Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.mso.anu.edu.au/pfrancis/roleplay/html_dir/firstlight.ps Äàòà èçìåíåíèÿ: Sun Aug 30 05:04:02 1998 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 10:23:34 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: supernova

Harley Wood Winter School 1998: Roseworthy
College: First Light in the Universe Exercise
Paul J. Francis
August 30, 1998
1 Introduction
This exercise was run at a winter school of Australian astrophysics grad stu­
dents, and was designed to acquaint them with current progress in the study
of high redshift galaxies. The class were divided into 11 groups. Their briefing
papers are listed below.
One group (representing the IAU) had the unenviable task of organising
and chairing the meeting: their briefing paper is different in format, and is
listed first below.
Please don't take offence at the stereotypes, please: they are simply de­
signed to make the exercise more memorable!
2 You are the International Astronomical Union
2.1 Research Interests
You have the unenviable job of chairing this meeting of the IAU `New Facilities'
committee. You have to get this fractious bunch of ego­maniacs to come to a
consensus on which new facilities should be funded, to advance our knowledge
of galaxy formation and evolution. You are a group of astronomers from various
small institutions in Europe, Asia and South America.
Everybody here will have a pet instrument or telescope they want to pro­
mote. They will also have some particular science questions that appeal to
them, and some special expertise in particular areas of high redshift research.
Your job is to get them to focus on the science: get them to think not
purely about getting their own instrument or facility funded, but about what
we, the human race, want to know about the early universe, and how we can
best use our limited resources to learn these things.
You have to get this group to come up with a ranked shortlist of the five
best proposals. If you just take a vote at first, everyone will probably vote for
1

their own ideas: every proposal will get one vote, and the whole meeting will
collapse in acrimonious failure. You will have to think of how best to manage
the process of getting a consensus. You could let people vote at the end, but
not for their own project. You might try and force the different groups to
join in consortia to propose particular shortlists. You might try a preferential
voting system, or simply try and steer the discussion with assertive chairing.
Perhaps you should assemble sub­groups made up of representatives from the
different institutions to discuss separate issues (eg. infra­red vs. optical, or
ground vs. space).
In the final session, you may with to get representatives from each group to
stand up and make brief presentations, before calling for some sort of decision.
This might be dangerous, however, as it could lead to a hardening of negoti­
ating positions, and make finding a compromise acceptable to all impossible.
You should talk to all the various groups repeatedly as they work on their
proposals, and keep tabs on any consensus that might emerge. Hopefully you
can spot fights before they start and pour oil on the troubled waters.
2.2 Proposed Agenda
Bear in mind that, once this meeting has decided a priority list, it then has
to go the world governments for funding. There is no point in coming up
with a scientifically wonderful proposal that you cannot sell to the politicians.
Something exciting, thrilling, and easy to explain to numb­skull politicos is
what you need.
Secondly, you have to ensure that virtually everyone is happy with the
compromise ranking you come up with. All it takes in one irate astronomer
to go to their congressional representative and tell them that the proposal
is a big mistake, and you can say goodbye to your funding for any of these
projects. Politicians (unreasonably) require that the astronomical community
be completely uniform in its proferred views, before they will give us any
money.
You have two personal biasses, and if possible, you should steer the meeting
towards a conclusion that reflects these biasses. Firstly, history shows over and
over again that the biggest results from any new telescope or facility are not
the ones that were expected when it was built. Most of these facilities will not
see first light for years: by then, the whole field of astronomy will have moved
on. You have to make sure that whatever is funded has enough flexibility to
cope with this.
Secondly, progress in astronomy depends on having a wide range of intelli­
gent and imaginative astronomers. You have to make sure that these facilities
can be used by young researchers, researchers at minor institutions, and from
third world countries. And you have to make sure that people with unfashion­
able, maverick ideas, can get time to try them out. If, for example, you only
2

