https:/.../30c3-5305-en-Breaking_Baryons_h264-iprod.mp4

Michael Büker: Yes, alright, thank
you very much, okay. I’m glad

that you all found your way here
and it’s been mentioned already,

this is Comic Sans, which as you
know is the official type-font

of awesome particle physics stuff.
laughter

But in the interest of our mental
sanity, I will keep it to other fonts.

So from here on Comic Sans
is just a bad memory.

Okay, two things: First the
title, Breaking Baryons,

which of course is an allusion
to Breaking Bad, was inspired

by the wonderful talk from last year which
was called “How I Met Your Pointer”.

And which was also very successful
and you can check out that talk,

I got the link there. And this
talk goes especially well

with another talk that we’ll have
tomorrow by a real particle physicist,

at least a bit more than myself.

And it’s called “Desperately Seeking
SUSY” which is about particle theories

and the real cutting edge physical
questions. This is going to be

happening tomorrow. Allright, so
we’re going to start out with my talk

and I’m going to be talking about the
questions of “what are we doing?”,

“why?” and “what kind of stuff do we
use?”. And I’m gonna spend some time

on explaining this last part
especially. What is it that we do

and how does this work? So, what
we do is we give a very high energy

to small particles which
we call accelerating.

But from a certain level of energy
this doesn’t really make sense,

because we don’t actually make them go
faster. Once they reach the speed of light

they can’t go any faster. We just
turn up the energy and the speed

doesn’t really change. This is technically
useful but it also gives rise

to doubts about the term accelerating,
but anyway, we just call it ‘accelerate’.

There’s 2 basic types of devices that
you see there, you have storage rings,

which are the circular facilities that
most of you know. And then there is

linear accelerators which are in
comparison very boring, so I’m

not going to be talking about them
a lot. We make the particles collide

which is the reason for giving them high
energies, we want them to smash head-on.

And then this last part which is about
the most difficult thing is we just

see what happens. Which is not
at all as easy as it might sound.

So why are we doing this? You all
know this formula but I’m going to try

and put it in terms which are
a little bit closer to our hearts,

as we are here at Congress.
I might postulate that

parts, like electrical parts, building
parts, are actually the same as a device.

Now this is not quite wrong but it
doesn’t feel exactly right, either.

I mean, if you have some parts and then
build a device from it, it’s not the same.

It’s made from the same thing but you do
require a certain amount of conversion.

You have a building process, you have
specific rules how you can assemble

the parts to make a device and
if you do it wrong it will not work.

And this is actually pretty similar to the
notion of energy being equivalent to mass,

because energy can be converted into mass
but it’s not at all easy and it follows

a lot of very strict rules. But
we can use this principle

when we analyze how particle
reactions are used to take a look at

what mass and what energy forms
there are. Now suppose we are

thinking about a device
which is very, very rare,

such as a toaster that runs Net-BSD.
laughter

Now as you can see from the photo
and the fact that you see a photo,

I’m not making this shit up. There
is a toaster that runs Net-BSD but

that’s beside the point. Now if we
are particle physicists and we want

to research this question, we know
that parts are the same as a device,

so if we just get enough parts and
do the right kind of things to them,

there might just turn out, out of
nowhere a toaster that runs Net BSD.

So let’s give it a try. We produce
collisions with technical parts

and if we do enough of it, and if we
do it right, then there is going to be

this result. Now from these pictures
you can see, that doesn’t seem

to make a lot of sense. You will not
get a toaster from colliding vehicles.

laughter
But as particle physics go,

this is the best we can do. We
just smash stuff into each other

and we hope that some other stuff
comes out which is more interesting.

And that’s what we do. So to
put it in the technical terms,

we use storage rings which are this
one circular kind of accelerator

to produce collisions. Lots
of them with high energy.

And then we put some enormous
experimental devices there

and we use them to analyze what
happens. Now first let’s talk about

these storage rings. This schematic
view is what a storage ring is

mostly made of, and you can see right
away, that it’s not actually a circle.

And this is true for any storage ring.
If you look at them closely they are

not a perfect circle, you always
have acceleration parts which are

not actually curved. So we
have the 2 basic elements

of a curved part which is just “the
curve” and then you have a straight part

which is there for acceleration. Now you
have this separation, it would be nicer

to have a ring but it’s much more easy
this way. You have the acceleration

where it is straight and because it is
straight you don’t need to worry about

making the particles go on a curved
path. So you can just leave out

the magnetic fields. We
need magnetic fields

to keep them on a curve, but we need
electrical fields to accelerate them.

Now we could try and assemble these
into one kind of device. A device

that uses an electric field to accelerate
the particles and at the same time

uses a magnetic field to keep them on
a curved path. Now this is the first thing

that was tried. These kinds of
accelerators where called cyclotrons,

but they were very inefficient, you
couldn’t go to high energies, it was

very difficult. So the evolution went to

this way where you just
physically separate the 2 tasks.

You have a straight part for acceleration,
you have a curved part for the curve

and then that’s much more easy.
Okay, so let’s take a look at the

acceleration part of things. You
may know computer games

where you go racing about and then
you have some kind of arrows

on the ground and if you go over them in
the right direction they make you faster.

This is a kind of booster if you will.

If you happen to go around the wrong
way and you go onto these arrows,

they will slow you down, which makes sense
because you’re going the wrong way,

you shouldn’t be trying that. And this is

the same effect we can think of when we
think about what an electrical field does

to a charged particle. If a charged
particle moves through an electrical field

in the ‘right’ direction so to speak
it will speed the particle up,

taking energy from the field and to the
particle making it go faster. But if you

go the wrong way, then this particle
will slow down and it will

give off energy. If we where to try and…

let’s say we have a level editor,
right? And we can edit this level

where this little vehicle is going and
we want to make it go really fast.

So what do we do? We just take this
acceleration path, we just take

these arrows and we put them in a long
line. Let’s put 4, 5, 10 of them

in a row, so if we go over them
we’ll be really fast at the end.

Now suppose the level editor
does not allow this. It’s just

by the rules of the game it’s not possible
to put a bunch of arrows in a row.

Which sucks, because then we can’t
really make them go really fast.

But then we just ask an engineer
who’s got this shit together.

And what is he going to suggest?
You know what he’s going to suggest.

Can I hear it? Come on, “inverse the
polarity”, that’s what he always does!

laughter and applause

So we inverse the polarity. And we are
going to make our track look like this.

