it's a great pleasure for me to welcome
Dave here Dave kid and I we collaborated

for last 10 years on Veritas and most
recently also on CTA so we were both

involved in building the fluoroscope
that's da mineira zona here they want to

use these telescopes to do still and
Reformatory

to measure the the size of of stars so
we'll talk about that but before he does

that I was say a few words about where
he came from

so I did as PhD in Pennsylvania right
University of Pennsylvania and one of

his three supervisors was Raymond Davis
jr. named that maybe Orion is a bell in

the tree nose down and mine detecting
them for the first time Ava's PhD moved

on to the immersive Utah as an assistant
professor and then he climbed up through

the ranks the foot professor and now he
is also be the Dean of credit students

at the University of Utah and that's
where he is now I would like to tell you

one anecdote that he told me when when
we had one of those many lunches what

stuff you could do when you were a kid
back in the days so I was the time when

they just invented that lays on when you
say hey I want to build one laser - so

he sent an email the email was pen and
paper what did he call him was the phone

already invented it was so he made bail
apps to send him the cavity the

resonance committee that he need to
build a laser and they did so they'd

have to pay for that he got it for free
when he built the other surrounding

electronics I need to get this thing to
lace and hello did you get it to lace

so who of us build a laser from scratch
not many are so they first how many of

these these stories where it tells you
all sorts of very interesting and

dangerous things that he was doing when
he was a kid that in these days with all

you know parents all these days right
would not be possible all right so I

hand it over to you Dave and I think
this will be very nice talking I hope I

should actually annotate that story a
little bit more so I was it was 1972

labs on the phone I convinced him to
send me his laser back then it was worth

about $400 I remember an ad and say you
can buy a Volkswagen bug for under

$2,000 this is pretty good that I
actually got these guys descent is

twelve year old and you can just
convince them to send it to me my dad

got the phone bill back then phone bills
were expensive and it cost like fifty

bucks that call and talk these guys for
a half an hour

my dad is furious about this like but
Dad it's a laser you know so I had to go

and give him my paper route money to pay
him back for this because I wasn't

supposed to be doing this type of thing
so we get in trouble for these things

any what it was it was a fun adventure
okay what I want to talk today thank you

for having me today I want to talk today
about some work that were they're

developing a new type of astronomy base
is reinventing an old technique but what

we want to do is we want to try to go
for very high-resolution astronomical

imaging using this array of telescopes
that leap amok and I work on it's called

Veritas Goff telescope array the short
of it is if you think about where we've

been doing astronomy since Galileo
forded years ago everything really has

been
with is looking at photons we

astronomies very simple we point
something at the sky we measure a rate

of photons they measure their frequency
and maybe measure how they met their

with time but that's about it that's
everything we know about the universe

comes from just measuring some
frequencies measuring the time that they

arrive and or whether they flare up and
down and and some position things but

other than that we haven't really
exploited very much of the properties of

the photons actually there's more
information in photons than just that

and so that's why I want to talk today
about some of the other properties that

we might exploit and some of the things
that you may wind up with it from that I

should show some of them my group at the
University of Utah I'm the PI of of this

work stuff on the book he was a Veritas
collaborator and he also wrote some of

the original papers about ten years ago
about trying to do this type of

technique so he still works with us on
this project and graduate student Owen

Matthews he's been doing a lot of the
heavy lifting on actually doing

observations out at this telescope they
have called Starbase and this fall he's

gonna be coming down to Veritas and
doing we're deploying some the

instruments down at Veritas he'll be
doing quite a bit of the observations

there we also have a couple
undergraduates so they have worked with

us as well over the last few years so
this work is a lot of it is associated

with them and our group there is some
additional work that comes out of the

strength of telescope or a Michael
Daniel in particular at the Smithsonian

Astrophysical Observatory a good deal of
work has been done in coincidence with

him as well some of the CTA stuff okay
let's go back 400 years ago and optical

Astronomy whether that look like this is
1609 and now if you know what this is

this is Galileo's original telescope and
so that's that's what optical Astronomy

consisted of you put your eye through it
and you look at something and and then

make some hand sketches so here's you
know it's just refractive optics over

here and you take an image and then you
write down what you see in some obscure

language and now today you know
2017-2018 is a Hubble and it's not much

different really in terms of what
happens it's more sophisticated but it's

the same sort of idea
you take white you concentrated it and

you put on some kind of a focal the
technical setup using your eye you use a

you just record the light and take an
image of it using the camera that's

there the different frequency cameras
though being up above the by the

atmosphere I put this picture in here
that uh you know Galileo is very

optimistic about his first telescope and
he says he thinks showing it off here to

the Vatican thinking this is all gonna
go over very well from some of those

observations they didn't realize what
type of trouble he would wind up I hope

we don't wind up in trouble and the
stuff I this is observations of Jupiter

he made so you go to the log book and
this was a high-speed data recording

here's picture Jupiter and a couple
stars and you watch as time goes by and

you actually can watch the stars of
moons you can watch the moon's moving

around so he just had a notebook he just
sketched it in there and typed it up and

