#1 some more info on Universe@Home
Posted: Tue Jan 13, 2015 2:47 am
http://universeathometest.info/universe/universe.pdf
UNIVERSE@HOME: ULX’S, GR SOURCES, SNIA PROGENITORS
PI: Krzysztof Belczynski (Warsaw University)
This proposal aims to create the first database of the simulated stellar content of the Universe,
from the earliest stars to the most exotic black hole binaries. This comprehensive stellar library will
be a world-wide scientific resource for astronomers, and an educational asset for the public. With
a state-of-the-art suite of computational tools, we can calculate the physical properties of individual
stars, evolve them in time, and track how they interact with each other as they age. With this library
of stellar populations that spans the known range of star formation histories and initial elemental
abundances, it is possible to tackle fundamental science questions that have been out of reach in
general relativity, astronomy, and cosmology. As a group we will use this database to characterize the
population of supernova progenitor stars that have been used to discover the mysterious dark energy
that accelerates the expansion of the Universe; we can calculate how (if at all) binary evolution
facilitates the formation of ultra-luminous X-ray sources, and we can model the number and physical
properties of neutron stars and black holes that strongly emit gravitational waves that are at the
heart of research into gamma-ray bursts – the most energetic explosions known-to-date. The data
that we create can be mined not only to match current observations, but to predict how the sky
will appear to future astronomical observatories as new windows to the Universe open. Altogether,
this proposal will allow us to consolidate a strong and highly visible research group; an independent
computational center that trains a new cohort of highly educated students, and communicates with
the public via innovative science venues, and offer a fresh look at the Universe.
1. OBJECTIVES
1.1. Context
Stars are the basic building blocks of galaxies that,
in turn, make up the bulk of the baryon content of
the Universe. The first stars formed very early—
roughly 100 Myr after the Big Bang —out of the pri-
mordial hydrogen-helium mix. These stars, referred
to as Population III stars, were likely extremely mas-
sive (M 1000M⊙) and produced the first heavy
elements through nuclear burning and supernova ex-
plosions. These first supernovae created black holes
with masses between 10−1000M⊙ [1,2,3]. Thus, these
first stars provided not only the seed black holes for
the eventual formation of supermassive black holes
(Mbh 106 − 109M⊙), but also the metals that we
see in the second generation of stars. These stars,
known as Population II stars, began forming after the
Population III era from gas that was already polluted
with heavy elements [4]. Still, the Sun has 104 times
more metals than this population ( Z > 10−4Z⊙).
Population II stars most likely generated the photons
necessary to re-ionize the Universe, and set the chem-
ical composition of the Universe we observe today.
At about 1-2 Gyr after the Big Bang, Population
I stars began forming. These stars have metallicities
comparable to that of the Sun, and presently domi-
nate the overall star formation in the Universe [5,6].
Although most stars today are in this category, this
population is in no way a uniform sample. For exam-
ple, the metallicity varies widely: about 50% of star
formation in the past Gyr occurs at Z 0.2Z⊙, while
the remaining half is found at solar composition or
higher (Z > Z⊙) [7]. In the bottom panel of Figure
1, we show a quantitative picture of star formation
history with its uncertainties [8].
The environment of Population I stars varries as
well. Typically, stars most commonly form as single
objects or in binaries, although higher multiplets are
known to exist. Even the dynamics of these stars are
different: In galaxies, most stars do not directly inter-
act with other stars – these are the field population.
However, a small number of stars form in dense stellar
clusters, and here dynamical interactions such as col-
lisions are frequent [9,10]. The fact that today’s stars
form and evolve in such varied circumstances means
that they produce a wealth of information that probes
an extraordinary range of of physics. Stars emit light
across the entire electromagnetic spectrum, from ra-
dio to hard gamma rays, and even generate gravi-
tational radiation. Various stages of stellar evolution
are connected with explosive events: X-ray flares from
magnetic low mass stars [11]; luminous blue variable
eruptions from massive single stars [12]; type Ia su-
pernovae from binary white dwarfs [13]; X-ray bursts
from binary neutron stars [14]; gravitational radiation
bursts from mergers of neutron stars and black holes
[15]; long gamma-ray bursts (GRBs) from massive,
rapidly rotating stars [16], and short GRBs from dou-
ble neutron stars and black hole-neutron star mergers
[17].
A number of stellar phenomena are still poorly un-
derstood and repeatedly defy our attempts to explain
1
2
them. Our library of models will help us to solve, or
at least constrain these mysteries. At the most basic
level, astronomers still lack a full description of star
formation [18], convection [19], mass loss [20] and su-
pernova explosions [21]. Since the amount of heavy el-
ements released in stellar winds, supernovae and com-
pact object mergers change how stars interact with
each other and their environment, these uncertainties
propagate into modeling the chemical evolution of the
Universe. We will develop physics that models these
phenomena which we will publish to facilitate com-
parisons with current and future observations.
As just one example of the power of this database,
consider the following to put things into perspective.
Although there is a consensus that the collapse of
a massive stellar core powers a long GRB, the issue
of why some stars produce GRBs and others do not
is still debated [22,23]. Currently, we do not know
which stars produce these powerful explosions [24].
The problem is even more pronounced for the case of
short GRBs where there are literally tens of alterna-
tives to the favored merger model. Short GRBs are
thought to be very distant; one is the most distant ob-
ject in the Universe [25,26]. It is commonly assumed
that these most distant GRBs originate from Popula-
tion III stars, and yet the debate ensued on whether
they were the result of NS-NS/BH-NS mergers or col-
lapsars. Using our synthetic stellar database, we can
generate stellar populations at high redshifts and in-
corporate the uncertainties of primordial star forma-
tion, which will provide astrophysically-based answers
for the origin and nature of these distant explosions
(see Fig. 1, top panel).
1.2. Project Overview
We propose to generate synthetic data on stars and
stellar remnants. The data will be the result of nu-
merical simulations that will include detailed stel-
lar evolutionary calculations and population synthe-
sis predictions. The data will cover known popula-
tions (e.g., high- and low-mass X-ray binaries, double
neutron stars) that could be used for calibration and
comparisons, as well objects that are yet undetected
(e.g., Population III stars and double black holes).
The project will be divided into three specific science
tasks/categories that will be conducted in parallel,
complementing and guiding each other. Each task
will be performed with one or more young team mem-
bers and external advise from a senior scientist from
abroad.
In the following section we outline three key is-
sues that will be studied within the framework of the
Synthetic Universe library during the funding pe-
riod. Of course, we expect the library to extend well
beyond the funding time. The database will be em-
ployed by small and large collaborations, that may or
may not include the local Warsaw team, and our ex-
ternal collaborators around the world. The data that
0 5 10 15 20 25
GRB 090423
GRB 080913
0 5 10 15 20 25
Fig.1 – Bottom panel: Overall star formation
rate (SFR) history: all stars. We also show
the rate we obtained for PopII stars only: the
population is rather different for a slow (solid)
and fast (dashed line) metallicity evolution
model. The models bracket the uncertainty
on metal enrichment in the Universe. PopIII
(first) stars and PopI (most recent) stars form
to the right and left of PopII, respectively. It is
clearly seen that independent of the metallicity
evolution, the majority of the stars are PopII
at z < 10 – this is the redshift of the farthest
spectroscopically confirmed objects in the Uni-
verse: GRB 080913 (z = 6.7) and GRB 090423
(z = 8.1). Top panel: Predicted Swift detection
rates of short and long GRBs. Rates are for
GRBs originating exclusively from PopII stars.
GRB 080913 and GRB 090423 have PopII pro-
genitors, and are both long bursts resulting
from collapsars (Belczynski et al. 2010 [8]).
we will obtain for any of the science projects will
be available in the database for any group to perform
3
their own analyses and studies.
The projects within the database will not be, by
any means, limited to the list discussed below, but
represent specific interests of the team members. At
the moment, the initial results from our library are
already being used by the Einstein Telescope team
for mock data challenges (Tania Regimbau, University
of Nice, France), to study nucleosynthetic yields and
subsequent Galactic metal enrichment from neutron
star mergers (Stephan Rosswog, Albanova University,
Sweden; Thomas Janka, Max Planck Institute for As-
trophysics, Germany) and for studies of Galactic mi-
crolensing on black hole binaries (Jeremy Schnittman,
NASA, USA).
1.3. Ultraluminous X-ray Sources
The ULX is defined as non-nuclear (not a cen-
tral galactic object) X-ray point source that has an
isotropic X-ray luminosity higher than a few times
1039 erg/s. This definition has roots in early X-ray
observations (Einstein, ROSAT, ASCA) that have
revealed a population of unexpectedly bright extra-
galactic sources [27,28]. Since ULXs are point sources
that are not associated with their host galaxy cen-
ters, accreting neutron stars or black holes in close
binary systems seem to be the most natural candi-
dates. However, the X-ray luminosity of these sources
is so high that it exceeds the critical Eddington lumi-
nosity of what was believed to be the maximum mass
of a stellar origin black hole. The Eddington lumi-
nosity can be written as LEdd = 1.3 × 1038(M/M⊙)
erg/s, and it denotes the maximum bolometric lumi-
nosity attained by an accreting object of mass M. For
accreting neutron stars and black holes, most of the
power comes in the X-ray band and thus X-ray lu-
minosity is a good approximation of bolometric lumi-
nosity. The first known stellar black holes were found
in our Galaxy and were determined to have masses
within the range 5−15M⊙ [29]. Therefore, the maxi-
mum luminosity that can be provided by an accreting
black hole (15M⊙) of the stellar origin could reach
only LEdd = 2 × 1039 erg/s. Any accreting source
that exceeds this limit is presently classified as an ex-
ceptional source under the ULX category.