So we have an arrow which gives us a boost
in the right direction and then there’s

an arrow in the wrong direction.
If we go over the track in this way,

we’ll speed up and slow down and speed
up and slow down. And in the end

we won’t win anything. But here is where
Geordi comes into play, because

we’ll be switching polarities at just the
right moment and if we switch polarities

at the precise moment that we are
in between two of these fields,

then the next one will be an accelerating
field. And it goes on and on like this,

we always switch the direction
of the arrows at the right moment

when we are in between the two. And
from the point of view of the vehicle

it will look like there is an accelerating
field followed by an accelerating field,

followed by an accelerating field.
Which is the same as we tried to build

but which the game, or in the case
of real accelerators the universe

just wouldn’t allow. So we’re tricking
the universe by using Geordi’s tip

and inversing the polarity at just the
right moments. And this is what is done

in particle accelerators and this is
called Radio Frequency Acceleration.

Now this kind of device that you see
there is the device that is used

for this actual process in actual
accelerators. It’s about as big

as a human child, but it
weighs a bit more, it weighs

several hundred kilograms.
And in contrast to a child

it’s made of a metal called Niobium.
Now Niobium is a rare metal,

but it’s not super rare, and it fulfills

3 basic requirements that
we have for these devices.

It’s ductile, which means you can
easily shape it, because you see

that this shape is really weird, you got
these kind of cone things going on,

and they must be very precise. If these
cones on the inside of the cavity

are off by just micrometers the whole
thing won’t work. So you need a metal

which can be formed very well.

Then you must be able to make it
superconductive, to cool it down

to a temperature where it will
lose its electrical resistance.

The electrical resistance will go down
to almost zero, some nano-Ohms

is what’s left. So that’s the second
requirement for this metal,

and the third one is: it shouldn’t
be ‘super’ expensive. I guess

you could use platinum or something but
then you couldn’t pay for the accelerator

and as we are going to see, the
accelerator is expensive enough as it is.

So Niobium is what is used
for this kind of device and

as I said, we cool it down to about
4 Kelvins, which is -269°C

or 4°C above absolute Zero.
And at this temperature,

the electrical resistance of the metal
is almost zero which we need

for the high frequency
fields that we put in.

What we used to cool these things is
liquid helium, so when they’re in use

inside the accelerator they’re not
naked, exposed like you see here,

they are enclosed by huge tanks
which are super tight and must

hold on to large pressures and
be super temperature efficient,

very well insulating
because these must keep

the liquid helium inside. But on
the outside there is the tunnel

of the accelerator and that’s where people
walk around. Not while the accelerator is

running, but people walk around to do
maintenance and stuff. So you must have

a temperature differential between room
temperature next to the accelerator

and 4 Kelvin inside the tank
where this cavity is sitting.

So you have a temperature difference
of 300 degrees, which this tank

around the cavity must keep. So that’s
a very hard job, actually cooling

is one of the more difficult things

from an engineering point of view.
The thing which feeds the fields

– the actual changing electrical
fields are polarity switched –

into these cavities are called klystrons.
There’s a picture of a klystron,

it’s the longish device sitting on the
bottom. And they’re usually about

as big as a refrigerator or two.
And these klystrons produce

radio waves not very much unlike that

which you hear in your car when you just
turn on the radio. It’s not modulated

in the same way, so there’s no
sound information encoded,

but it’s extremely strong.
You can see on the bottom

that one of these klystrons as it is in
use at the LHC has a transmitting power

of 300 Kilowatts. Now if you think of the
transmitting power of the Fernsehturm

like the Hertz-Turm which is right next
- no, that way -

which is right next to the conference
center, or even the Fernsehturm in Berlin.

It has about half the transmitting
power of one of these klystrons.

Now for the LHC accelerator
16 of them are used.

So that’s a lot of transmitting power.
And because the power is so high

we don’t actually use cables.
Usually you transfer your…

when you have some oscillator and
you’re checking out some signals,

you just put cables between
your source and your device.

This is not what’s used here, because
cables get way too complicated

when you have these high energies.

So what is used, is waveguides and that
is what you can see on the top there

in this picture. It looks like an
air duct, it looks like there’s some

sort of air conditioning system and the
air moves through. That’s not what it is.

It is a waveguide which is designed
to have the radio waves inside

radiate in a certain direction.
Think of a series of mirrors,

long rectangular mirrors and you put
them all with the mirroring area inside.

So you have a tube which is mirroring
inside. And then at one side

you shine in a bright light. Now the
light can’t escape anywhere and it

always hits the mirrors so it
goes on in a straight path.

You’ve built yourself a waveguide
for light. Now this here,

this clunky looking metal
part is a waveguide

but for high frequency, high energy radio
waves which are fed into the cavities.

And that’s how acceleration happens.
Now let’s talk about the curves.

This is where it gets less
fidgety and more… boom!

So these devices you see here, there’s
2 devices sitting next to each other,

identical devices. These
are the cryo-dipoles.

Again, they have the word “cryo” in
them because they are also cooled

by liquid helium down to
a temperature of about -270°C.

They’re 40 meters long, they weigh
35 tons and each of these babies

costs about half a million Swiss Francs.

And as you can see one line above that,
there’s 1200 of these curve dipoles

in the LHC. So there you have
a cost of 1.5 to 2 billion dollars

in the curve magnets alone.
We’re not talking acceleration,

we’re not talking about power use, we are
not talking about the helium that you need

for cooling or the power that you need for
cooling. It’s just building these things,

just building the curve, 27 kilometers.

And that’s what you have there as a
cost. Now what they do is, they make

a huge magnetic field, because in
a magnetic field a charged particle

will go on a curve. That’s
what we want, right? But

to make these particles with a very high
energy and keep them on a tight curve…

now in particle physics’ terms
let’s say that 27 kilometers

to go around one way is a tight curve.

We need a current of 12,000 amps.
Which is a large current

that goes through these dipoles.
Which is the reason why we have them

superconductingly cooled, because
otherwise you put 12,000 amps

through a piece of metal and it just melts
away. You don’t get a magnetic field,

maybe for a microsecond or 2.
But you want to sustain a stable field

of 8.5 Tesla to make these
protons go around on a curve.

