SLIDE 1 New UX Ori ri type pe candi ndida dates de detect cted d us using ng Gaia DR2 and nd Ma Machi chine ne Learni rning ng
Miguel Vioque
University of Leeds
- R. D. Oudmaijer (University of Leeds, UK), M. Schreiner (Desupervised, Denmark),
- D. Baines (ESAC, Spain), and R. Pérez-Martínez (Isdefe, Spain)
The UX Ori type stars and related topics, 1st of October, 2019
SLIDE 2 Hi High gh-Ma Mass Star r Form rmation
Intermediate-mass T-Tauri stars
Spectral types ~F5 to B0 Mass: 2 − 10𝑁⨀
Herbig Ae/Be stars T-Tauri stars
Alecian, et al. (2013), Villebrun, et al. (2019)
SLIDE 3 Hi High gh-Ma Mass Star r Form rmation
Intermediate-mass T-Tauri stars
Spectral types ~F5 to B0 Mass: 2 − 10𝑁⨀
Around 260 known to date
Herbig Ae/Be stars T-Tauri stars
Alecian, et al. (2013), Villebrun, et al. (2019)
SLIDE 4
Hi High gh-Ma Mass Star r Form rmation
SLIDE 5 Vioque, et al. (2018)
All known Herbig Ae/Be stars
Hi High gh-Ma Mass Star r Form rmation
Gaia DR2
SLIDE 6 Vioque, et al. (2018)
All known Herbig Ae/Be stars
Hi High gh-Ma Mass Star r Form rmation
Gaia DR2
SLIDE 7 Gaia vari riabi bility
Vioque et al. 2018
𝑊
) ~ 𝜏 𝐺,
𝑂./0 𝐺,
Deason et al. 2017
SLIDE 8 Gaia vari riabi bility
Vioque et al. 2018
𝑊
) ~ 𝜏 𝐺,
𝑂./0 𝐺,
Deason et al. 2017
𝑊
) > 2
All known UXORs in the sample (17)
With variabilities larger than 0.5mag …
SLIDE 9 Gaia vari riabi bility
Vioque et al. 2018
𝑊
) ~ 𝜏 𝐺,
𝑂./0 𝐺,
Deason et al. 2017
𝑊
) > 2
All known UXORs in the sample (17)
With variabilities larger than 0.5mag …
Proposed 31 new UX Ori candidates
SLIDE 10 Looking for new Pre-Main Sequence (PMS) objects in Gaia! Main characteristics of PMS objects:
- Infrared excesses
- H𝛽 emission
- Photometric variability
SLIDE 11 Main characteristics of PMS objects:
- Infrared excesses
- H𝛽 emission
- Photometric variability
High mass PMS objects (Herbig Be stars) are very similar to Classical Be stars Looking for new Pre-Main Sequence (PMS) objects in Gaia!
... and supergiants, B[e] stars, …
SLIDE 12 Main characteristics of PMS objects:
- Infrared excesses
- H𝛽 emission
- Photometric variability
High mass PMS objects (Herbig Be stars) are very similar to Classical Be stars
Perform an homogeneous selection, distance and position independent!
Looking for new Pre-Main Sequence (PMS) objects in Gaia!
... and supergiants, B[e] stars, …
SLIDE 13 Before training:
- Training Set
- Set of characteristics
- Set of categories
Algorithm is trained with known labeled data
After generalizing:
a probability
algorithm The best architecture is selected
Neur ural Network rk
SLIDE 14 Selection of the characteristics:
𝑿𝟐, 𝑿𝟑, 𝑿𝟒, 𝑿𝟓 𝑲, 𝑰, 𝑳𝒕
AllWISE (WISE+2MASS)
IPHAS VPHAS+
𝒔 − 𝑰𝜷
Gaia
𝑪𝒒, 𝑯, 𝑺𝒒 2 variability indicators
Neur ural Network rk
SLIDE 15 Selection of the characteristics:
𝑿𝟐, 𝑿𝟑, 𝑿𝟒, 𝑿𝟓 𝑲, 𝑰, 𝑳𝒕
AllWISE (WISE+2MASS)
IPHAS VPHAS+
𝒔 − 𝑰𝜷
Gaia
𝑪𝒒, 𝑯, 𝑺𝒒 2 variability indicators
Neur ural Network rk
Create all possible colours
Distance independent!