Recent years have brought wealth of information.
Targeted programs with new generation satellites
(Chandra, XMM-Newton, Suzaku, Swift) delivered
detailed ULX population X-ray spectral and timing
properties and initiated multi-band searches for their
counterparts [30]. Recent catalogs contain more than
500 point sources with X-ray luminosities as high as
1039−1042 erg/s [31,32,33]. Some ULXs exhibit strong
X-ray variations on timescales of minutes confirming
their compact nature [34,35]. The combination of
the luminosity and the compactness arguments im-
plies that accreting black holes within the mass range
10 − 10, 000M⊙ are powering ULXs. Although, some
small contamination of the ULX catalogs by other
sources cannot be excluded without extensive obser-
vational follow up. The most likely contamination is
expected from background AGNs (can be found in
entire ULX luminosity range; [36]), young supernovae
(Type IIn may reach 1040 erg/s; [37]) and young X-ray
pulsars (may slightly exceed 1039 erg/s; [38]). That
said, most observational time and theoretical effort is
dedicated to study the majority of sources that are
powered by accreting black holes. On one hand, ob-
servers provide luminosity functions [39], host galaxy
information [32], correlations between the ULX num-
bers and star formation/metallicity [40,41] in addition
to the multi-wavelength temporal spectral character-
istic of sources [42,43,44]. On the other, theorists de-
velop sophisticated models of accreting black holes
that include supercritical inflows, beamed emission,
and outflows that can interact with the surround-
ing medium producing shocked nebulae [45,46,47,48].
However, despite the observational and theoretical ad-
vances the nature of ULX sources remains unknown.
The most straightforward interpretation of the ul-
tra high X-ray luminosities comes from the Eddington
limit. If the limit is in fact not violated it means that
ULXs host black holes with masses in the range 10-
10,000 M⊙. If it is further assumed that stellar black
holes can only reach mass of 10 − 20M⊙ as observed
in the Galaxy, it follows that ULXs host an interme-
diate mass black hole population of unspecified (most
likely dynamical formation) origin [49,50,51]. This
idea became very popular, as it would provide the
missing link between the stellar mass black holes and
supermassive black holes residing in center of various
galaxies. There are three major caveats to this inter-
pretation. First, there are models of â€leaky†accre-
tion disks with photon bubbles that can violate the
Eddington limit by a factor of 10 [52]. Second, the
ULX X-ray luminosity may be overestimated by an-
other factor of 10 due to emission anisotropies caused
by the beaming expected for high accretion rate bi-
naries [53]. Third, the stellar mass black holes can
significantly exceed the typical mass assumed in the
above discussion. It was recently demonstrated that if
updated wind mass loss rates are employed the stars
in low metallicity environment can form black holes
as massive as 80M⊙ [54]. This gives another factor of
about 10 in potential increase of luminosity for stellar
mass black holes. If all the above factors are combined
(increase by factor of 1000), it is potentially pos-
sible, that stellar mass black holes, can explain even
the brightest ULXs (1039 ! 1042 erg/s).
Some of the above ideas were already utilized in
the ULX studies. It was argued for stellar mass black
4
hole origin combining the beaming and leaky disks
to explain the ULX sources with luminosities upto
1041 erg/s [55]. They could not reach the most ex-
treme luminosities (1042 erg/s), since at the time of
their study stellar mass black holes were not known to
have masses over 10 − 20M⊙. It is worth noting that
the same year the so far most massive known stellar
black hole (30M⊙) was discovered in low metallicity
galaxy IC10 [56,57]. A few years later the formation
of this most massive known black hole was explained
and used to predict the maximum mass of stellar black
holes (80M⊙) by our group [54]. A number of ULX
papers followed. In the first approximation, it was
postulated that if a very massive stellar origin black
hole accretes from a binary companion all (or almost
all) ULX sources can be explain without invoking
intermediate mass black holes [40,58,59]. However,
these results were challenged by the detailed popu-
lation synthesis calculations implying that the chance
for the formation of a ULX source consisting of a very
massive stellar origin black hole and a mass transfer-
ring companion is close to nil [60,61]. This brings
back the heated conundrum: do the most luminous
ULX sources have to be powered by intermediate-
mass black holes or could they potentially host stellar
mass black holes?
Here, we propose the final solution to this issue.
We will test whether such evolutionary channels that
produce the brightest ULX sources powered by stellar
origin black holes exist. We will employ the only pop-
ulation synthesis code (StarTrack [62,63]) that at this
moment allows self-consistently for the formation of
the very massive stellar origin black holes. We will in-
corporate the relevant beaming and leaky disk physics
into the underlying physical model and test whether it
is possible to form a stellar black hole ULX with X-ray
luminosity reaching 1042 erg/s. The negative answer
will indicate that indeed the most luminous ULXs,
like 1042 erg/s source HLX-1 in ESO234-49 [64], must
be powered by an intermediate-mass black hole. The
positive answer will allow for the possibility that even
the brightest ULXs may be powered by stellar origin
black holes. Obviously, our proof-of-principle study
will not be able to rule out the intermediate-mass
black hole ULX origin. However, if the answer to
our hypothesis is positive, we will provide a detailed
ULX formation rate estimate. Comparison with the
observed number of ULXs in the local Universe with
our predicted number of stellar origin ULXs will in-
dicate what fraction (potentially all?) of sources are
powered by stellar mass black holes.
The final comment is of somewhat political and of
a sensitive nature. However this point needs to be
made in order to justify the proposed line of the re-
search. There are claims [60,61] that it is unlikely for
a very massive stellar black holes to form a ULX that
are based on StarTrack population synthesis calcula-
tions. These calculations were performed on a sample
of several million massive binaries. Such a sample
corresponds to about 10% of the stellar content of a
large galaxy (e.g., Milky Way). Such a sample may
be considered relevant for most population synthesis
studies, provided that a considered population con-
sists of tens of sources per a large galaxy. This is sat-
isfied for most known stellar populations and such an
approach is readily justified for such cases. However,
this is fundamentally different in the case of ULXs.
For one, there is no ULX in our Galaxy, and it takes
many galaxies to produce just one. In other words,
ULXs are extremely rare objects. Therefore, any pop-
ulation synthesis study that is focused on ULX pop-
ulation, needs to be performed on a very large stellar
sample. This is to ensure that, if in fact ULXs form
from binary stars, the very rare formation channels
are properly identified. One cannot simply perform
calculations on a typical population synthesis sam-
ple (although it may seem very large), find no bright
ULXs (as expected) and then extrapolate the results
to the local Universe. This is why PI of this proposal
(K.Belczynski) removed his name from the author list
of [60]. Here, we propose a suite of massive popula-
tion synthesis calculations that will encompass a num-
ber of large galaxies, and therefore it will be ensured
that the conclusions are statistically and scientifically
meaningful.
We are also planning to undertake a study of very
massive stars within Population I/Population II stars.
Recent observations [65,66] demonstrate that stars
with significant amount of metals (30%Z⊙; LMC) can
reach mass upto 300M⊙. Such a possibility was
never considered in ULX nor gravitational radiation
studies as it defies the current theoretical understand-
ing of star formation (maximum star mass of about
100M⊙). We will extend the Initial Mass Function
(IMF) up to (or over) the currently observed limits to
study the effects of very massive stars on ULX and
double compact object populations. The initial es-
timates, performed within a large international col-
laboration (Duncan Brown: co-chair of LIGO coa-
lescence group, Cole Miller: LIGO/VIRGO oversight
committee, Chris Fryer: Los Alamos astrophysicist;
Alessandra Bunanno: the black hole relativity key-
person + Warsaw Observatory group) indicate that
despite the very low numbers (steep IMF) these very
massive stars forming intermediate-mass black holes
may totally alter the consensus on the impact of the
stellar origin of black holes [67].
1.4. Gravitational Radiation Signature
5
Double compact objects, binaries consisting of black
holes, neutron stars and white dwarfs in various com-
binations, are of special interest in context of current
ground-based (LIGO/VIRGO) and near future (DE-
CIGO, Einstein Telescope, eLISA) space based gravi-
tational radiation observatories. Double compact ob-
jects are the primary sources for ground based inter-
ferometers while they will constitute the foreground
noise for some space instruments, limiting observa-
tions of other objects. In the last decade, several
groups were working on rate predictions and char-
acterization of double compact object populations
[62,68,69]. Without exception, the gravitational wave
predictions were thus far based on observations of
stars in our Galaxy. In particular, the predictions
were based on stars with solar chemical composition –
typical for Milky Way thin disk population. However,
even the currently operating instruments can reach far
beyond Milky Way. In this local volume, recent deep
observations of tens of thousands of nearby galaxies
demonstrated that as much as 50% of local star
formation is occurring in low metallicity environments
( 20%Z⊙ [7]). This fact is not yet fully recognized in
the community, but the implications of this finding are
critical. As discussed earlier, metallicity has a strong
impact on the final mass of compact objects, espe-
cially black holes (see Fig.2). If half of the star form-
ing mass is producing heavy black holes the predicted
detection rates can increase by more than an order of
magnitude [70] and direct detection of gravitational
waves may be much easier than previously believed.
