arXiv:1912.09992v1 [astro-ph.IM] 18 Dec 2019
Astro2020 White Paper
Theoretical Astrophysics 2020-2030
Thematic Areas:
Ground-Based Project
Space-Based Project
⊠
Infrastructure Activity
⊠
Technological Development Activity
⊠
State of the Profession Consideration
Principal Author:
Name: Juna A. Kollmeier
Institution: Carnegie Observatories, 813 Santa Barbara St
reet, Pasadena, CA 91101, USA
Email: jak@carnegiescience.edu
Co-authors:
Lauren Anderson, Flatiron Institute, landerson@flatironi
nstitute.org, Andrew Benson, Carnegie
Observatories, abenson@carnegiescience.edu, Tamara Bog
danovi ́c, Georgia Tech,
tamarab@gatech.edu, Michael Boylan-Kolchin, UT Austin, m
bk@astro.as.utexas.edu, James S.
Bullock, UC Irvine, bullock@uci.edu, Romeel Dav ́e, Univer
sity of Edinburgh, rad@roe.ac.uk,
Federico Fraschetti, University of Arizona, frasche@lpl.
arizona.edu, Jim Fuller, Caltech,
fuller@tapir.caltech.edu, Phil Hopkins, Caltech, ph@tap
ir.caltech.edu, Manoj Kaplinghat, UC
Irvine, mkapling@uci.edu, Kaitlin Kratter, University of
Arizona, kkratter@email.arizona.edu,
Astrid Lamberts, Observatoire de la Cˆote d’Azur in Nice, M.
Coleman Miller, University of
Maryland, miller@astro.umd.edu, James E. Owen, Imperial C
ollege London,
james.owen@imperial.ac.uk, E. Sterl Phinney, Caltech, es
p@tapir.caltech.edu, Anthony L. Piro,
Carnegie Observatories, piro@carnegiescience.edu, Hans
-Walter Rix, Max Planck Institute for
Astronomy, rix@mpia.de, Brant Robertson, UC Santa Cruz, br
ant@ucsc.edu, Andrew Wetzel,
UC Davis, awetzel@ucdavis.edu, Coral Wheeler, Caltech/Ca
rnegie, coral@caltech.edu, Andrew
N. Youdin, University of Arizona, youdin@email.arizona.e
du, Matias Zaldarriaga, Institute for
Advanced Study, matiasz@ias.edu
1
1 Key Issue and Overview of Impact on the Field
1.1 Issue
The past two decades have seen a tremendous investment in obs
ervational facilities that promise to
reveal new and unprecedented discoveries about the univers
e. These machines are indeed marvels
of technology and engineering and their large pricetags (al
l nearing or in excess of $1B USD)
are testament to the difficulty and expense of creating this e
quipment. However, over the same
period, no similarly large investment in theoretical work h
as taken place
even when accounting for
computational infrastructure.
In this white paper, we argue that in order for the promised cr
itical breakthroughs in astro-
physics over the next decade and well beyond, the national ag
encies must take a serious approach
to investment in theoretical astrophysics research. We pro
vide a multi-level strategy, from the
grassroots to the national, to address the current under inv
estment in theory relative to observa-
tional work.
1.2 Impact on the Field
But happily, hypotheses, even if full of error, fail us not. —
F.W. Argelander
Theory plays several key roles within astronomy and astroph
ysics. First and foremost, theory
guides insight into the physical nature of the universe
even when the theory is wrong
. Indeed, it
is the
difference
between theory and observation that has ushered in the most d
ramatic revolutions
in physics and astronomy. Examples abound ranging from the d
eviation of mercury’s orbit relative
to Newtonian dynamics that ultimately yielded General Rela
tivity, to the variation in the expected
cosmic microwave background radiation and the observed ver
y smooth cosmic microwave back-
ground (CMB) that ultimately gave way to our current concord
ance model for the cosmos.
In the 20th century, it was perhaps sufficient to provide a sma
ll number of exceptional astro-
physicists access to pencils and paper. This approach no lon
ger suffices in the 21st century. The-
oretical research has become increasingly complex and the c
ommunity seeks to test these ideas
at increasingly challenging observational regimes. Many t
heorists perform heavily computational
work, which, thus far, has been encouraged primarily in an ad
-hoc, grass-roots level in individual
departments and universities.
