Exploiting TESS data to
Understand Young Stars
Adina Feinstein
NSF Graduate Research Fellow
Advisors: Jacob Bean & Benjamin Montet
!1
Candidacy I September 26, 2019
Slide 2
Slide 2 text
The Transiting Exoplanet Survey Satellite (TESS) is a
four+ year, nearly all-sky survey monitoring millions
of stars.
!2
Slide 3
Slide 3 text
The Transiting Exoplanet Survey Satellite (TESS) is a
four+ year, nearly all-sky survey monitoring millions
of stars.
!2
27 days
54 days
81 days
108 days
189 days
351 days
Slide 4
Slide 4 text
!3
The entire 96° x 24° sector is observed
in the TESS Full-Frame Images (FFIs).
Slide 5
Slide 5 text
Predictions indicate the detection of thousands of
exoplanets which will not be covered in the 2-minute data.
!4
(Barclay et al. 2018)
Slide 6
Slide 6 text
!5
Although TESS’s primary objective is to find
exoplanets, it is a great data set for other fields.
Exoplanet
Solar System
Extragalactic
Stellar Astrophysics
Slide 7
Slide 7 text
!6
Outline
1. Developing eleanor, a tool for light curve extraction
from TESS
2. Robustly identifying and characterizing flares of young
stars
3. Connecting flares and spots to understand stellar
activity
Slide 8
Slide 8 text
!7
Outline
1. Developing eleanor, a tool for light curve extraction
from TESS
2. Robustly identifying and characterizing flares of young
stars
3. Connecting flares and spots to understand stellar
activity
Slide 9
Slide 9 text
Introducing eleanor, an open-source tool
for light curve extraction from the FFIs.
!8
Feinstein, A. D., Montet, B. T., Foreman-Mackey, D., et al. 2019, PASP, 131, 094502
https://GitHub.com/afeinstein20/eleanor
Slide 10
Slide 10 text
Introducing eleanor, an open-source tool
for light curve extraction from the FFIs.
!8
Feinstein, A. D., Montet, B. T., Foreman-Mackey, D., et al. 2019, PASP, 131, 094502
https://GitHub.com/afeinstein20/eleanor
Slide 11
Slide 11 text
!9
All FFIs for a given sector are about 1TB
of data, which is not user friendly.
Slide 12
Slide 12 text
!9
All FFIs for a given sector are about 1TB
of data, which is not user friendly.
Slide 13
Slide 13 text
!10
Slide 14
Slide 14 text
!10
Slide 15
Slide 15 text
!11
From the “postcards” we create Target Pixel Files
(TPFs) which we perform aperture photometry on.
(Feinstein et al. 2019)
Slide 16
Slide 16 text
!12
We provide four different types of
light curves for the community to use.
(Feinstein et al. 2019)
Slide 17
Slide 17 text
!13
We completed a quick search of Sectors 1 and 2
for new planet candidates.
(Feinstein et al. 2019)
Normalized Flux
0.8
1.0
1.2
0.99
1.00
1.01
0.99
1.00
1.01
0.98
1.00
0.98
1.00
1.02
Time [BJD-2457000] Time from Mid-Transit [Days]
1325 1330 1335 1340 1345 1350 -0.2 -0.1 0.0 0.1 0.2
TIC 350844139
TIC 394340349
TIC 139771134
TIC 159835004
TIC 38907808
Slide 18
Slide 18 text
!14
The eleanor package is a powerful tool for those
not interested in stars and exoplanets as well.
(Feinstein et al. 2019: See also Fausnaugh et al. 2019; Vallely et al. 2019)
Time [BJD-2457000]
Flux
SN2018fhw
SN2018eph
SN2018exc
MOA 2018-LMC-002
MOA 2018-LMC-003
60
50
40
140
130
120
140
130
120
2900
2800
2700
300
200
100
1330 1340 1350 1360 1370 1380
Slide 19
Slide 19 text
!15
Combining the FFIs and the two-minute data can
allow for further confirmation of exciting systems.
Planet Transit
Stellar Eclipse
(Kostov, Welsh, Feinstein et al. in prep)
!15
Slide 20
Slide 20 text
!15
Combining the FFIs and the two-minute data can
allow for further confirmation of exciting systems.
Planet Transit
Stellar Eclipse
(Kostov, Welsh, Feinstein et al. in prep)
!15
Slide 21
Slide 21 text
!15
Combining the FFIs and the two-minute data can
allow for further confirmation of exciting systems.
Planet Transit
Stellar Eclipse
(Kostov, Welsh, Feinstein et al. in prep)
!15
Slide 22
Slide 22 text
!15
Combining the FFIs and the two-minute data can
allow for further confirmation of exciting systems.
Planet Transit
Stellar Eclipse
(Kostov, Welsh, Feinstein et al. in prep)
!15
Slide 23
Slide 23 text
!15
Combining the FFIs and the two-minute data can
allow for further confirmation of exciting systems.
Planet Transit
Stellar Eclipse
(Kostov, Welsh, Feinstein et al. in prep)
!15
Slide 24
Slide 24 text
!16
The eleanor data products are being transferred
to Mikulski Archive for Space Telescopes (MAST),
but the software is available now.
Slide 25
Slide 25 text
!17
The eleanor software has been
widely used by the community.
Slide 26
Slide 26 text
!17
The eleanor software has been
widely used by the community.
Slide 27
Slide 27 text
!17
The eleanor software has been
widely used by the community.
Slide 28
Slide 28 text
!18
Future plans including using eleanor to find
young planets for Rossiter-McLaughlin studies
with MAROON-X.