This is supported further by empirical estimates of
BH-BH merger rates [71] and have rather deep im-
plications the instrument (LIGO/VIRGO) develop-
ment. Motivated by our estimates [70,71] Ligo Sci-
entific Collaboration have already designed and car-
ried out a specific signal search on the last (S6) initial
LIGO/VIRGO science run [72]. At the moment both
observatories are being upgraded to a higher sensi-
tivity, and our results indicate that the engineering
runs (ones that are planned before reaching targeted
sensitivity) will either result in detections or provide
robust constraints on stellar evolution. This calls for
a different approach to data analysis for the first ad-
vanced LIGO/VIRGO observations.
Our theoretical predictions are supported by re-
cent empirical estimates [71,73,74,75]. At the moment
these are the only estimates that take into account
a realistic BH/NS mass spectrum, supernovae explo-
sions, common envelope evolution and properly con-
sider the metallicity distribution in the local Universe
[76]. Based on our calculations, we argue that the
science that can be done with gravitational wave de-
tections shifts toward a new direction and we have al-
ready begun providing some very clear-cut examples
of what can be learned from the first gravitational
wave detections. In particular, we have demonstrated
that the detection of a single NS-NS binary within
Fig.2 – The initial-remnant mass relation for
a StarTrack single star evolution for two wind
prescriptions: the previously used Hurley et
al. winds and newly adopted (modified) Vink
et al. winds. Top panel: Solar metallicity.
This corresponds to the stellar field popula-
tions in the Galaxy. The predicted maximum
black hole mass Mbh
,
max 15M⊙ for new and
10M⊙ for old winds is consistent with the
most massive stellar black holes observed in
our Galaxy (e.g., in GRS 1915 Mbh = 14±4M⊙).
Middle panel: Moderate metallicity. This cor-
responds to stellar populations in galaxy IC10,
which hosts the most massive known stellar
black hole (Mbh = 23 − 34M⊙). Note that the
predicted maximum black hole mass Mbh
,
max
30M⊙ for the new winds is consistent with the
measurement in IC10, while Mbh
,
max 15M⊙
obtained for old winds appears to be signif-
icantly too small. Bottom panel: Very low
metallicity. This corresponds to stellar popu-
lations of Galactic globular clusters, or proto-
galaxies. The maximum black hole mass may
reach Mbh
,
max 80M⊙ for the newly adopted
wind prescriptions (Belczynski et al. 2010
[54].)
the first 10 detections will put very strong constrains
on BH natal kicks [77]. Additionally, the available
electromagnetic data on known BH and NS masses is
6
inconclusive as to the existence of the infamous â€mass
gap†(the dearth of compact objects in 2−5M⊙ mass
range). It is not clear whether this gap is the effect
of some observational biases or rather reflects on the
physics of the supernova explosion engine [78]. In a
recent study, we have identified which part of the su-
pernova explosion physics may be responsible for the
mass gap and we have indicated that the final answer
will be provided by the unbiased gravitational wave
measurements of compact object masses [79].
We will provide a qualitatively new description of
double compact object populations. The synthetic
populations of double compact objects formed at vari-
ous host galaxy environments (range of galaxy masses,
metallicities and star formation histories) will become
part of our library. Our model allows for evolution of
metal rich stars (Population I and II; [63]) as well as
first metal-free stars (Population III; [80,81]). Syn-
thetic data will be provided not only as a function of
a given host galaxy type, but also as a function of red-
shift under an adopted cosmological model that incor-
porates chemical evolution [8]. The data will include
not only basic parameters for double compact ob-
jects (component masses or separations/eccentricities
at formation) but also very specific parameters (spins:
magnitude and their direction [106] or position within
the host galaxy [82]) that can allow for a number of
applications not only within gravitational wave as-
tronomy but also crucial for gamma-ray burst or X-
ray binary studies. At the moment, a number of US-
based scientists are using StarTrack models to inter-
pret Chandra observations [83,84,85,86]. At the same
time I am leading efforts together with Harvard (Edo
Berger) and Boulder JILA groups (Rosalba Perna) to
characterize populations of transients (Palomar Tran-
sient Factory) and Gamma-ray bursts (Swift) with the
use of our models.
Our results will allow for rate/parameter estimates
specific to a given gravitational wave observatory and
will not be limited to LIGO/VIRGO but will be di-
rectly applicable to any future observatory that is des-
ignated to search for double compact objects at arbi-
trary redshifts. This approach will facilitate the un-
derstanding of what science can be done with gravita-
tional radiation detections with existing and planned
observatories.
1.5. Type Ia Supernovae
Supernova explosions are the fate of all massive
stars: these generate the core-collapse supernovae
(SNe). SN explosions are also the final stage for a less
massive binary system after one of the stars evolves
into a carbon-oxygen white dwarf. These generate
Type Ia SNe (SNe Ia). All SNe play a critical role
in driving the galactic gas ecosystem, in star forma-
tion by heating and compressing interstellar gas, and
in chemically enriching galaxies. SNe are responsi-
ble for releasing radioactive and heavy nuclei into in-
terstellar space and spreading the building blocks of
other stars, planets, and ultimately, life. Additionally,
observations of SNe Ia have led to the breakthrough
discovery that the expansion rate of the Universe is
accelerating, due to a mysterious “Dark Energy†[87].
Despite their use as the most important “standard
candle†distance indicators, the origin of SNe Ia re-
mains unknown. The long-standing paradigm [88] is
that their progenitors consist of a white dwarf that ac-
cretes enough matter from its binary companion such
that its mass exceeds the Chandrasekhar mass limit
(1.4 M⊙). However, the nature of the companion
star, the efficiency of matter accumulation onto the
white dwarf, and the explosion mechanism are all thus
far unclear [89].
To date, the two most favored scenarios thought
to lead to SNe Ia are the Single Degenerate Scenario
(stable mass accretion on to a white dwarf from a
normal star [90,91]) and the Double Degenerate Sce-
nario (the merger of two massive white dwarf stars
[92,93]). However, other promising formation scenar-
ios exist [94], and support for more than one forma-
tion scenario is substantiated by recent SNe Ia obser-
vations. Several different sub-classes of SNe Ia have
been identified [95], and recent advances in theoret-
ical SN modeling indicate that there is likely more
than one formation scenario (and explosion mecha-
nism) contributing to the observed sample of SNe Ia
[96]. This has far-reaching implications for the the-
ory of Dark Energy: if the explosion mechanism that
leads to the formation of a SN Ia is dependent on
evolutionary scenario and/or metallicity, this could
potentially pose a challenge for cosmological studies
by introducing systematic errors [97]. Thus, deter-
mining the nature of SN Ia progenitors – both in the
local and the most distant Universe – is one of the
most important long-standing puzzles in astrophysics
that remains to be solved.
The delay time distribution (DTD) is the distri-
bution of times in which SNe Ia explode following a
(hypothetical) burst of star formation. Knowing the
DTD gives the age of the progenitor, which places
strong constraints on the different proposed formation
scenarios. If the SN Ia rate is known in addition, then
it becomes possible to rule out theoretically-predicted
evolutionary channels.
Today, we are uniquely primed to address the ques-
tion of SN Ia origin. This has already been done
to some extent for Local Universe-like stellar popu-
lations [13]. Excitingly, now we are poised to deter-
mine how well SNe Ia can be considered standard can-
dles [98], and quantify to what degree their rates and
7
delay times vary as a function of redshift and metal-
licity. In order to accomplish this, in the next step
we will combine the theoretical rates (quantity) and
delay times (feedback timescale) with nucleosynthetic
yield predictions for the leading SN Ia formation sce-
narios as a function of redshift (metallicity). This
requires generating a suite of binary evolution models
for a grid of stellar metallicities, computing explo-
sion models with the most up-to-date nucleosynthetic
yield predictions, and calculating synthetic spectra
and light-curves. A complete pipeline is already in
place with the collaborators in our existing network:
from binary star birth/evolution of SN Ia progeni-
tors (A. Ruiter+K.Belczynski) to explosion model-
ing (MPA, Wurzburg, Heidelberg) to synthetic spec-
tra calculation (MPA, Mt. Stromlo, LANL). Such
a pipeline has already been extensively tested, and
shown successful in validating the importance of some
of the most promising SNe Ia formation scenarios
[99,100].
