Rachel C. Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Rachel C. Zhang has a total of 25 co-authors. Here are the top co-authors by number of shared publications:
- Chase Kimball
Northwestern University
2 shared publications - Giacomo Fragione
Northwestern University
2 shared publications - Adam Ingram
Newcastle University
1 shared publication - Andrew C. Fabian
University of Cambridge
1 shared publication Anne Lohfink
10 Department of Physics, Montana State University, P.O. Box 173840, Bozeman, MT 59717-3840, USA
1 shared publication- Bing Zhang
Nanjing University
1 shared publication
Latest Publications
- Open Access
- 11 December 2025
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 995 (2) , 136
Predicting Stellar Parameters of Massive Stars from Light Curves with Machine Learning
High-resolution spectroscopic measurements of OB stars are important for understanding processes like stellar evolution but require labor-intensive observations. In contrast, photometric missions like the Transiting Exoplanet Survey Satellite (TESS) can monitor hundreds of thousands of stars with a range of temporal resolutions but do not provide such detailed measurements. With surveys like the Legacy Survey of Space and Time promising unprecedented photometric coverage over the next 10 yr, it is increasingly important to develop methods that connect large-scale time-series photometry with the detailed stellar parameter measurements typically derived from spectroscopy. In this paper, we test whether machine learning can recover such parameters by combining TESS light curves with spectroscopic measurements from the IACOB project, using a sample of 285 light curves from 106 unique O stars. Using both multilayer perceptrons and convolutional neural networks, we demonstrate that (1) O star light curves contain sufficient information to meaningfully infer stellar parameters, and (2) periodograms derived from light curves capture substantially more information than previously identified correlation parameters. Our best model achieves moderate success in predicting both spectroscopic luminosity ( R 2 = 0.64 1 − 0.167 + 0.107 ) and effective temperature ( R 2 = 0.44 3 − 0.234 + 0.056 ), key stellar parameters for determining positions of stars on the spectroscopic Hertzsprung–Russell diagram, despite the small dataset size. Further progress will require expanded datasets of matched photometric and spectroscopic observations.
- 11 December 2025
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 995 (2) , 136
Open AccessPredicting Stellar Parameters of Massive Stars from Light Curves with Machine Learning
High-resolution spectroscopic measurements of OB stars are important for understanding processes like stellar evolution but require labor-intensive observations. In contrast, photometric missions like the Transiting Exoplanet Survey Satellite (TESS) can monitor hundreds of thousands of stars with a range of temporal resolutions but do not provide such detailed measurements. With surveys like the Legacy Survey of Space and Time promising unprecedented photometric coverage over the next 10 yr, it is increasingly important to develop methods that connect large-scale time-series photometry with the detailed stellar parameter measurements typically derived from spectroscopy. In this paper, we test whether machine learning can recover such parameters by combining TESS light curves with spectroscopic measurements from the IACOB project, using a sample of 285 light curves from 106 unique O stars. Using both multilayer perceptrons and convolutional neural networks, we demonstrate that (1) O star light curves contain sufficient information to meaningfully infer stellar parameters, and (2) periodograms derived from light curves capture substantially more information than previously identified correlation parameters. Our best model achieves moderate success in predicting both spectroscopic luminosity ( R 2 = 0.64 1 − 0.167 + 0.107 ) and effective temperature ( R 2 = 0.44 3 − 0.234 + 0.056 ), key stellar parameters for determining positions of stars on the spectroscopic Hertzsprung–Russell diagram, despite the small dataset size. Further progress will require expanded datasets of matched photometric and spectroscopic observations.
- Open Access
- 21 August 2023
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 954 (1) , 23
On the Likely Dynamical Origin of GW191109 and Binary Black Hole Mergers with Negative Effective Spin
With the growing number of binary black hole (BBH) mergers detected by LIGO/Virgo/KAGRA, several systems have become difficult to explain via isolated binary evolution, having components in the pair-instability mass gap, high orbital eccentricities, and/or spin–orbit misalignment. Here we focus on GW191109_010717, a BBH merger with component masses of 65 − 11 + 11 and 47 − 13 + 15 M ⊙ and an effective spin of − 0.29 − 0.31 + 0.42 , which could imply a spin–orbit misalignment of more than π/2 rad for at least one of its components. Besides its component masses being in the pair-instability mass gap, we show that isolated binary evolution is unlikely to reproduce the proposed spin–orbit misalignment of GW191109 with high confidence. On the other hand, we demonstrate that BBHs dynamically assembled in dense star clusters would naturally reproduce the spin–orbit misalignment and masses of GW191109 and the rates of GW191109-like events if at least one of the components were to be a second-generation BH. Finally, we generalize our results to all events with a measured negative effective spin, arguing that GW200225 also has a likely dynamical origin.
