A Systematic Search for Main-Sequence Dipper Stars with ZTF

Anastasios (Andy) Tzanidakis

in Collaboration with LINCC Frameworks and UW DiRAC

atzanida@uw.edu // andytza.github.io //      AndyTza

Characterizing Stellar Variability Across the HR Diagram

  1. The underlying goal is to understand stellar variability within the context and bounds of the HR diagram.
     
  2. Stellar variability tree and classifications.
     
  3. What is the rarest variability throughout the HR diagram?

Image Credit: ESA

ESA / Gaia

Andy Tzanidakis / atzanida.github.io

Boyajian Star

Andy Tzanidakis / atzanida.github.io

  • During the primary Kepler mission, KIC 8462852 exhibited non-periodic complex ~3-10 day long dimming events (Boyajian et al. 2016)
     
  • Montet & Simon 2016 + showed a long-term secular dimming trend, beyond the typical amplitude range of stellar cycles
     
  • Evidence for size scale <1um based on post-Kepler dimming (Boyajian et al. 2018)

Boyajian Star

Andy Tzanidakis / atzanida.github.io

FGK Dwarfs

Largest Systematic Search for MS Dippers

  • Make an initial selection with Gaia eDR3 StarHorse (Anders +2022)
    • Obtain stellar parameters and distances for FGK dwarf selection
    • Select ~100M candidates
  • Crossmatch to Zwicky Transient Facility (ZTF) survey object catalog
    • Remove spurious photometry and bad light curves
    • N~63M stars (~126M g+r light curves)

Next Generation Time Domain Tools

StarHorse 2.0
(Andersell et al.)

LSDB

Nested Pandas

Object (1)

(catalogs)

ZTF

PS1

(object)

(object)

(object)

(object)

LSDB (crossmatching)

ZTF Zubercal DR14

(photometry)

Nested Pandas DataFrame

existing HATS obj

existing HATS obj

F(\mathcal{D})

Data Pipeline

Master

+

Selections + Hierarchical Adaptive Tiling Scheme

 

Full ZTF DR14 Photometry

Identify Event

Scoring

Tzanidakis et al. (2025; in-prep)

Evaluate score:

\text{Score} = \frac{1}{ln(N+1)} \sum_N d_i \times \Phi \times n

Compute features:

\text{Statistics}: J, K, ..., \Theta

Filtering criteria:

\text{(per band)}
  • SNR threshold cuts
  • ZTF bogus threshold cuts
  • ...

Performance

Time

LSDB crossmatching
and HATS partitioning

minutes

Full computation
120 million sources with over 25 columns...

~12-14 hrs

Scalability

ZTF-g

ZTF-r

Discovery of ~80 New Rare MS Dippers

Using injection-recovery methods with LSDB & Nested, we're able to measure the score threshold of bonafide dippers

CMD Findings

Absent IR Variability

2MASS/WISE photometry and CMD positions indicate that candidates are not consistent with Class I/II YSO's...

Absent Ha Variability

Photometry via IPHAS DR2 shows that at least ~40% of our sample has no significant H-alpha emission...

PS1 griz

PS1 griz

Tzanidakis et al. 2025 (in-prep)

DBSP, P200"

DBSP, P200"

Our systematic search through >63 million FGK dwarfs has demonstrated that MS dippers are exceptionally rare objects ~1/M rate.

New Dipper Demographics

User Feedback

Easy Pandas API makes tools accessible to the user interacting with catalogs... Nested Pandas makes it very easy to apply Python time-series analysis...

The scalability of these tools has been my favorite part of this project. It has helped me expand my understanding of data pipelines. The barrier to access the bulk data doesn't feel difficult anymore

1.

2.

3.

I still want to better understand how Dask handles memory allocation, how to properly allocate CPU and memory clients...

Future Projects

Student collaboration: Tessa Ward, and Cassidy Johnson (University of Washington): 

  • Interested in searching for MS dippers in open clusters for clues of robust stellar age...
  • Plan on using a similar workflow for their analysis

Optical + IR searches for Giant Impacts

The first volume-limited systematic search for signatures of Giant Impacts to probe terrestrial planet formation!

Conclusions

Special thanks to Neven Caplar,  Doug Branton, Wilson Beebe, Andy Connolly, and the LINCC Frameworks team for making this project possible!

We are conducting the largest systematic search for main-sequence/Boyajian star analogs with ZTF and Gaia.

Both Large Survey DataBase (LSDB) and Nested Pandas, are pivotal tools for the success of this study - providing: data exploration, discovery, and rapid analysis

1.

2.

Project can be found on GitHub: github.com/dirac-institute/ZTF_FG_BoyajianSearch

Andy Tzanidakis// andytza.github.io // atzanida@uw.edu

3.

Our preliminary pipeline using LSDB + NP has already revealed over 80 new and exciting main-sequence dipper candidates!!

Backup Slides

Crossmatch

Workflow

Pipeline

Nested Table

Object
Data

Source
Data

Final data output & merge with Nested

Source

(photometry)

Object (1)

(catalogs)

Object (1)

(catalogs)

Object (1)

(catalogs)

Object (N)

(catalogs)

ZTF Systematics

Kepler Statistics

Jackson & Wyatt (2012)

Giant Impact Debris

Dask Dashboard

Slow Dippers

Slow and longterm secular trends can be found in simple light curve statistics

Typical activity cycles are in the 10-30 mmag

Mascareño + 2016

args = ImportArguments(
    ra_column="RA_ICRS",
    dec_column="DE_ICRS",
    lowest_healpix_order=2,
    highest_healpix_order=7,
    file_reader="parquet",
    input_file_list=["/nvme/users/atzanida/tmp/starhorse_query.parquet"],
    output_artifact_name="StarHorse_hats", # output file name
    output_path="/nvme/users/atzanida/tmp/",
    resume=False,
)

with Client(n_workers=4) as client:
    pipeline_with_client(args, client)

StarHorse 2.0

StarHorse 2.0
HATS

sample = ztf.crossmatch(starhorse_hips, radius_arcsec=1)

Code Pipeline

Andy Tzanidakis / atzanida.github.io

Infrared Properties?

  • The IR locus does not seem consistent with young (Type I/II YSOs) but could be more consistent with WTT's or, transition disks...
     
  • Re-radiated dust from just the stellar photospheres could explain why we don't see any dust excess or strong IR variability.
     
  • Surprisingly, we did not find any significant IR variability on the ~100-day timescale (W1 and W2)

D/S Type

Symbiotic Binaries

Newly discovered
Sb Binary!

Probing Circumstellar Material

Debris

Planets, Comets, and Stars

Planet Englufment

Young Stars

Old(er) Stars

Interacting Stars

Characterizing Stellar Variability Across the HR Diagram

  1. The underlying goal is to understand stellar variability within the context and bounds of the HR diagram.
     
  2. Stellar variability tree and classifications.
     
  3. What is the rarest variability throughout the HR diagram?
f(\mathcal{D})

Gaia DR3

Customizable Python time-series evaluation routines (e.g. fitting, time-series features, ...)

ZTF (src) DR14

ZTF DR14

LSDB

Fast object catalog operations and querying 

TAPE

Bridges object

and source catalogs

Large Survey DataBase

Timeseries Analysis & Processing Engine

Main Sequence Dippers with ZTF - LINCC Summary

By Anastasios Tzanidakis

Made with Slides.com

Main Sequence Dippers with ZTF - LINCC Summary

  • 17

More from Anastasios Tzanidakis