that's that's how you did astronomical
observations back then interesting thing

about that picture this is what Jupiter
look like it's look in his book this is

the edge of the field of view of that
eyepiece and then he sees Jupiter as a

bunch of something fuzzy there and this
is what we call high-resolution

astronomical imaging back in 1609 but it
was enough that he could actually see

the the moons around Saturn that's
pretty cool or on Jupiter so it's just

Jupiter just with that you couldn't see
those with your your bare eyes so all of

the stuff that we know since then as
first of all this all relies on some

very simple classical optics remember a
few classically you can treat photons

like a wave double slit image and you
actually have these interference

patterns over here and there's an
optical there's a really criterion for

what sort of resolution you can get D
sine theta equals mmm is that order the

interference times lambda and that tells
you what type of resolution you have so

if I have a telescope with a certain
diameter D here and it's a certain

wavelength I can tell you right away
what sort of angular resolution I'm

gonna gonna have from that classically
from all those measurements that have

been subsequently made using essentially
classical description of light we

understand how stars evolve and how the
universe was created we're underwater

from this so here's the main sequence of
stars we Mabel do spectral

classifications and then understand how
they move off the main sequence into the

supergiant sequence and eventually you
make these wolf-rayet stars they burn

off their outer shells and then become
these these white dwarves so a very

sophisticated understanding of how the
Sun works but remember what they're

looking at that is looking at points of
light in the sky and they measure a

color and some spectrum they don't take
an image of anything it's amazing how

much we know about stars because we've
never taken a picture of anything except

for our own we've got that but but
beyond that it's amazing that this

picture actually works that we've got
such a detailed picture and that works

really well they seem to describe
everything okay so angular scales and

optical Astronomy as we know these are
half a degree across 30 arc minutes for

the Sun in the moon that's why you have
an eclipse planets are thirty arc

seconds and it turns out that if you put
your so this guy if you look at your

really criterion up here for a single
lens or telescopes on a baseline it's

the same criterion this is the
resolution that you get based upon the

size of the telescope and the wavelength
of the light so human eye is one

centimeter and you put this into here
with visible light and you can just

barely see on it so you can see you
can't resolve them but do you see them

as point sources but you can't see the
Sun of that moon obviously if you have

an optical telescope a ten meter
telescope it turns out you can resolve

these no problems at all but you can't
actually do stars with a ten meter

telescope there is a proposal to build a
thirty meter telescope that that'll

happen in the next 10 to 15 years if
everything goes well and that will just

start to approach where you can image
nearby stars with 30 meter baselines but

right now everything is a point source
as far as we're concerned Hubble in all

the telescopes that we know well
everything looks like a point source in

their unresolved if I want to actually
resolve pictures of stars and these are

actually pictures of some nearby stars
these are 30 Milla arc seconds and this

is done through interferometry so we can
do is we make an array of telescopes

separated by several hundred meters and
then the light from the Telex

basically it's like taking a very large
dish but we're masking off everything

except for a few regions we take the
light and then we beam it down into some

kind of central chamber and then we do
an interference with it when we measure

the amount of light in an interference
plane and reconstruct that we're going

to the details right now about that when
you can get baselines in the infrared

using classical Michelson interferometer
II of about 300 meters that's feasible

they're starting to push maybe to 4 to 4
to 500 now but it's in the infrared the

reason why it's in the infrared is the
atmosphere is not stable in the visible

light it has too much variation in index
refraction over short timescales and so

you can't maintain the consistent path
length from the star going through the

atmosphere the stars twinkle too much
and so we can't actually have a stable

in or Mach interferometry going on if
the light path keeps on the time delay

keeps on changing on the order of tens
or hundreds of wavelengths every few

milliseconds so they can do this in the
infrared but they really can't do

anything in the green or the ultraviolet
bands typical bright star is on the

other hand around one millisecond that's
even by beyond enter from Raj equipment

so there are some stars that we can see
like this and these are actually and

really interesting most bright stars are
1mil arcs ok so that's something where

you can't do that with a typical you
can't do things with radio images the VL

VA has baselines of a thousand
kilometers but you have to compensate

that with the fact that the the
wavelengths much longer as well see what

the ratio in your gain

this is this shows a classical Michelson
interferometer Shar it's a up on Mount

Wilson in California and sexually in
Georgia Georgia State actually as part

of sharra they have a several different
one meter diameter telescopes at some

distance and
white in these vacuum tubes into a

Central Station there's a lot of
sophisticated electronics and optics in

here you can reconstruct an image this
is Altair it's about twice as wide as

the Sun the analogy is of them looking a
hundred miles away with this telescope

you can actually read a newspaper in one
letter and hundred miles away with image

if you look at that star that's actually
what it really looks like it doesn't

look round it looks kind of like a like
somebody sat on a ball and that's

because it's rapidly rotating that's the
cool thing about it in fact there's some

interesting surface features if you
notice this this is done the in the

infrared but it is hotter here and this
is colder around the equator and the

reason that happens is if I have
something that's rapidly rotating just

like the earth the equator bulges out
and the poles push inwards you have

shorter distance between the center
where the that the core of the Sun or

the core of the star where the fusion is
happening to the outer surface at the

poles and you do at the equator so
radiation can escape from the center to

the poles much easier so it's a couple
thousand degrees difference in

temperature at the poles sources

it's out a little better that sounds
better okay so like I was saying the

core the distance between this where the
fusion is going on the center of the

star it's less to the surface of the
star at the pole than it is in the and

so it takes longer for photons to go to
the outside you have more opacity so

therefore you get heating that's
actually called the Banzai pole effect

was predicted in 1920 1919 for rapidly
rotating stars and then so they actually

took a picture they can see it turns out
that the distribution that they expected

analytic for the Von's eiffel effect it
explains that a proxy but it's not

exactly correct they're not sure why it
doesn't match right now but it's an open

mystery as to what's what's going on
this with a star and why it doesn't

match that effect okay so well I want to
tell was a different type of astronomy

we use the quantum properties of light
and I'll mention two different examples

of this one of them has to do with the
fact that photons are bosons and

remember that bosons have to have a Bose
Einstein statistics so they have to

group together and both in space in time
when you make the wave functions they

have to both have a if you have two
photons coming to you from a source you

have a wave function for the one wave
function of the other and you actually

have one where they're there if they're
identical you can do the cross exchange

as well and so the total wave function
has to be symmetric with the exchange of

two particles when you do that that
cross term has a positive sign and it

results in bunching of the photon both
in space and in time so this is a sort

of a quantum description as you've
probably known you're you're a modern

physics class you can also talk about a
double slit experiment and individual

photons going through these things if
you try to fool it and figure out which

slot the photon went by by trying to
measure it then you'll lose the you lose

the interference pattern over here but
of course the thing that always was

interesting that you can slow the number
of photons down as much as possible so

he's seeing one photon at a time but you
still see interference patterns here

over on this screen even so the photons
sort of self interferes with itself it's

a basic property of the
quantum properties of the photon and

it's not it doesn't have a classical
analog as to why this happens but it

comes out of this the quantum mechanics
so the two photon waves that has to be

symmetric these are non classical
effects this is also can be referred to

as hbt interferometry or its intensity
interferometry is the tournament so

Hanbury Brown first used this type of
punching in the 1960s to make the first

images of diameter measurements of stars
well we'll talk about that in a minute

but if you want something that's a
little bit more intuitive as to what's

going on here with white photons
themselves if I if I ask you if I look

at a a white screen and I have a
projector putting white light on the

screen and I say well how are the
photons distributed on that screen say

well you know he's just they're painted
uniformly they arrive uniformly but if

you go very very quick on the order of
femtoseconds 10 to minus 15 seconds

you're gonna see individual photons
arriving on the screen so then if I ask

you how are those distributed say oh
they're uniform and they're not uniform

actually they bunched together in space
and also in time so they like to come

together in pairs and like to be grouped
together you've seen something like this