- 21 August 2023
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 954 (1) , 23
Open AccessOn the Likely Dynamical Origin of GW191109 and Binary Black Hole Mergers with Negative Effective Spin
With the growing number of binary black hole (BBH) mergers detected by LIGO/Virgo/KAGRA, several systems have become difficult to explain via isolated binary evolution, having components in the pair-instability mass gap, high orbital eccentricities, and/or spin–orbit misalignment. Here we focus on GW191109_010717, a BBH merger with component masses of 65 − 11 + 11 and 47 − 13 + 15 M ⊙ and an effective spin of − 0.29 − 0.31 + 0.42 , which could imply a spin–orbit misalignment of more than π/2 rad for at least one of its components. Besides its component masses being in the pair-instability mass gap, we show that isolated binary evolution is unlikely to reproduce the proposed spin–orbit misalignment of GW191109 with high confidence. On the other hand, we demonstrate that BBHs dynamically assembled in dense star clusters would naturally reproduce the spin–orbit misalignment and masses of GW191109 and the rates of GW191109-like events if at least one of the components were to be a second-generation BH. Finally, we generalize our results to all events with a measured negative effective spin, arguing that GW200225 also has a likely dynamical origin.
- Open Access
On the Likely Dynamical Origin of GW191109 and of Binary Black Hole Mergers with Negative Effective Spin
With the growing number of binary black hole (BBH) mergers detected by LIGO/Virgo/KAGRA, several systems have become difficult to explain via isolated binary evolution, having components in the pair-instability mass gap, high orbital eccentricities, and/or spin-orbit misalignment. Here, we focus on GW191109\_010717, a BBH merger with component masses of $65^{+11}_{-11}$ and $47^{+15}_{-13}$ $\rm M_{\odot}$, and effective spin $-0.29^{+0.42}_{-0.31}$, which could imply a spin-orbit misalignment of more than $π/2$ radians for at least one of its components. Besides its component masses being in the pair-instability mass gap, we show that isolated binary evolution is unlikely to reproduce the proposed spin-orbit misalignment of GW191109 with high confidence. On the other hand, we demonstrate that BBHs dynamically assembled in dense star clusters would naturally reproduce the spin-orbit misalignment and the masses of GW191109, and the rates of GW191109-like events, if at least one of the components were to be a second-generation BH. Finally, we generalize our results to all the events with a measured negative effective spin, arguing that GW200225 also has a likely dynamical origin.Open AccessOn the Likely Dynamical Origin of GW191109 and of Binary Black Hole Mergers with Negative Effective Spin
With the growing number of binary black hole (BBH) mergers detected by LIGO/Virgo/KAGRA, several systems have become difficult to explain via isolated binary evolution, having components in the pair-instability mass gap, high orbital eccentricities, and/or spin-orbit misalignment. Here, we focus on GW191109\_010717, a BBH merger with component masses of $65^{+11}_{-11}$ and $47^{+15}_{-13}$ $\rm M_{\odot}$, and effective spin $-0.29^{+0.42}_{-0.31}$, which could imply a spin-orbit misalignment of more than $π/2$ radians for at least one of its components. Besides its component masses being in the pair-instability mass gap, we show that isolated binary evolution is unlikely to reproduce the proposed spin-orbit misalignment of GW191109 with high confidence. On the other hand, we demonstrate that BBHs dynamically assembled in dense star clusters would naturally reproduce the spin-orbit misalignment and the masses of GW191109, and the rates of GW191109-like events, if at least one of the components were to be a second-generation BH. Finally, we generalize our results to all the events with a measured negative effective spin, arguing that GW200225 also has a likely dynamical origin. - Open Access
- 1 November 2022
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 939 (2) , 109
X-Ray Reverberation Mapping of Ark 564 Using Gaussian Process Regression
Ark 564 is an extreme high-Eddington narrow-line Seyfert 1 galaxy, known for being one of the brightest, most rapidly variable soft X-ray active galactic nuclei (AGN), and for having one of the lowest temperature coronae. Here, we present a 410 ks NuSTAR observation and two 115 ks XMM-Newton observations of this unique source, which reveal a very strong, relativistically broadened iron line. We compute the Fourier-resolved time lags by first using Gaussian processes to interpolate the NuSTAR gaps, implementing the first employment of multitask learning for application in AGN timing. By simultaneously fitting the time lags and the flux spectra with the relativistic reverberation model reltrans, we constrain the mass at 2.3 − 1.3 + 2.6 × 10 6 M ⊙ , although additional components are required to describe the prominent soft excess in this source. These results motivate future combinations of machine learning, Fourier-resolved timing, and the development of reverberation models.