Testing theoretical work is often used to motivate observin
g proposals and in justification for
large missions. Nevertheless, the support for this work is w
oefully lacking compared to the obser-
vations themselves. Who is to make these predictions and who
is to vet them when they increas-
ingly rely on calculations that take months to years on large
computers to reproduce?
Rather than continuing to give lip service to the need for the
orists in the community, we call
on the National Academy to take a strong stand in support of th
eory. We call on NAS not only to
acknowledge the critical role theory plays in leading the pa
thway to fundamental breakthroughs
in dark matter and dark energy, but to endorse a specific progr
am to support theory and thus keep
astronomy a vibrant field of truth seeking, rather than a back
drop for technological research and
development.
In addition to the obvious role of theory in making the nature
of the universe manifest in clear,
predictable, and manipulatable form, theoretical researc
h serves two more practical functions. The
2
first is in the
Guidance of Observational Programs
and the second is in the
Interpretation of New
Results
. We discuss each of these in turn below.
1.3 Guidance for Observations
Some of the most successful observational programs of all ti
me, particularly in recent decades,
were guided by theory. These include the CMB, large scale str
ucture of galaxies, gravitational
lensing, gravitational wave emission, just to name a few. Ne
arly all major observational facilities
are developed to test theory. Notable examples in physics in
clude the LHC and LIGO, both of
which confirmed theoretical predictions (the Higgs boson an
d gravitational waves) and generated
Nobel prizes.
In many cases, extensive theory is needed just to measure a si
gnal. For instance, the signals of
merging black holes measured by LIGO would not be detected wi
thout matched filtering against
theoretically generated waveform templates. At present, w
e are still limited in our ability to detect
many types of signals (e.g., eccentric mergers) due to the la
ck of available templates and a lack of
theorists, funding, and computational resources to genera
te them. If our community continues to
invest so heavily into such facilities, shouldn’t we also be
funding the theoretical research necessary
to extract the most science from our investments?
1.4 Interpretation of Observations
Theory also is vital to learn from observations. Consider th
e CMB. This relatively simple data set
(a measurement of temperature across the sky) has fueled muc
h of the field of cosmology. But none
of our current understanding (e.g., the big bang, cosmic nuc
leosynthesis, the
Λ
-CDM paradigm)
would exist without extensive theoretical efforts to inter
pret CMB data.
Most fields in astrophysics have not received as much theoret
ical attention, and enormous ad-
vancement could be achieved (with data already obtained) if
more support for theoretical analysis
was available. This need will only grow in the coming decades
with the explosion of unprecedent-
edly large datasets that will require the trained eye of theo
rists to find order and physical insight.
We note though that we are not advocating for a horde of theori
sts that are tasked to simply
“match” the data from upcoming missions. Rather, the goal sh
ould be a deeper understanding. A
successful investment in theory will reinvigorate the field
, particularly in areas where observations
are at present too rudimentary to test predictions.
2 Strategic Plan
The strategy for the future needs to be as bold and visionary i
n the theoretical domain as it is in the
observational and instrument domains. One can cite many inc
redible machines/missions for the
future: JWST, TMT, GMT, LSST, Lynx, HABEX, OST, LUVOIR, WFIR
ST, CMB-S4. The list
goes on. Missing from this long catalog is any explicit suppo
rt for theoretical astrophysics. Indeed,
over this same period of tremendous growth, support for theo
ry has declined precipitously. Even
the highly successful NASA Astrophysics Theory Program has
seen dramatic cuts, just as it was
created. To deny the need for this is to pretend that progress
comes from experiment in isolation.
The consequence of such thinking will be the intellectual de
ath of our field.
3
2.1 National Theory Program
The only nation in the world with a National program in theore
tical astrophysics is Canada, the
Canadian Institute for Theoretical Astrophysics (CITA). C
ITA has long been regarded as an in-
tellectual jewel of Canada and we should take lessons from th
is successful program from our
Northern neighbor. The US investment in theoretical astrop
hysics should be proportionate to the
US investment in national observational facilities and sho
uld thus be expanded greatly and reflect
our national commitment to astrophysics research. The nati
onal program should have a central
hub, but should serve all theory groups in the nation. As part
of this, we should use the National
Laboratory infrastructure to continue and expand support f
or theoretical astrophysics research at
these sites and at their Leadership Computational Faciliti
es to serve U.S. national science interests
and capabilities.