had one, highly over­subscribed mega­facility, most of the time will tend to
go to people with political clout, doing projects which conform to the current
orthodoxy, and which have a very high probability of success. Original science,
and new researchers might be squeezed out.
Good luck!
3 You are the Anglo­Australian Observatory
3.1 Research Interests
Your main research interests concern redshift surveys.
Why are people so fixated with redshifts above 3? Over half the history
of the universe has occurred since redshift one. If we observe out to redshift
1.5, that is around 70% of the whole history of the universe! And we don't
need fancy new­fangled techniques to work at these redshifts: tried and tested
techniques will work just fine, and the galaxies we find and study are close
enough that we can really learn something about them.
You are world authorities at redshift surveys. You start off by getting
deep images of some part of the sky. Then, using the multi­fibre or multi­slit
instruments you have pioneered, you obtain spectra of them all. This sounds
simple, but the scientific pay­off has been huge, especially when combined with
Hubble Space Telescope imaging of the galaxies, to determine their colours and
morphologies.
Little evolution in the numbers of elliptical galaxies is seen out to redshift
one. The major evolution is in a vast population of small blue galaxies, the
so­called `faint blue galaxies', which were abundant at redshift one, but which
have mostly disappeared by now. They have probably faded into obscurity, as
their initial burst of star formation dies away.
The nett effect of all of this is that the star formation rate per unit co­
moving volume was much higher at redshift one than it is today, which actually
agrees well with theoretical calculations.
Galaxy clusters have also changed: at redshifts around one, clusters seem to
contain more spiral galaxies, fewer S0s and the same number of giant elliptical
galaxies as we see today. This is presumably because the spirals are merging
to form S0s. One puzzle: the velocity dispersions, weak gravitational lensing
measurements and X­ray fluxes of these high redshift clusters imply that they
are just as massive as clusters today, which is in strong contradiction to most
current theoretical models. If this holds up with better data, it implies that
something may be wrong with theories of structure formation in the universe.
You are now working on enlarging your samples, and actually measuring
velocity dispersions for galaxies at redshifts around 1, to derive their masses.
This will really help nail down the evolution of different types of galaxy.
3

3.2 Proposed Facility
All of this is good, solid science, unlike the flaky stuff the high redshift people
are doing. These techniques use well­understood techniques, have good statis­
tics and allow us to trace the different galaxy populations in detail as they
evolve. It might not be as trendy and `Gee Wizz' as going after fuzzy blobs at
redshift 5, but it is slowly, securely extending our knowledge of the universe.
The main challenge is to extend this work into the infrared. All the main
emission­lines of typical galaxies disappear into the IR beyond redshift one,
which puts an upper limit on what this technique can do. What is needed,
therefore, is an infra­red multi­object spectrograph on an 8­m class telescope
(or preferably several of them). With this, you can push all the good solid
work you've been doing at lower redshifts out to redshift 2 or 3 and beyond,
and come to really understand, in detail, the evolution of different classes of
galaxies throughout the history of the universe.
You see your role in this meeting as bring all the crackpot high redshift
people down to Earth (figuratively speaking). All their proposals may sound
exciting, but the aim is not to produce excited astronomers but to learn about
the history of the universe: anything that won't produce good solid inter­
pretable data should be thrown out.
4 You are the Australia Telescope National Facility
(ATNF)
4.1 Research Interests
Your main research interests concern neutral hydrogen.
In the early universe, most galaxies have presumably not yet formed. That
means that there must be enormous quantities of gas lying around, waiting
for the chance to collapse under its own weight into lumps and form stars
and galaxies. If we could actually observe this gas, as it moves around and
assembles itself into lumps, we would be directly observing the process of
galaxy formation. If we look at the amount of gas as a function of redshift, it
should drop as galaxies form, telling us the era of galaxy formation.
Optical astronomers seem to think that they can observe galaxy formation
by taking optical images. What emits visible light? Stars. And where do stars
live? In galaxies. So you can only see visible light once a galaxy has formed.
So how do they propose to study galaxy formation using optical telescopes?
This has always puzzled you. Furthermore, even once a galaxy has formed
its first stars, it will probably be dusty for a while, just like local starburst
galaxies, and dust is remarkably good at blocking optical radiation.
In the local universe, gas can be studied in enormous detail using 21cm
radiation. This has been responsible for a large fraction of our knowledge of
4