So, yeah, that’s a big thing.
There’s also niobium in there,

not the big clunky parts like the cavity
we saw, but thin niobium wires,

actually half niobium, half titanium
most of the time. But since

there are so many magnets and it’s
so long a curve, there is 600 tons

of atomic niobium in this
entire accelerator thing.

And this was a fourth of the
world production of niobium

which comes mostly from Brazil by the way.
This was a fourth of the world production

of niobium for 5 years.
So that’s where it all went.

It just went into the accelerator.
And now if we have this running,

we have it up, we have it cooled, we have
a large current going, we got our nice

big magnetic fields. And
there is energy stored.

I mean we put in a lot of power and the
magnetic fields are up and they’re stable

and that means that there’s magnetic
energy stored in this. And the amount

of energy that is stored in the curve
magnets alone of the LHC when it’s running

is 11 gigajoules. Sounds like a lot,

let’s compare it to something: If we
have an absurdly long freight train

with let’s say 15,000 tons. I hear that
normal freight trains in Germany

or England have about 5000 tons.
So let’s take a big freight train

and multiply it by 3. If this
freight train goes at 150 km/h,

then the kinetic energy, the
movement energy of this train

is equivalent to the magnetic
energy that is stored in the LHC.

And that is why we don’t want
any problem with the cooling.

laughter

Because if we get a problem with
the cooling, bad things happen.

This is a photograph of what at CERN
at the LHC they just call “the incident”.

laughter

Which was a tiny mishap that
happened just a few weeks

after the LHC was taken into
operation for the first time in 2008.

And it shut the machine
down for about 8 months.

So that was a bad thing. It’s
a funny story when they where

constructing these magnets; now
what you see here is the connection

between 2 of these magnets. I told you
that each of them weighs 35 tons.

So here you have a connection between
2 parts that are 35 tons in weight each.

And they’re shifted by almost half
a meter. So it takes a bit of boom.

So what happened was: the cooling broke
down and the helium escaped and

the sheer force of the helium expanding,
because if you have liquid helium

and it instantly evaporates into gaseous
helium then the volume multiplies

by a very large amount.
And what they had was…

what I hear is that the tunnel of the
LHC, which has a diameter of about

let’s say 6 or 7 meters was
filled with nothing but helium

which pushed away the air
for about 100 meters

around this incident. So the helium
evaporated, it pushed everything away,

it made everything really cold, some
cables broke and some metal broke.

And the funny thing now is, the
engineers that built the LHC,

before they did that, visited
Hamburg. Because here there is

a particle accelerator which is
not quite as large. The LHC

has 27 kilometers; here in Hamburg we
have a particle accelerator called HERA

which had 6.5 kilometers. So it’s
the same ballpark, it’s not as big.

And in HERA they had a safety system
against these kinds of cryo failures,

they’re called quenches.
They had a protection system,

which protects this exact part.
Now we’re talking about “Yeah,

how should we build this? Should
we have a quench-protection

at the connection between the dipoles?”
And the HERA people in Hamburg said:

“Well we have it, it’s a good thing,
you shouldn’t leave it out,

if you build the LHC.” Well,
they left it out. laughter

They ran out of time, they ran out of
money, the LHC project was under pressure.

Because they had promised to build
a big machine by that time and

they weren’t really finished, so they
cut some edges. Well this was

the edge they cut and it cost them
8 months of operation. Which says

that they really should have listened to
the people of Hamburg. Okay, so,

in summary of the operations of
a storage ring we can just say this:

They get perfectly timed kicks
with our polarity switching

at just the right moment by radio waves
generated in these large klystrons

from the funny looking metal
tubes that we called cavities.

And some big-ass superconducting
magnets keep them on a curve

when they are not being accelerated.
Now the trick is, one of these kicks

like moving through the cavity once, may
not give you all the energy you want,

in fact it doesn’t. But if you
make them go round in the ring,

they come by every couple of
nanoseconds. So you just have them

run through your acceleration all the
time. Which is the big difference

between the storage ring and a linear
accelerator. A linear accelerator

is basically a one shot operation but
here, you just give them an energy kick

every time they come around, which
is often, we’re going to see that.

So that’s the summary of what
the storage rings do. Now,

the machine layout, if you
look at a research center

which has a bunch of accelerators,
it almost always goes like this:

You have some old, small storage
rings and then they built

newer ones which were
bigger. So this is just

a historical development, first
you build small machines, then

techniques get better, engineering gets
better, you build bigger machines. But

you can actually use that, it’s very
useful because the older machines,

you can use as pre-accelerators.
For a variety of reasons it’s useful

to not put in your particles with
an energy of zero and then

have them accelerated up to the energy you
want. You want to pre-accelerate them,

make them a little faster at a time.
That’s what you do, you just

take the old accelerators. And if
we look at the accelerator layout

of some real world research centers,
you can actually see this. On the left

you have CERN in Geneva and on the
right you have DESY here in Hamburg.

And you can see that there are smaller
accelerators, which are the older ones,

and you have bigger accelerators
which are connected to them.

And that’s this layout of the machines.
Okay, now let’s talk about collisions.

This is a nice picture of a collision.
It’s not actually a proton collision

but a heavy-ion collision, which
they do part of the time in the LHC.

They are extremely hard to produce, we’re
going to see that, but still we make

an awful lot of them.
So let’s see, first of all

let’s talk about what the beam looks like,
because we’re going to be colliding beams.

So what are these beams? Is it
a continuous stream of particles?

Well it’s not. Because the acceleration
that we use, these radio frequency,

polarity shifting mechanisms, they
make the particles into bunches.

So you don’t have a continuous stream,
you have separate bunches.

But how large are these bunches?
Is there one particle per bunch?

You’ve got a particle, you wait
a while, there’s another particle?

Well, it’s not like that.
Because if it were like that,

if we had single particles coming after
one another, it would be impossible

to hit them. You have to aim
the beams very precisely.

I mean, think about it. One comes
around 27 kilometers around the ring.

The other comes around 27
kilometers going the other way.

And now you want them to hit. You have
to align your magnets very precisely.

You can think of it like this:
You have a guy in Munich

and you have a guy in Hamburg and
they each have a rifle. And the bullets

of the rifle are let’s say one centimeter
in size. So the guy in Hamburg

shoots in the air and the guy in Munich
shoots in the air, and they are supposed

to make the bullets hit in the
middle, over, let’s say Frankfurt.

Which they’re not going to manage.
And which is actually way too simple.