These efforts will enable us to assess the viability of
various explosion models, and gain insight into the ori-
gin of different subclasses of SNe Ia: ‘normal’ SN Ia,
sub-luminous and super-luminous SNe Ia, and other
peculiar objects (e.g., 2002cx-likes). The Synthetic
Universe library will contain detailed information on
the formation history of the most promising SN Ia
progenitors as computed by StarTrack. Addition-
ally, delay times and formation histories will also be
recorded for core-collapse SNe; some of which are ex-
pected to be synonymous with GRBs. The synthetic
data on delay times and rates of the various SNe will
serve as input for the NuGrid collaboration (see 3.2) to
obtain estimates of chemical enrichment scenarios for
various galaxy types. Such a long-term project would
involve convolving theoretical delay times for all stel-
lar explosions with galaxy mass distribution functions
as a function of redshift, and imposing a galaxy mass-
metallicity dependence [5]. We will thereby provide
important input data for chemical evolution models
by supplying the relative rates (and DTDs) of various
stellar explosions on cosmological scales.
2. DISCIPLINE CONTRIBUTION
We will deliver the first library of synthetic stel-
lar populations to the astro community. At first
we will provide the data on the populations of GR
sources (NS-NS, BH-NS, BH-BH), ULX binary can-
didates (BH + stellar type companion) and potential
progenitors of Type Ia supernovae (various configu-
rations of accreting white dwarfs in binary systems).
The database will be maintained beyond the period
of funding and new projects will be added as the need
arises. Our contribution will extend beyond the scien-
tific community. We will actively involve the public in
our research. We will provide a platform for the pub-
lic to get involved in our numerical simulations and
at the same time we will invest in education and our
project advertisement to a wider audience through the
Internet.
3. SIGNIFICANCE
We propose to tackle three very specific stellar re-
search projects at the frontier of stellar astrophysics.
We will provide a proof-of-principle study of the ori-
gin of ultraluminous X-ray sources and we will show
whether stellar mass black holes can or cannot ex-
plain the observed populations of these objects. We
will fully describe the populations of double com-
pact objects (NS-NS, BH-NS, BH-BH) that are the
most promising sources for gravitational-wave detec-
tors. LIGO and VIRGO observatories will start tak-
ing data around 2015 and we will provide a scientific
rationale behind the near-future detection of double
compact object mergers and help to guide the signal
search techniques. We will study various potential
progenitors of Type Ia supernovae in a quest to finally
identify which binaries are specifically responsible for
these explosions. Our synthetic models will help to
eliminate some of the proposed models, while other
models will gain in significance, driving progenitor re-
search. We will compare our predictions with the ob-
served supernovae rates, delay times, light curves and
spectra.
4. WORK PLAN
4.1. Database Development
The database will be placed at a local server main-
tained by our group, and will be visible and accessible
from the outside through the Synthetic Universe
website. The database will consist of electronic data
tables accompanied by detailed descriptions of each
table’s content. For each simulation from either the
evolutionary code or population synthesis code we will
post the raw data with a full description of the sim-
ulation. A detailed code manual will be placed up
front for each code, as well as a general overview of
the database and usage instructions. The raw data
files will serve as consistency checks and comparisons
for groups that prefer to do the post-processing on
their own.
Post-processing involves a number of computational
tasks that are very often required to present synthetic
data in a useful way. For example, it is important
to generate a synthetic spectrum to compare to an
observed galaxy. To proceed, we will use a popula-
tion synthesis code to obtain the entire stellar pop-
ulation in this galaxy. For each star, we will then
obtain a spectrum from one of the available stellar
spectra libraries, and co-add the spectra. These tasks
8
all require extra calculations, and external libraries.
These tasks will be performed by analysis codes that
we will create and distribute to address our specific
data needs. In this example, the final product would
be the spectrum of a synthetic galaxy that could be
downloaded from our server with the full description
of how it was generated.
As the database expands, we will adapt and im-
prove our analysis codes to serve as interactive re-
search tools. In the above example, a need may arise
to combine stellar population data to model unfore-
seen future observations. Since we will have a full
library of raw stellar population data within the first
years of opening the database, we will also provide
interactive software that can allow users to combine
the data in novel ways. This can include, but is not
limited to: building galaxy models and changing cos-
mological parameters. Finally, we will provide visual-
ization tools that will specifically work on information
contained within the database, requesting time snap-
shots, imaging, and rapid sampling of the synthetic
data. The first pilot project is already available on-
line at: http://www.syntheticuniverse.org
4.2. Specific Schedule
1) First year. We will start preparations for all
three science projects. That will include search of
the available literature, contacting other groups that
work on the similar subjects and we will establish
background for our calculations. We will prepare
our physical model (extra input physics, calibrations
and updates) for all three projects. In particular, we
will use the MESA code to updated criteria on com-
mon envelope development, we will implement the
primary effects of rotation into StarTrack based on
the Geneva group results, we will add various, less
thoroughly-explored though promising SN Ia models
(e.g., sub-Chandrasekhar mass merger models [101],
violent white dwarf mergers [102], Symbiotic channel
[103], double-detonation scenarios [104], thick wind
prescription [105]) and we will expand our code for rel-
evant ULX physics (leaky disks, anisotropic emission,
very massive stars). We will create and test inter-
nal (Warsaw Observatory) computing network within
the framework of the Universe@home program. We
will apply for computer time at various open Polish
computational centers.
2) Second year. We will run numerical population
synthesis simulations relevant to all three projects.
The data will be stored at first on our local disks. At
the same time we will work on the development of our
websites (www.syntheticuniverse.org) and we will
get it ready for the arrival of the synthetic data. The
simulations will be run at US and Polish Research In-
stitution computer clusters, and (if granted) in Polish
open computational centers. And we will launch our
Universe@home program and open it for the public.
We will begin assessing the changes imposed by effects
of rotation on relevant stellar populations.
3) Third year. The numerical simulations will con-
tinue. We expect that more and more computa-
tional resources will be acquired with time via our
Universe@home program. The first data will be
placed in our www.syntheticuniverse.org database
and made publicly available for other research groups.
We will carry out the first analysis of our data for all
three projects and write up and submit our initial
conclusions to international journals. For practical
reasons, the GR papers will be submitted to Astro-
physical Journal, SN Ia papers to Monthly Notices of
the Royal Astronomical Society and ULX studies to
Astronomy & Astrophysics. We will start incorpo-
rating MESA models into the StarTrack code as an
optional tool for comparative simulations.
4) Fourth year. The numerical simulations will con-
tinue and at the end of this year the simulations will
be completed. The full data sets will be placed and
described in our library. We will start the final analy-
sis and progress with the three research projects. We
will continue incorporating the Geneva models into
our population synthesis. Furthermore, we will obtain
the first population synthesis simulations that include
full effects of rotation from the MESA code.
5) Fifth year. Available computer power will be
used to obtain extra simulations that we will deem
required during our studies or that are requested by
the astro community or reviewers of our submitted
manuscripts. The entire group will focus on detailed
analysis of our results and comparisons of our sim-
ulations with (available at this stage) observations.
We will also compare our results obtained with ro-
tating and non-rotating stellar models. We will be
writing our findings into a series of papers. Students
and graduate students will be using these results to
progress their theses. At the same time we will be
already planning further future extensions of our li-
brary. By that time we expect to have major feedback
from various groups working on the related subjects
and this will help us guide our future database devel-
opment.
5. METHODOLOGY
5.1. Stellar Modeling
Although the theory of stellar evolution is quite
a mature subject, it has still to face big questions.
Among them: How do massive stars form? How does
stellar evolution influence star formation? What was
the nature and the characteristics of the first genera-
tions of stars in the Universe? What was the role of
these first generations in the reionization of the Uni-
9
verse? What physics drives the variations in the stel-
lar populations observed in different environments?
The answers to all these questions require improv-
ing stellar models. These improvements are obtained
by including important effects, neglected until re-
cently, such as the effects of the stellar winds, the
internal transport processes induced by axial rotation
or by other instabilities such as the thermohaline mix-
ing, the transport of angular momentum by the grav-
ity waves and/or by a magnetic field. All these effects
have been implemented in the Geneva and Bill Paxton
stellar models and have been tested by a whole series
of observational features. These comparisons all sup-
port the view that extra mixing processes do indeed
occur in stars, and that models accounting for them
can much better reproduce the observations than the
models neglecting them. Including these transport
mechanisms have deep impacts on the stellar popula-
tions expected at a given age and metallicity as well
as on the nucleosynthesis of the elements. Many of
these effects can be tested using the new windows
which are now opening on stellar interiors through
astroseismology (e.g., COROT, KEPLER). Additionally,
the huge number of stars that will be observed by the
satellite GAIA will allow to probe very short evolution-
ary phases which so far were only in the realm of the-
ory. Heavy mass loss of material by stellar winds, like
in the Luminous Blue Variable phase, may be good
example of short phenomena that may very deeply
affect the end of the stellar life, the nucleosynthetic
outputs, the supernova type and the nature of the
stellar remnant. GAIA observations will lead to strong
revisions of the stellar evolution in many parts of the
Hertzsprung-Russel diagram.
We will require two type of codes (i) population syn-
thesis code and (ii) detailed stellar evolutionary code.
Though these codes can work independently, ideally
they should work together, delivering feedback from
one code to another. The end product of such an
approach will be a comprehensive and self-consistent
picture of stellar populations at a brand new level,
both quantitatively in terms the of a data library, and
qualitatively in terms of evaluating the input physics.