galaxies in the local universe. Alas, with current technologies it is impossible
to use this technique at even moderate redshifts: telescopes simply aren't
sensitive enough.
The only way we can study gas in the early universe at present is through
QSO absorption­line spectroscopy. If the gas lies directly in front of a bright
background QSO, then high resolution optical spectroscopy of the absorption
it causes in the spectrum of the background QSO can tell us many things
about the gas. This technique has been brilliantly sucessful: a wide variety
of exciting and important things have been learned. But QSO absorption­line
work is forever restricted to statistical sampling: all you can ever measure is
what happens to lie in front of a background QSO: something even a milli­
arcsec away is completely unknown. Thus the shapes, sizes, geometries and
velocity structures of whatever gas clouds are causing the absorption are almost
completely unknown.
4.2 Proposed Facility
Clearly, what we need is a really large and sensitive radio telescope that can
observe gas in the high redshift universe, using 21cm radiation. This will not
be easy: the signals will be very faint and susceptible to interference, as the
redshifted wavelengths of 21cm are crowded with commercial transmissions.
The solution is to go to a very large collecting area: one square kilometer,
and to use fancy nulling techniques to screen out interference. This will not
be easy or cheap, but a one square kilometer telescope (called the 1kT) would
be an awesome instrument for studying the processes of galaxy formation. It
promises to be the only technique that can study the motions of gas in the
very early stages of galaxy formation, before the first stars light up.
Few of the people at this meeting seems to be radio astronomers, and so
they may need to be persuaded of the manifest advantages of radio telescopes
over optical in this type of research. Go out there and persuade them!
5 You are CALTECH
5.1 Research Interests
Your main research interests lie in Lyman­break galaxies.
You are the best, and you know it. Every one of you gets more 8­m class
telescope time than the whole of Australia combined. You lead the world: you
have the smartest faculty, the hardest working students, the biggest telescopes,
and the smoggiest atmosphere. How irritating that all these lesser minds from
other, low­grade institutions, get to vote on which facilities should be funded:
by rights, the choice should be yours alone!
5

Until a few year ago, almost no galaxies had ever been found in the high
redshift universe. You changed all that: you pioneered the Lyman­break tech­
nique, which finds galaxies by searching for the break in their spectra caused
by Ly­limit absorption. You take a series of broad­band images: say U , B and
R. Galaxies that show up as blue in B and R but are very faint in U will
be the ones which have the Ly­limit at around 4000 š A: ie. redshift 3 galaxies.
With the two Keck telescopes for follow­up spectroscopy, you have identified
several hundred galaxies between redshifts 2 and 4. This technique only finds
young, dust­free galaxies, but what other sorts of galaxies would you expect
in the early universe?
These galaxies (when observed with the Hubble Space telescope) appear
as tiny blue specks. Their spectra show that they are young, compact blue
galaxies, forming stars at rates of a few solar masses per year. You are finding
many of these galaxies per square arcmin: they appear to simply be normal
galaxies in the early universe.
What are these galaxies? You hypothesise that they are the ancestors of
the bulges of normal spiral galaxies today. These galaxies seem to be forming
slowly and steadily: not in single monumental bursts, as some people once
thought.
In collaboration with those moderately intelligent people at Lick Observa­
tory and the Space Telescope Science Institute, you used the Hubble Space
Telescope observations to extend this technique to fainter galaxies and differ­
ent redshifts. The Hubble Deep Field (an enormously long pointing at a blank
bit of sky) allowed you to find Ly­limit galaxies at redshifts ranging from 1­6.
Keck spectra confirmed the redshifts of the brightest galaxies. Piero Madau
used your data to calculate the star formation rate of the universe as a function
of redshift: he found that it peaks at redshift 1, declining slowly on either side
of this redshift.
You are now busily measuring the clustering of these galaxies. They seem to
be remarkably strongly clustered. You hope to use this clustering to constrain
the geometry of the universe and the nature of dark matter.
5.2 Proposed Facility
What this work needs is a good space telescope to get high resolution images
of these galaxies, and some really big ground­based telescopes equipped with
low resolution, multi­object spectrographs, to measure their redshifts. So,
the money should be spent on keeping the Hubble Space telescope going (or
launching a replacement), and on a few more Keck telescopes. Or maybe a
private 50m telescope for Caltech would be a good idea. . .
Rumor has it that some low individuals, jealous at your great success, are
claiming that you have missed most of the galaxies in the early universe: that
these galaxies are too red or too dusty for the Ly­limit technique to detect.
6

