Title Text

University of Delaware

Department of Physics and Astronomy

federica bianco

Biden School of Public Policy and Administration

Data  Science Institute

Title Text

University of Delaware

Department of Physics and Astronomy

federica bianco

Biden School of Public Policy and Administration

Data  Science Institute

Rubin Observatory Construction Project - Deputy Project Scientist

Rubin Science Collaborations - Coordinator

Rubin Transients and Variable Stars Science Collaboration - Co-Chair           

slides available at

https://slides.com/federicabianco/ucsc22

This is a living land acknowledgement developed in consultation with tribal leadership of Poutaxet, what is now known as the “Delaware Bay,” including: the Lenape Indian Tribe of Delaware, the Nanticoke Indian Tribe, and the Nanticoke Lenni-Lenape Tribal Nation in 2021. We thank these leaders for their generosity.

The University of Delaware occupies lands vital to the web of life for Lenni Lenape and Nanticoke, who share their ancestry, history, and future in this region. UD has financially benefited from this regional occupation as well as from Indigenous territories that were expropriated through the United States land grant system. European colonizers and later the United States forced Nanticoke and Lenni Lenape westward and northward, where they formed nations in present-day Oklahoma, Wisconsin, and Ontario, Canada. Others never left their homelands or returned from exile when they could. We express our appreciation for ongoing Indigenous stewardship of the ecologies and traditions of this region. While the harms to Indigenous people and their homelands are beyond repair, we commit to building right relationships going forward by collaborating with tribal leadership on actionable institutional steps.

Time domain astrophysics at a glance

Rubin LSST

Leading the Rubin LSST army of volunteers

Rubin LSST projects at FASTLab

At the intersection of astrophysics and Public Policy: From Light Echoes to Pollution Plumes

Data skills for good: COVID response support in Delaware

Time domain astrophysics at a glance

Rubin LSST

Leading the Rubin LSST army of volunteers

FASTLab projects in data-driven astrophysics

At the intersection of astrophysics and Public Policy: From Light Echoes to Pollution Plumes

Data skills for good: COVID response support in Delaware

Time domain astrophysics at a glance

Rubin LSST

Leading the Rubin LSST army of volunteers

Rubin LSST projects at FASTLab

At the intersection of astrophysics, Public Policy, and AI:                                From Light Echoes to Pollution Plumes

Data skills for good: COVID response support in Delaware

Time domain astrophysics at a glance

Rubin LSST

Leading the Rubin LSST army of volunteers

Rubin LSST projects at FASTLab

At the intersection of astrophysics, Public Policy, and AI:                                From Light Echoes to Pollution Plumes

Data skills for good: COVID response support in Delaware

Time domain astrophysics at a glance

The immutable skies

Ivezič (w Bianco) et al 2019

circa 1900

Will we discover new physics?

A survey of Machine Learning features and methods to discover unique objects and rare classes in regularly and irregularly sampled time series

Galarza, Bianco et al. 2021

2D t-SNE projection of feature space

Evenly sampled Kepler time series

.

Sparse, unevenly sampled Kepler time series

Will we discover new physics?

Li, Ragosta, Clarkson, Bianco 2021

A comparative assessment of LSST potential surveys in the discovery of unknown unknowns

Vera C. Rubin Observatory:

Ushering a New Era of TDA

LSST Science Drivers

Probing Dark Energy and Dark Matter

image credit ESO-Gaia

LSST Science Drivers

Mapping the Milky Way and Local Volume

via resolved stellar population

An unprecedented inventory of the Solar System from threatening NEO to the distant Oort Cloud

LSST Science Drivers

LSST Science Drivers

Exploring the Transients and Variable Universe

10M alerts every night shared with the world

60 seconds after observation

single image depth ~24

10-year stack image depth ~26

image resolution 0.2'' (seeing limited)