Because if the bullet is really
one centimeter in size,

then the equivalent distance that the two
shooters should be away from each other,

if we want to make it the same
difficulty as these protons,

would not be between Hamburg and Munich.
It would be from here to fucking Mars.

laughter and applause
I calculated that shit.

applause

We don’t even have rifles on Mars
anyway. laughter

So what we got is, we got large
bunches, very large bunches.

And in fact there’s 10^11
protons per bunch, which is

100 Billion. This is where I called Sagan
“ you going Millions of Millions“

Okay, so you got 100 Billion
protons in one bunch.

And the bunches go by one after the other.
Now, if you stand next to the LHC

and you were capable of observing these
bunches, you would see one fly by

every 25 nanoseconds. So you go “there’s
a bunch, now it’s 25 nanoseconds,

there is the next one”. And there’s about
7.5 meters between the bunches.

Now, 7.5 meters corresponds to
25 nanoseconds, you see that

the speed is very big and indeed
it’s almost the speed of light.

Which is just, we accelerate them
and at some point they just go

with the speed of light and we just push
up the energy, we don’t make them

go any faster actually. And if you
were to identify the bunches,

which actually you can, you would
see that there are 2800 bunches

going by; and then when
you have number 2809,

that’s actually the first one that you
counted which has come round again.

Per direction! So in total
we have over 5000 bunches

of 100 Billion protons each. So
that’s the beam we are dealing with.

Oh, and a funny thing: you get charged
particles moving, it’s actually a current,

right? In a wire you have
a current running through it,

there’s electrons moving or holes moving
and you get a current. If you were

to measure the current of the
LHC, it would be 0.6 milliamps,

which is a small current, but
we’re doing collisions anyway

and not power transmission,
so that’s fine. laughter

This is a diagram of what the actual
interaction point geometry looks like.

You get the beams from different
directions, think of it like the top one

coming from the right, the bottom
one coming from the left;

and they are kicked into intersecting
paths by magnets. You have

very complicated, very precise
magnetic fields aligning them,

so that they intersect. And it’s
actually a bit of a trying-out game.

I’ve heard this from
accelerator operators.

You shift the position of the beams
relative to each other by small amounts

and you just see where the collisions
happen. You go like: “Ah yeah, okay,

there’s lots of collisions, ah, now
they’re gone, I’m going back.”

And you do it like that. You can save the
settings and load them and calculate them

but it’s actually easier
to just try it out.

If we think of how much stuff we’ve
got going on: you got a packet,

a bunch of 100 Billion
protons coming one way,

you got another packet of 100 Billion
protons coming the other way.

Now the interaction point area is as small
as the cross section of a human hair.

You can see that, it’s one hundredth
of a square millimeter.

Now how many collisions do
you think we have? We’ve got…

Audience: Three!
Michael laughs

Michael: …it’s actually not that bad.
We got about 20 in the LHC.

And the funny thing is, people
consider this a bit too much.

The effect is called pile-up. And the
bad thing about pile-up is you’ve got

beams intersecting, you’ve got bunches
‘crossing’ – that’s what we call it.

And there’s not just one collision which
you can analyze, there is a bunch of them,

around 20. And that makes that
more difficult for the experiments,

we’re going to see why. Well, and if we
have 20 collisions every bunch crossing

and the bunches come by every
25 nanoseconds, that gives us a total

of 600 Million collisions per
second. Per interaction point.

Which we don’t have just one of. We
have 4 experiments, each experiment

has its own interaction point. So
in total, we have about 2 Billion

proton-proton collisions happening
every second when the LHC is running.

Now let’s look at experiments.
laughs

Yeah, this is a photograph of one part of
the ATLAS experiment being transported.

And as for the scale of this thing, well,
in the physics community, we call this

a huge device.
laughter

I have a diagram of the experiment
where this is built in and

you’re going to recognize the part
which is the one I’ve circled there.

So the real thing is even bigger.
And down at the very bottom,

just to the center of the
experiment, there’s people.

Which if I check it like this,
they’re about 15 pixels high.

So that’s the scale of the experiment.

The experiment has the interaction point
at the center, so you got a beam line

coming in from the left, you got the other
beam line coming in from the right.

And in the very core of the experiment
is where the interactions,

where the collisions happen. And then
you got the experiment in layers,

like an onion, going around
them in a symmetrical way.

Inside you have a huge magnetic
field which is almost as big

as the curve magnets we were talking about
when I was describing the storage ring.

This is about 4 Teslas,
so it’s also a very big field.

But now we got a 4 Tesla field
not just over the beam pipe

which is about 5 centimeters in diameter,
but through the entire experiment;

and this thing is like 20-25 meters.
So you’ve got a 4 Tesla field

which should span more than 20 meters.

And, just for shits and giggles,
it’s got 3000 kilometers of cables.

Which is a lot; and if you just
pull some random plug

and don’t tell anyone which one it
was you’re making a lot of enemies.

So the innermost thing is what we
call the inner tracking. It is located

just centimeters off the beam line,
it’s supposed to be very very close to

where the actual interactions happen.

And this thing is made to leave the
particles undisturbed, they should just

fly trough this inner tracking detector.
And the detector will tell us

where they were, but not actually
stop them or deflect them.

This gives us precise location data,
as to how many particles there were,

what way they were flying,
and, from the curve,

what momentum they have. Outside
of that we’ve got calorimeters.

Now these are supposed to be stopping
the particles. A particle goes through

the inner tracking without being disturbed
but in the calorimeter it should stop.

And it should deposit all its energy there
and which is why we have to put around it

the inner tracking. You see, if we put the
calorimeter inside, it stops the particle,

outside of that nothing happens. So we
have the calorimeters outside of that.

And then we got these funny wing things
going on. That’s the muon detectors.

They are there for one
special sort of particle.

Out of the… 50, let’s say 60
– depends on the way you count –

elementary particles that we
have. These large parts are

just for the muons. Because the
muons have the property,

the tendency to go through all sorts of
matter undisturbed. So you just need to

throw a huge amount of matter
in the way of these muons, like:

“let’s have a brick wall and then
another one”. And then you

may be able to stop the muons,
or just measure them.