They will doubtless be pushing for money to buy expensive toys to search for
these non­existent red galaxies. It would be sad if it wasn't a threat to your
research: you have to make sure that these poor deluded individuals don't warp
peoples minds and persuade them to vote for these hair­brained schemes.
6 You are from The University of Durham
6.1 Research Interests
Your main research interests concern the computer modelling of galaxy forma­
tion.
How very unpleasant: here you are stuck in a meeting with a pack of
observers and instrument builders! New telescopes and detectors are all very
good: they have led to a revolution in observational astronomy. The observers
think that they are smart, but it is only thanks to these new telescopes that
all the exciting recent progress has been made in the study of the high redshift
universe. Even more dramatic advances in computer power have had a similar
effect on theoretical studies of the early universe. For the first time, it is
now becoming possible to simulate a representative chunk of the universe in a
computer.
You have long been pioneering this approach, using a combination of ana­
lytic maths, n­body simulations and hydrodynamics codes to simulate galaxy
and cluster formation. You start your simulations with a uniform gas­filled
section of the universe, rippled with tiny primordial fluctuations. You then
let the code run: slowly, the overdense regions suck matter (both dark and
baryonic) in towards them. At redshift z ¸ 1000, many tiny (¸ 10 6 M fi ) mat­
ter concentrations collapse: to stop these from forming vast numbers of stable
globular clusters, you add a line to your computer code, which means that the
first stars that form in these tiny concentrations go supernova and blow the
remaining gas out of the gravitational potential well.
As time continues, the matter starts to form sheets and filaments: the
exact shapes and densities determined by the cosmology and form of primordial
fluctuations you chose in this simulation. By redshift ten, some knots (where
filaments and sheets join) are getting dense enough to form galaxies, and the
first light in the universe switches on. You had to cheat here a bit: you have
no idea what densities are needed to trigger galaxy formation: you just chose
a number that made your simulation come out right at redshift zero.
By redshift 5, clusters of galaxies are centred on these knots. These clusters
have not yet assembled much matter, but maybe enough has assembled to
power quasars.
By redshift 2, more galaxies have formed, filling in the sheets and filaments.
Galaxy formation rates peak at redshift one, after which most of the gas has
been consumed, and the universe quietens down, and the galaxies just drift
7

around, slowly being sucked into the high dark matter concentrations in the
sheets (`great walls') and knots (` superclusters').
The remarkable thing is that the predictions from these simulations quite
accurately match the observations of Ly­break galaxies and QSO absorption­
line systems. The galaxies seen at high redshifts are indeed small, strongly
clustered, and forming stars slowly and gently, just as you predict. Just as you
predict, the star formation rate in the universe seems to peak around redshift
one.
6.2 Proposed Facility
These computer models are doing very well, but not yet well enough. The
problem is dynamic range: any simulation that models a large enough piece of
the universe to be representative (a cube of 100 Mpc on a side seems sufficient),
cannot have the resolution to accurately follow what happens on galactic scales
(¸ kpc). At present, each point in the n­body code represents around 10 8 M fi ,
and you cannot model star formation and galaxy dynamics with only ¸ 1000
particles. This means that ad­hoc rules have to be introduced to model the
formation of stars and galaxies, eg. you could assume that a galaxy forms
whenever the density passes some arbitrary threshold, with a luminosity pro­
portional to its baryonic mass.
This is unfortunate: the computers can accurately simulate things we can­
not observe, such as the large­scale distribution of dark matter, but not things
that we can, like the galaxy luminosity functions, star formation rates and
clustering.
So surely the best thing to spend money on is faster supercomputers! With
a network of massively parallel modern supercomputers, the simulations will
become so good that they can accurately predict observable quantities without
the necessity of making so many ad­hoc approximations. The supercomputers
will be welcomed by scientists in many other disciplines, as well as by other
astronomers. Forget obtaining more data: a relatively modest investment in
computer power will allow us to better understand the data we already have.
7 You are the Royal Observatory, Edinburgh
7.1 Research Interests
Your main research interests lie in starburst galaxies.
From your base, high on a rocky crag overlooking the capital of Scotland,
you have long been studying starburst galaxies in the local universe. Perhaps
it is the howling wind and lack of insulation in your offices that does it, but
the glowing warmth of a galaxy violently forming vast numbers of stars has a
great attraction to you.
8