18,000 sq degrees 

839 images over 10 years in 6 filters

2-3 images per night

each fields reobserved within ~days

5 fields observed to higher cadence and more images

survey specification

~25.9, 26.8, 26.8, 26.3, 25.6, 24.8

~56, 74, 184, 187, 166, 171

8m Telescope - 9.6 deg.sq. field of view

Observatory

The DOE LSST Camera - 3.2 Gigapixel

3.2 Gigapixel camera 

378 4K ultra-high-definition TV

S. Ritz (UCSC), Camera Project Scientist, now Rubin Construction Project Scientist

Aaron Rodman (SLAC), LSST Camera Program Lead

2020

18 months of COVID have caused a 24 months delay in construction

interlocking requirements have superlinear effects

Survey start 2024

2020

Survey Start 2024 (probably)

Data Products

world public!

world public

data right holders

Astronomy meets Data Science

At this level of precision,everything is variable, everything is blended, everything is moving.

 https://ls.st/srd

SDSS

LSST

http://faculty.washington.edu/ivezic/talks/NASAseminar.pdf

Ivezič+19

SDSS 2x4 arcmin sq griz

MYSUC (Gawiser 2014) 1 mag shallower than LSST coadds

At this level of precision,everything is variable, everything is blended, everything is moving.

 https://ls.st/srd

http://faculty.washington.edu/ivezic/talks/NASAseminar.pdf

Rubin will see ~1000 SN every night!

A lot of them will be too faint to study with traditional means, particularly spectra. 

Lots of emphasis in new analysis techniques that rely on "Big Data"

Ivezič (w Bianco) et al 2019

Survey

design

Rubin LSST survey design

distributions of time gaps in 76 OpSims

Bianco et al. 2021 https://iopscience.iop.org/article/10.3847/1538-4365/ac3e72

Rubin LSST survey design

Rubin LSST survey design

Bianco et al. 2021

https://iopscience.iop.org/article/10.3847/1538-4365/ac3e72

https://www.lsst.org/content/charge-survey-cadence-optimization-committee-scoc

Because the Rubin LSST data is open to all US scientists and to a broader yet community worldwide, to truly make it a survey of and for and of the people, Rubin Observatory called the community to design its survey -

this is a uniquely "democratic" process!

single document

9 chapters ​

25 science cases

14

46 papers

467 unique authors

16

39 notes

218 unique authors

173

date

goal

target community

response

available simulation

2015-17

2018

2021

23

23

24.6

24.6

23

26.8

28

26.9

28.1

5σ depth

5σ depth

coadd 5σ depth

coadd 5σ depth

survey specification

(current baseline)

g band

r band

source http://astro-lsst-01.astro.washington.edu:8080/?runId=2

0.576

0.52

intranight gap

hours

15

internight gap

days

15

3

median internight gap

days

50

5

median internight gap

days

any filter

r band

survey specification

source http://astro-lsst-01.astro.washington.edu:8080/?runId=2

0.576

intranight gap

hours

Parallax error @ 24 mag (mas)

Gaia DR2: 0.1 mas @ mag 17

https://www.cosmos.esa.int/documents/29201/1770596/Lindegren_GaiaDR2_Astrometry_extended.pdf

survey specification

source http://astro-lsst-01.astro.washington.edu:8080/?runId=2

survey specification

(current baseline)

the Rubin LSST

Science Collaborations

The Rubin Organization is almost as complex as the Universe it will explore!

Rubin LSST Science Collaborations

The Rubin Organization is almost as complex as the Universe it will explore!

8 SCs - 6 continents - 2000 people - 25 countries

Rubin LSST Science Collaborations

8 SCs - 6 continents - 2000 people - 25 countries

Rubin LSST survey design

40+ submissions

85% led by SC members

and virtually all including SC members

The SCs were core to this process

Rubin Focus Issue of ApJS on Rubin Survey Strategy Optimization

https://iopscience.iop.org/journal/0067-0049/page/rubin_cadence

A focus issue of ApJS to publish LSST cadence proposals work

Rubin LSST Science Collaborations

8 SCs - 6 continents - 2000 people - 25 countries

Chairs of the Transients and Variable Stars Science Collaboration

Science Collaborations Coordinator

Chairs of the Transients and Variable Stars Science Collaboration

Science Collaborations Coordinator

Member of the SAC

Member of the Contribution Evaluation Committee

Member of the LSSTC Executive Board

Member of the Research Inclusion Working Group

We study rare and unusual transients with machine learning and probabilistic inference