Slide 29
Slide 29 text
!19
Outline
1. Developing eleanor, a tool for light curve extraction
from TESS
2. Robustly identifying and characterizing flares of young
stars
3. Connecting flares and spots to understand stellar
activity
Slide 30
Slide 30 text
!20
What is the relationship between stellar
flare energy and age? Spectral type?
Slide 31
Slide 31 text
!21
What is the relationship between stellar
flare energy and age? Spectral type?
Slide 32
Slide 32 text
!22
Similar studies with Kepler demonstrated a relationship
between spectral type, rotation period, and flare energy.
(Davenport, 2016)
1.66 1.13 0.86 0.73 0.62 0.47 0.28
Mass [MSun
]
Maximum Log(Flare Energy [ergs])
32
33
34
35
36
37
38
39
40
Slide 33
Slide 33 text
!22
Similar studies with Kepler demonstrated a relationship
between spectral type, rotation period, and flare energy.
(Davenport, 2016)
1.66 1.13 0.86 0.73 0.62 0.47 0.28
Mass [MSun
]
Maximum Log(Flare Energy [ergs])
32
33
34
35
36
37
38
39
40
Slide 34
Slide 34 text
!22
Similar studies with Kepler demonstrated a relationship
between spectral type, rotation period, and flare energy.
(Yang & Liu, 2019)
(Davenport, 2016)
1.66 1.13 0.86 0.73 0.62 0.47 0.28
Mass [MSun
]
Maximum Log(Flare Energy [ergs])
32
33
34
35
36
37
38
39
40
Slide 35
Slide 35 text
!22
Similar studies with Kepler demonstrated a relationship
between spectral type, rotation period, and flare energy.
(Yang & Liu, 2019)
(Davenport, 2016)
1.66 1.13 0.86 0.73 0.62 0.47 0.28
Mass [MSun
]
Maximum Log(Flare Energy [ergs])
32
33
34
35
36
37
38
39
40
Slide 36
Slide 36 text
!23
What is the relationship between stellar
flare energy and age? Spectral type?
Slide 37
Slide 37 text
!24
We have completed a literature search for young
moving group members with ages 1-750 Myr.
Slide 38
Slide 38 text
!25
What is the relationship between stellar
flare energy and age? Spectral type?
Slide 39
Slide 39 text
!26
Previous flare studies have relied on sigma
clipping methods to identify flares.
Slide 40
Slide 40 text
!26
Previous flare studies have relied on sigma
clipping methods to identify flares.
(Chang, Byun, & Hartman, 2015)
Slide 41
Slide 41 text
!26
Previous flare studies have relied on sigma
clipping methods to identify flares.
(Chang, Byun, & Hartman, 2015)
:(
Slide 42
Slide 42 text
!27
What is the most complete method for
detecting flares of all energies?
Slide 43
Slide 43 text
!28
Machine learning can be used when searching
for signals with a characteristic shape.
Slide 44
Slide 44 text
!28
Machine learning can be used when searching
for signals with a characteristic shape.
(Pearson et al. 2017)
Slide 45
Slide 45 text
!28
Machine learning can be used when searching
for signals with a characteristic shape.
(Pearson et al. 2017)
Slide 46
Slide 46 text
We can use machine learning techniques
to automate finding flares in TESS data.
!29
Slide 47
Slide 47 text
!30
We simulate spot modulation and flares of
different amplitudes to pass in as the training set.
Slide 48
Slide 48 text
!31
Every light curve receives a label that the neural
network learns over several epochs.
Slide 49
Slide 49 text
!32
Feeding in a test training set assigns a
probability that the object is or is not a flare.
Slide 50
Slide 50 text
!33
We determine the probability that each
data point is a part of a potential flare.
Slide 51
Slide 51 text
!33
We determine the probability that each
data point is a part of a potential flare.
Slide 52
Slide 52 text
!34
The first results of applying the neural network to
a TESS two-minute target are promising.
Slide 53
Slide 53 text
!35
https://GitHub.com/afeinstein20/stella
Zooming in, we can see the confidence of the
neural network with identifying large flares.
Slide 54
Slide 54 text
!36
Outline
1. Developing eleanor, a tool for light curve extraction
from TESS
2. Robustly identifying and characterizing flares of young
stars
3. Connecting flares and spots to understand stellar
activity
Slide 55
Slide 55 text
!37
In the same year, two papers presented conflicting
results on where flares occur with relation to spots.
Slide 56
Slide 56 text
!37
In the same year, two papers presented conflicting
results on where flares occur with relation to spots.
(Doyle et al. 2018)
Slide 57
Slide 57 text
!37
In the same year, two papers presented conflicting
results on where flares occur with relation to spots.
(Doyle et al. 2018)
Slide 58
Slide 58 text
!37
In the same year, two papers presented conflicting
results on where flares occur with relation to spots.
(Doyle et al. 2018) (Roettenbacher et al. 2018)
1-5% increase
> 5% increase
Slide 59
Slide 59 text
!37
In the same year, two papers presented conflicting
results on where flares occur with relation to spots.
(Doyle et al. 2018) (Roettenbacher et al. 2018)
1-5% increase
> 5% increase
Slide 60
Slide 60 text
!37
In the same year, two papers presented conflicting
results on where flares occur with relation to spots.
(Doyle et al. 2018) (Roettenbacher et al. 2018)
1-5% increase
> 5% increase
Sample size: 32 Sample size: 119
Slide 61
Slide 61 text
!38
The take-away figure will show the location of flares
with respect to the phase of spot modulation.
Slide 62
Slide 62 text
!39
Summary
• We have created an open-source software package for
extracting light curves from the TESS FFIs.
• Light curve data products will be hosted on MAST for
community use.
• Currently, I am exploring flare properties of young (1-750
Myr) stars observed by TESS using a neural network, with
the hopes of answering how flares are related to starspots.