Over the last few years, you have watched with incredulity as group after
group has claimed to detect new­born galaxies in the rest­frame UV! If there
is one thing you learn from low redshift galaxies, it is that star formation
always takes place accompanied by dust. And dust is really extremely good
at blocking UV light. In our own galaxy, star formation takes place in giant
molecular clouds, and you won't get very far studying them in the UV. All
the nearby galaxies that are forming stars at reasonable rates are very dusty:
most of the radiation from their young stars and supernovae is absorbed by
dust and re­radiated in the far­IR.
So it is no wonder to you that the Caltech mafia and their friends are
only finding wee little galaxies forming stars at some pathetically gentle rate.
Any decent, strapping starbursts will be so dusty that the little dears won't
see a thing with their fancy UV Lyman­limit techniques. There could be any
number of new­born galaxies in the early universe, forming stars at enormous
rates, and the Ly­limit people would never know.
How do you find and study starbursts in the local universe? You look in
the far­IR, where these galaxies stand out like sore thumbs (they are called
ultra­luminous IRAS galaxies for a reason, you know). Another possibility is
to use mm­wave telescopes to look for molecular emission from these galaxies.
You have just finished building a new instrument, called SCUBA (Sub­
mm ContinUum Bolometer Array), which goes on the 15­m James Clerke
Maxwell Telescope (JCMT) that you run on Mauna Kea. With this, you
could, in principle, detect redshifted far­IR thermal radiation from dusty high­
z galaxies. And this is just what you are finding: many galaxies, which are
almost certainly at high redshifts, with enormous amounts of hot dust. These
things must be forming stars at colossal rates: far more actively than those
pathetic Ly­limit galaxies. Clearly everyone has been looking in the wrong
wavelengths to find new­born galaxies.
7.2 Proposed Facility
With current telescopes, finding dusty starbursts in the early universe is hard
work. What you really need is a mm­wave telescope with really good resolution
and collecting area on a high, dry site. That is why you are interested in
proposals to build a mm­array: a network of lots of mm­wave telescopes, with
oodles of collecting area, built at 5000m elevation at a site in the altiplano
of northern Chile, near the Bolivian border. Using aperture synthesis, this
telescope would have fantastic angular resolution. Not only could you find high
redshift dusty galaxies, but you could use their molecular emission lines to map
them, measure their velocity fields, temperatures and chemical compositions.
It will be awesome! It should also be able to search for galaxies at redshifts
beyond five, which are very hard to find with normal optical telescopes.
It is obvious to you that this is the best way to study the early universe
9

and the formation of galaxies. Some of your colleagues seem to persist in the
bizarre notion that you can study galaxy formation in the optical: you must
gently show them the error in their ways.
8 You are the NASA Goddard Space­Flight Center
8.1 Research Interests
Your main research interests lie in the inter­galactic medium.
For years, you have been fascinated by what lies in the gaps between the
galaxies. Your colleagues often derided your fixation with empty space, but you
remained convinced that this was an interesting and important field. After all,
galaxies only occupy a tiny tiny fraction of the universe: if there is anything at
all, however tenuous, in all those vast inter­galactic spaces, it could dominate
the mass budget and dynamics of the whole universe.
The first big breakthrough occurred in the late 1960's, when high redshift
QSOs were first discovered. If intergalactic space was full of neutral hydro­
gen, then all wavelengths shortward of Lyff (1216 š A) in the rest­frame of the
QSOs should have been absorbed (the Gunn­Peterson effect). It was not: this
indicated either that intergalactic space at high redshifts is completely empty
(which was very hard to understand theoretically), or that the gas was ex­
traordinarily hot, and hence too higly ionised to absorb at the wavelengths of
Lyff. Note that these absorption­lines occur in the UV: you can only observe
this phenomenon in the early universe, where cosmological redshifts shift the
relevant absorption­lines into the near­UV or optical.
This observation posed great problems: what is the energy source that
heated up all this gas? Even at redshift 5, the inter­galactic medium is almost
completely ionised: this means that a lot of very powerful ionising sources,
hitherto undetected, most have existed way back then.
The second big breakthrough occurred about three years ago: Peter Jakob­
sen used the Hubble Space telescope to look for absorption due to He II in the
far­UV (304 š A). This would only show up if the inter­galactic medium was in­
deed very hot: strong absorption was seen, showing that there is a hot medium
out there. This absorption has now been seen in four QSOs.
Very recently, HST observations of the He II Gunn­Peterson effect have
shown that this hot gas is patchy: every now and then, voids are found in
the hot gas: regions which may really be completely empty, or in which the
intergalactic gas is either cold, or very very hot, so that He II is not found.
So: exciting days. The mystery of what lies between the galaxies is slowly
being cracked, and the answer is turning out to be complex: remarkably hot
gas, with a total mass probably exceeding that in all galaxies and denser gas
clouds combined, and with these mysterious voids all over the place.
10