Olague et al 2017

@akumadog

Brain Programming and the Random Search in Object Categorization

Convolutional NN

The visual cortex learns hierarchically: first detects simple features, then more complex features and ensembles of features

The visual cortex learns hierarchically: first detects simple features, then more complex features and ensembles of features

Convolutional NN

https://www.kdnuggets.com/2016/11/intuitive-explanation-convolutional-neural-networks.html/2

multilayer perceptron

output

input layer

hidden layer

output layer

1970: multilayer perceptron architecture

Defines a 5-fold metric to evaluate completeness of LSST observations along features relevant to transients

Li et al 2021

explosions in the sky

explosions in the sky

Astronomy’s Discovery Chain

Discovery Engine

10M alerts/night

Community Brokers

target observation managers

the astronomy discovery chain

LSST data will be insufficient for individual transients

Can we study unusual SNe with Rubin data alone?

Data Driven Templates for rare classes of supernovae arising from massive stars: can we tell them apart from sparse LSST data lightcurves?

Creating templates from over 1000 photometry of Stripped Envelope Supernovae

Khakpash, Bianco, et al. in prep

Willow Fox Fortino's next paper

Do we really need a spectrograph??

Umer Zubair's

master's thesis

fast transients with Rubin LSST

Khakpash, Bianco, et al. in prep

Data from Anna Ho et al 2021 - ZTF

Presto Color observing strategy

proposing an LSST strategy collecting 3 images in 1 night in 2 filters Bianco et al. 2019

also:

Andreoni et al.

w FBB  2019

Margutti et al.

 w FBB 2018

Andreoni et al.

w FBB and Xiaolong Li 2021

A probabilistic classifier for unusual transients within a single night of observations

Mr. Ming Lian Master Thesis

Sampling 40,000 transients

for each of 22 classes

in a hypercube in 6 dimensions:

- DT1

- DT2

- Filter 1

- Filter 2

- color

- magnitude

Sampling 40,000 transients

for each of 22 classes

in a hypercube in 6 dimensions:

- DT1

- DT2

- Filter 1

- Filter 2

- color

- magnitude

Punusual~0.95

Punusual~0.01

A probabilistic classifier for unusual transients within a single night of observations

Mr. Ming Lian Master Thesis

Difference Image Analysis

Improving the efficiency of transient detections with Neural Networks: can we avoid template subtraction altogether?

search

template

Acero-Cuellar et al. DESC submitted

Improving the efficiency of transient detections with Neural Networks: can we avoid template subtraction altogether?

search

template

difference

-

=

Acero-Cuellar et al. DESC submitted

Improving the efficiency of transient detections with Neural Networks: can we avoid template subtraction altogether?

search

template

difference

-

=

Acero-Cuellar et al. DESC submitted

Improving the efficiency of transient detections with Neural Networks: can we avoid template subtraction altogether?

search

template

difference

-

=

Saliency maps: what pixels matter? 

search

template

difference

95% accurate

Acero-Cuellar et al. DESC submitted

Improving the efficiency of transient detections with Neural Networks: can we avoid template subtraction altogether?

search

template

difference

-

=

95% accurate

Acero-Cuellar et al. DESC submitted

Improving the efficiency of transient detections with Neural Networks: can we avoid template subtraction altogether?

search

template

difference

-

=

95% accurate

Acero-Cuellar et al. DESC submitted

Saliency maps: what pixels matter?

Acero-Cuellar et al. DESC submitted

original paper : Karen Simonyan, Andrea Vedaldi, Andrew Zisserman 2013

Saliency maps: what pixels matter?