UNIVERSE@HOME: ULX’S, GR SOURCES, SNIA PROGENITORS
PI: Krzysztof Belczynski (Warsaw University)
This proposal aims to create the first database of the simulated stellar content of the Universe,
from the earliest stars to the most exotic black hole binaries. This comprehensive stellar library will
be a world-wide scientific resource for astronomers, and an educational asset for the public. With
a state-of-the-art suite of computational tools, we can calculate the physical properties of individual
stars, evolve them in time, and track how they interact with each other as they age. With this library
of stellar populations that spans the known range of star formation histories and initial elemental
abundances, it is possible to tackle fundamental science questions that have been out of reach in
general relativity, astronomy, and cosmology. As a group we will use this database to characterize the
population of supernova progenitor stars that have been used to discover the mysterious dark energy
that accelerates the expansion of the Universe; we can calculate how (if at all) binary evolution
facilitates the formation of ultra-luminous X-ray sources, and we can model the number and physical
properties of neutron stars and black holes that strongly emit gravitational waves that are at the
heart of research into gamma-ray bursts – the most energetic explosions known-to-date. The data
that we create can be mined not only to match current observations, but to predict how the sky
will appear to future astronomical observatories as new windows to the Universe open. Altogether,
this proposal will allow us to consolidate a strong and highly visible research group; an independent
computational center that trains a new cohort of highly educated students, and communicates with
the public via innovative science venues, and offer a fresh look at the Universe.
1. OBJECTIVES
1.1. Context
Stars are the basic building blocks of galaxies that,
in turn, make up the bulk of the baryon content of
the Universe. The first stars formed very early—
roughly 100 Myr after the Big Bang —out of the pri-
mordial hydrogen-helium mix. These stars, referred
to as Population III stars, were likely extremely mas-
sive (M 1000M⊙) and produced the first heavy
elements through nuclear burning and supernova ex-
plosions. These first supernovae created black holes
with masses between 10−1000M⊙ [1,2,3]. Thus, these
first stars provided not only the seed black holes for
the eventual formation of supermassive black holes
(Mbh 106 − 109M⊙), but also the metals that we
see in the second generation of stars. These stars,
known as Population II stars, began forming after the
Population III era from gas that was already polluted
with heavy elements [4]. Still, the Sun has 104 times
more metals than this population ( Z > 10−4Z⊙).
Population II stars most likely generated the photons
necessary to re-ionize the Universe, and set the chem-
ical composition of the Universe we observe today.
At about 1-2 Gyr after the Big Bang, Population
I stars began forming. These stars have metallicities
comparable to that of the Sun, and presently domi-
nate the overall star formation in the Universe [5,6].
Although most stars today are in this category, this
population is in no way a uniform sample. For exam-
ple, the metallicity varies widely: about 50% of star
formation in the past Gyr occurs at Z 0.2Z⊙, while
the remaining half is found at solar composition or
higher (Z > Z⊙) [7]. In the bottom panel of Figure
1, we show a quantitative picture of star formation
history with its uncertainties [8].
The environment of Population I stars varries as
well. Typically, stars most commonly form as single
objects or in binaries, although higher multiplets are
known to exist. Even the dynamics of these stars are
different: In galaxies, most stars do not directly inter-
act with other stars – these are the field population.
However, a small number of stars form in dense stellar
clusters, and here dynamical interactions such as col-
lisions are frequent [9,10]. The fact that today’s stars
form and evolve in such varied circumstances means
that they produce a wealth of information that probes
an extraordinary range of of physics. Stars emit light
across the entire electromagnetic spectrum, from ra-
dio to hard gamma rays, and even generate gravi-
tational radiation. Various stages of stellar evolution
are connected with explosive events: X-ray flares from
magnetic low mass stars [11]; luminous blue variable
eruptions from massive single stars [12]; type Ia su-
pernovae from binary white dwarfs [13]; X-ray bursts
from binary neutron stars [14]; gravitational radiation
bursts from mergers of neutron stars and black holes
[15]; long gamma-ray bursts (GRBs) from massive,
rapidly rotating stars [16], and short GRBs from dou-
ble neutron stars and black hole-neutron star mergers
[17].
A number of stellar phenomena are still poorly un-
derstood and repeatedly defy our attempts to explain
1
2
them. Our library of models will help us to solve, or
at least constrain these mysteries. At the most basic
level, astronomers still lack a full description of star
formation [18], convection [19], mass loss [20] and su-
pernova explosions [21]. Since the amount of heavy el-
ements released in stellar winds, supernovae and com-
pact object mergers change how stars interact with
each other and their environment, these uncertainties
propagate into modeling the chemical evolution of the
Universe. We will develop physics that models these
phenomena which we will publish to facilitate com-
parisons with current and future observations.
As just one example of the power of this database,
consider the following to put things into perspective.
Although there is a consensus that the collapse of
a massive stellar core powers a long GRB, the issue
of why some stars produce GRBs and others do not
is still debated [22,23]. Currently, we do not know
which stars produce these powerful explosions [24].
The problem is even more pronounced for the case of
short GRBs where there are literally tens of alterna-
tives to the favored merger model. Short GRBs are
thought to be very distant; one is the most distant ob-
ject in the Universe [25,26]. It is commonly assumed
that these most distant GRBs originate from Popula-
tion III stars, and yet the debate ensued on whether
they were the result of NS-NS/BH-NS mergers or col-
lapsars. Using our synthetic stellar database, we can
generate stellar populations at high redshifts and in-
corporate the uncertainties of primordial star forma-
tion, which will provide astrophysically-based answers
for the origin and nature of these distant explosions
(see Fig. 1, top panel).
1.2. Project Overview
We propose to generate synthetic data on stars and
stellar remnants. The data will be the result of nu-
merical simulations that will include detailed stel-
lar evolutionary calculations and population synthe-
sis predictions. The data will cover known popula-
tions (e.g., high- and low-mass X-ray binaries, double
neutron stars) that could be used for calibration and
comparisons, as well objects that are yet undetected
(e.g., Population III stars and double black holes).
The project will be divided into three specific science
tasks/categories that will be conducted in parallel,
complementing and guiding each other. Each task
will be performed with one or more young team mem-
bers and external advise from a senior scientist from
abroad.
In the following section we outline three key is-
sues that will be studied within the framework of the
Synthetic Universe library during the funding pe-
riod. Of course, we expect the library to extend well
beyond the funding time. The database will be em-
ployed by small and large collaborations, that may or
may not include the local Warsaw team, and our ex-
ternal collaborators around the world. The data that
0 5 10 15 20 25
GRB 090423
GRB 080913
0 5 10 15 20 25
Fig.1 – Bottom panel: Overall star formation
rate (SFR) history: all stars. We also show
the rate we obtained for PopII stars only: the
population is rather different for a slow (solid)
and fast (dashed line) metallicity evolution
model. The models bracket the uncertainty
on metal enrichment in the Universe. PopIII
(first) stars and PopI (most recent) stars form
to the right and left of PopII, respectively. It is
clearly seen that independent of the metallicity
evolution, the majority of the stars are PopII
at z < 10 – this is the redshift of the farthest
spectroscopically confirmed objects in the Uni-
verse: GRB 080913 (z = 6.7) and GRB 090423
(z = 8.1). Top panel: Predicted Swift detection
rates of short and long GRBs. Rates are for
GRBs originating exclusively from PopII stars.
GRB 080913 and GRB 090423 have PopII pro-
genitors, and are both long bursts resulting
from collapsars (Belczynski et al. 2010 [8]).
we will obtain for any of the science projects will
be available in the database for any group to perform
3
their own analyses and studies.
The projects within the database will not be, by
any means, limited to the list discussed below, but
represent specific interests of the team members. At
the moment, the initial results from our library are
already being used by the Einstein Telescope team
for mock data challenges (Tania Regimbau, University
of Nice, France), to study nucleosynthetic yields and
subsequent Galactic metal enrichment from neutron
star mergers (Stephan Rosswog, Albanova University,
Sweden; Thomas Janka, Max Planck Institute for As-
trophysics, Germany) and for studies of Galactic mi-
crolensing on black hole binaries (Jeremy Schnittman,
NASA, USA).
1.3. Ultraluminous X-ray Sources
The ULX is defined as non-nuclear (not a cen-
tral galactic object) X-ray point source that has an
isotropic X-ray luminosity higher than a few times
1039 erg/s. This definition has roots in early X-ray
observations (Einstein, ROSAT, ASCA) that have
revealed a population of unexpectedly bright extra-
galactic sources [27,28]. Since ULXs are point sources
that are not associated with their host galaxy cen-
ters, accreting neutron stars or black holes in close
binary systems seem to be the most natural candi-
dates. However, the X-ray luminosity of these sources
is so high that it exceeds the critical Eddington lumi-
nosity of what was believed to be the maximum mass
of a stellar origin black hole. The Eddington lumi-
nosity can be written as LEdd = 1.3 × 1038(M/M⊙)
erg/s, and it denotes the maximum bolometric lumi-
nosity attained by an accreting object of mass M. For
accreting neutron stars and black holes, most of the
power comes in the X-ray band and thus X-ray lu-
minosity is a good approximation of bolometric lumi-
nosity. The first known stellar black holes were found
in our Galaxy and were determined to have masses
within the range 5−15M⊙ [29]. Therefore, the maxi-
mum luminosity that can be provided by an accreting
black hole (15M⊙) of the stellar origin could reach
only LEdd = 2 × 1039 erg/s. Any accreting source
that exceeds this limit is presently classified as an ex-
ceptional source under the ULX category.
