I'm fascinated just by that idea
of going to work at thistelescope in this remote location.
Aside from looking atthe data as it comes in,
what do you do when you'reon top of the mountain?
- So generally, there's a 4:00 PM meeting
where you get on Zoomwith people at Fermilab
who, like, manage a lot ofthe telescope operations,
and you check in about, like,what the plan is for the day,

(41:28):
get everything set up,
you go eat dinner in theastronomers' cafeteria,
you come back, you get, like,the various scripts queued up
that you're gonna run,
and then you just have towait for the sun to go down.
And so, like, kind of partof your job is to go, like,
"Well, there's nothing wecan do in the control room,
we're gonna go..."
Everyone goes and watchesthe sun set over the ocean,
and you're on a mountainthat's somewhat taller
than all the other mountains,

(41:49):
and usually it's very clearand it's just very beautiful.
And there's also these littlerodents called viscachas.
They look like rabbits with squirrel tails
that also seem to comeout and watch the sunset
so you're always kinda looking for those.
(Colin laughs)
And then, yeah, during the night,
you're kind of keeping an eyeon the images as they come in,
making sure that nothing's going wrong.
You also are supposed to monitor

(42:10):
how much cloud cover there is,
and it can be detected tosome extent with instruments.
But, like, part of your job
that you do sort of a little report
is you're supposed to step outside
and let your eyes adjust tothe dark once every hour.
So as you would expect from somewhere
where you put a telescope,
like, that's some of the most stars
I've ever seen in my life.
So you can see theMilky Way super clearly,
especially when the moon is down,

(42:30):
you can see the Magellanic Clouds,
and it's just like
you're kind of like aloneon a windy mountaintop,
it makes you feel very small.
- I wanna go back to asking you
about the way you summarizethis result that has recently
come out of this DarkEnergy Survey collaboration.
You said this,
I think you said the Lambda-CDMmodel survives another day,
or maybe another way to say that
is some relatively simple model

(42:53):
passes another series of tests.
And, you know, maybe on the surface,
this result could seem not so exciting
'cause we're not announcingsomething big and new
that we couldn't expect.
But I think it must be pretty incredible
to think that all ofthis observation time,
all of this noise and dust and clouds
that you had to account for

(43:14):
with so many people over so much time,
all of that was done and, in the end,
something pretty simplecan describe all of that,
and I'm just curious to getyour perspective on that.
Do you find that simplicity exciting?
Or do you find yourselfwanting to find something new?
- It is both excitingand frustrating because,
so we have the simplest model,
so, yeah, Lambda-CDM

(43:36):
is sort of the maybe somewhat jargony name
that we often refer to this,like, simplest model as.
So Lambda is the symbolthat we usually use
to represent the cosmological constant,
so this simplestdescription of dark energy.
CDM stands for cold dark matter,
which is, you know, this matter
that doesn't interact with light
but clumps up under theinfluence of gravity.
It is a real achievement of the field

(43:57):
that we have this model
that we can use todescribe pretty accurately
basically all of the observations
we've made of the universe.
There's a few exceptions that are debated,
but as I said earlier,it's not the whole story.
Like, we don't know what dark energy is
and we don't what dark matter is,
and together, they make up 95%of the stuff in the universe.
There are a lot of different models
or descriptions that people consider

(44:19):
that, you know, could darkenergy be like this or that,
or might dark matter have alittle bit of interaction,
or what kind of particle makes it up.
For neither of these things,
there is not a, like, clear front-runner,
like, oh, this must be it.
And so there's a lot of,like, very important work
being done on the theory
and to think of different possibilities,
But, ultimately, on the dataend, what we're looking at

(44:40):
is trying to make more andmore precise measurements
of this simplest model Lambda-CDM
and kind of look for,like, cracks in the facade
or places where the predictionsof the simplest model
don't match our observations
because if we find a mismatch
that holds up as ourdata get more precise,
maybe holds up ifdifferent teams measure it
and make different,

(45:00):
like, there's all these ways
that, I think if we start seeing hints,
we'll wanna really makesure what we're seeing
is a hint of physics
and not of some modeling assumption
we don't understand well.
But ultimately, we'relooking for mismatches
that will give us a clue
for how to build a morefundamental understanding
of 95% of the universe.
So it's frustrating thatthe results match that

(45:20):
because it'd be very exciting
if we found, like, aclear hint for something,
but, you know, it's allpart of the process.
Like, we can narrow in on,like, what kinds of models
are allowed or not allowed
or at least, like, what are the ranges
of the size of effects
that deviations from general relativity
on large scales might have.
In my mind, a concrete example
is, like, one of the commonthings you can sort of study

(45:41):
if you're looking for deviations
from the prediction of general relativity
is that theory will giveyou a specific relationship
between the way that light interacts
with the gravitational potential,
so causing that gravitational lensing,
and the way that gravity affects matter,
like particles with mass,
so the galaxies and darkmatter clustering up.