This is to give you an idea of the
complexity of the instrument

on the inside. This is the inner tracking
detector, it’s called a pixel detector;

and you see guys walking around in
protective suits. That is not for fun

or just for the photo, this is a very,
very precise instrument. But it’s sitting

inside this huge experiment which – again,

I calculated that shit – is about
as large as a space shuttle

and weighs as much as the
Eiffel Tower. And inside

they’ve got electronics, almost a ton
of electronics which is so precise

that it makes your smartphone
look like a rock. So there you go,

it’s a very, very complicated sort of
experiment. Let’s talk about triggering,

because as I said there’s 600 Million
events happening inside this.

That’s 40 Million bunch crossings.
Now: how are we going to analyze this?

Is there a guy writing everything
down? Obviously not.

So this experiment with all the tracking
and the calorimeters and the muons

and everything has about
100 Million electronic channels.

And one channel could be the measurement
of a voltage, or a temperature

or a magnetic field or whatever. So
we’ve got 100 Million different values,

so to speak. And that makes
about 1.5 Megabytes per crossing,

per every event readout. Which
gives us – multiplied by 40 Million –

gives us about 60 terabytes
of raw data per second.

That’s bad. I looked it up, I guess

the best RAM you can do is about
1 terabyte per second or something.

So we’re obviously not going to tackle
this by just putting in fast hardware,

because it’s not going
to be fast enough. Plus,

the reconstruction of an event is done
by about 5 Million lines of C++ code.

Programmed by some 2000-3000
developers around the world.

It simulates for one crossing
30 Million objects, which is

the protons and other stuff flying around.

And it is allocated to take 15 seconds
of one core’s computing time.

To calculate it all, you would
need about 600 million cores.

That’s not happening. I mean,
even if we took over the NSA

laughter
and used all of their data-centers

for LHC calculations, it still wouldn’t be
enough. So we have to do something

about this huge mass of data. And
what we do is, we put in triggers.

The trigger is supposed to reduce the
number of events that we look at.

The first level trigger looks at
every collision that happens.

And it’s got 25 nanoseconds
of time to decide:

Is this an interesting collision?
Is it not an interesting collision?

We tell it to eliminate
99.7% of all collisions.

So only every 400th collision
is allowed for this trigger to go:

“Oh, yeah, okay that looks interesting,
let’s give it to Level 2 trigger”.

So then we end up with about 100,000
events per second. Which get us

down to 150 Gigabytes per second. Now
we could handle this from the data flow,

but still we can’t simulate it. So
we’ve got another level trigger.

This is where the two
experiments at the LHC differ:

the CMS experiment has just a
Level 2 trigger; does it all there.

The ATLAS experiment goes the more
traditional way, it has a Level 2 trigger

and a Level 3 trigger. In the end these
combined have about 10 microseconds

of time, which is a bit more and it gives
them a chance to look at the events

more closely. Not just, let’s say:
“Was it a collision of 2 protons

or of 3 protons?”; “Were there
5 muons coming out of it

or 3 electrons and 2 muons?” This is
the sort of thing they’re looking at.

And certain combinations the triggers
will find interesting or not.

Let’s say 5 muons, I don’t give a shit
about that. “3 muons and 2 electrons?

Allright, I want to analyze it”. So
that’s what the trigger does.

Now this Level 2 and 3 trigger,
again, have to kick out about

99.9% of the events. They’re
supposed to leave us with

about 150 events per second. Which
gives a data volume of a measly

300 Megabytes per second and that’s
something we can handle. We push it

to computers all around the world.
And then we get the simulations going.

This is a display, this is
what you see in the media.

If you take one of these events – just
one of the interesting events which

actually reach the computers – because
those 40 million bunch crossings… well,

most of them don’t reach the computers,
they get kicked out by the triggers.

But out of the remaining 100 or 200
events per second, let’s say this is one.

It’s an actual event and it’s been
calculated into a nice picture here.

Now, normally they don’t do that, it’s
analyzed automatically by code

and it’s analyzed by the physics data.
And they only make these pretty pictures

if they want to show something to
the press. To the left you have

what’s called a Feynman Diagraph.
That’s just a fancy physical way

of saying what’s happening there. And
it involves the letter H on the left side,

which means there’s a Higgs involved.
Which is why this event was particularly

interesting to the people
analyzing the data at the LHC.

And you see a bunch of tracks, you see
the yellow tracks all curled up inside,

that’s a bunch of protons hitting
each other. The interesting thing is

what happens for example above
there with the blue brick kind of things.

There’s a red line going through
these bricks. This indicates a muon.

A muon which was created in
this event there in the center.

And it went out and the
bricks symbolize the way

the reaction was seen by the experiment.

There was actually just a bunch of bricks
lighting up. You got, I don’t know,

500 bricks around it and brick 237
says: “Whoop, there was a signal”.

And they go: “Allright, may have been
a muon moving through the detector”.

When you put it all together you
get an event display like this. Okay,

so we got to have computers analyzing
this. And with all the 4 experiments

running at the LHC, which is not just
CMS and ATLAS I mentioned but also

LHCb and ALICE, they produce about
25 Petabytes of data per year.

And this cannot be stored at CERN alone.
It is transferred to data centers

around the world by what is called
the LHC Optical Private Network.

They’ve got a network of fibers going from
CERN to other data-centers in the world.

And it consists of 11 dedicated
10-Gigabit-per-second lines

going from CERN outwards. If we
combine this, it gives us a little over

100 Gigabits of data
throughput, which is about

the bandwidth that this congress has.

Which is nice, but here it’s dedicated
to science data and not just porn

and cat pictures.
laughter and applause

applause

From there it’s distributed outwards
from these 11 locations to about

170 data centers in all the
world. And the nice thing is,

this data, these 25 Petabytes
per year, is available

to all the scientists working
with it. There’s about… well,

everybody can look at it, but there’s
about 3000 people in the world

knowing what it means. So all these
people have free access to the data,

you and I would have free access to the
data, just thinking it’s cool to have

a bit of LHC data on your harddrive maybe.
laughter

All in all, we have 250,000
cores dedicated to this task,

which is formidable. And about
100 Petabytes of storage

which is actually funny, because
25 Petabytes of data are accumulated

per year and the LHC has been
running for about 4 years.

So you can see that they buy the
storage as the machine runs. Because

100 Petabytes, okay, that’s what we have
so far. If we want to keep it running,

we need to buy more disks. Right! Now,

what does the philosoraptor
say about the triggers?

If the triggers are supposed to eliminate
those events which are irrelevant,

which is not interesting, well,
who tells them what’s irrelevant?