Recent years have brought wealth of information.
Targeted programs with new generation satellites
(Chandra, XMM-Newton, Suzaku, Swift) delivered
detailed ULX population X-ray spectral and timing
properties and initiated multi-band searches for their
counterparts [30]. Recent catalogs contain more than
500 point sources with X-ray luminosities as high as
1039−1042 erg/s [31,32,33]. Some ULXs exhibit strong
X-ray variations on timescales of minutes confirming
their compact nature [34,35]. The combination of
the luminosity and the compactness arguments im-
plies that accreting black holes within the mass range
10 − 10, 000M⊙ are powering ULXs. Although, some
small contamination of the ULX catalogs by other
sources cannot be excluded without extensive obser-
vational follow up. The most likely contamination is
expected from background AGNs (can be found in
entire ULX luminosity range; [36]), young supernovae
(Type IIn may reach 1040 erg/s; [37]) and young X-ray
pulsars (may slightly exceed 1039 erg/s; [38]). That
said, most observational time and theoretical effort is
dedicated to study the majority of sources that are
powered by accreting black holes. On one hand, ob-
servers provide luminosity functions [39], host galaxy
information [32], correlations between the ULX num-
bers and star formation/metallicity [40,41] in addition
to the multi-wavelength temporal spectral character-
istic of sources [42,43,44]. On the other, theorists de-
velop sophisticated models of accreting black holes
that include supercritical inflows, beamed emission,
and outflows that can interact with the surround-
ing medium producing shocked nebulae [45,46,47,48].
However, despite the observational and theoretical ad-
vances the nature of ULX sources remains unknown.
The most straightforward interpretation of the ul-
tra high X-ray luminosities comes from the Eddington
limit. If the limit is in fact not violated it means that
ULXs host black holes with masses in the range 10-
10,000 M⊙. If it is further assumed that stellar black
holes can only reach mass of 10 − 20M⊙ as observed
in the Galaxy, it follows that ULXs host an interme-
diate mass black hole population of unspecified (most
likely dynamical formation) origin [49,50,51]. This
idea became very popular, as it would provide the
missing link between the stellar mass black holes and
supermassive black holes residing in center of various
galaxies. There are three major caveats to this inter-
pretation. First, there are models of â€leaky†accre-
tion disks with photon bubbles that can violate the
Eddington limit by a factor of 10 [52]. Second, the
ULX X-ray luminosity may be overestimated by an-
other factor of 10 due to emission anisotropies caused
by the beaming expected for high accretion rate bi-
naries [53]. Third, the stellar mass black holes can
significantly exceed the typical mass assumed in the
above discussion. It was recently demonstrated that if
updated wind mass loss rates are employed the stars
in low metallicity environment can form black holes
as massive as 80M⊙ [54]. This gives another factor of
about 10 in potential increase of luminosity for stellar
mass black holes. If all the above factors are combined
(increase by factor of 1000), it is potentially pos-
sible, that stellar mass black holes, can explain even
the brightest ULXs (1039 ! 1042 erg/s).
Some of the above ideas were already utilized in
the ULX studies. It was argued for stellar mass black
4
hole origin combining the beaming and leaky disks
to explain the ULX sources with luminosities upto
1041 erg/s [55]. They could not reach the most ex-
treme luminosities (1042 erg/s), since at the time of
their study stellar mass black holes were not known to
have masses over 10 − 20M⊙. It is worth noting that
the same year the so far most massive known stellar
black hole (30M⊙) was discovered in low metallicity
galaxy IC10 [56,57]. A few years later the formation
of this most massive known black hole was explained
and used to predict the maximum mass of stellar black
holes (80M⊙) by our group [54]. A number of ULX
papers followed. In the first approximation, it was
postulated that if a very massive stellar origin black
hole accretes from a binary companion all (or almost
all) ULX sources can be explain without invoking
intermediate mass black holes [40,58,59]. However,
these results were challenged by the detailed popu-
lation synthesis calculations implying that the chance
for the formation of a ULX source consisting of a very
massive stellar origin black hole and a mass transfer-
ring companion is close to nil [60,61]. This brings
back the heated conundrum: do the most luminous
ULX sources have to be powered by intermediate-
mass black holes or could they potentially host stellar
mass black holes?
Here, we propose the final solution to this issue.
We will test whether such evolutionary channels that
produce the brightest ULX sources powered by stellar
origin black holes exist. We will employ the only pop-
ulation synthesis code (StarTrack [62,63]) that at this
moment allows self-consistently for the formation of
the very massive stellar origin black holes. We will in-
corporate the relevant beaming and leaky disk physics
into the underlying physical model and test whether it
is possible to form a stellar black hole ULX with X-ray
luminosity reaching 1042 erg/s. The negative answer
will indicate that indeed the most luminous ULXs,
like 1042 erg/s source HLX-1 in ESO234-49 [64], must
be powered by an intermediate-mass black hole. The
positive answer will allow for the possibility that even
the brightest ULXs may be powered by stellar origin
black holes. Obviously, our proof-of-principle study
will not be able to rule out the intermediate-mass
black hole ULX origin. However, if the answer to
our hypothesis is positive, we will provide a detailed
ULX formation rate estimate. Comparison with the
observed number of ULXs in the local Universe with
our predicted number of stellar origin ULXs will in-
dicate what fraction (potentially all?) of sources are
powered by stellar mass black holes.
The final comment is of somewhat political and of
a sensitive nature. However this point needs to be
made in order to justify the proposed line of the re-
search. There are claims [60,61] that it is unlikely for
a very massive stellar black holes to form a ULX that
are based on StarTrack population synthesis calcula-
tions. These calculations were performed on a sample
of several million massive binaries. Such a sample
corresponds to about 10% of the stellar content of a
large galaxy (e.g., Milky Way). Such a sample may
be considered relevant for most population synthesis
studies, provided that a considered population con-
sists of tens of sources per a large galaxy. This is sat-
isfied for most known stellar populations and such an
approach is readily justified for such cases. However,
this is fundamentally different in the case of ULXs.
For one, there is no ULX in our Galaxy, and it takes
many galaxies to produce just one. In other words,
ULXs are extremely rare objects. Therefore, any pop-
ulation synthesis study that is focused on ULX pop-
ulation, needs to be performed on a very large stellar
sample. This is to ensure that, if in fact ULXs form
from binary stars, the very rare formation channels
are properly identified. One cannot simply perform
calculations on a typical population synthesis sam-
ple (although it may seem very large), find no bright
ULXs (as expected) and then extrapolate the results
to the local Universe. This is why PI of this proposal
(K.Belczynski) removed his name from the author list
of [60]. Here, we propose a suite of massive popula-
tion synthesis calculations that will encompass a num-
ber of large galaxies, and therefore it will be ensured
that the conclusions are statistically and scientifically
meaningful.
We are also planning to undertake a study of very
massive stars within Population I/Population II stars.
Recent observations [65,66] demonstrate that stars
with significant amount of metals (30%Z⊙; LMC) can
reach mass upto 300M⊙. Such a possibility was
never considered in ULX nor gravitational radiation
studies as it defies the current theoretical understand-
ing of star formation (maximum star mass of about
100M⊙). We will extend the Initial Mass Function
(IMF) up to (or over) the currently observed limits to
study the effects of very massive stars on ULX and
double compact object populations. The initial es-
timates, performed within a large international col-
laboration (Duncan Brown: co-chair of LIGO coa-
lescence group, Cole Miller: LIGO/VIRGO oversight
committee, Chris Fryer: Los Alamos astrophysicist;
Alessandra Bunanno: the black hole relativity key-
person + Warsaw Observatory group) indicate that
despite the very low numbers (steep IMF) these very
massive stars forming intermediate-mass black holes
may totally alter the consensus on the impact of the
stellar origin of black holes [67].
1.4. Gravitational Radiation Signature
5
Double compact objects, binaries consisting of black
holes, neutron stars and white dwarfs in various com-
binations, are of special interest in context of current
ground-based (LIGO/VIRGO) and near future (DE-
CIGO, Einstein Telescope, eLISA) space based gravi-
tational radiation observatories. Double compact ob-
jects are the primary sources for ground based inter-
ferometers while they will constitute the foreground
noise for some space instruments, limiting observa-
tions of other objects. In the last decade, several
groups were working on rate predictions and char-
acterization of double compact object populations
[62,68,69]. Without exception, the gravitational wave
predictions were thus far based on observations of
stars in our Galaxy. In particular, the predictions
were based on stars with solar chemical composition –
typical for Milky Way thin disk population. However,
even the currently operating instruments can reach far
beyond Milky Way. In this local volume, recent deep
observations of tens of thousands of nearby galaxies
demonstrated that as much as 50% of local star
formation is occurring in low metallicity environments
( 20%Z⊙ [7]). This fact is not yet fully recognized in
the community, but the implications of this finding are
critical. As discussed earlier, metallicity has a strong
impact on the final mass of compact objects, espe-
cially black holes (see Fig.2). If half of the star form-
ing mass is producing heavy black holes the predicted
detection rates can increase by more than an order of
magnitude [70] and direct detection of gravitational
waves may be much easier than previously believed.
