I løpet av de siste par tiårene har kosmologiske observasjoner blitt så nøyaktige at vi kan bestemme alderen og sammensetninga av universet med usikkerheter på prosentnivå. Men samme hvor nøyaktige måleinstrumenter som blir bygd, er det en ting vi ikke slipper unna når vi prøver å måle noe langt borte, nemlig lokal forurensing i form av stråling fra melkeveien. For å skille mellom kosmologiske data og lokal forurensing er vi avhengig av det som kalles astrofysisk komponentseparering. Dette er viktig innen kosmologien i dag, og er noe som Universitetet i Oslo er internasjonalt ledende på.
Det opprinnelige målet med INTPART-finansierte Global Component Separation Network (GCSN) var å bringe sammen internasjonale forskere fra fire fremtidsrettede eksperimenter (COMAP, LiteBIRD, PASIPHAE og SPIDER) for utveksle erfaring og kunnskap og lære opp neste generasjon kosmologer i astrofysisk komponentseparering. Fordelen med felles analyse av komplementære datasett er at de bryter hverandres degenerasjoner, og vi får en best mulig beskrivelse av både de forskjellige intrumentene, astrofyikken i vår egen galakse og de underliggende kosmologiske parametrene vi er på jakt etter.
Betydninga av dette arbeidet har blitt stadig klarere for det kosmologiske og astronomiske fagmiljøet, og i 2019 finansierte Det Europeiske Forskningsrådet (ERC) et stort forskningsprosjekt kalt Cosmoglobe, leda av prosjektlederen for GCSN. Siden den gang har Cosmoglobe og GCSN vokst til et integrert forskningsprosjekt der målet er felles global analyse av flest mulige dataset i en felles himmelmodell, der Cosmoglobe dekker grunnforskning og GCSN dekker nettverksbygging og opplæring av neste generasjon forskere. Den første store data-releasen fra Cosmoglobe fant sted i mars 2023, og inkluderte verdens første felles rådata-analyse av to ledende kosmisk bakgrunnsstrålings-satellitter, nemlig NASA's WMAP og ESA's Planck. Denne analysen har revolusjonert måten denne typen data vil analyseres på i framtida, og eksperimenter, kollaborasjoner og forskere over hele verden slår seg nå sammen innad i Cosmoglobe for å skape en felles modell av hele universet.
Global Component Separation Network inkluderte opprinnelig 8 internasjonale universiteter og forskningsinstitusjoner (Caltech, IUCAA, KwaZulu-Natal, Kavli IPMU/Tokyo, Oslo, Princeton, South African Astronomical Observatory og Toronto), men har i løpet av prosjektperioden vokst til å inkludere 15 institusjoner i 6 land. Sammen med Cosmoglobe har antall dedikerte eksperimenter vokst fra 4 til 9, mens minst 18 eksperimenter og flere hundre forskere har deltatt på møtene våre. Vi har utvikla et permanent intensivkurs i kosmologisk komponentseparering og bayesiansk dataanalyse ved Universitetet i Oslo, med flere internasjonale deltakere enn vi har plass til, og vi har tilrettelagt for student- og forskerutveksling samt arrangert mer enn ti internasjonale møter - noen av dem digitale pga pandemien.
We have built up an intensive course in cosmological component separation and Bayesian data analysis at University of Oslo (UiO). This is an integrated part of UiO's course portfolio with course code AST5240 for master students and AST9240 for phd students, and with a mechanism for giving official ECTS from UiO to external students. Here students get an introduction to the cosmic microwave background (CMB), cosmological experiments, instrumentation and data analysis, and the astrophysics of the Milky Way, as well as hands-on tutorials to component separation and Bayesian data analysis in general. More importantly, students with very diverse backgrounds (geographic, scientific, and experiment-wise) come together to work on a common project. We have during the project period continuously developed the course based on feedback, current research and new ideas, and the overarching theme of each installment of the course is now a joint large project work that leads to a joint publication. Anonymous course evaluations show that students find the course useful and fun.
The collaborative cutting-edge nature of the course naturally leads to continued collaborations, and the course has become the most important recruitment channel for scientific collaborators. Together with community-wide meetings, experiment-related smaller meetings and intermediate workshops, the yearly intensive course is a cornerstone of the global community that Cosmoglobe and the Global Component Separation Network (GCSN) have built to jointly tackle joint global analysis. This initiative now has around 50 core collaborators whereof more than half are international, while several hundred have attended some of our activities.
Moving from per-experiment analysis to joint global analysis can be viewed as a paradigm shift in the community and the GCSN was one of the main forerunners for this to happen. The latest and most visible example of this is the joint data analysis of the two last CMB satellite missions, Planck and WMAP, being submitted in March 2023; for the first time the maps from these two satellite experiments are consistent at large scales, a problem that has puzzled the community since the first Planck release in 2013. Joint analysis of two or more experiments break each others degeneracies, and for Planck and WMAP this means that the excess signal from unconstrained instrumental systematics is gone and we get a clearer view of the underlying cosmology like when the first stars were born.
The INTPART-funded Global Component Separation Network has already contributed positively to the funding of Cosmoglobe from the ERC, and the results and collective expertise of GCSN and Cosmoglobe will hopefully lead to even more EU funding and ideally a Center of Excellence in the future. These may then be utilized to complete the vision of truly global analysis of all cosmological datasets, and serve as a training ground and opportunity network for the next generation of cosmologists.
During the last two decades, new high-precision cosmological
observations have revolutionized our understanding of the early
universe, to the point at which we are today able to pin-point the age
and the composition of the universe to percent accuracy. However, as
the precision of each experiment increases, there is one source of
uncertainty that eventually will affect all new measurements, namely
contaminating radiation emitted from our own Milky Way. To remove
these contaminants from the data, one has to perform a process called
astrophysical component separation, which is a major branch of
contemporary computational cosmology. Furthermore, it is a field in
which the University of Oslo (UiO) currently plays an internationally
leading role.
In this proposal we propose to establish a network of eight
internationally leading education and research institutions (Caltech,
IUCAA, KwaZulu-Natal, Kavli IPMU/Tokyo, Oslo, Princeton, South African
Astronomical Observatory, and Toronto) and four state-of-the-art
cosmology experiments (COMAP, LiteBIRD, PASIPHAE, and SPIDER), and
employ this network to enhance both teaching and research in each
individual member institution. The main focus of this network lies on
education and teaching, and is thus complementary to and will enhance
on-going efforts dedicated to the scientific aspects of each
experiment. Specifically, we will 1) host a bi-annual summer school
for partner students; 2) create a new course in astrophysical
component separation at the University of Oslo; 3) facilitate student
exchange between parter institutions; and 4) organize work meetings,
cross-experiment meetings, and an international conference targeting
young scientists, PhD and master students.
The long-term goal of the project is to establish this network as an
internationally recognized hub for astrophysical component separation
effort and computational cosmology. As such, it will consolidate
Norway's leading position in the field in the coming decade.