Annual Report 1998  

MPA Homepage > Scientific Research > Research Groups > Galaxy Formation > Euro. Network (Galaxy Formation) > Annual Report 1998

Go to:

-1 Radio Galaxies
-2a Evolution of Galaxy Clusters
-2b Evolution of Field Clusters
-3 QSO Absorption Lines
-4 Population synthesis models
-5 Semianalytic modelling
-6 Simulations of galaxies and clusters
-7 The Hubble deep field south

TMR Research Network Contract No. ERBFMRX--CT96--0086

Activity in the Network continued at its full level throughout the year resulting in 53~joint publications for refereed journals, and 24 conference reports. As for the first year, the scientific report below is split according to the major research areas detailed in our original proposal. An additional major activity during 1997-98 is, however, listed separately since it does not neatly fall under any single heading. This is the Network's activities in preparation for, and to exploit the Hubble Deep Field South programme, a Director's observing programme on the Hubble Space Telescope.

The Network's first set of postdoctoral fellows reached the end of their 2-year terms in September 1998. Earlier in the year the Network therefore carried out a joint hiring exercise based on a common internationally circulated advertisement and joint consideration of the 40 or so applications which arrived in response to it. This exercise resulted in one of the Network Fellows being retained for a further two years and the other five being replaced. All five who moved on were successful in obtaining permanent or long-term research posts in astrophysics, four in their home country and the fifth in Mexico.

The annual meeting of the Network was held over the period 31 August - 5 September at Castle Poelgeest in Leiden. This was a very lively and interactive meeting attended by 71 scientists from the six Network sites. It led to substantial progress on existing joint programmes as well as to a number of new collaborations. A number of smaller meetings were held at various sites during the year and are detailed in the scientific report below.


1 Radio galaxies top

Many of the most interesting targets uncovered by radio surveys are optically faint, and as a result ESO's VLT (available from April 1999), with its huge collecting area and diverse battery of instrumentation offers a great opportunity for European astronomers. In consequence, in addition to continuing their fruitful multi--wavelength imaging and spectroscopic campaigns, Network radioastronomers have this year concentrated on preparing samples of radio galaxies for VLT studies. The Leiden-Cambridge collaboration has obtained and reduced completely data for half of the 92~cm ``Westerbork survey In the Southern Hemisphere'' (WISH), the first deep large-area radio survey at low frequencies in the southern hemisphere. In addition, spectroscopic redshifts have been obtained for the first large and complete sample of powerful southern radio galaxies.

Scientific highlights

  • Network astronomers obtained the first solid detection of CO emission from a distant radio galaxy, 1909+722.
  • Sub-millimetre studies established that the most distant radio galaxies contain more dust than their low redshift counterparts.
  • Deep HST infrared imaging showed that, unlike their low redshift counterparts, the highest redshift radio galaxies are not associated with well formed ellipticals.
  • Deep spectroscopy of powerful redshift one radio galaxies demonstrated that the interstellar medium is dramatically affected by the passage of the radio source shocks.

2a Evolution of cluster galaxies: top

A key objective is to understand how galaxies evolve in rich cluster environments. We have investigated the luminosity evolution of cluster galaxies, concentrating in particular on the spheroidal galaxies, using a newly-constructed catalog of morphologically-classified Hubble Space Telescope images of 13 clusters with redshifts $0.17
A deep, multi-colour imaging survey of a sample of high X-ray luminosity clusters (z=0.22-0.28) shows that the integrated luminosity of the redder cluster galaxies is well correlated with cluster X-ray temperature, and hence with cluster mass. Furthermore, the typical restframe UV-optical colours of the luminous ellipticals exhibit a remarkably small cluster-to-cluster scatter, indicating that this population varies little between clusters. However, at fainter magnitudes we observe a marked increase in the range of mid-UV colours. In the light of the apparent decline in the population of S0 galaxies seen in distant, $z>0.4$, clusters we propose that they may be the progenitors of the dominant S0 population of local rich clusters, caught in the final stage before they become completely quiescent.

The colour-magnitude relation for early-type galaxies has been used to constrain the formation of rich cluster galaxies up to redshifts $\simeq1$. The bulk of the stellar populations of these galaxies has been found to be old, formed at redshifts greater than 2. We have also developed a new Bayesian classifier for photometric redshifts, and studied its application to the identification of new high redshift clusters.

2b Evolution of field galaxies: top

The study of distant star-forming galaxies has witnessed remarkable recent progress. Several network researchers have very successfully used the new observing capabilities provided by the SCUBA bolometer array at the James Clerk Maxwell Telescope (JCMT), which has improved the performance of submillimetre-wave continuum observations substantially. Observations of dust-enshrouded star formation in extremely distant galaxies has opened a new window on galaxy formation and evolution. The faint objects detected have been used to estimate a major contribution to the global star formation history of the Universe. Their HST morphologies indicate that interacting and disturbed galaxies predominate, suggesting that interactions remain important for triggering star formation bursts at high redshift. Faint, compact galaxies may represent a later evolutionary stage in these mergers, or be more centrally-concentrated starbursts. The Network team has obtained the deepest sub-mm galaxy number counts published so far.