before if you take a laser and you shine
it on the wall and see a speckle pattern

that's what it would look like if your
eyes were very fast if you look that

black body light coming from the Sun or
from a star we'd actually see a speckle

pattern on the wall and it would be just
changing extremely rapidly because your

eyes are slow it looks like time
averages just uniform but in fact there

is a speckle pattern going on now the
question is to look at the speckle from

our laser this there's a kind of an
average size of these speckles they're

all kind of the same they're not all all
over the place in terms of their

dimension they all have about the same
type of size maybe a little bit of

variation here and there but never
there's something that makes them all

kind of the same and that's because
they're because of these correlations

what sets the the size of these
correlations this correlation distance

between them it depends on the
wavelength and also on the angular size

of whatever's emitting at the the laser
hasn't half

but you're associated with it and how
much the aperture is how much its

collimated coming out of that so you
have the same optical theorem here those

d equals one point two lambda over theta
where theta is the collimation of it in

the lambdas the frequency that tells you
what the spacing you expect to see on

the on the screen associated with that
laser beam or were the or if it was a

white light source or a black body light
so it's like the sun that tells you the

spacing that you'd expect to see on the
ground and very quick time scales okay

so this this gives you a little bit more
mathematics behind this you can think

about the speckle interpretation as
there's a bunch of sources up here they

have some kind of a angular diameter
here the light scatters on it and then

you get these correlations you can see
that over short distance is correlated

and so if I calculate from one photon
and another like look at the histogram

of separations you actually see
something where I'm short scales it's

bunched and then along scales it's just
random this g2 factor is uncorrelated

here this must be one and that wants it
being two so g2 is defined as the

intensity of one times the intensity the
other at some other scale divided by the

average intensity if they're
uncorrelated then I 1 times i2 equals I

1 I 2 and you just get one but uh if I
squared is greater than the average of I

squared and you get a correlation net at
at small distance scale so that's the

correlation that you're seeing due to
the fact there they're bunched together

g2 is just a measure of the bunching I
wanted to give a quick analogy to

something that perhaps people have seen
before you see the same effect in atomic

and molecular physics so if you take it
up you can make a group of atoms put

them at a low temperature and put them
in a very small volume and have them

suspended in the air then remove the
suspension that's let them drop to the

ground so that's what we're gonna do
here we've got a silicon detector here

we have a bunch of atoms that are all
these are going to be helium atoms and

we're gonna put them in a small size can
find them and this let them drop so this

is a paper there's a nature paper in
2007 which did this again so if a photon

or a helium atom comes over
here from one place it can go that

straight directory go that way officers
just crossed her and that shows up and

you have to write down the wave of
function as the sum of the straight

terms and the cross terms so it's gonna
be a symmetric wave function what do you

what you see is if you have helium four
it's a boson and so therefore they

actually bunch and if you simply change
helium four to helium three in this

thing then you actually get anti
bunching because they're fermions it

reverses the sign behind it so again
this is something that you understand

from this particles will do this they'll
bunch together on the ground this is the

typical separation that you get between
the particles on the ground and the

shape where this depends on the size of
the size of the region to which there

can find in fact in this paper they they
wind up putting different sizes of

confinement regions and you actually can
get different function you actually

reconstruct the size of how its confined
by measuring things on the ground so we

have the same situation with stars
everything we want to know about the

star actually we actually have that
information is hitting us on the ground

all the time what we want to do is you
want to look at how photons are bunched

on the ground from a particular star and
if they're grouped together and we

measure that size of that grouping we
can actually say okay then it must have

come from this diameter the story no the
wavelength I know the size of the the

grouping on the ground all I need to do
is grab those pieces and reconstruct

them and generate the image of the
original star it's not what we typically

do in astronomy tipping astronomy just
measure all the photons and you're not

looking at a correlation we have to look
over very short timescales and look for

these correlations over nanoseconds or
faster depending on the bandwidth were

looking at so it's not something that
anybody's ever done before except for

one group had done it previously but
generally astronomy consists of opening

up my CCD camera integrating for 10 or
15 minutes taking the picture and I'm

done you're not measuring how photons
are correlated within the camera it's a

different type of astronomy

okay so this is the basic idea behind
intensity fer ama tree we have two

telescopes if separated by some baseline
here's my star over here it goes through

the atmosphere and atmosphere can
actually distort things a little bit but

for this type of correlation it doesn't
matter I measure the intensity in one

telescope multiplied by the intense and
the other divided by the average and

that gives me this one plus this
correlation factor squared and this is

my my observable for my inner front
metric observable this was first done in

1960 set the 1960s and Hanbury Brown
published his first results on this in

the 1960s using telescopes in Austria a
10 meter large telescope so if I if I'm

using invisible like 1 meter will get me
point one arcsecond if I have a hundred

meter baseline I can get 10 millisecond
this is where my I want to get to so I

can start seeing stars I've got to have
a hundred meter baseline in the visible

light so go backward in history a little
bit Henry Brown came up with this idea

of taking two telescopes putting
photomultiplier tubes this we met just

measuring the current in the two
photomultiplier telescopes and then

having a photomultiplier or a vacuum
tube here where they take the two

currents and you multiply the two
currents together and look at the

correlation between the two over
sometimes go this is about a 40

megahertz bandwidth that he did this
measurement over the idea was that if

you this was already had been done in
radio telescopes that take two radio

telescopes you move them together over
small separations you actually see a

correlation in the radio signal you see
the same sort of intensive infer amah

tree and that tells you the scale of
whatever is emitting the radio emission

so he figured that he must have the
exact same thing going on in the optical

if it happens in the radio their photon
so they forgot to behave the same way

and so he thought he would try this this
is before any of this quantum

description about what happening what
was happening would go on when he first

tried this and he saw a result many
people were extremely skeptical

it only took it took about 20 years
later for the for the theory to catch up

and for people that realize that fact
this is a real effect that you would

expect

this is what his original experiment
looked like in Australia so they had

these two ten meter dishes over here and
there on this Narrabri on a railroad

track they Park them in this building
over here and there's rate there's these

cables over here so they would have the
two telescopes look at a star and they

would have them continually moving
they're trying to project the same angle

to within maybe ten or fifteen meters so
that way they don't have to correct for

the time delay between them and just
track a star for 20 minutes half an hour

the way that they read these things out
if you read the original paper it's

actually scary that they actually was
correct because I had all these worries

about it and I read it the
photomultiplier tube they didn't have

electronics back then for computers or
things like that so instead they would

take the current from the
photomultiplier tube and they would run

it into a servo motor a servo motor they
have a little clock that would go around