(46:02):
If you're assuming generalrelativity is part of your model
as you are in Lambda-CDM,
putting those different kindsof measurements together
lets you really get precise constraints
on the parameters or theproperties of that model.
But if you relax thatassumption a little bit,
you can say, all right,
we're looking at the same sortof structures in the universe
and we're seeing how they affect light
and how they affect matter,

(46:22):
and we can use that to test
whether or not they havethe expected relationship.
And like a weak lensing survey like DES,
and particularly, we'remaking both measurements
of the lensing and the clustering,
lets us make the most precise version
of that kind of test available.
General relativity seemsto be doing very well.
(Jessie laughs)(Colin laughs)
- Yeah, it seems to be standingup to a lot of the tests

(46:42):
that it's being put under,
which is pretty amazingfor a century-old theory.
- Very much so, yeah.
- I was looking around your website,
learning about the Dark Energy Survey
and your role and your past,
and I have to say I enjoy,
on your website there's atab that just says Cartoons,
and you click Cartoons
and there's theseillustrations that you've made
of some pretty cool scientific concepts

(47:05):
in a really sort of fun,bright, engaging way.
And one I keep thinkingof as you're talking
is there's a person at a desk in a room,
I'm assuming maybe it's you,
maybe it's, you know, it could be anybody,
but they're wearing, like, VR goggles.
What they see is thisbeautiful expanse of galaxies
and swirls and stars and things,

(47:25):
but, really, they're at a desk in a room
and there's a cat sleepingon the bed nearby.
And so I wondered, A, if that's you,
and B, more generally, canyou tell us about your artwork
and how, you know, I thinkyou're the first person
whose academic website I've gone on
and it has a tab that says Cartoons
for all their artwork.
How did that come to be?- I have spent a lot of time

(47:46):
in the last couple years
working from home with a catsleeping on my bed so that
is an accurate representation.- OK, so that one's accurate.
Is that a self portrait,the person in the VR helmet?
- No, not necessarily,
but it was inspired by my roommate
who I shared an apartmentwith during the pandemic
who would play a lot ofVR games in his room.
So, yeah, that cartoonwas part of a series

(48:06):
that I did with some collaborators in DES.
We released sort of the firstround of the cosmology results
from the galaxy clusteringand weak lensing measurements
from the first three years of DES data.
So I guess that'ssomething I didn't mention
when talking about the project before.
We've analyzed the first threeof six years of observations,
and we're just gettingstarted on the next round now.

(48:26):
Yeah, when we were releasingthose cosmology results,
there's the main cosmology paper,
but there's also like 30 other papers
documenting all the workand tests and things
that go into making thatmeasurement possible.
And we were talking about how,
you know, we've gotthe Dark Energy Survey,
like, Twitter account and things,
like it'd be fun to try and,like, highlight these works
and try and figure out a way
to make them a bit moreaccessible to the general public

(48:48):
even if, you know, peoplearen't gonna go open up a PDF
of a very technical paper
about measuring, like, galaxydistances or something.
A couple of years ago, mycolleague Chihway Chang,
who's now a professor at Chicago,
she had done this seriesof, like, one cartoon a week
about different science concepts,
and so we decided it'dbe fun to revive that
to illustrate these like30 different papers.
So we kind of split themup and got the authors

(49:10):
to help us write sort of alittle, like, blurb description
of each of the papers,
and then we tried to figureout ways to illustrate them.
So that cartoon that you'rementioning was the one I drew
for a paper describingsome simulated analyses,
so the idea that we kindof used simulated data,
analyzed it as, like, atest run for our analysis.
And so partly 'cause myroommate during the pandemic