Or to put it in the terms
of Conspiracy-Keanu:

“What if the triggers throw away the
wrong 99.something % of events?”

I mean, if I say: “If there’s an event
with 5 muons going to the left,

kick it out!”. What if that’s actually
something that’s very, very interesting?

How should we tell? We need to
think about this very precisely.

And I’m going to tell you about
an example in history where

this went terribly wrong, at least for
a few years. We’re talking about

the discovery of the positron.
A positron is a piece of anti-matter;

it is the anti-electron. It was
theorized in 1928, when

theoretical physicist Dirac put up a bunch
of equations. And he said: “Right,

there should be something which is like
an electron, but has a positive charge.

Some kind of anti-matter.” Well,
that’s not what he said, but that’s

what he thought. But it was
only identified in 1931.

They had particle experiments back then,
they were seeing tracks of particles

all the time. But they couldn’t
identify the positron for 3 years,

even though it was there on paper.
So what happened? Well,

you see the picture on the left. This
is the actual, let’s say baby picture

of the positron. I’m going to
build up a scheme on the right

to show you a bit more, to
give you a better overview of

what we are actually talking about.
In the middle you’ve got a metal plate.

And then there’s a track which is bending
to the left, which is indicated here

by the blue line. Now if we analyze
this from a physical point of view,

it tells us that the particle
comes from below,

hits something in the metal plate
and then continues on to the top.

So the direction of movement
is from the bottom to the top.

The amount by which its curvature
reduces when it hits the metal plate

tells us it has about the mass of
an electron. Okay, so far so good.

But then it has a positive charge.
Because we know the…

we know the orientation of the magnetic
field. And that tells us: “Well,

if it bends to the left, it
must be a positive particle.”

So we have a particle with the mass of
an electron, but with a positive charge.

And people were like “Wat?”.
laughter

So then someone ingenious came
up and thought of a solution:

‘They developed the picture
the wrong way around!?’

laughter and applause

applause

It’s what they thought. Well it’s wrong,
of course, there’s such a thing as

a positron. And it’s like an electron,
but it’s positively charged. But…

to put it in a kind of summary maybe:
you can only discover that

which you can accept as a result.
This sounds like I’m Mahatma Gandhi

or something but it’s just what we call
science. laughter

Okay, so to recap: What have we
seen, what have we talked about?

We saw from the basic principle,
that if we have energy in a place,

then that can give rise to other forms of
matter, which I called ‘parts = a device’.

You got your little parts, you do
some stuff, out comes a device.

We have storage rings which give
a lot of energy to the particles

and in which they move around in huge
bunches. Billions of billions of protons

in a bunch and then colliding. Which
gives in the huge experiments

that we set up an enormous amount of data
ranging in the Terabytes per second

which we have to program triggers
to eliminate a lot of the events

and give us a small amount of data which
we can actually work with. And then

we have to pay attention to the
interpretation of data, so that

we don’t get a fuck-up like with the
positron. Which is a very hard job.

And I hope that I could give you
a little overview of how it’s fun.

And it’s not just about building
a big machine and saying:

“I’ve got the largest accelerator of
them all”. It’s a collaborative effort,

it’s literally thousands of people working
together and it’s not just about

two guys getting a Nobel Prize. You
see this picture on the top left, that’s

about 1000 people at CERN watching
the ceremony of the Nobel Prize

being awarded. Because everybody felt
there’s two people getting a medal

in Sweden, but it’s actually an
accomplishment… it’s actually an award for

everybody involved in this enormous thing.
And that’s what’s a lot of fun about it

and I hope I could share some of this
fascination with you. Thank you a lot.

huge applause

Before we get to Q&A, I’m going to be
answering questions that you may have.

My name is Michael, I’m @emtiu on
Twitter, I’ve got a DECT phone,

I talk about science, that’s
what I do. I hope I do it well.

And you can see the slides and
leave feedback for me please

in the event tracking system. And
tomorrow, if you have the time

you should go watch the “Desperately
seeking SUSY” talk which is going to be

talking about the theoretical side of
particle physics. Okay, that’s it from me,

now on to you.
Herald: Okay, if you have questions,

please line up, there’s a mic there and
a mic there. And if you’re on the stream,

you can also use IRC and
Twitter to ask questions. So

I’m going to start here,
please go ahead.

Question: Thanks a lot, it was a very
fascinating talk, and nice to listen to.

My question is: Did HERA
ever suffer a quench event

in which the quench protection
system saved the infrastructure?

Michael: No, actually it didn’t. There
were tests where they provoked

a sort of quench event in order to
see if the protection worked. But

even if this test would have failed it
would not have been as catastrophic.

But there were failures in the
operation of the HERA accelerator

and there was one cryo failure. Which
is actually a funny story. Which is

where one part of the
helium tubing failed

and some helium escaped
from the tubing part

and went into the tunnel. Now what
happened was that the air moisture,

just the water in the
air froze at this point.

And the Technical Director of the HERA
machine told us this: at one point

he sat there with a screwdriver and
a colleague, picking off… the ice

off the machine for half the night before
they could replace this broken part.

So, yeah, cryo failures
are always a big pain.

Herald: Do we have questions
from the internet? …Okay.

Signal Angel: We have
one question that is:

“How are the particles
inserted into the accelerator?”

Michael: They mostly start
in linear accelerators.

Wait, we’ve got it here. So you
got the series of storage rings

there at the top in the middle and
you have one small line there.

That’s a linear accelerator. To get
protons is actually very easy.

You buy a bottle of hydrogen which
is just a simple gas you can buy.

And then you strip off the electrons.
You do this by ways of exposing them

to an electric field. And what you’re left
with is the core of the hydrogen atom.

And that’s a proton. Then you
accelerate the proton just a little bit

into the linear accelerator and from there
on it goes into the ring. So that means

basically at the start of these colliding
experiments is just a bottle of helium

that somebody puts in there. And
at the LHC it’s about, you know,

a gas bottle. It’s about this big and it
weighs a lot. At the LHC they use up

about 2 or 3 bottles a year for
all the operations, because

a bottle of hydrogen
has a lot of protons in it.

Herald: You please, over there.

Question: Actually I have
2 questions: One part is,

you said there are 2 beams
moving in opposite directions.

And you explained the way where you
switched polarity. How can this work

with 2 beams opposing each other?

Michael: That’s a good question. Now, if
I show you the picture of the cryo dipole,

you will see that these 2 beams
are not actually in the same tube.