The evolution of field ellipticals has been studied using data from the Hubble Deep Field. Even at redshifts as high as 2-3, these galaxies lie on the fundamental plane, apparently in a Virial equilibrium. However, at these same redshifts a significant lack of large galaxies is observed, indicating that the formation of an important fraction of present-day ellipticals was not complete at these early epochs.

Several Network researchers in our network have been studying the host galaxies of QSOs, reporting the first unambiguous detection of the host galaxy of a normal radio-quiet high-redshift QSO at 2.2 microns (K-band). The luminosity of the host is about 35\% of the total in the observed K-band, implying that the host is about 100 times brighter than an unevolved L* galaxy placed at the QSO redshift (z=2), and about 25 times brighter than a passively evolved L* galaxy at the same redshift. The luminosity of the host galaxy is thus similar to the highest found in radio-loud QSOs and radio-galaxies at the same redshift.

At lower redshifts, we have made substantial progress towards understanding why Lyman-alpha emission is very weak or absent in the UV spectrum of HII galaxies. We have observed 8 HII galaxies with low metallicity with HST and detected Lyman alpha with a strong P-Cygni profile in half of them, while broad Lyman alpha absorption with no emission was detected in the other half. This result supports the hypothesis that Lyman alpha is destroyed in these systems by multiple scattering in gas at the same velocity as the emitting gas, and that Lyman alpha photons can only escape in the line wings. The difference between detecting or not the Lyman alpha emission may be related to the presence, or not, of the high velocity scatterers, i.e., a galactic wind.

Network researchers from Paris and Cambridge have detected molecular gas in a third QSO at redshift larger than 4, the lensed radio quiet quasar BRI~0952-0115 at redshift of $z=4.43$. further papers are in preparation concerning their latest results on the mm/submm detection of dust emission in high redshift QSOs using the IRAM 30m, SCUBA/JCMT and Caltech/CSO. In order to select new sources to search for molecular lines in absorption at high redshift, the group has also performed a millimeter continuum survey of a large number of radiosources. They have detected 1.3mm emission in several hundred sources.

3 QSO absorption lines top

A group including a number of Network researchers has used the adaptive optics system on the Canada-France-Hawaii telescope to study intermediate and high redshift quasar-hosts. In particular, they observed the $z$~=~3.87 broad absorption line quasar APM~08279+5255 (Ledoux et al. 1998). The object is found to be a double source. The separation of the two images is 0.35~$\pm$~0.02 arcsecs and the intensity ratio $I_{\rm north}$/$I_{\rm south}$ = 1.21~$\pm$~0.25 in the $H$-band. No other image is detected down to $H$(5$\sigma$)~=~21.3 within 10$''$ from the double image. Strong support for the lensing hypothesis comes from the uniformity of the quasar spectrum as a function of spatial position in the image obtained with the integral field spectrograph OASIS at CFHT. From the 2D-spectroscopy, narrow-band images are reconstructed over the wavelength range 5600-6200~\AA\ to search for emission-line objects in a field of 15$''\times$12$''$ around the quasar. We find no such object to a limit of 6$\times$10$^{-17}$ erg cm$^{-2}$ s$^{-1}$. We use the images centered on the deepest absorption lines of the to dim the quasar and to increase the sensitivity closer to the line of sight. One of the images, centered at 5766.4~\AA , exhibits a 3$\sigma$ excess 1.5$''$ from the quasar to the north-east. \par\noindent Further Network collaboration has been developed as part of HdFS project (see below). This is a unique opportunity to study the relation between the Ly$\alpha$ forest and metal absorption systems detected in the QSO spectrum and the galaxies in the field at $z$~$\sim$~1.2. A VLT proposal has been submitted.

4 Population synthesis models top

Charlot (Paris) and Kauffmann (Munich) have shown that a powerful way to constrain galaxy formation theories is to analyze the evolution of the $K$-band luminosity function, since this avoids uncertainties in relating galaxy colours to morphology. Based on presently available samples, the evolution of the $K$-band luminosity function appears to be inconsistent with traditional models, but similar to that expected in hierarchical models of galaxy formation.

A major effort in Padova concentrated on interfacing the new Padova multi-component population synthesis code with the Durham semi-analytic galaxy formation model. Very encouraging results from this effort were presented at the 1998 annual meeting and are the subject of two papers under preparation. An INTERNET interactive section providing the new models is under construction. The Padova stellar populations group is also collaborating with Cambridge (star formation indicators in galaxies), Paris (photometric redshifts and narrow band indices in early--type galaxies), and Munich (stellar evolution and population synthesis).