(49:32):
was doing a lot of flightsimulators on VR in his room
during the pandemic,
and so that was kind ofthe inspiration there.
I'm just kinda trying
to think of fun things.- Yeah, my first thought
was flight simulators,
and even earlier in this conversation
when you were describingthe simulation process
and why you do it,
I thought, well, it's similar
to why pilots take flight simulators
'cause you don't wannacrash the real plane

(49:53):
unless you know what you're doing, right?
- Exactly.- You do the simulations
to figure it out.
There was one other thatI have to ask about.
There's one other cartoonof two volleyball players.
One is setting the ball,
the other one's about tospike it over the net.
And I didn't fully graspthe science behind it,
but I think, you know, these things,
they're meant to invitepeople in and and learn more,

(50:14):
so can you tell us what thevolleyball players are doing?
- That was to illustrate one of the papers
that starts combining thesedifferent types of measurements.
So we've got the map of galaxy shapes,
we've got the map of galaxy positions.
You can either look atpairs of galaxy positions,
pairs of galaxy shapes,or the cross-correlation,
pairs where you have ashape and a position.

(50:34):
These statistical thingsI'm talking about,
we call them correlation functions,
that's the technical term.
That was meant to illustrate
that analyzing these typesof measurements together
gives you informationthat you wouldn't get
by analyzing them separately,
so it's this kind ofcombined probe analysis idea.
- Team sport.- And so the volleyball thing
is to say they're working together,
it's teamwork to get the ball over the net

(50:54):
or to tell us what darkenergy is acting like.
- I don't wanna ask you
to describe your art in words too much
'cause I know everyoneshould also go look at it,
but I also have to ask youabout the platypus comic.
(Lauren laughs)(Colin laughs)
- One of these cartoons
is a little, like, three-panelcomic-looking thing
that has a bulletin board like you'd see

(51:14):
in, like, a detective movie,
so you've got photos onit with, like, string.
So the scenario is you're trying to learn
about what an animal is
by getting, like, photos ofdifferent parts of the animal.
You know, you have a photo of a foot
that's like a webbed foot,
and you have a photo ofa nose, which is a beak,
and so the working model,
sort of the simplest model,Lambda-CDM, is that it's a duck.

(51:35):
Then you go, and a lot ofwhat we're doing in cosmology
is going and making eithermore precise measurements,
which I guess would be like
a less blurry picture of your duck
or imaging differentaspects of the animal.
So the second panel of the comic
is the detective gets aphoto of the animal's tail,
and instead of looking like a duck tail,

(51:56):
it looks like a beaver tail.
If the new data doesn'tmatch your expectations
of the model given your previous data,
that might be a hint that youneed to develop a new model
for your description of the universe
or, like, what animal you're looking at.
And so in this case, thenew model is a platypus,
which has a duck-like beak and webbed feet
and a tail that looks like a beaver tail.

(52:18):
So that's sort of the analogyfor kind of what we're doing
and trying to test Lambda-CDM
by looking for sort of mismatches
between its predictionsand our measurements.
- Has it been usefulto you as a researcher
to take these long papers
and try to condense theminto these short comics?
- Yeah, I think so.
It's definitely a funbrainstorming process.
You know, with this set of like 30 papers,
like, everyone's working together,

(52:39):
but there's definitely some
that I contribute moredirectly to than others.
And so for doing illustrationsfor all of these,
it was kind of fun to navigate the project
and try and help authors come up
with, all right, what isthe one- or two-sentence
sort of hopefully accessible description
we can come up with?
So it helps me have aclearer understanding
of, like, the core concept
behind a number of my colleagues' papers

(53:00):
that are very important for my work
but I might not be, like, deeplyfamiliar with the details.
And then for things
that are more closelyrelated to what I work on,
so, like, model testingby looking for mismatches
between model and data, orplatypus hunting, I guess,
it's just kind of fun to think through
and, like, come up withanalogies like that.
And, I mean, it was also,like, one of my goals

(53:21):
over the last couple years
was to learn how to dodigital art on an iPad,
and this was was a very good project
for learning how to do that.
As an added benefit,
I now use a lot of thesecartoons when I give talks.
- Have you always beenartistically inclined?
Have you always expressedyourself through drawing as well?
- I've definitely had it more as a habit
at some times in my life than others,
but, yeah, I always liked to draw.