There we go. You see a cryo dipole and

on the inside of this blue tube, you
see that there’s actually 2 lines.

You can’t see it very well but
there’s 2 lines. So they are

inside the same blue tube, but then
inside that is another small tube,

which has a diameter of just about
a Red Bull bottle. Say 5 or 6 centimeters

in diameter. And this is where the beam
happens. And they are just sitting

next to each other. So the beams
are always kept separate

except from the interaction points
where they should intersect.

And the acceleration happens
obviously also in separate cavities.

Herald: You had a second question?

Question: The second question is: The
experiments, where are they placed,

on the curve or on the acceleration part?

Michael: The interaction points are
placed between the acceleration

on the straight path. Because, again,
it’s much easier if you had the protons

going straight for 200m; then you
can more easily aim the beam.

If they come around the curve then they
have – you know they have a curve motion,

you need to cancel that. That
would be much more difficult.

Herald: And the left, please.

Question: Okay, so you got yourself
a nice storage ring and then

you connect it to the power plug
and then your whole country

goes dark. Where does the power come from?

Michael: Well, in terms of power
consumption of, let’s say

households, cities, or aluminum plants:

accelerators actually don’t
use that much power. I mean

most of us don’t run an aluminum
plant. So we’re not used to this

sort of power consumption. But’s it’s not
actually all that big. I can tell you about

the HERA accelerator that we had here
in Hamburg, which I told you is about

6.5 kilometers, not the 27, so you
can sort of extrapolate from that.

It used with the cryo and the
power current for the fields

and everything – it used about
30 MW. And 30 Megawatts is a lot,

but it’s not actually very much in
comparison to let’s say aluminum plants,

our large factories. But in fact,
the electricity cost is a big factor.

Now you see the LHC is located at the
border between Switzerland and France.

It gets most of its power from France.

And you always have an annual shutdown of
the machine. You always have it off about

1 or 2 months of the year. Where you do
maintenance, where you replace stuff,

you check stuff. And they always
take care to have this shutdown

for maintenance in winter. Because
they get their power from France.

And in France many people use
[electrical] power for heating.

There’s not Gas heating or Long
Distance heat conducting pipes

like we have in Germany e.g. The people
just use [electrical] power for heat.

And that means in winter the electricity
price goes up. By a large amount. So

they make sure that the machine is off in
winter when the electricity prices are up.

And it’s running in the summer where
it’s not quite as bad. So it’s a factor

if you run an accelerator. And you
should tell your local power company

if you’re about to switch it on!
laughter

But actually, it won’t make the grid off,
even a small country like Switzerland

break down or anything.

Herald: Do we have more questions from
the internet? Internet internet, no,

no internet. Okay. Then just
go ahead, Firefox Girl.

Question (male voice): So you see a lot
of events. And I guess there’s many

wrong ones, too. How do you select if
an event you see is really significant?

Michael: Well, you have different kinds
of analysis. Like I told you there is

100 Mio. channels you can pick from.

With the simplest trigger that
you have, the Level 1 trigger,

it can’t look at the data in much
detail. Because it only has 25 ns.

But as you go higher up the chain,
as the events get more rare,

you can look at them more closely. And
what we end up in the end, these 100,

maybe 200 events per second, you can
analyze them very closely. And they get…

they get a full-out computation. You
can even make these pretty pictures

of some of them. And then it’s basically,
well, theoretical physicists’ work,

to look at them and say: “Well, this
might have been that process…”, but

still a lot of them get kicked out. When
the discovery of the Higgs particle

was announced, it was ca. 1 1/2 years ago…

Well, the machine had been running
for 2 1/2 years. And, like I told you,

there’s about 2 Billion proton collisions
per second. Now the number of events

that were relevant to the discovery
of the Higgs – the Higgs events –

it was not even 100.
Out of 2 Billion per second.

For 2 1/2 years. So you have to sort out
a lot. Because it’s very very, very rare.

And that’s just the work of
everybody analyzing, which is why

it’s a difficult task,
done by a lot of people.

Herald: The right, please.

Question: What I’m interested in: You
say ‘one year of detector running’.

How much time in this year does
this detector actually run…

…is it actually running?

Michael: Well, yeah, like I said, we
have the accelerator off for about

1 or 2 months. Then if something
goes wrong it will be off again.

But you want to keep it running
for as long as possible, which…

in the real world… let’s say it’s
9 months a year. That’s about it.

Question: Straight through?

Michael: Straight through – ah, well,
not in a row. But it’s always on

at least for a week. And then you
get maybe a small interruption

for a day or two, but you can also have
a month of straight operation sometimes.

Herald: Internet, please!

Signal Angel: Yeah, another question:
what would happen if they actually find

what you are looking for?
Michael laughs

Do we throw the LHC in the
dumpster or what do we do?

Michael: That’s a good question!
It would be one hell-of-a waste

of a nice-looking tunnel! laughs
You might consider using it for

– I don’t know – maybe swimming
events, or bicycle racing.

Well, but actually that’s a very good
question because the tunnel

which the LHC sits in, this 27 km
tunnel, it was not actually dug,

it was not actually made just for the LHC.
There was another particle accelerator

inside before that. It had less energy,
because it didn’t accelerate protons

but just electrons and positrons.
That’s why the energy was a lot lower.

But they said: “Well, okay, we’re going
to build a very large accelerator,

does anyone have a
30 km tunnel, maybe?”

and then someone came up with:
“Yeah, well, we got this 27 km tunnel

where this LEP accelerator is sitting in.
And when it’s done with its operations

in…” – I don’t know, by that time,
let’s say in – “…10 years, we’re going

to shut it off. Why don’t we put the next
large accelerator in there?” So you try

to reuse infrastructure, but of course
you can’t always do that. The next big,

the next huge accelerator, if we get the
money together as a science community,

because the politicians are
being a bitch about it…

if we get the money it’s going to be
the International Linear Collider.

And that’s supposed to have
100 km of particle tubes

and, well, you need to build
a new tunnel for that, obviously.

Question: First off, couldn’t
you use it in something

like material sciences, like
example with DESY?

Well okay, if you are done with
leptons you can still use it

for Synchrotron Laser
or something like this.

Michael: That was thought of. The HERA
accelerator at DESY was shut off

and people were thinking about if they
could put a Synchrotron machine inside it.

But the problem there is the HERA
accelerator is 25 m below the ground.