SLIDE 16
Selection of the categories: PMS category Classical Be category Other sources
SLIDE 17 Selection of the Training Set: PMS category Classical Be category Other sources
AllWISE IPHAS VPHAS+ Gaia
+ + = 4,151,538
sources
SLIDE 18 Selection of the Training Set: PMS category Classical Be category Other sources
+ + = 4,151,538
sources
- 848 Pre-Main Sequence
- bjects (163 Herbig
Ae/Be)
- 775 Classical Be stars
- 471,111 random sources
AllWISE IPHAS VPHAS+ Gaia
SLIDE 19
Training the Neural Network
SLIDE 20 Trained Neural Network
AllWISE IPHAS VPHAS+ Gaia + + = 4,151,538
sources
SLIDE 21
Proba babi bility Ma Map
SLIDE 22
Proba babi bility Ma Map
636 Classical Be candidates 1266 either 8452 PMS candidates 4,140,629 other
SLIDE 23
Proba babi bility Ma Map
636 Classical Be candidates 1291 either 8452 PMS candidates 4,140,629 other PMS Completeness 𝟖𝟗. 𝟗 ± 𝟐. 𝟓% Classical Be Completeness 𝟗𝟔. 𝟔 ± 𝟐. 𝟑% Evaluation on Test Set
SLIDE 24 ¡1 1 2 3 4 Bp ¡ Rp ¡4 ¡2 2 4 6 8 10 12 14 MG [mag] ¡1 1 2 3 4 Bp ¡ Rp ¡4 ¡2 2 4 6 8 10 12 14 MG [mag]
Gaia HR di diagram
¡1 1 2 3 4 Bp ¡ Rp ¡4 ¡2 2 4 6 8 10 12 14 MG [mag] ¡1 1 2 3 4 Bp ¡ Rp ¡4 ¡2 2 4 6 8 10 12 14 MG [mag]
PMS candidates > 50% PMS candidates > 65% PMS candidates > 80% PMS candidates > 95%
SLIDE 25
Co Coordi dina nates
SLIDE 26
Co Coordi dina nates
SLIDE 27
Co Coordi dina nates
SLIDE 28 UX Ori ri candi ndida dates
Vioque et al. 2018
SLIDE 29 Vioque et al. 2018
UX Ori ri candi ndida dates
Proposed 31 new UX Ori candidates
SLIDE 30 UX Ori ri candi ndida dates
0:5 0:6 0:7 0:8 0:9 1:0 Probability PMS 50 100 150 200 250 300 350 Gaia Variability
Vioque et al. in prep
3436 UX Ori candidates
SLIDE 31 UX Ori ri candi ndida dates
0:5 0:6 0:7 0:8 0:9 1:0 Probability PMS 50 100 150 200 250 300 350 Gaia Variability
Vioque et al. in prep
3436 UX Ori candidates ~40% of the PMS candidates
SLIDE 32 H𝛽 emission Reddening Main Sequence
PMS Candidates
Ca Caveats
SLIDE 33 PMS Candidates
H𝛽 emission Reddening Main Sequence
Ca Caveats
SLIDE 34 PMS Candidates
H𝛽 emission Reddening Main Sequence Planetary Nebulae!
Ca Caveats
SLIDE 35 Fut Futur ure work rk
Populate HR diagram Past and future observations
INT 2.2m Calar Alto NTT
+ +
𝟑𝟕𝟏 objects + ~𝟒𝟏𝟏𝟏 objects
¡1 1 2 3 4 Bp ¡ Rp ¡6 ¡4 ¡2 2 4 6 8 10 12 MG [mag] 0:55 0:60 0:65 0:70 0:75 0:80 0:85 0:90 0:95 Probability PMS ¡1 1 2 3 4 Bp ¡ Rp ¡6 ¡4 ¡2 2 4 6 8 10 12 MG [mag]
SLIDE 36
- We retrieve 8452 new PMS
- candidates. 3131 potential
Herbig Ae/Be stars.
- We retrieve 636 new Classical
Be stars candidates. Completeness 𝟖𝟗. 𝟗 ± 𝟐. 𝟓%
Resul ults
Completeness 𝟗𝟔. 𝟔 ± 𝟐. 𝟑% We retrieve 3436 new UX Ori type stars candidates
IDS/INT 𝐼M line spectra Gaia 2204517656901678848 𝐻 = 14.0mag, 𝑒 = 940pc