This is supported further by empirical estimates of
BH-BH merger rates [71] and have rather deep im-
plications the instrument (LIGO/VIRGO) develop-
ment. Motivated by our estimates [70,71] Ligo Sci-
entific Collaboration have already designed and car-
ried out a specific signal search on the last (S6) initial
LIGO/VIRGO science run [72]. At the moment both
observatories are being upgraded to a higher sensi-
tivity, and our results indicate that the engineering
runs (ones that are planned before reaching targeted
sensitivity) will either result in detections or provide
robust constraints on stellar evolution. This calls for
a different approach to data analysis for the first ad-
vanced LIGO/VIRGO observations.
Our theoretical predictions are supported by re-
cent empirical estimates [71,73,74,75]. At the moment
these are the only estimates that take into account
a realistic BH/NS mass spectrum, supernovae explo-
sions, common envelope evolution and properly con-
sider the metallicity distribution in the local Universe
[76]. Based on our calculations, we argue that the
science that can be done with gravitational wave de-
tections shifts toward a new direction and we have al-
ready begun providing some very clear-cut examples
of what can be learned from the first gravitational
wave detections. In particular, we have demonstrated
that the detection of a single NS-NS binary within
Fig.2 – The initial-remnant mass relation for
a StarTrack single star evolution for two wind
prescriptions: the previously used Hurley et
al. winds and newly adopted (modified) Vink
et al. winds. Top panel: Solar metallicity.
This corresponds to the stellar field popula-
tions in the Galaxy. The predicted maximum
black hole mass Mbh
,
max 15M⊙ for new and
10M⊙ for old winds is consistent with the
most massive stellar black holes observed in
our Galaxy (e.g., in GRS 1915 Mbh = 14±4M⊙).
Middle panel: Moderate metallicity. This cor-
responds to stellar populations in galaxy IC10,
which hosts the most massive known stellar
black hole (Mbh = 23 − 34M⊙). Note that the
predicted maximum black hole mass Mbh
,
max
30M⊙ for the new winds is consistent with the
measurement in IC10, while Mbh
,
max 15M⊙
obtained for old winds appears to be signif-
icantly too small. Bottom panel: Very low
metallicity. This corresponds to stellar popu-
lations of Galactic globular clusters, or proto-
galaxies. The maximum black hole mass may
reach Mbh
,
max 80M⊙ for the newly adopted
wind prescriptions (Belczynski et al. 2010
[54].)
the first 10 detections will put very strong constrains
on BH natal kicks [77]. Additionally, the available
electromagnetic data on known BH and NS masses is
6
inconclusive as to the existence of the infamous â€mass
gap†(the dearth of compact objects in 2−5M⊙ mass
range). It is not clear whether this gap is the effect
of some observational biases or rather reflects on the
physics of the supernova explosion engine [78]. In a
recent study, we have identified which part of the su-
pernova explosion physics may be responsible for the
mass gap and we have indicated that the final answer
will be provided by the unbiased gravitational wave
measurements of compact object masses [79].
We will provide a qualitatively new description of
double compact object populations. The synthetic
populations of double compact objects formed at vari-
ous host galaxy environments (range of galaxy masses,
metallicities and star formation histories) will become
part of our library. Our model allows for evolution of
metal rich stars (Population I and II; [63]) as well as
first metal-free stars (Population III; [80,81]). Syn-
thetic data will be provided not only as a function of
a given host galaxy type, but also as a function of red-
shift under an adopted cosmological model that incor-
porates chemical evolution [8]. The data will include
not only basic parameters for double compact ob-
jects (component masses or separations/eccentricities
at formation) but also very specific parameters (spins:
magnitude and their direction [106] or position within
the host galaxy [82]) that can allow for a number of
applications not only within gravitational wave as-
tronomy but also crucial for gamma-ray burst or X-
ray binary studies. At the moment, a number of US-
based scientists are using StarTrack models to inter-
pret Chandra observations [83,84,85,86]. At the same
time I am leading efforts together with Harvard (Edo
Berger) and Boulder JILA groups (Rosalba Perna) to
characterize populations of transients (Palomar Tran-
sient Factory) and Gamma-ray bursts (Swift) with the
use of our models.
Our results will allow for rate/parameter estimates
specific to a given gravitational wave observatory and
will not be limited to LIGO/VIRGO but will be di-
rectly applicable to any future observatory that is des-
ignated to search for double compact objects at arbi-
trary redshifts. This approach will facilitate the un-
derstanding of what science can be done with gravita-
tional radiation detections with existing and planned
observatories.
1.5. Type Ia Supernovae
Supernova explosions are the fate of all massive
stars: these generate the core-collapse supernovae
(SNe). SN explosions are also the final stage for a less
massive binary system after one of the stars evolves
into a carbon-oxygen white dwarf. These generate
Type Ia SNe (SNe Ia). All SNe play a critical role
in driving the galactic gas ecosystem, in star forma-
tion by heating and compressing interstellar gas, and
in chemically enriching galaxies. SNe are responsi-
ble for releasing radioactive and heavy nuclei into in-
terstellar space and spreading the building blocks of
other stars, planets, and ultimately, life. Additionally,
observations of SNe Ia have led to the breakthrough
discovery that the expansion rate of the Universe is
accelerating, due to a mysterious “Dark Energy†[87].
Despite their use as the most important “standard
candle†distance indicators, the origin of SNe Ia re-
mains unknown. The long-standing paradigm [88] is
that their progenitors consist of a white dwarf that ac-
cretes enough matter from its binary companion such
that its mass exceeds the Chandrasekhar mass limit
(1.4 M⊙). However, the nature of the companion
star, the efficiency of matter accumulation onto the
white dwarf, and the explosion mechanism are all thus
far unclear [89].
To date, the two most favored scenarios thought
to lead to SNe Ia are the Single Degenerate Scenario
(stable mass accretion on to a white dwarf from a
normal star [90,91]) and the Double Degenerate Sce-
nario (the merger of two massive white dwarf stars
[92,93]). However, other promising formation scenar-
ios exist [94], and support for more than one forma-
tion scenario is substantiated by recent SNe Ia obser-
vations. Several different sub-classes of SNe Ia have
been identified [95], and recent advances in theoret-
ical SN modeling indicate that there is likely more
than one formation scenario (and explosion mecha-
nism) contributing to the observed sample of SNe Ia
[96]. This has far-reaching implications for the the-
ory of Dark Energy: if the explosion mechanism that
leads to the formation of a SN Ia is dependent on
evolutionary scenario and/or metallicity, this could
potentially pose a challenge for cosmological studies
by introducing systematic errors [97]. Thus, deter-
mining the nature of SN Ia progenitors – both in the
local and the most distant Universe – is one of the
most important long-standing puzzles in astrophysics
that remains to be solved.
The delay time distribution (DTD) is the distri-
bution of times in which SNe Ia explode following a
(hypothetical) burst of star formation. Knowing the
DTD gives the age of the progenitor, which places
strong constraints on the different proposed formation
scenarios. If the SN Ia rate is known in addition, then
it becomes possible to rule out theoretically-predicted
evolutionary channels.
Today, we are uniquely primed to address the ques-
tion of SN Ia origin. This has already been done
to some extent for Local Universe-like stellar popu-
lations [13]. Excitingly, now we are poised to deter-
mine how well SNe Ia can be considered standard can-
dles [98], and quantify to what degree their rates and
7
delay times vary as a function of redshift and metal-
licity. In order to accomplish this, in the next step
we will combine the theoretical rates (quantity) and
delay times (feedback timescale) with nucleosynthetic
yield predictions for the leading SN Ia formation sce-
narios as a function of redshift (metallicity). This
requires generating a suite of binary evolution models
for a grid of stellar metallicities, computing explo-
sion models with the most up-to-date nucleosynthetic
yield predictions, and calculating synthetic spectra
and light-curves. A complete pipeline is already in
place with the collaborators in our existing network:
from binary star birth/evolution of SN Ia progeni-
tors (A. Ruiter+K.Belczynski) to explosion model-
ing (MPA, Wurzburg, Heidelberg) to synthetic spec-
tra calculation (MPA, Mt. Stromlo, LANL). Such
a pipeline has already been extensively tested, and
shown successful in validating the importance of some
of the most promising SNe Ia formation scenarios
[99,100].