(53:41):
I mean, I like drawing in general
and find it relaxing and enjoy doing it.
I think a thing Istruggle with especially,
I think we all, in thepast couple of years,
have a little bit of, like,pandemic-related burnout
so it's a little hardto, like, find motivation
or ideas during downtime.
And I think particularly this,like, science cartoon project
was very nice 'cause it wasa little bit collaborative

(54:02):
and then it sort ofseeds a bunch of ideas.
And, like, once I have an idea,
like, the sort of typeof mental energy used
to, like, plan and figure out a drawing,
it's like a form of problem solving,
but it's a differentkind of problem solving
than, you know, workingon a scientific analysis
or a calculation.
So it's kind of fun to bringthose things together a bit
and to, like, get to share them
with both collaboratorsand the general public.

(54:24):
- Colin talked about how unique
this Cartoons tab is on your website.
I wanted to tell you something else
that stood out to me on your website,
which is that right on your homepage,
you start by giving, youknow, a brief description
of your research,
and then right after that, you write,
"I'm also interested in science outreach
and in making STEM fields more accessible
and welcoming to everyone."

(54:45):
And we actually had a question sent in
about this sentence on your website.
- Matt Duschenes, a PhDstudent at Perimeter.
I'm wondering what barriershave you experienced
while trying to makescience more accessible
and more diverse?
- So the main way Ihave engaged with this,
it's varied depending ondifferent stages of my career,
and sort of recognizingthe existence of barriers

(55:07):
and the ways that those can manifest
was definitely a progression.
Like, you know, I look backat being an undergrad student,
and I had several classes
where I was, like, oneof two women in the room.
And at that point, I don't think
I would've identified anythingnecessarily as a barrier.
The social dynamics, I thinkI mostly experienced that,
and then a bit during my master's
is just being a littlebit of like an isolation.

(55:28):
There are more concreteand more abstract ways
that that can manifest,
and, you know, they impactdifferent people differently.
Like on one hand, I may have been one
of the only couple womenin my physics classes
while also recognizing that Iwas being supported partially
by my parents in undergrad
and so I could go work in a physics lab
and not have to, you know,work other jobs after class.

(55:49):
You know, so there are someways that isolation can crop up
and can become barriers.
Definitely have had atleast a couple interactions
with professors assumingI knew less than I did,
almost certainly a gendered point of view.
But, you know, there are other ways
in which I, you know, was privileged
and had this access to,say, this research program

(56:10):
and had the support to, like,go to Europe for a summer
and do physics research.
So there are ways I've faced barriers,
but also ways that I have not had barriers
that other people might have.
And I think in grad school,
I had a big learning experience with this
in that I helped organize theSociety for Women in Physics
at the University of Michigan
for most of my grad school career.

(56:31):
A big focus of that
was, you know, justbuilding sort of a community
within the department forsupport and mentoring,
which, honestly, I think canbenefit everyone in academia,
but especially people whomight feel a bit isolated
or face some challenges.
And I think a big part ofthat learning experience
was, often, we would also communicate with
and work jointly with otherstudent groups on campus.
For me, it's an ongoinglearning experience

(56:53):
of recognizing ways
in which, you know, Imight have faced barriers
or ways which people mightface barriers that aren't me.
So, like, things like making sure
that these kind of summer programs
have enough, like, financial support
that a student who mightotherwise need to work a job
can, like, participate
or trying to set up programs
where, you know, you don'thave to be in the know

(57:14):
to go seek out a research experience
that, like, might change thetrajectory of your career.
So I think that kindof thing is important.
And, you know, also thinking through
these collaboration dynamics
of, like, if you have abunch of stressed out people
who are trying to pay attentionto too many things at once,
that's, like, a prime environment
for well-intentioned peopleto make others feel excluded,

(57:36):
which I know I have been guilty of
and, you know, I think we'reall trying to work on it,
and so there's a lot of discussion
within, you know, Dark Energy Survey
and other collaborations
of, like, how can we make sure
people who are new to the experiment
or people who are not white
or women or other gender minorities,
like, can feel supported,can find community,

(57:56):
know who to ask for advice
and, you know, can feelheard in conversations,
recognizing that not everyonecommunicates in the same way.
- And I know here at Perimeter,
you've become pretty involved in outreach
and in mentoring and supervising students
at more junior stages.
What motivates you to beinvolved in that kind of work?
- I mean, kind of selfishly, I enjoy it.