2.1.1 NAS Fellows in Theoretical Astrophysics
The establishment of a national fellowship in theoretical a
strophysics is a priority for 2020-2030.
In contrast to programs like the extremely successful Hubbl
e Fellowship program (or Chandra
and Sagan fellowships) that are tied to a specific mission or a
specific domain area, the national
fellowship in theoretical astrophysics should be open to al
l theorists regardless of whether their
theoretical program makes direct contact with a specific obs
ervational initiative or not. It is indeed
critical to foster theory both within
and
outside of the auspices of a specific observational mission
or program. Support of theory not limited to current mission
s could lead to the inspiration of
future missions and facilities that would not be thought of o
therwise. These fellowships should be
directed at
both
graduate and post-graduate programs.
2.1.2 NAS Symposia In Theoretical Astrophysics
Theorists need to argue. Unlike engineering, where there ar
e multiple solutions, some more or less
elegant than others but many that “do the job”, theoretical w
ork is either correct or incorrect. It is
thus critical that theorists have a forum for vigorous argum
ent. Meetings such as the AAS do not
provide a forum for theorists as the speaking format is only c
onducive to the presentation of results
and not the argument about them.
The symposia should be of general interest and should focus o
n several topics each year on
a rotating basis. This high-interaction format has been suc
cessfully implemented at the Kavli
Institute for Theoretical Physics in California and the Asp
en Institute in Colorado. Those venues
could (and should) be directly supported for the NAS Symposi
a, but other venues could (and
should) be considered. The essential point is that theorist
s, in particular, need a venue for discourse
that is not overly stifled by the current conference format wh
ich is focused on presentation rather
than confrontation
1
.
2.1.3 Theory-Observation Co-Investment
Successful investment portfolios require a combination of
assets to ensure future wealth and in-
sure against market uncertainties. While many observation
al programs are extremely ambitious,
1
We note that this recommendation is highly subfield dependen
t
4
we recognize that the cost and schedule of these programs are
difficult to predict accurately. With
increasingly long lead times for next generation programs,
it is important that the community ex-
plicitly adopt a program of co-investment in theory alongsi
de observational missions. The level
of this co-investment may vary depending on the nature and sc
ale of the program, but the current
default (of 0%) is unhealthy and exposes the national facili
ties to tremendous risk – both in under-
utilized scientific potential as well as community atrophy a
nd apathy over the long periods required
to make the discovery machines of the future.
As part of this recommendation, we recommend a Summer School
program for theoretical
apprentices interested in making predictions for upcoming
facilities. There currently exists a very
large number of summer schools for young researchers that ar
e observation focused, especially
in emerging fields, that train apprentices in data reduction
and analysis techniques. By contrast,
there are almost no summer schools associated with theoreti
cal tools with the notable exceptions
of the MESA Summer School and the Kavli Summer Program in Astr
ophysics. Where are the
opportunities for students to learn the tools of theory, in p
articular, in areas that require more than
mathematics and physics domain knowledge (for example, num
erical hydro, radiative transfer,
dynamics). We recommend that funding be allocated for summe
r schools for theorists to support
these flexible and targeted training grounds.
NAS ACTION
•
The National Academy should form, or recommend the formatio
n of, a National Program
in Theoretical Astrophysics with multiple nodes and repres
entatives across the nation.
•
The National Academy should establish, or recommend the est
ablishment of, Theory Fel-
lowships, distinct from Hubble, Einstein, or other fellows
hip programs, that are exclu-
sively for theoreticians at graduate and postdoctoral stag
es.
•
The National Academy should establish or recommend the esta
blishment of Theory Sym-
posia
•
The National Academy should explicitly recommend a theory c
omponent for all large
facilities recommended as part of Astro 2020
2.2 Infrastructure for Theory
There are several ways in which Theory infrastructure must b
e supported. The first, of course, is
personnel. While NAS cannot recommend the wholesale manufa
cture of FTE lines at universities,
it can support infrastructure at the local and national leve
ls to leverage the academic workforce.