These efforts will enable us to assess the viability of
various explosion models, and gain insight into the ori-
gin of different subclasses of SNe Ia: ‘normal’ SN Ia,
sub-luminous and super-luminous SNe Ia, and other
peculiar objects (e.g., 2002cx-likes). The Synthetic
Universe library will contain detailed information on
the formation history of the most promising SN Ia
progenitors as computed by StarTrack. Addition-
ally, delay times and formation histories will also be
recorded for core-collapse SNe; some of which are ex-
pected to be synonymous with GRBs. The synthetic
data on delay times and rates of the various SNe will
serve as input for the NuGrid collaboration (see 3.2) to
obtain estimates of chemical enrichment scenarios for
various galaxy types. Such a long-term project would
involve convolving theoretical delay times for all stel-
lar explosions with galaxy mass distribution functions
as a function of redshift, and imposing a galaxy mass-
metallicity dependence [5]. We will thereby provide
important input data for chemical evolution models
by supplying the relative rates (and DTDs) of various
stellar explosions on cosmological scales.
2. DISCIPLINE CONTRIBUTION
We will deliver the first library of synthetic stel-
lar populations to the astro community. At first
we will provide the data on the populations of GR
sources (NS-NS, BH-NS, BH-BH), ULX binary can-
didates (BH + stellar type companion) and potential
progenitors of Type Ia supernovae (various configu-
rations of accreting white dwarfs in binary systems).
The database will be maintained beyond the period
of funding and new projects will be added as the need
arises. Our contribution will extend beyond the scien-
tific community. We will actively involve the public in
our research. We will provide a platform for the pub-
lic to get involved in our numerical simulations and
at the same time we will invest in education and our
project advertisement to a wider audience through the
Internet.
3. SIGNIFICANCE
We propose to tackle three very specific stellar re-
search projects at the frontier of stellar astrophysics.
We will provide a proof-of-principle study of the ori-
gin of ultraluminous X-ray sources and we will show
whether stellar mass black holes can or cannot ex-
plain the observed populations of these objects. We
will fully describe the populations of double com-
pact objects (NS-NS, BH-NS, BH-BH) that are the
most promising sources for gravitational-wave detec-
tors. LIGO and VIRGO observatories will start tak-
ing data around 2015 and we will provide a scientific
rationale behind the near-future detection of double
compact object mergers and help to guide the signal
search techniques. We will study various potential
progenitors of Type Ia supernovae in a quest to finally
identify which binaries are specifically responsible for
these explosions. Our synthetic models will help to
eliminate some of the proposed models, while other
models will gain in significance, driving progenitor re-
search. We will compare our predictions with the ob-
served supernovae rates, delay times, light curves and
spectra.
4. WORK PLAN
4.1. Database Development
The database will be placed at a local server main-
tained by our group, and will be visible and accessible
from the outside through the Synthetic Universe
website. The database will consist of electronic data
tables accompanied by detailed descriptions of each
table’s content. For each simulation from either the
evolutionary code or population synthesis code we will
post the raw data with a full description of the sim-
ulation. A detailed code manual will be placed up
front for each code, as well as a general overview of
the database and usage instructions. The raw data
files will serve as consistency checks and comparisons
for groups that prefer to do the post-processing on
their own.
Post-processing involves a number of computational
tasks that are very often required to present synthetic
data in a useful way. For example, it is important
to generate a synthetic spectrum to compare to an
observed galaxy. To proceed, we will use a popula-
tion synthesis code to obtain the entire stellar pop-
ulation in this galaxy. For each star, we will then
obtain a spectrum from one of the available stellar
spectra libraries, and co-add the spectra. These tasks
8
all require extra calculations, and external libraries.
These tasks will be performed by analysis codes that
we will create and distribute to address our specific
data needs. In this example, the final product would
be the spectrum of a synthetic galaxy that could be
downloaded from our server with the full description
of how it was generated.
As the database expands, we will adapt and im-
prove our analysis codes to serve as interactive re-
search tools. In the above example, a need may arise
to combine stellar population data to model unfore-
seen future observations. Since we will have a full
library of raw stellar population data within the first
years of opening the database, we will also provide
interactive software that can allow users to combine
the data in novel ways. This can include, but is not
limited to: building galaxy models and changing cos-
mological parameters. Finally, we will provide visual-
ization tools that will specifically work on information
contained within the database, requesting time snap-
shots, imaging, and rapid sampling of the synthetic
data. The first pilot project is already available on-
line at: http://www.syntheticuniverse.org
4.2. Specific Schedule
1) First year. We will start preparations for all
three science projects. That will include search of
the available literature, contacting other groups that
work on the similar subjects and we will establish
background for our calculations. We will prepare
our physical model (extra input physics, calibrations
and updates) for all three projects. In particular, we
will use the MESA code to updated criteria on com-
mon envelope development, we will implement the
primary effects of rotation into StarTrack based on
the Geneva group results, we will add various, less
thoroughly-explored though promising SN Ia models
(e.g., sub-Chandrasekhar mass merger models [101],
violent white dwarf mergers [102], Symbiotic channel
[103], double-detonation scenarios [104], thick wind
prescription [105]) and we will expand our code for rel-
evant ULX physics (leaky disks, anisotropic emission,
very massive stars). We will create and test inter-
nal (Warsaw Observatory) computing network within
the framework of the Universe@home program. We
will apply for computer time at various open Polish
computational centers.
2) Second year. We will run numerical population
synthesis simulations relevant to all three projects.
The data will be stored at first on our local disks. At
the same time we will work on the development of our
websites (www.syntheticuniverse.org) and we will
get it ready for the arrival of the synthetic data. The
simulations will be run at US and Polish Research In-
stitution computer clusters, and (if granted) in Polish
open computational centers. And we will launch our
Universe@home program and open it for the public.
We will begin assessing the changes imposed by effects
of rotation on relevant stellar populations.
3) Third year. The numerical simulations will con-
tinue. We expect that more and more computa-
tional resources will be acquired with time via our
Universe@home program. The first data will be
placed in our www.syntheticuniverse.org database
and made publicly available for other research groups.
We will carry out the first analysis of our data for all
three projects and write up and submit our initial
conclusions to international journals. For practical
reasons, the GR papers will be submitted to Astro-
physical Journal, SN Ia papers to Monthly Notices of
the Royal Astronomical Society and ULX studies to
Astronomy & Astrophysics. We will start incorpo-
rating MESA models into the StarTrack code as an
optional tool for comparative simulations.
4) Fourth year. The numerical simulations will con-
tinue and at the end of this year the simulations will
be completed. The full data sets will be placed and
described in our library. We will start the final analy-
sis and progress with the three research projects. We
will continue incorporating the Geneva models into
our population synthesis. Furthermore, we will obtain
the first population synthesis simulations that include
full effects of rotation from the MESA code.
5) Fifth year. Available computer power will be
used to obtain extra simulations that we will deem
required during our studies or that are requested by
the astro community or reviewers of our submitted
manuscripts. The entire group will focus on detailed
analysis of our results and comparisons of our sim-
ulations with (available at this stage) observations.
We will also compare our results obtained with ro-
tating and non-rotating stellar models. We will be
writing our findings into a series of papers. Students
and graduate students will be using these results to
progress their theses. At the same time we will be
already planning further future extensions of our li-
brary. By that time we expect to have major feedback
from various groups working on the related subjects
and this will help us guide our future database devel-
opment.
5. METHODOLOGY
5.1. Stellar Modeling
Although the theory of stellar evolution is quite
a mature subject, it has still to face big questions.
Among them: How do massive stars form? How does
stellar evolution influence star formation? What was
the nature and the characteristics of the first genera-
tions of stars in the Universe? What was the role of
these first generations in the reionization of the Uni-
9
verse? What physics drives the variations in the stel-
lar populations observed in different environments?
The answers to all these questions require improv-
ing stellar models. These improvements are obtained
by including important effects, neglected until re-
cently, such as the effects of the stellar winds, the
internal transport processes induced by axial rotation
or by other instabilities such as the thermohaline mix-
ing, the transport of angular momentum by the grav-
ity waves and/or by a magnetic field. All these effects
have been implemented in the Geneva and Bill Paxton
stellar models and have been tested by a whole series
of observational features. These comparisons all sup-
port the view that extra mixing processes do indeed
occur in stars, and that models accounting for them
can much better reproduce the observations than the
models neglecting them. Including these transport
mechanisms have deep impacts on the stellar popula-
tions expected at a given age and metallicity as well
as on the nucleosynthesis of the elements. Many of
these effects can be tested using the new windows
which are now opening on stellar interiors through
astroseismology (e.g., COROT, KEPLER). Additionally,
the huge number of stars that will be observed by the
satellite GAIA will allow to probe very short evolution-
ary phases which so far were only in the realm of the-
ory. Heavy mass loss of material by stellar winds, like
in the Luminous Blue Variable phase, may be good
example of short phenomena that may very deeply
affect the end of the stellar life, the nucleosynthetic
outputs, the supernova type and the nature of the
stellar remnant. GAIA observations will lead to strong
revisions of the stellar evolution in many parts of the
Hertzsprung-Russel diagram.
We will require two type of codes (i) population syn-
thesis code and (ii) detailed stellar evolutionary code.
Though these codes can work independently, ideally
they should work together, delivering feedback from
one code to another. The end product of such an
approach will be a comprehensive and self-consistent
picture of stellar populations at a brand new level,
both quantitatively in terms the of a data library, and
qualitatively in terms of evaluating the input physics.