a little meter and it would turn around
at a certain rate depending on how much

current was going into it so that was
for one telescope and he had one for the

other telescope they turn around they
multiply the two together in the vacuum

tube that went to a third servo motor
and that would tell the product of the

tomb so observations consisted of point
them at the telescope at the starry

wanted to you'd set the things to zero
and then you on piece of paper you'd

count how many times they went around
until you finished up your observations

and what they found it when the two
telescopes were together the middle

needle went around faster then you push
them apart in the middle we needle went

around slower it's very scary turns out
that he got the right answer but when

you first think about how many variables
are involved and what could go wrong

it's extremely scary it turns out they
took a long time to do it too if you

know one of his later stories he talks
about all the problems they ran into

with weather and nor he spoke up they
didn't have cell phones back then thank

God because that would have been bad
when we tried to in the snow lab that's

that was what was really killing us was
people walking around and cell phones

and radios
in the middle of Austria in the 1960s

there were no cell phones or even radio
towers that's why they were doing radio

store on me there
it had a very low level of radio

emission so that was why it was possible
trying to do a modern thing in a lab

when people are downstairs sputtering
materials with RF generators we find out

all the all the interesting things that
happen when you try to do that at the

University so this is just a list he
actually measured 32 stellar diameters

visual 92.5 and 0.4 milla arcseconds
23.2 four Milla arcseconds in diameter

this was really important before this
measurement there was a theory of

stellar evolution which he showed before
the main sequence and there was an

explanation about the relationship
between the color the temperature of the

star the mass of the star the lifetime
of the star this had all been worked out

analytically using cross-sections that
people had measured for some of the

process some of the fusion processes
once that was worked out and temperature

profiles and estimating what the
polytope was for the star but there one

of the key things you can measure would
be the diameters of stars as the mass

goes up you look in different spectral
classification or measurements nobody

had ever seen you know at one point he
had the Sun and that was it there were

no other stellar measurements out there
so this was really important to make

this measurement it turns out he
couldn't make it using amplitude infer

Amma tree that was way too sophisticated
to do Michelson because of this path

length correction so this was the only
game in town and that's why he got

wanted to be able to do it and it turns
out if you look nowadays what people

have done subsequent measuring he got
most of them pretty close he's won about

10 or 15 percent I'm most of these so
it's pretty good so subsequently

confirmed he actually measured all the
starters that he could give him the

signal noise that he had from the size
of his telescope the quantum efficiency

was photomultiplier tubes the bandwidth
he's using for his optical bandwidth the

sampling rate all those things fit into
the signal noise and also how long he

can stay on the source so it turns out
there were only 32 of them that you can

measure in the southern hemisphere and
then he was done after that has to do

with what most of us in astronomy do you
have to submit a proposal to build a

better experiment and spent a lot more
money and he did that in 1972

down by the Australian government they
thought that this was nice but Michelson

interferometer was going to take over
him and it was going to be the wave of

the future because it was much more
sophisticated and so that's exactly what

happened
nobody pursued this after 1971

this is shows what the typical curve
looks like so here's my correlation as a

function of distance between the two
telescopes fairly large error bars but

you fit a you fit a Bessel function here
and you can get a stellar diameter for

that so and the first time we looked at
this and I was studying the paper I got

really worried that I can fit a lot of
things to that and how well do you how

well do you know that this works so
that's why there's quite a bit of

uncertainty this is also done at a
fairly crummy weather it's not really a

great there's like dust in the
atmosphere near sea level so it's not

one would call a good good place to do
astronomy but because of the technique

using this intense interferometry
technique it turns out that those

atmosphere fluctuations cancel out and
it's really remains relatively

insensitive to those type of details

okay so the signal of noise hearing when
you're doing this intensity

interferometry the hbt interferometry
Sigma noise depends on the telescope

areas the if I have two telescopes that
geometric mean of the telescope's

depends also in my quantum efficiency
square root of integration time and also

my electronic bandwidth so if I'm
sampling at 40 megahertz if I decide to

sample at 400 megahertz I get a square
root a square root of 10 increase in

signatories or a reduction in a factor
of 10 in terms of the amount of

observation time I need to do to be able
to see that same level of sitting the

noise since 1971 there have been a lot
of improvements in all of these things

we have much larger telescope areas in
it the Tector quantum efficiency back

then for a photomultiplier tube was
about 15 percent 10 to 15 percent you

can buy regular vacuum photomultiplier
tubes now with 50 percent 50 40 to 50

percent quantum efficiency and you can
buy silicon ones that approach

80% quantum efficiency so you have major
gains there integration time we can

certainly do a better job with that but
that bandwidth also we have better

mirrors and we can get bandwidth
up to a gigahertz now so again you can

gain factors of 10 20 30 in some noise
so whenever Bennett's spy conferences

SPIE conference is talking about this a
Michelson interferometer II people will

say well the sing of noise is bad so why
would you want to do it and my response

is if I go back to 1971 Michael said
interferometry the signal noise was

pretty bad then too but things have come
a long way and so that's why it's a good