(58:18):
I think I'm happiest doing science
when I'm, like, chattingwith other people about it.
You know, all these labs that I worked in,
I also did a little bit ofgalaxy cluster cosmology
in undergrad as well.
And, like, all of theprofessors I worked with
or more senior undergrads or grad students
that, like, helped melearn how to do computer...
You know, it's a learningprocess along the way,
and different mentors have made an impact

(58:40):
on the trajectory of my career,
and so the idea of being able
to, like, support andintroduce other people
and help them feelsupported feels important.
- That trajectory of your career,
where do you see or hope it's headed next?
You know, this is ongoingwork with the DES.
- Well I'm gonna be on the job market
for faculty jobs in thenext couple years so.
(Jessie laughs)(Colin laughs)

(59:01):
Yeah, I would like to keepdoing cosmology research.
I would like to be able to teach as well
and keep mentoring students.
This analysis team thatI've been co-leading
with Agnes Ferte, who'sanother postdoc in DES,
we've led this analysis
extending the year threeanalysis, which we call it,
to extended cosmological models,
models beyond the simplest one.
The analysis of the full sortof legacy data set for DES,

(59:23):
the year six analysis is ramping up.
I'm gonna be taking a littlebit more of a backseat.
Like, I'm still gonna be contributing
to different pieces of validation
for, like, the Lambda-CDM analysis
as well as the extended models.
Some people who are on our team
during this year threeanalysis are stepping up
and are gonna have a chanceto lead the group as well now.
And part of this analysis,
there'd been a lot ofpatches where we realized,

(59:45):
like, oh, this modeling toolthat we would need to do this
just doesn't exist,
and so we, you know,kinda have to find ways
to work around that,
and so there are a couple of these things
that were not workable on thetime scale of that analysis,
but with a little bit more work, I think,
are gaps we can fill to letus do a more precise analysis
of the data we already have
and also get it readyfor our next analysis.

(01:00:07):
So one of the students I'm supervising
here at PI as a summer student,
we're working on one of these projects.
And I was just at a meeting
where I was discussingplans with a grad student
about sort of extending oneof these other analyses,
so there's sort of moredirect spinoff projects
and then I also want to geta little bit more involved
in sort of the next-generation survey,

(01:00:27):
which is called the VeraRubin Observatory LSST.
- That's sort of the next evolution
in precision or in power?
- Yeah, so it's gonna be turning on
I think in the next year or so.
It's, like, on the next mountain over
from where the Dark Energy Camera is.
Many times we've heard, overthe course of six years,
that the LSST is gonna image
as much of the sky as you can image

(01:00:47):
without the Milky Waygetting too in the way.
It's also on the ground,
so the half of the sky it has access to,
like, basically everynight or every two nights.
Like, it has an even biggerfield of view than DES
and will be able to get more precise data
looking at fainter galaxies
and making more precise measurements
of shapes and other things.
You know, maybe outsideof survey science as well,
you know, if I'movercounting my free time,

(01:01:09):
look for more theoretical projects
looking for, like, what areother ways we can use this data
or, like, the fact thatI'm interested in theory
and have this experienceworking with data,
compared to your average theorist,
I have a good sense of the ways
that which data is messy and tough
and so, like, when you try tobring those things together,
things that you don'twanna have to care about,

(01:01:30):
you might have to care about.
So I'll probably continueworking at the interface of that,
both, you know, looking for ways
we can get more informationout of data we already have
and also making sure that when we do that,
we're doing it carefully and robustly.
- Well, thank you so much fortaking us on this journey.
There's so many things I didn't know about
and so many things thatI just find fascinating
and at scales that are just mind boggling.

(01:01:51):
And I hope you'll keep us posted
on the next stages of thisexperiment and the ones after.
- Yeah, that would be great.It was great talking to you.
(bright music)
- Thanks so much for listening.
Be sure to subscribe
so you don't miss anyof our conversations.
We've interviewed somany brilliant scientists
whose research spans fromthe quantum to the cosmos,

(01:02:12):
and we can't wait for you to hear more.
And if you like what you hear,
please rate and review our show
on your preferred podcast platform.
Great science is for everyone,
so please help us spread the word,
and thanks for being part of the equation.
(bright music continues)