I am a theorist who studies the fundamental building blocks of astrophysics: stars. My work focuses on modeling stellar and binary evolution, asteroseismology, and tidal physics in binaries. I am a core developer of two widely used public codes, GYRE (an open-source stellar oscillations code) and POSYDON (a single- and binary-star population synthesis code), and I serve as a science advisor to MESA (the field-standard open-source stellar evolution code) with more than ten years of experience. My papers span exoplanets and massive stars, young and evolved stars, and compact objects.
Stellar Tides
Motivation and Challenges
Observations of close-in exoplanets and compact binaries show orbital decay and spin synchronization timescales that differ by one to three orders of magnitude from the predictions of simplified prescriptions such as the constant-Q model. These models cannot reproduce the observed inner boundary of hot-Jupiter systems or the rapid orbital decay seen in WASP-12 system.
Stellar tides involve a coupled process of forced oscillations and energy dissipation.
-
In the radiative core, the response is dominated by internal gravity waves and damped by radiative diffusion.
-
In the outer convective envelope, the quasi-static equilibrium tide is dissipated through turbulent viscosity.
Numerically, the problem requires solving the non-adiabatic oscillation equations with inhomogeneous forcing terms and boundary conditions at each tidal forcing frequency. Achieving stable and accurate solutions across the frequency spectrum, while also developing an open, parallel and maintainable code base, is a major technical challenge.
The GYRE-Tides Framework
Building on the GYRE stellar oscillation code, we developed and implemented GYRE-Tides. To our knowledge, it is the only publicly available code for frequency-dependent stellar tidal responses that is actively maintained by a team. The solver integrates the inhomogeneous, linear, non-adiabatic oscillation equations at each Fourier component of the tidal potential.
The open-source release converts a specialized numerical workflow into a reproducible community tool. It supports equilibrium and dynamical tides in aligned binaries.
Implementation, validation and applications are described in
Sun, Townsend, & Guo (2023, ApJ 945, 43), “GYRE-Tides: Modeling Binary Tides within the GYRE Stellar Oscillation Code.”
Showcase: The above figure shows the frequency-resolved tidal circularization timescale for a 5-solar-mass main-sequence star with a 1.4-solar-mass neutron-star companion. At each orbital period, the code solves the forced, non-adiabatic oscillation equations and computes the eccentricity damping rate. Sharp dips correspond to resonances with stellar modes. The labels indicate the mode type (g, p, or f), the harmonic index k, and whether the mode is prograde (p) or retrograde (r) in the corotating frame. Labels in bold mark overstabilized modes. Blue segments correspond to circularization (t_e < 0), and orange segments correspond to eccentricity growth (t_e > 0).
Before the public release of GYRE-Tides, I developed an internal prototype based on mode-by-mode tidal analysis and applied it to two problems.
1. Tidal dissipation and orbital decay in the WASP-12 system (Weinberg, Sun et al., 2017, ApJL, 849, 11).
WASP-12b is a typical hot Jupiter with an orbital period of about 1.1 days. Transit timing observations reveal that its orbital period is decreasing, implying an orbital-decay timescale of roughly 3.2 Myr. The observed stellar parameters allow two possible interpretations for the host star: a 1.3 solar mass main-sequence star or a 1.2 solar mass subgiant. In the main-sequence case, the tidal dissipation is too weak to explain the measured decay rate. In the subgiant case, nonlinear breaking of gravity waves in the radiative core can greatly enhance dissipation, producing an orbital-decay rate consistent with the observations. This scenario also naturally explains why the planet could survive for billions of years during the main-sequence phase but spirals inward rapidly once the star evolves off the main sequence. Given the current sample size, the chance of observing such a terminal-stage system is a few percent, low but not implausible.
2. Orbital-period distribution of evolved close binaries (Sun et al., 2018, MNRAS, 481, 4077).
This work targeted close binaries containing evolved stars. By computing both equilibrium and dynamical tides and coupling them with spin and orbital angular-momentum exchange, the model identified a critical orbital separation that depends on stellar mass and age. Systems below this separation lose orbital angular momentum efficiently and merge within a stellar evolutionary timescale; systems above it can survive for much longer. The model predicts a “survival boundary” below which few old binaries should exist. APOGEE data confirm that evolved binaries are indeed rare inside this limit. The result provides a quantitative inner boundary for long-term orbital stability in both binary and planet-hosting systems.
More recent works:
3. Tides in Massive Binaries: Numerical Solutions and Semi-Analytical Comparisons (Sun et al. 2025a, ApJ, submitted).
Using non-adiabatic forced-oscillation calculations for massive binaries, we compared the numerical results with the semi-analytic prescriptions commonly adopted in population-synthesis models. We found systematic differences between the two, with the default MESA tidal prescription tending to underestimate dissipation efficiency, especially for high-eccentricity systems. This work provides one of the first numerical benchmarks for such comparisons and shows that the commonly used analytic approximations can introduce systematic biases. In the currently testable cases, the frequency-dependent results from GYRE-Tides are closer to the observed decay rates, offering guidance for improving future prescriptions used in stellar-population simulations.