time to revisit and think about what can
we do with modern technology instead of

the way it was done back then one of the
peculiarities here is the signal noise

is actually independent of the optical
bandpass so all these measurements

you'll measure the photomultiplier to
current in a specific frequency plus or

minus Delta Lambda if I make that larger
or smaller it turns off the noise

doesn't depend on it you say why is it
happened if I go to narrower bandwidth

the correlation actually is better
because you photons are all in the same

quantum state and so they court more and
more than will be correlated but the

problem is I have fewer photons they
decreases so you have to wait longer so

when you multiply a two out together one
said being independent of bandwidth

having said that larger effects like if
they're qualm efficiency varies over the

bandwidth you can go to a narrow
bandwidth and you pick the highest

quantum efficiency you can gain so yeah
it's a little bit more subtle than just

saying I can choose whatever I want
you can also saturate your photo

detector if you have to larger bandwidth
you have too many photons coming in so

wasn't what a modern version of an
intensity Reformatory array would look

like we've got a set of telescopes for
the most by r2 up here and we measure

the currents from each of these
telescopes more than to measure an array

of them with an analog and digital
converter there's some kind of a

synchronizing clock here which keeps
these things so that we know the exact

time of each one of these things we
write these things to a disk and then

the next day we wind up using off the
offline correlator so the nice thing

about this is we can have an individual
telescope which just sits by

so take the data with it you write it to
a disk and we don't have to worry about

doing any correlation the next day we
grab the disk we walk it to a central

location and then he's a computer array
to do that do the analysis the next day

previously when he did the correlation
remember we had one photomultiplier tube

or sorry one vacuum to put the currents
into it then we had to record that data

stream if I had an array of a hundred
telescopes I'd have to have 99 times a

hundred of these correlation systems
nine nine nine hundred ninety of them

are nine thousand of these things to be
able to do all the correlations for all

the different pairs but if I have
software I don't I don't have to

actually make all those different
correlators I just have to run my

computer have a logic computers in
parallel doing the processing so I don't

have to be running cables all over the
place and trying to do this in real time

it actually makes the cost go way down
by by being able to record the data and

then do it the next day the other nice
thing is we can also apply filter into

the data we can get rid of electronic
noise you can do narrowband filtering on

the get rid of sixty Hertz noise or
things like that and post process the

data to clean it up so there's a lot of
nice things about it my focal point up

here you have light coming in goes to
some kind of a narrowband filter and

then we're envisioning that the optimal
thing we have a polarized beam splitter

here one photomultiplier - we'll look at
one polarization this guy will look like

a different photo polarization and this
could be a photo multiplier to you or us

it also could be a SPAD or an SI PM
there have to be a fast photo detector

to be able to do this so the next part
of us is we started working on this and

you want to demonstrate this actually
works so we built it a prototype system

called we call it star vase and it's
named after a a diving place in Salt

Lake City amazingly there's a place
called Sea Base you toss the highest

altitude deep water diving place you can
go the XF whop stirs and sharks and all

sorts of things here it's a hot spring
nearby Salt Lake City at five thousand

feet and the guy that owns it actually
is a friend of the physics department he

let us build these crazy array of
telescopes up here so you can see these

two telescopes separated by 25 meters
and these RF 1.0 telescopes point spread

function is about 0.2 degrees or 13
millimeters this is a good fit for a

photomultiplier tube they're identical
21 metres east-west baseline and so we

use this for testing things out for for
developing the technique and making sure

that we know what we're doing the
optical interface this is the focal

point what it looks like and it's kind
of hard to see here but here's our photo

detector here and there is a mirror here
that's what this thing is here so the

light actually hits this mirror and then
bounces into the photo mobile - and then

there are some elements here of a
there's a actually a slide back and

forth we can slide it in and out a
polarizer and turn the polarization on

and off so part of our tests are we
would have a polarization on one photo

more fire to this direction the other
one would either turn it on or turn it

off by pulling a long cable and opening
it or closing we try to keep all sorts

of motors and things like that away from
the photomultiplier tubes to keep the

noise down so they're really theirs
physically a long cable that we wind up

pulling kind of like you have on like a
choke on a kind of lawnmower it's the

same sort it's actually that's what we
got out from one of the choices we had

to make was what type of filter we use
we we actually found some really

interesting filters by this company
called SEM Rock these are 5 nanometer

filters this is what it typically looks
like the problem is that the wavelength

for a light going into one of these
interference filters actually shifts

depending on the angle that the light
comes in it so here's my original angle

and if you have something from angle
which is off axis coming at 10 or 15

degrees then it actually shifts the way
things that you're looking at so the

original wavelength actually consists of
a bunch of different wavelengths this is

from the center of the mirror this is
from an outer annulus and this is from

the outer ring of the mirror so you
actually have an effective bandwidth of

about 10 to 12 nanometers
although the filter itself has an

inherent bandwidth of 5 nanometers these
are special ones that we found they have

a high index of refraction and so you
can see this effective index refraction

over here goes in the denominator so the
higher you push that

then the smaller effect you have so the
12 nanometer bandwidth we can do pretty

well in doing this inter format your
technique it also cuts in the night sky

background down as well our digitization
system is just off-the-shelf electronics

it came from National Instruments 250
mega samples per second its phase locked

with an FPGA we have a correlator that's
done with the FPGA as well we stream

while our database disks a 12 terabyte
disk so we fill this thing up in a

couple hours with data and then do the
cross correlation to telescope the next

day we have a timing module that's what
it looks like in a single crate if you

buy the stuff off the shelf
it'll cost you about fifty or sixty

thousand dollars per telescope but if
you're careful and watch eBay he turns

out you can save about a factor of ten
in the cost and that's how you end up

doing this on the floor we pretty much
bought every thing for about five or six

thousand dollars per telescope so eBay's
our friend science with eBay we did some

initial tests in the laboratory just to
simulate stars so he had an arc lamp

here narrow bandwidth filter
here's accommodating lens and and then

we had a little pinhole his in my
neutral my my filter that I show that

samraat filter before a focal lens and a
pinhole the pinholes can consist of

different types of pinholes there's
basically pieces of tungsten you blast

them with a laser and you actually make
the size of the hole to be similar and

you push it away far enough so it
simulates what a star looks like we also

made binary pairs both stars that are
different diameters or the same diameter

so we could actually do imaging and
reconstruct binary systems see if you

can use this to reconstruct the binary
system so this has shows you the blue

light coming through here from the end
of the long tube here it comes out of

longer this is about 10 meters away
comes to a spit or splitter over here

this to this one 102 moves back and
forth on a robotic axis the other one

over here is once it being fixed and you
can see it's a little hole here on the

PMT so you can localize things and then
the data goes into this data acquisition

system over here so one of the ways that
we first demonstrated that we saw this

we that we saw that correlation
by changing the polarization so if we

measure data in the polarization mode
here they're both parallel to each other

for the light coming in you actually get
a signal which is this blue one and if I