8.2 Proposed Facility
Clearly, the burning need is for more data on this thrilling topic. Mid­UV spec­
tra of faint QSOs are the way to go, and this means a space mission. Some sort
of space telescope: a successor to the HST, with good spectrographs that work
in the UV is what you want. With such an instrument you could investigate
when and how the intergalactic gas became so hot, how it is distributed, and
what it is made of. With good coverage far into the UV, you could trace how
this gas changes as the universe evolves.
Other useful facilities would include smaller telescopes for finding lots of
nice high­redshift QSOs for you to study.
You will have to be careful: unless you can get funding for a big space
mission, your budget will probably be cut by Congress. Some of the other
people at this meeting will try and argue for other space missions or (horror
of horrors) ground­based telescopes. Fight them -- your jobs are on the line!
You must try and persuade other groups that a UV spectroscopic satellite
will be useful for them too: without their support you could be in serious
trouble. Bear in mind that this telescope could have other instruments on as
well: perhaps optical imaging cameras and spectrographs. This might be the
bait that lures the other astronomers into helping you!
9 You are the Institute d'Astrophysique de Paris
9.1 Research Interests
Your main research interests lie in QSO absorption­line systems.
The wretched anglo­saxons seem fixated on seeing things: unless they have
an images they can flog to the media, they are just not happy. You, however,
are more enlightened about these things (it comes naturally, being French).
The trouble with trying to see galaxies in the early universe is that all you
can see will generally be stars. What's wrong with that, the English­speaking
fools may ask? Well, most of the universe is made up of dark matter and
gas: stars are only a trace impurity, even today. And in the early universe,
before most of the stars have formed, stars must be an even less significant
component of the universe.
It clearly makes far more sense to study the gas in the early universe. Gas
is simple: we understand how it behaves. There is far more mass in gas than
stars. By looking at the dynamics, distribution and metallicities of the gas,
we can learn about dark matter, nucleosynthesis and galaxy formation in ways
that picture fixated astronomers will never do.
So, how do you study gas in the early universe? QSO absorption­lines, of
course. You have spent a decade obtaining high resolution spectra of distant
QSOs, studying the gas that lies along the sight­lines. This technique has
11

taught the world far more about the early universe than all others combined.
Highlights include:
ffl As stars form, they must use up gas. By watching the decline in the
space density of damped Lyff absorption­line systems (clouds of neutral
hydrogen with column densities of ? 10 20 cm \Gamma2 , Mike Fall and collabo­
rators were able to calculate the star formation rate of the universe as a
function of time: five years before anyone else came close.
ffl The tiny, low column density Lyff forest lines are by far the most common
denizens of the high redshift universe, and probably contain more baryons
than all galaxies and bigger objects combined.
ffl You yourselves were able to show, by comparing QSO absorption­line
data with deep images and spectra of galaxies near the sight­lines, that
huge halos of absorbing gas surround all galaxies out to at least redshift
one.
ffl Max Pettini and collaborators used QSO absorption­lines to measure the
average metalicity of the universe as a function of redshift, showing that
it steadily increases between z = 5 and today, giving a measure of the
metal (and hence star) formation rate of the universe as a function of
redshift.
9.2 Proposed Facility
Clearly, what we need is