4. Extending GYRE-Tides to Exoplanet Systems (Sun et al. 2025b, ApJL, submitted).
We applied GYRE-Tides to recompute the non-adiabatic, linear tidal response of the WASP-12 system. Our linear calculations that include radiative diffusion and turbulent-convective damping reproduce the level in previous work, but yield decay timescales longer than the observed orbital decay rate, indicating that an additional dissipation mechanism is needed. A plausible explanation is wave breaking in or near the fully damped regime (by nonlinear damping). These contributions can be evaluated within our workflow by estimating the wave luminosity at the radiative–convective boundary or by using wave-steepness proxies. As the only open-source, actively maintained tool that computes orbit evolution for exoplanet systems, GYRE-tides provides a benchmark calculation for WASP-12 and a scalable framework for the many short-period systems expected from upcoming Roman surveys.
Binary Evolution
Core Scientific Questions and Modeling Challenges
Binary evolution involves complex physics, including stable mass transfer, common-envelope evolution, and magnetic braking. Rapid population-synthesis codes rely on empirical recipes and therefore struggle to predict reliably the formation rates and property distributions of observed systems such as gravitational-wave sources and blue stragglers. The central difficulties are well known: the stability of mass transfer depends sensitively on internal stellar structure, common-envelope evolution lacks a self-consistent physical description, and the angular-momentum evolution mechanisms, including magnetic braking and tides, remain debated. Large binary grids balance physical fidelity and computational cost while covering the full parameter space and maintaining numerical robustness.
The POSYDON team includes more than fifty developers. I am one of the long-term core contributors responsible for development. We compute more than 500,000 binary evolution tracks with MESA. Beyond launching and managing jobs, the main contribution was to diagnose and overcome convergence failures by analyzing the physical and numerical behavior of each problematic run, then proposing remedies that are both physically justified and computationally efficient. This work produced a physically self-consistent grid that classifies evolutionary pathways across mass ratios and orbital periods and provides a unified framework for the formation of gravitational-wave sources, X-ray binaries, and blue stragglers. The resulting models also provide theoretical support for the ground based and space based gravitational-wave missions.
The above Figure shows the complicated mapping from initial binary parameters and metallicities to different evolutionary outcomes and formation histories of compact objects.
Before joining the POSYDON collaboration, I completed two foundational studies using MESA.
1. Evolution and Asteroseismology of Extremely Low-Mass White Dwarfs (Sun & Arras 2018, ApJ, 858, 14)
Recent discoveries of pulsating extremely low-mass (ELM) white dwarfs revealed, for the first time, acoustic (p-mode) oscillations in a white dwarf, opening a new window on stellar interiors. I constructed detailed evolutionary models with MESA, showing that long-term magnetic braking removes orbital angular momentum, triggers stable mass transfer, and leaves behind an ELM white dwarf once the donor star exhausts its envelope and detaches. I then computed oscillation modes for a large grid of ELM models and matched the theoretical periods to observations, thereby inferring the core-envelope structure.
2. Solving the Blue Straggler Formation Puzzle (Sun et al. 2021, ApJ, 908, 7; Sun & Mathieu 2023, ApJ, 944, 89)
In the open cluster NGC 188, I used the MESA binary module to model two benchmark blue straggler systems that represent opposite ends of the orbital-period distribution.
-
For WOCS 5379 (short period), a stable but non-conservative mass transfer naturally reproduces the observed blue straggler + helium-core white dwarf configuration.
-
For WOCS 4540 (long period), including wind mass transfer produces a luminous blue straggler with a 0.53 solar mass carbon-oxygen white dwarf companion and an orbital period of about 3030 days.
Together, these models link the short- and long-period populations through a reproducible modeling, consistent with the observational picture proposed by Mathieu & Geller (2009, Nature) that most blue stragglers in old open clusters form via stable mass transfer.
Independent Research within the POSYDON Framework
In addition to code development, I have also acted as a scientific advisor within POSYDON and carried out two independent studies that incorporate new physical processes into numerical workflows. Both combine physical modeling with observational calibration, extending the scope of POSYDON as a population-synthesis framework and laying the foundation for the next generation of physically self-consistent binary grids.
3. Wind Mass Transfer and the Formation of Blue Lurkers (Sun et al. 2024a, ApJ, 969, 8)
“Blue lurkers” are post-interaction stars that remain close to the main sequence on the color–magnitude diagram, show no significant brightening, yet rotate unusually fast. They are thought to be the low-mass counterparts of blue stragglers. Within the POSYDON framework, I implemented a one-dimensional description of wind mass transfer into MESA and performed a parameter study across mass and period for wide, low-mass binaries. The joint transfer of mass and angular momentum spins up the accretor to near-critical rotation. The rotation-enhanced winds from the accretor, which continuously remove angular momentum and maintain the star below critical rotation. Under this feedback, the accretor gains only about 2 percent in mass and shows little luminosity increase, thus appearing as a blue lurker while remaining a rapid rotator. This framework explains why blue lurkers rotate fast without significant brightening.
4. Magnetic Braking and Post-Mass-Transfer Spin Evolution (Sun et al. 2024b, ApJ, 971, 80)
This study was motivated by the observed rotation of post-mass-transfer accretors. Using POSYDON, I modeled binary systems spanning a wide parameter range and tested four major magnetic braking prescriptions. During mass transfer, the accretor spins up rapidly to near-critical rotation; afterward, magnetic braking governs its angular-momentum evolution. Comparing models with observations shows that the 3D MHD magnetic-braking model of Garaffo et al. (2018) best reproduces the observed time evolution of rotation periods in post-mass-transfer stars, especially those in intermediate and long-period binaries. The results suggest that the accretor’s rotation period, combined with its mass, can serve as a “clock” since the end of mass transfer. This explains anomalously rapid rotators in gyrochronology and provides an observationally testable benchmark for magnetic braking in binaries.