flip the thing like this and off you
actually get this green signal over here

so there is some kind of funky stuff
going on here it turns out that it's

it's just average noise pickup that you
pick up from from the data acquisition

electronics you subtract these two guys
you actually wind up with a nice

correlation peak at T equals zero and it
actually grows with time the way that

you expect it to it grows like the
square root of time so it's not

something that's a artifact that keeps
on adding over and over again it's a

random process so we can make it
disappear reappear bus by flipping the

polarization on it one of the things we
can also do that's the time correlation

between the two we can also make that
robotic arm moved back and forth and you

can see the you can see the the
correlation between the two photo

muffler has drop-off this looks very
similar to the HB T where they you move

the two telescopes apart and you
actually watch the correlation going

down if you take the the pinhole and use
a uniform disk model that to model what

happens especially with finite aperture
effects you actually can fit this thing

pretty well so we tested things over a
wide range of photon rates we also have

done it with different types of
correlator code we did it two different

ways as well
one is streaming data and then then the

next day doing the cross correlation but
we also developed the ability to

actually use the data acquisition system
those FPGAs to plug another one in the

other data go across the backplane into
another FPGA near the correlation in

real time and that was our big concern
we first started doing this was again we

figured it was gonna take several weeks
to process the data but now we know how

to do this in real time fairly
inexpensively this shows you what

happens when you have a binary system so
here are 2 is in my two holes here in my

and my pin Hall mask if I sweep my my
foot of up r2 in this direction here

this is called the UV plane
so I make a sweep on this direction you

get this peak here and the reason why
you only get a single peak is there's

only one link scale in this direction
and that's the diameter of the the

pinholes if I sweep it in this direction
there's two different link scales

there's a diameter of the pinholes but
there's also the separation between the

two and you want up getting an
interference between them and you wind

up being these secondary fringes and the
fitted function here actually is

associated with measuring both the
diameter and the separation between

these two guys so by measuring a
different phase angles you get different

information about how the light is
distributed and if you want to make a

full image reconstruction not only would
you measure it along this axis and this

axis you'd also measure along different
axis until about this thing this is

called the inter four metric plane you'd
measure you measure that out and then

you do it a Fourier transform of the of
the interfere metric plane to generate

the actual image so I wanted to talk
about this thing so here's here's wha

this is actually a reconstruction of two
different images that we did in the in

the lab one is a stellar disc and the
other one is a binary system so he

measured the UV plane and and and and
took an average of it and we're able to

reconstruct both the binary system as
well as the single star

using this thing in the lab so at that
point we said okay we're good we

actually know what we're doing with
noticed our base will start taking some

measurements so we started doing star
based measurements this past spring we

picked out two stars that are quite
bright magnitude 11.2 they didn't even

long exposures Regulus and also Alkaid
we're separated by this shows you what

the how you'd expect the correlation to
fall off versus the baseline separation

we're at 20 meters here so we start out
right around that distance there and as

you and as you track the star you
actually sample different places here we

picked out the wavelength filter here's
a different spectrum of the two stars we

wanted to avoid these spectral lines so
we're over here in this flat part over

so a 10 nanometer effective bandwidth
over where it's relatively flat we could

gain a little bit in Sigma noise by
going up over here with a narrower

filter you have to get a place to
exactly the right way and it varies from

start if start where those lines are so
so that's something to do in the future

if you try to get the signal noise up a
little bit more we're going to go to

this one
this is shows our signal - background

for both alkaline and reyga's regulus so
this is the time lag and looking for the

two-photon correlation what we see is
some random backgrounds about 10 hours

worth of data and then both of them have
a peak if we sum they - these guys

together you actually have a nice peak
here at like t equals 0 which is what

you expect the way that it concerns is
there's something else going on over

here there's some kind of negative
nasty-looking thing if you want to

explore that again this took about 12 12
hours of observation time we'd have to

at least double that and the question is
how much more time do you want to do I

took us about two weeks worth of
observations to get this data getting to

work correctly so for us it was a
question of either doing this or else

going on to the next set of telescopes
which is Veritas these are three meter

telescopes Veritas is a 12 metre
telescope and turns out that signal

noise goes like the fourth power of the
radius of the telescope so or is this

talk 11 hours they've taken on the order
of 10 minutes or something like that at

Veritas so at that point we said it'd be
probably better just to move on to

Veritas it is it we think that we're
seeing the signal and lets us go and put

it on something has a better chance of
actually seeing it

okay so beyond Starbase the next step is
to take this to a set of imaging air

shrunk off telescopes and our first goal
is to put something on in the next year

Veritas imaging air shrunk away this is
the four telescopes of Veritas separated

by about a 100 metre 80 100 metre
baseline and 12 metre diameter mirrors

but there are other possibilities from
this as well and paranoia is a very

large VLT I Very Large Telescope
interferometer

it's consist of these ten meter
telescopes on 100-meter baselines and

they're actually doing Michelson
interferometer II and so the cool thing

is you know that they're already looking
at stars that they can image things that

have the right diameter remember how the
mikaelsons work they actually have a

photon it gets collected by the mirror
it was up at the focal plane and then

they pipe it down in down below here in
this base underneath here and they have

the they have the path length correction
all the optics associated with that and

they do the finished prints tracking
down and the vacuum tubes down in the

basement they actually throw away the
visible light because there's no visible

band and the u-men because they can't do
anything with it it just has too much

variation due to atmospheric fluctuation
so they have a little beam dump that

just dumps it into a hole one of the
things we could do is put a white sensor

on the beam dump we just record the data
and we wouldn't affect the focal plane

at all so we're talking to them about
potential are you doing that nice thing

is we take a take a we hitch a ride on
the back of their coats because they're

already looking at interesting objects
that are resolved that's what they're

trying to do so we don't even have to
think about a science program we'll just

provide them with additional wavelengths
besides the infrared wavelengths we can

provide them with a visible or the other
UV coverage for those for those same

sources they're really interesting
things to go to this future shrunk off

telescope or a CTA and this that was the
first telescope we saw in the beginning

it's called a prototype s CT telescope
and use those telescopes over kilometer

baselines
when one goes to kilometer baselines you

get down to 10 to a hundred nano arc and
then a nano arc second resolution so

factors of ten better than ten to one
hundred better than you can do with any

other technique that's been done so far
and so at that point you start to look

at some very interesting things well
with these telescopes so they mentioned

CGA consists of one to two kilometer
baselines 2200 telescopes there's a

northern slightness of insight there are
these large outlet disease are called

LSTs they're 23 meter
mirrors there's also some medium-sized

telescopes these are 10 to 12 meter
class mirror mirrors and then there are

these little tiny ones out here these
are 3 meter telescopes and those are

called this SSD small size telescopes
that the layout that's being looked at

right now for CTA is an array in the
north this is Canary Islands of about as

for these MS for these LSTs in a larger
area the small telescope Southern array

will be for them small or the large ones
array of about 20 or 30 the medium ones

and a very larger area these smaller
ones and so with us you actually can

have a large number of base lines
mm base lines here to fill out the UV

point instead to imaging with these this
shows you that as a function of

wavelength this is where Hubble is
angular diameter they can observe this

is Miller arcseconds voi and Shari down
here one mil arc second we will be down

here at 10 to 100 nano arc seconds over
here here is the VL VA over here in the

radio they can just start to get down to
there at a few specific wavelengths but

most of the like almonds up here at the
10 mil arc second level

I'm gonna jump over a couple things here
because I know we're running out of time