2.2.1 Local Infrastructure
As national facilities have grown to exa-scale, they effect
ively require local mid-scale computing
infrastructure to enable their use. Mid-scale facilities a
re needed for code development, testing,
scaling tests, and a huge range of analysis projects (e.g. ra
diation transfer, structure finding, etc.)
which are too expensive to run on a single workstation but can
not be deployed in “production”
calculations without prior work. Without local facilities
, it is difficult if not impossible to demon-
strate the readiness level of code development, scaling, an
d testing required for large allocations on
5
national facilities. National facilities on larger scales
are also increasingly unable to accommodate
small testing protocols (e.g., analysis). Moreover, large
national facilities typically guarantee data
storage for only the duration of the main proposal – any usage
of the data generated beyond the
grant period requires local mid-scale facilities.
These resources are generally funded in one of two ways: eith
er (1) groups directly pay for
equipment (purchase of a mid-scale compute cluster) and sup
port (paying administration costs
as salaries or fees, which are usually comparable in cost to t
he equipment buy over a few-year
equipment lifetime), or (2) increasingly, as universities
centralize compute support to leverage
economies of scale and cloud computing becomes more widely u
seful, groups pay a bundled cost
to the university or external supplier for computing (eithe
r paying ’up front’ via a subscription
or buy-in fee, or paying as-you-go per compute hour). Howeve
r the current grants infrastructure,
especially for small (e.g., single-PI) theory grants and su
per-computing grants, provides no support
for either of these infrastructures. Option (2) in particul
ar, which is rapidly becoming the norm in
both business and academic systems, does not even fit into the
existing budget categories of most
grant agencies. It is imperative that grants supporting the
huge time and effort, for both scientific
analysis and supercomputing time, provide a category to fina
ncially support the required local
infrastructure expenses, especially the payment of admini
strative or per-hour compute costs.
2.2.2 Legacy Support for Theory Products
Theory grants and national super-computing centers spend m
illions of dollars and tremendous
effort and resources supporting the generation of theoreti
cal models and predictions. A major sci-
entific goal of these is, without exception, providing predi
ctions or comparisons to current and
future observational data. But there is no support for the ar
chiving or access of such data. With
super-computing proposals, data management plans are ofte
n required, but there is no infrastruc-
ture provided by any of the funding agencies for making that d
ata available in useful form after the
grant period expires. There is also no way to ensure theoreti
cal data products are made public, in
the way that is now viewed as an obvious (and tremendously imp
ortant) requirement for observa-
tional products. Given that there is already a large infrast
ructure in place to support observational
data sets, it is a natural and straightforward step to suppor
t mock theoretical data products of said
data. This requires no new fundamental infrastructure, sim
ply support for additional data sets, the
creation of which is already being funded – essentially, sim
ply providing support for access of
said products (after financial and computing resources are a
warded for their creation) in the same
way as is already done for their observational equivalents.
This would also tremendously increase
the impact of theoretical products, by making them not just m
ore readily accessible to observers,
but accessible via the same interactive data archival syste
ms which they already use to access the
equivalent observational products, catalogues, images, s
pectra, etc.
NAS ACTION
•
Grant support for local infrastructure and maintenance sho
uld be recommended as a pri-
ority for 2020-2030 within the AST and NASA grant programs di
rectly.
•
Explicit competitions for replacement infrastructure for
astrophysics theory should be
recommended and regularly conducted along with mid-scale i
nfrastructure calls.
6
2.3 Program Support
Both NSF and NASA have the resources to direct toward this vis
ion if they have the community
support to do so. Investment in theory should be made an expli
cit priority and should be done in
a proportion
α
to the investment in the observatories of the future. We do no
t speculate what
α
should be in this document. We suggest NAS undertake a study t
o determine this number, but we
argue it should be no less than 10%.
3 Conclusions
The national investment in observational infrastructure i
s growing without bound, but in sharp con-
trast, the national investment in theory has remained const
ant for decades. This situation threatens
the health and vitality of our field. The goal of astronomy and
astrophysics is not just to catalog the
universe, but to understand how it works. Without broad supp
ort of theoretical work, ranging from
theoretical predictions that suggest new observations or i
nstruments, to theoretical interpretation
and modelling of existing data, astronomy will be reduced to
an expensive and aimless cosmic
census.
7