Observation and Astrophysics

Black-hole astrophysics connects measurable light and gravitational waves to hidden properties: absorption, geometry, birth mass, growth history, and formation environment. A good project makes one part of that connection testable. These six directions offer different entry points for students with analytic, computational, or observational interests.

The records assess evidence through 2026-08-20. Their early entry projects describe proposed reproductions and possible extensions; they do not promise a discovery or establish that an extension is new. Use the dated evidence below each project to learn which interpretation your calculation could actually change.

Compare the six directions

Direction and finite question	A manageable starting point	The limiting inference
Horizon versus surface: which surface responses are excluded?	Reconstruct a thermal-emission bound	Accretion, equilibration time, and reradiation assumptions
Imaging tests of Kerr: how does an observed ring constrain geometry?	Fit and inject a simple visibility-domain model	Plasma, variability, calibration, and mass-to-distance information
Massive-black-hole seeds: what can descendants tell us about their seeds?	Reproduce growth tracks reaching one endpoint	Duty cycle, radiative efficiency, source classification, and selection
Pair-instability gap: what remnant feature survives into a binary population?	Reproduce a component-mass distribution	Stellar uncertainty, mixtures, mass ordering, and detection efficiency
Binary counterparts: is a transient associated with a merger?	Recompute one spatial-overlap statistic	Flare background, trials, delay model, and survey coverage
Formation channels: which causal histories are distinguishable?	Test a synthetic mixture with an omitted component	Incomplete channel libraries and uncertain formation physics

Choose an entry route

With analytic preparation and a laptop, thermal-surface bounds and seed-growth histories are good places to start. You can derive the energy budget or growth equation and propagate published uncertainty without first reducing raw observations. The hard part is deciding which physical assumptions the observation tests. A more detailed model is useful only if it predicts something that could distinguish alternatives.

With Fourier analysis and inverse-problem experience, horizon imaging offers a direct connection between instrument sampling and physical inference. Begin with calibrated visibilities and synthetic data. A collaboration’s image is already a reconstruction; fitting its pixels as though they were independent raw measurements does not reproduce the likelihood. An interferometry supervisor is especially valuable before using station gains, closure statistics, or source variability.

With Bayesian statistics and population interests, the mass gap and formation channels share tools but ask different questions. The former follows a stellar mechanism into a mixed population; the latter tests whether causal models predict the population adequately. Released summary products offer a first exercise. A new hierarchical fit needs event likelihoods, compatible selection injections, and Monte Carlo diagnostics. Large posterior archives also make storage a real planning consideration.

With time-domain astronomy support, counterpart association combines gravitational-wave localization with transient surveys. A one-event overlap is accessible; a rate claim needs survey exposure and a background population. Establish access to those inputs before choosing a catalog-wide project.

How the questions connect

1. For compact-object identity, read the surface-evidence background, then follow Broderick–Narayan → EHT Paper VI in the surface project. Extract the energy-conservation and equilibration assumptions. Continue to the imaging project and identify the separate steps from visibility to brightness ring to metric.
2. For origins and growth, begin with massive seeds for a growth-degeneracy exercise, or pair instability for a stellar-remnant exercise. From the named primary papers, record the latent mass being inferred and every process between that mass and the data. Read the formation-channel project next for a practical test of model inadequacy.
3. For environmental inference, follow the counterpart project’s Ashton → Morton comparison, extracting the changed waveform, prior, and background assumptions. Then study the formation-channel project’s Zevin → Passenger sequence. Decide what independent host, eccentricity, or transient observable could distinguish channels that overlap in mass and spin.

Design a project whose answer will remain useful

Specify the observation you will reproduce, the physical parameter you want to learn, and the nuisance effect most likely to defeat that inference. Choose a finite output: one spectral bound, one recovered ring diameter, one family of growth tracks, one population summary, or one association statistic. The leaf projects give concrete inputs and conditions for stopping or narrowing each calculation.

For a thesis-scale extension, establish a working relationship with the relevant community: radiative transfer, VLBI, stellar evolution, galaxy formation, gravitational-wave populations, or survey transients. Access to a new dataset alone does not supply a calibrated likelihood. If the missing ingredient is a selection function or physical forward model, making that ingredient explicit can be a stronger project than adding another fit with unconstrained assumptions.