but I saw a couple examples of things
you can do so this is vo of Ti

this is that was that middle imaging or
middle Michelson interferometer II and

ADA Carini is a very bright source we'd
like to look at this is what they would

measure this what they resolve an ADA
Queen it's actually a binary system and

it shows their their resolution is on
the order of 30 au they can resolve this

thing we get CTA we actually this little
red dot here and red dot here you

actually can see the binary partners on
here this shows you the uv-plane

coverage of a Veritas and this is the
same picture here if I look at the

separation pre telescopes that that Alma
has it's on these sort of scales here

and Veritas fills in these other other
distance scales over here and this shows

the instantaneous coverage in the
uv-plane there the 48 image plane so

it's densely sampled you can do a really
good job getting an image here I

mentioned this before I'm gonna jump
over this is the Devon cycle effect with

altair there is a interesting a very
famous star known as Vega which

everybody uses for optical Astronomy
that's the one that used for your

reference star pretty much all
astronomical observations since about

1870 or 1880 1890 they would do that
we're using Vega as the reference star

so you look at Vega you put some filters
on you'd measure some magnitude and you

go to the star that you want measure
their magnitudes and the different

filters that come back to Vega measured
again and you know what the Vega

spectrum is supposed to look like and so
you can correct for perhaps changing the

atmospheric conditions and then you can
extract in the star you're looking at

its actual true spectral uh it's
probably much expectrum out of it

one thing that was discovered by char
about ten years ago and in fact it's a

poll on rotator so it's actually hot in
the center and

the edge edges here are actually cooler
they're supposed to be so in fact it's

been spectrally misclassified the most
important star in the northern

hemisphere has been misclassified in the
past 150 years but and still you know

they know what the spectrum is and they
use it but it's in terms of the HR

diagram where it belongs it's actually
blue or the work of where it's supposed

to be given its mass in minutes given
its size that actually brings up an

interesting question as to you know
there are these things called blue

stragglers that people see in globular
clusters clusters of stars where all the

stars were made at the same time and so
as they evolved in the HR diagram this

is temperature here and this is their
magnitude so remember if we had the

main-sequence over here stars at the
larger masses here evolve faster and

they move over the giant branch and
looking at where this turn office tells

you the age of this glob or their
cluster there are these blue stragglers

here that people see in their globular
clusters and they should have died a

long time ago they should actually be
gone given their mass but they're still

that's why they're called Ruby strikers
have done a lot of discussion as to what

causes them perhaps they're in binary
systems and they've been spun up and

mass has been transferred to them but it
also could be that they're simply just

like Vega maybe their end on rotators
and they actually are over here but

they're just observing the poles on them
breath and observing their equators so

again these are kind of interesting
things you can do with being able to

image a large number of stars and being
able to see how they evolve you don't

get this from just looking at the
spectrum itself you have to look at the

the imaging of those stars okay I know
we're about out of time this is just a

simulation that was done in a couple of
papers that were published a little

while ago that's part of our white paper
we put binary images in here and

actually then your reconstruction this
is what the UV plane looks like when you

sample it using CTA and you can actually
generate pristine images from it so what

I think I'm gonna
jump the one last thing that I think is

really interesting in terms of quantum
proxy light and we're not intending to

do anything on this right now but there
are groups in the world that are looking

at this property of light we had
mentioned photon bunching but there are

other properties of light which people
are not using as well and one was called

the orbital angular momentum of light so
most undergraduates if you ask them what

spin does a photon have how much spin
can it carry how much angular momentum

say well it's a spin one take one it's a
spin one particle so you can get h-bar

or minus h-bar associate but turns out
that's incorrect that's just a spin

component there's an angular momentum
component as well and you can actually

put them in a spun up State you can have
the orbital angular momentum carry 10

h-bar or 50 h-bar 100 H bar if you want
so nobody's ever looked at that before

we just measure a photon so we don't
look at the quantum state of the photons

there was a group in Italy that's
actually looking at at this stuff it is

expected that in some Astrophysical
planet plasmas and some shocked shocked

environments they who actually can spend
things up this various much like a maser

emission but nobody's ever looked at it
so Mark Martin or a paper in 2003 about

this stuff rotating black holes might
have it may be in the Cassini might also

have these type of photons as well
pulsars can also you generate this type

of light so it wasn't telling you
something about the origin of the light

by looking at the quantum properties
rather than looking at the spectrum in

the intensity gives you more information
about the source so okay this is the my

last slide here no worry we're out of
time it's four o'clock so some people

ask you know what is 100 100 nano arc
second resolution live like and so this

is what it would look like this is a
star that would be stellar derivative

one point seven solar master seller
massive one point seven solar masses

something that's 2 parsecs away so it's
a 6 ml arc second diameter so it's quite

large this is much larger than what we
were then then our angular resolution of

CT a but if we had a Jupiter going
across

image and you could actually resolve the
shadow of it that's what you would see

actually now you see the shadow jury
whatever had rings like Saturn you

actually see the Rings and these guys
are actually the moons these are this is

true the angular extent of the moons
that you would get around Jupiter you'd

be able to pick those up as well from a
star that's nowhere near us but you know

several parsecs away from us that's
pretty cool remember Galileo started out

by looking this and he saw some moons
around Jupiter nearby us you know with a

little bit more work you may be actually
be able to go to other stars and be able