Acero-Cuellar et al. DESC submitted

search

template

difference

Saliency maps: what pixels matter?

search

template

difference

template

search

Acero-Cuellar et al. DESC submitted

Saliency maps: what pixels matter?

search

difference

template

search

Acero-Cuellar et al. DESC submitted

Light Echoes

Light Echoes

η-Carinae light echoes

Rest et al. (w Bianco) 2012Natur.482..375R​

Light Echoes

η-Carinae light echoes

Frew 2004, Smith & Frew 2011

Light Echoes

η-Carinae light echoes

Light Echoes

η-Carinae light echoes

Li et al. submitted

AILE: the first AI-based platform for the detection and study of Light Echoes

NSF Award #2108841

Detecting and studying light echoes in the era of Rubin and Artificial Intelligence

P.I. Bianco

Pessimal AI problem:

• small training data
• inaccurate labels
• imbalance classes
• diverse morphology
• low SNR

AILE: the first AI-based platform for the detection and study of Light Echoes

YOLO3 + "attention" mechanism

precision 80% at 70% recall with a training set of 19 light echo examples! 

NSF Award #2108841

Detecting and studying light echoes in the era of Rubin and Artificial Intelligence

P.I. Bianco

Li et al. submitted

Multi-city Urban Observatory Network

Dobler, Bianco, et al 2021, in Remote Sensing

Studying cities as complex systems through imaging data

Multi-city Urban Observatory Network

Dobler, Bianco, et al 2021, in Remote Sensing

Studying cities as complex systems through imaging data

• energy demand and consumption
• ecology of flora and fauna
• urban metabolism
• circadiem rhythms

Multi-city Urban Observatory Network

From Light Echoes to Polluting Plumes

Ian Heffner, UD MSDS

​improved image

subtraction through PCA

Pessimal AI problem:

• small training data
• inaccurate labels
• imbalance classes
• diverse morphology
• low SNR
• complex BG

From Light Echoes to Polluting Plumes

Ian Heffner, UD MSDS

​improved image

subtraction through PCA

Pessimal AI problem:

• small training data
• inaccurate labels
• imbalance classes
• diverse morphology
• low SNR
• complex BG

Plumes and heat in NIR

Plumes in hyperspectral imaging

Text

COVID-19 - an opportunity to be useful or an opportunistic exploitation of a serious problem?

Is there a need?

dont overestimate the resources that are available

"Christiana care is seeking someone with Python expertise..."

surge plans rely on mutual aid between States

Is there a need?

• Delaware is a small state (unlike Washington State, CA, NY)
• 5 hospitals only one has an epidemiological team... of 1
• There are limited computational resources in people and in hardware

"Christiana care is seeking someone with Python expertise..."

Can I fulfill that need?

• Is there a team of domain experts you can work with?

"Christiana care is seeking someone with Python expertise..."

"ChristianaCare is seeking someone with Python expertise..."

to work with the CC Value institute epidemiologist and software engineer

DE COORDINATION

EPIDEMIOLOGY

INTEGRATION W CC NEEDS AND RESOURCES

COVID-19 - through Eric Best we have the most accurate dataset on COVID of all of the USA

Eric Best works tirelessly to harvest and curate data from each DE hospital (and now  potentially PA) and demographic data from DEMA

Dr. Eric Best, UD Biden School, now Ass. Professor at PennState

    . what are the impacts of DE’s unique demographic distribution on hospitalization?

    . how will policy changes affect disease progression and hospitalization?

    . will there be a suppression of disease propagation in future summer months?

    . how do we incorporate behavioral properties of the in- and out-of-state sub-populations?

    . what level of resurgence might we expect to see if social distancing is relaxed (e.g. UD repoen)?

    . if re-infection is a possibility, what is the efficacy of vaccines?

    . new virus variants?

DE Schools closing

murder of

George Floyd

US election campaign

Vaccines

Delta Variant,

pessimistic prediction

MSDS 2020

Jonathan Clifford

Farid Qmar

Engineering &

Public Policy PhD

 NIH The Effects of Subpopulations and Policy Change on COVID19 Hospital Demand Models  P.I Dobler, CoP.I. Bianco

dashboard now being used by other states 

e-DHARMA: a DHARMA emulator

Manjula Ibrahim

MSDS Thesis 2021

 NIH The Effects of Subpopulations and Policy Change on COVID19 Hospital Demand Models  P.I Dobler, CoP.I. Bianco

Replacing MCMC with a Neural Network to speed up the predictions

This framework allows for substantial flexibility to include effects such as:

        · time-dependent social distancing

        · mask mandate compliance

        · person-to-person contact rates

        · age, gender, race, and ethnicity  

        · individual susceptibilities

        · comorbidities

        · probabilistic hospital length of stay

        · asymptomatic subpopulations

        · vaccine availability and use

        · closing and reopening

        · hospital discharge and intake operations

"It has been invaluable to have predictions for the next week instead of having to panic about the next shift"

DE Hospital Director

A UD COVID-19 hackathon 

Health Care and Recovery

Jinny Sawyer-Morris's thesis

in Human Development and Family Science

longitudinal surveys of 120 individuals;

61% women

8 sober living homes

Exploring Gender-Specific Differences in Substance Use Disorder Recovery Capital:

Research Inclusion: sonification of LSST lightcurves

Riley Clarke, UD grad student Sid Patel, UD undergrad summer research project

Sonification: Data → Sound

New way of understanding data

• Can be complementary to visualizations

• Gives access to people who cannot
  interpret data visually

• Sounds cool! Good for public outreach

Research Inclusion: sonification of LSST lightcurves

Presented at 2021 Rubin Project Community Workshop (300 ppl) 

Started a new Rubin Working Group

Rubin Rhapsodies

Rubin Observatory LSST 

Rubin Observatory LSST 

Rubin LSST will deliver 10M alerts each night about anything that changed in the southern hemisphere sky: supernovae, merging stars, transiting planets, gravity bendind light and magnifying stars...

to anyone in the world by a set of selected "brokers" that will work with Rubin to accomplish this incredible goal!

the data belongs to the people!

within 60 second of detection!

Rubin Observatory LSST 

The LSST Science Platform

Rubin Observatory LSST 

Rubin and MMA

Rubin Observatory LSST 

LIGO/VIRGO area of localization ~100deg square

 Ursa Minor contains 255.86 square degrees

S190425z 18% of the sky localization

Rubin Observatory LSST 

Rubin can find the electromagnetic counterpart of Gravitational Wave and Neutrino discoveries better than any survey!

LIGO/VIRGO area of localization ~100deg square

 Ursa Minor contains 255.86 square degrees

Rubin FoV 10 deg

S190425z 18% of the sky localization

Rubin Observatory LSST 

 AT 2017gfo

optical counterpart we have identified near NGC 4993 is associated with GW170817

Evidence that mergers of NS are significant sources of r-process elements heavier than iron, including gold and platinum, which was previously attributed exclusively to supernova explosions

Rubin Observatory LSST 

Rubin Observatory LSST 

MARGUTTI+ 2018

Rubin Observatory LSST 

MARGUTTI+ 2018

Rubin Observatory LSST 

Rubin and LEOsats

Rubin Observatory LSST 

MMA and LEOsats

Iridium satellite number 35 lit up the predawn sky west of Boston at 5 a.m. EST on February 1, 1998,  Sky & Telescope

Satellite flares

Time domain Rubin LSST science

can be mitigated:

- orientation of satellite,

- directing flares away from observer

- knowing coordinates to associate them to alerts

if not mitigate there would be bogus alerts and images ruined by saturating flares

Hainaut & Williams 2020

https://arxiv.org/abs/2003.01992

Tyson+ 2020

https://arxiv.org/abs/2006.12417

modification to planned schedule carry long lasting consequences on the program

ToO enabled for GW need to scan ~100 sq deg footprint upon trigger - will be compromised

Time domain : the LSST observing strategy

Static Sky: correlated noise and cross talk

Cosmology probes are systematic dominated

mitigation: simulation of the non-linear crosstalk to measure the effect on precision cosmology and effectiveness of removal algorithms

(image credit: Canada-France Hawaii Telescope)

Tyson et al. 2020

https://arxiv.org/abs/2006.12417

Rubin Observatory LSST 

Diversity Equity Inclusion

Rubin LSST Science Collaborations

We aspire to be an inclusive, equitable, and ultimately just group and we are working with renewed vigor in the wake of the recent event that exposed inequity and racism in our society to turning this aspiration into action.

Rubin LSST Science Collaborations

what's in a name?

The first ground-based national US observatory named after a woman, Dr. Vera C. Rubin 

VRO

In the first 10 years of its life Rubin will conduct the Legacy Survey of Space and Time or LSST