- 1 November 2022
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 939 (2) , 109
Open AccessX-Ray Reverberation Mapping of Ark 564 Using Gaussian Process Regression
Ark 564 is an extreme high-Eddington narrow-line Seyfert 1 galaxy, known for being one of the brightest, most rapidly variable soft X-ray active galactic nuclei (AGN), and for having one of the lowest temperature coronae. Here, we present a 410 ks NuSTAR observation and two 115 ks XMM-Newton observations of this unique source, which reveal a very strong, relativistically broadened iron line. We compute the Fourier-resolved time lags by first using Gaussian processes to interpolate the NuSTAR gaps, implementing the first employment of multitask learning for application in AGN timing. By simultaneously fitting the time lags and the flux spectra with the relativistic reverberation model reltrans, we constrain the mass at 2.3 − 1.3 + 2.6 × 10 6 M ⊙ , although additional components are required to describe the prominent soft excess in this source. These results motivate future combinations of machine learning, Fourier-resolved timing, and the development of reverberation models.
- Open Access
- 1 January 2022
- Published by American Astronomical Society in The Astrophysical Journal Letters
- Vol. 924 (1) , L14
The CHIME Fast Radio Burst Population Does Not Track the Star Formation History of the Universe
The redshift distribution of fast radio bursts (FRBs) is not well constrained. The association of the Galactic FRB 200428 with the young magnetar SGR 1935+2154 raises the working hypothesis that FRB sources track the star formation history of the universe. The discovery of FRB 20200120E in association with a globular cluster in the nearby galaxy M81, however, casts doubts on such an assumption. We apply the Monte Carlo method developed in a previous work to test different FRB redshift distribution models against the recently released first CHIME FRB catalog in terms of their distributions in specific fluence, external dispersion measure (DME), and inferred isotropic energy. Our results clearly rule out the hypothesis that all FRBs track the star formation history of the universe. The hypothesis that all FRBs track the accumulated stars throughout history describes the data better but still cannot meet both the DME and the energy criteria. The data seem to be better modeled with either a redshift distribution model invoking a significant delay with respect to star formation or a hybrid model invoking both a dominant delayed population and a subdominant star formation population. We discuss the implications of this finding for FRB source models.
- 1 January 2022
- Published by American Astronomical Society in The Astrophysical Journal Letters
- Vol. 924 (1) , L14
Open AccessThe CHIME Fast Radio Burst Population Does Not Track the Star Formation History of the Universe
The redshift distribution of fast radio bursts (FRBs) is not well constrained. The association of the Galactic FRB 200428 with the young magnetar SGR 1935+2154 raises the working hypothesis that FRB sources track the star formation history of the universe. The discovery of FRB 20200120E in association with a globular cluster in the nearby galaxy M81, however, casts doubts on such an assumption. We apply the Monte Carlo method developed in a previous work to test different FRB redshift distribution models against the recently released first CHIME FRB catalog in terms of their distributions in specific fluence, external dispersion measure (DME), and inferred isotropic energy. Our results clearly rule out the hypothesis that all FRBs track the star formation history of the universe. The hypothesis that all FRBs track the accumulated stars throughout history describes the data better but still cannot meet both the DME and the energy criteria. The data seem to be better modeled with either a redshift distribution model invoking a significant delay with respect to star formation or a hybrid model invoking both a dominant delayed population and a subdominant star formation population. We discuss the implications of this finding for FRB source models.