This is not enough space.
With particles accelerating

you just need a small tube. But for
Synchrotron experiments you need

a lot of space. So you would have
to enlarge the tunnel by a lot,

and this was not worth it, in the case of
the HERA accelerator. But interestingly,

one of the pre-accelerators of HERA,
one that was older is now used

for Synchrotron science, which is
PETRA. Which used to be just an

old pre-accelerator, and now it’s one of
the world’s leading Synchrotron machines.

So, yeah, you try to reuse things
because they were expensive.

Question: And may I just
ask another question?

You said you get… you use just the matter

from a bottle of hydrogen
or a bottle of helium.

Well, most helium or hydrogen is protons

or, in the case of helium, helium-4. But

you have a little bit helium-3 or deuterium.

And well, you are looking for
interesting things you don’t expect.

So how do you differentiate if it’s really

something interesting or: “Oh, one of
these damn deuterium nuclides, again!”

Michael: You don’t get wrong isotopes
because you just use a mass spectrometer

to sort them out. You have a magnetic
field. You know how large it is. And

the protons will go and land – let’s say
– 2 micrometers next to the deuterons,

and they just sort them out.

Question: I have 2 questions. One is:

I guess you mentioned that
basically once the experiment

runs at speed of light you
just put more energy into it.

But what is actually the meaning
of the energy that you put into it?

What does it change in the experiment?
Like the Higgs was found

at a particular electron volt…

Michael: Yeah, it was
found at 128 GeV. Well,

it’s more of a philosophical question.
There is a way of interpreting

the equations of special relativity where
you say that, when you don’t increase

the velocity you increase the mass.
But that’s just a way of looking at it.

It’s more precise and it’s more
simple to say: you raise the energy.

And at some low energies that
means that you raise the velocity.

And at some high energies it means
the velocity doesn’t change anymore.

But overall you add more energy.
It’s one of the weird effects

of special relativity and there
is no very nice explanation.

Question: Let’s assume there is
an asteroid pointing to earth.

Michael laughs
Could you in theory point this thing

on the asteroid and destroy it,
or would it be too weak?

laughter

applause

Michael: I’m going to help you out.
Because it wouldn’t actually work

because between the accelerator and the
asteroid there’s the earth atmosphere.

And that would stop all the particles.
But even if there were no atmosphere:

no, it would be much too weak. Well,

you’d have to keep it up for a long time
at least. There was this one accident

at the HERA accelerator where the
beam actually went off its ideal path

and it went some 2 or 3 cm
next to where it should be.

And it hit a block of lead – just,
you know, the heavy metal lead –

and the beam shot into this
lead thing and the entire beam,

which was a couple of Billions of
protons, was deposited into this lead

and some kilograms of lead
evaporated within microseconds

and there was a hole like pushed by
a pencil through these lead blocks.

So, yeah, it does break stuff apart. But
even if you managed to hit the asteroid

you would make a very small hole.
But you wouldn’t destroy it.

It would be a nice-looking asteroid then.
laughter

Question: Before you turned on the LHC
the popular media was very worried

that you guys were going
to create any black holes.

Did you actually see any black holes
passing by? Michael laughs

Michael: Well, there may have been
some, but they were small, and

they were insignificant. The interesting
thing is… sorry, I’m going to recap, yeah.

The interesting thing is that whatever
we can do with the LHC – where

we make particles have large energies
and then collide – is already happening!

Because out in space there is black
holes with enormous magnetic fields

and electrical fields. And these
black holes are able to accelerate

electrons to energies much, much
higher than anything we can produce

in any accelerator. The LHC
looks like a children’s toy

in comparison to the energies that
a black hole acceleration can reach. And

the particles which are accelerated in
these black holes hit earth all the time.

Not a lot, let’s say one of these
super-energetic particles they come around

about once a year for every
square kilometer of earth.

But still, they’ve been hitting
us for Millions of years.

And if a high-energy particle
collision of this sort were able

to produce a black hole that swallows
up the earth it would be gone by now.

So: won’t happen.
applause

Question: Maybe more interesting
for this crowd: you talked about

the selection process of the events.

So I guess these parameters
are also tweaked to kind of

narrow down like what
a proper selection procedure.

Is there any kind of machine
learning done on this to optimize?

Michael: Not that I know of. But there is
a process which is called ‘Minimum Bias

Data Collection’. Where you
actually bypass all the triggers

and you select a very small portion
of events without any bias.

You just tell the trigger: “Take
every 100 Billionth event”

and you just pass it through no matter
what you think. Even if you think

it’s not interesting, pass it through.
This goes into a pool of Minimum Bias Data

and these are analyzed especially in order
to see the actual trigger criteria

are working well. So yeah,
there is some tweaking. And

even for old machines
we have data collected

and sometimes we didn’t know what we
were looking for. And some 20 years later

some guy comes up and says: “Well,
we had this one accelerator way back.

There may have been this and that
reaction. Which we just theorize about.

So let’s look at the old data and see
if we see anything of that in there

now, because it’s limited because
it goes through all the filters”.

You can’t do this all the time with
great success. But sometimes,

in very old data you find new
discoveries. Because back then

people weren’t thinking about looking
for what we are looking now.

Question: I always asked myself about
repeatability of those experiments.

Seeing as the LHC is the biggest one
around there, so there’s no one out there

who can actually repeat the
experiment. So how do we know

that they actually exist, those particles?

Michael: That’s a very good question.
I told you that there is 2 main

large experiments. Which is the CMS
experiment and the ATLAS experiment.

Now these both sit at the same ring.
They have some 10 km between them

because they’re on opposite ends
of the ring. But still, obviously,

they’re on the same machine. But these 2
groups, the ATLAS and the CMS experiment,

operate completely separately. It’s not
the same people, not the same hardware,

not the same triggers,
not even the same designs.

They build everything up from scratch,
separate from each other. And

it’s actually funny because when you
look at a conference and here is CMS

presenting their results and here is
ATLAS presenting their results,

they pretend like the other
experiment is not even there.

And that’s the point of it: they’re
not angry at each other. It must be

2 separate experiments because obviously
you can’t build a second accelerator.

So you try to have redundancy in order

for one experiment to confirm
what the other finds.

Herald: Okay. It’s midnight
and we’re out of time.

So please thank our awesome speaker!
applause

Subtitles created by c3subtitles.de
in the year 2016. Join and help us!