5 Semianalytic modelling top

Kauffmann (Munich) and Charlot (Paris) have combined semi-analytic models of galaxy formation in a hierarchical universe with new metallicity-dependent spectral synthesis models to compute the chemical enrichment, colors, line indices and mass-to-light ratios of early-type galaxies as a function of redshift. The success of their models in reproducing the slope and scatter of the color-magnitude and Mg$_2-\sigma$ relations of elliptical galaxies challenges the traditional view in which early-type galaxies form in a single monolithic collapse at high redshift and undergo closed-box chemical evolution. They further predict that the metallicity of the intracluster medium must evolve very little out to $z>1$.

Paris and Durham researchers are collaborating to implement the theoretical IR/submm spectra of Devriendt et al. (1998) into the semi-analytic model of Baugh et al. (1997), to obtain specific predictions for IR galaxy counts. Paris and Padova scientists are working to produce multi-wavelengths models of Active Galactic Nuclei in a simple semi-analytic scheme. They will implement simple recipes for the formation and evolution of starbursts in the simple semi-analytic scheme of Cavaliere \& Vittorini (1998), which takes into account QSO formation and rejuvenation through interactions. This model will give multi-wavelength predictions for galaxies and AGNs, in particular, for AGN counts at IR/submm wavelengths.

The Durham-Copenhagen group have used their semi-analytic galaxy formation model to calculate the size distribution of galaxy disks and how this evolves with redshift. They find that in a Lambda CDM model, the predicted present-day distribution of disk scalelengths at different galaxy luminosities agrees very well with the observational one derived by De Jong \& Lacey (1998). The evolution of the disk size distribution to z=1 is predicted to be quite weak at given luminosity, except for the most luminous galaxies, in qualitative agreement with what is found from the CFRS redshift survey. They have also extended the model to calculate the global HII region emission line properties of galaxies. The results give a good fit to the present-day H-alpha luminosity function of galaxies, and can reproduce the trend that the fraction of present-day galaxies with large OII equivalent widths increases with decreasing luminosity. The effects of dust on the emission line luminosity densities commonly used as star formation tracers are found to be large, around 1 magnitude of extinction for H-alpha and 2 magnitudes of extinction for the OII line over the range z=0-5. Finally together with Padova scientists they have made a combined stellar population/dust model which can predict galaxy spectra from the UV to the sub-mm, including both absorption and re-emission by dust. The result gives a reasonable match to the observed FIR (60 micron) luminosity function of galaxies at z=0, and predicts a steep rise in the FIR luminosity density of the universe with increasing redshift. This agrees qualitatively with inferences from recent SCUBA observations (see 2b).

6 Simulations of galaxies and clusters top

Substantial progress was achieved in this area. Activity took place at Durham, Munich, Paris and Padova, resulting in 12 joint refereed publications and 7 conference proceedings involving these institutions, particularly the first two. In order to take advantage of the unique opportunities offered by the network's access to one of the largest parallel supercomputers in Europe, the Cray-T3E in Garching, we deviated from our original programme. Thus, during 1998 we carried out the ``Hubble volume'' simulations described below under ``scientific highlights.'' This required developing and implemented a new, highly efficient parallel N-body code. We also invested substantial effort analysing the set of dark matter simulations which the network and external collaborators had carried out the previous year as described in our 1997 report. These simulations are the first to resolve individual galactic halos in volumes large enough to provide reliable clustering statistics. A variety of studies were carried out using these data. Highlights include the most accurate study so far of the clustering statistics of the dark matter (Jenkins et al 1998); detailed analyses of the internal structure (Thomas et al 1998), velocity distributions (Colberg et al 1998a), formation history (Tormen, Frenk \& White 1998), and large-scale environment of galaxy cluster halos (Colberg et al 1998b); a study, using novel techniques, of the topology of the dark matter distribution and a comparison with recent galaxy surveys (Springel et al 1998, Canavezes et al 1998); analyses of the strong (Bartelmann et al 1998) and weak (Jenkins \& Bartelmann 1998) gravitational lensing properties of clusters. Taken together, these studies represent the most comprehensive analysis ever of a high precision set of dark matter simulations and represent an important step towards the achievement of our year 3 milestone.

Our year 2 milestone (understanding the requirements for disk galaxies to form with the observed properties) proved to be more difficult to achieve than expected. Nevertheless, the TMR postdoc at Durham, Dr. Governato, and other network researchers made progress in the subject, further developing serial and parallel N-body/gasdynamic codes, including those written by young network reseachers Springel and Carraro. These codes are being applied not only to the formation of disks, but also to the formation of ellipticals and to the effects of galaxy mergers. Results from this work are already coming in.