to see the same same type of moons
around eggs EXO moons which nobody's

ever seen before and egg so rings those
would be kind of cool so got a long way

to get to get to that point but if we
can get to that anger resolution that's

the type of thing that you can do with a
kilometer baseline no no you can't see

the whole 80 or magnitude limited and it
has to do with a spot size of the

telescope quantum efficiency and so
whenever you you're measuring with this

white bucket you're going to be pulling
in starlight and you mean integrating

over and that actually sets a limit for
you so one of the questions people

always ask is Chi sees something at
another galaxy go to go to go to

Andromeda and do any of it you can't
actually you know the magnitude limit

their winds up being about like 14th or
15th magnitude you have to get to and we

would be barely able to get to ninth
magnitude if you push all the limits so

I don't think you're ever going to be
able to image something in in another

galaxy it's just uh not have enough
photos if you waited forever you can do

it but you're not gonna
they do it in a reasonable observing

time I did have a lot back here
someplace

quickly so there is a catalogue of
nearby stars like jumped over this

called the J M MC catalog and so it
depends on the temperature that you're

looking at here this is my angular
resolution so point six milliseconds

0.03 here's my visual magnitude so it
depends on how long you look for a CTA

if you look for a hundred hours you
could find everything down below this

this thing here so there's still a lot
of stars there's hundreds of stars here

this catwalk has three thousand stars so
even though these would fall within the

the diameter that you could see it
depends on the magnitude and how long

you're going to observe as well but on
the other hand there's this shows you

basically the spectral classification
over here it's a very small number and

they're always type stars these are
actually very easy to see but there's

not very many of them these ones are
harder to see and because the

temperature is lower but but anyway
there are there are quite a lot of them

nearby that you can see you so when we
receive some funding from the National

Science Foundation instrument Veritas
and part of it was we promised that we

would image 30 stars by end of 2019 it
looks like we're gonna be able to do

that obviously huh and one of the
questions the reviewers asked was how

many stars are out there and it's on the
order of several hundred stars that we

could look at so that actually brings up
an interesting opportunity people will

say well how much time you can be
spending using that imaging error and

cough telescope for this this optical
Astronomy I mean it was built to do

gamma-ray astronomy what are you doing
doing optical Astronomy with it we're

gonna be doing this during full moon
when the telescope's are not being used

for gamma astronomy but there's also the
possibility that if you in the central

pixel of the camera if you just
instrument that pixel you could do

gamma-ray observations at the same time
because there's so many sources that are

out there you just pick the gamma
resource you want to look at and then

shift the the field of view over to go
to one

these stars that you're interested in
and do the observations at the same time

so you may be able to pick up more than
just a few days per month you might be

able to put a whole program together and
look for several hundred of these stars

that depends on how noise we can keep
the electronics that's that's gonna be

the catch there so you say you just have
to go to so you can calculate yourself

you take the 0.1 millisecond resolution
how many radians that is you know what

the stellar diameter of stars are so you
can measure the one solar radius or 10

solar radii or so and then you then you
put that into the formula and that tells

you the distance you can see it - it's
nothing more than that you do have to

worry about the photon flux how long do
you have to wait so if it wants it being

a hot enough star it turns out the
spectral density is very high and you

can do it very quickly but if it's a
cooler star so I could go look at a red

giant but the problem is it's you know
3,000 degrees the photon density and

then you and the V band is quite low so
those are pretty hard to observe even

though they're large and if a big
luminosity there's not a lot in the in

the band you're looking at it's all down
in the sort of the red band so there's

those are more difficult so that that's
where these if you go to a number of

different references for this there's a
white paper that we're going to be

putting on the on the archive which
actually goes through the calculation of

all those different trade-offs but you
can get to tens of parsecs 20 parsecs

just using you know fairly simple
electronics for the punch off the shelf

if you go to more custom so custom
atronics you can go a bit farther you

get better signatories and as to how
much observation time obviously dude

that's that's another axis you have to
observe longer and longer for things

that are further

you mentioned using this technique we
were looking in the visible band in the

u-bend so think about it so so in
stellar physics we've got a couple of

interesting questions that you want to
know one of them is you know how does

how does the stuff that envelope
drop-off how does the the mass drop-off

as I go is further out that's part of my
my solar model depends on opacity it

depends on the temperature of the star
they're measuring the in the infrared

and so the the cross-section is lower so
you get a bigger star you should go to

higher if you go to longer wavelengths
the cross-section goes up the star

actually shrinks and so you actually
have information there in fact if you

talk to people that do stellar a strong
said they would love to have

measurements of diameters as resolve to
different wavelengths because then you

actually have a model that you can fit
to for that there are other things that

are really interesting to look at
sunspots or star spots so star spots are

much cooler than the rest of the star if
I go to if I'm in the infrared Bay

member that the blackbody curve from
6000 degrees Kelvin goes like this and

if it was at 4,000 it goes like this
the infrared bands are the they just

parallel to each other but they're not
really distantly separated if I go to in

the visible band one drops off the other
one's keeps going so you have a huge

contrast there so if you want to look
for something like star spots or things

like that you have a big advantage in
going into the U and the V band and

that's a it's a really big open question
as to we know that our son has star

spots do o class stars have them do em
class there's no actually you know

nobody knows there are theories out
there let's say perhaps the large stars

don't have it because the magnetic field
is different but nobody's ever seen one

so we don't know so I think that's
actually one the most interesting things

is there okay let's look at M Class
stars which we know have ours has an

11-year cycle or
11 year or twenty20 cycle are some 30

what makes the cycle and that's actually
not a computationally tractable problem

right now nobody has a model to actually
even do our own son because given the

complexity of the magneto hydrodynamics
and number of particles you have to

follow nobody has a viable model for
doing that if you want to try to make a

violent well you need to have more than
one point to be able to compare the

model against so things are a lot you
can learn about how stars of all then

you just have I mean this is what TMT
ones that are right there they they will

have visible coverage but they're it's
just a single dish is what they're doing

so you have to maintain phase coherence
over a 30 meter dish and so you have the

actual readers in the back of the
mirrors and they're automatically

adjusting for those wavelengths things
for something where it's separated by a

hundred meters like you or something I
guess you could do that but the stand

where they do what they measure they do
fringe tracking in one wavelength and

then they measure in another wavelength
so they're basically using the the

properties of the star in a more stable
band and then they correct for it in a

less stable band for the for the time
delays I haven't done Michelson

interferometer myself I'm learning about
about it myself

but I believe that's how it is you use
two different bands a short band and a

long band to do the fringe tracking