Scientific highlights

The "Hubble volume'' simulations are the largest calculations ever attempted of the formation of structure in the expanding Universe. They represents one of the most intensive applications of supercomputing to a civilian research project. With a billion particles, the Hubble volume exceeds the previous largest simulation by over an order of magnitude in size. It models the growth of clusters and larger structures over virtually the entire observable universe. It allows study, for the first time, of fundamental statistical properties of galaxy clusters, one of the primary tools for cosmological research. The achievement represented by this simulation attracted considerable media attention in Europe and the USA.

7 The Hubble Deep Field South top

The observation by HST of a ``Deep Field" in the northern hemisphere has stimulated enormous activity world-wide in areas relevant to the ``Formation and Evolution of Galaxies". As soon as the Space Telescope Director decided to conduct a similar effort in the southern hemisphere, we decided to take steps to ensure that European participation in this program would be substantial. Because European optical astronomy in the south is carried out primarily through the European Southern Observatory (ESO), we decided at our 1997 annual meeting to initiate a ground-based follow-up effort of the HST observations through ESO, with its main emphasis on spectroscopy.

A working group, coordinated by M. Dennefeld (Paris) and including members of most of the participating institutes, defined a step-by-step program to obtain spectra for as many as galaxies as possible around the Hubble Deep Field-South (HDF-S). To ensure that to complete this work as rapidly as possible, this ground-based program was started before the actual HST observations (taken in October 1998 and release November 23 1998) and a successful proposal was submitted to ESO resulting in imaging of the full field and spectroscopy of the spectroscopy of the 200 brightest galaxies being obtained at the ESO NTT in September and October 1998.

Simultaneously, coordination was set-up with ESO, who intended to do the corresponding ground-based deep imaging, to ensure that the appropriate images and colours would be available for the next step, deep spectroscopy with the VLT. Indeed, selection of the galaxies to be observed with the VLT will rely on photometric redshifts derived from these deep images using a tool developed within the Network. The corresponding scientific program was defined during a special meeting held in Padova in June 1998, and refined at the annual Network meeting in Leiden in September 98.

Three major programmes were identified which capitalise on the specific capabilities of the HST and VLT instrumentation, and a proposal for the VLT time was submitted in each area (decision expected December 1998). The first will measure the cosmic star formation history directly in the redshift range where it is supposed to peak ($~ 1 3.5,$ with the B drop-out technique) and use the FORS instrument on the VLT to derive their redshifts and star formation rates. This will determine whether star formation activity truly declines at high redshift. The third will investigate the nature and redshift of the faint galaxies which are close to the QSO in the STIS field of the HDF-S. Here FORS will be used to identify the galaxies responsible for the absorption lines in the QSO spectrum, and to estimate the metallicity of the gas component of galaxies at redshift 2. \vglue1truecm Networking Activities


An international workshop, "The most distant radio galaxies'', was held in Amsterdam on 15-17 October 1997, involving participants from 5 network sites. A 500 page book containing the proceedings of the workshop is in press.

The annual meeting of the network was held in Leiden on 31st August to 5 September 1998, with 71 participants from the 6 network sites. Details of the programme are given in the WWW pages.

A meeting was held at Cambridge during April/23-24/1998 to discuss progress and future plans in the cosmological simulations of galaxy formation. In addition to network participants, this meeting also included external colleagues (from the Virgo consortium) who often collaborate with network participants. The discussions were far ranging and included coordination of the science to be done with the ``Hubble volume'' simulations as well as detailed planning for our programme of gas-dynamical simulations of galaxy formation.

Training of young researchers

There have been substantial numbers of exchanges of young researchers between sites. Currently one PhD student (A. Helmi from Leiden) is carrying out her thesis work under the supervision of a scientist at another site (S. White in Garching). Network funds are also supporting an Italian student (S. Marri from Firenze) to pursue doct oral studies in Garching as part of a Firenze-Garching collaboration.

Of the researchers employed as Network postdoctoral Fellows, Philip Best will remain in Leiden for 2 more years, Fabio Governato has left Durham to take up an academic post in Milan, Bianca Poggianti has left Cambridge for an academic post in Padova, Malcolm Bremer has left Paris for a lectureship in Bristol, Itziar Aretxaga has left Garching to take up a long-term research post in Puebla (Mexico), and Yuen Ng has left Padova for a post at the Space Research Organisation of the Netherlands. The new Network Fellows are: in Cambridge Pierre-Alain Duc (a Frenchman previously at ESO, Garching), in Durham Sergio Gelato (an Italian previously in Copenhagen), in Garching Tom Theuns (a Belgian previously at Cambridge), in Padova Herve Aussel (a Frenchman previously in Paris) and in Paris, Steve Hatton (an Englishman from Durham). Almost all these researchers already have active collaborations in more than one Network site.

Comments to: Naoki naoki@mpa-garching.mpg.detop