For students

If you want to work on this, start with the papers, then pick a problem that is still open.

Reading

Mazur and Mottola, Universe 9, 88 (2023), arXiv:gr-qc/0109035
Mottola, Springer chapter (2023), arXiv:2302.09690
Mottola, Phys. Rev. D 111, 104018 (2025), arXiv:2502.02519
The effective theory behind that last paper: Mottola, JHEP 11, 037 (2022), arXiv:2205.04703

The rest of the publication list is on Google Scholar.

Open problems

Rotation. The effective-theory construction is under control for the non-rotating case. A Kerr analog, with a well-posed surface, frame dragging, and a discrete spectrum, is what can actually be compared with LVK data.
Formation. Collapse with a nucleating de Sitter core is no longer an empty subject. There is, for example, a 2026 Phys. Rev. D letter on forming a gravastar from dust. How generic that is, and what happens when the progenitor is spinning, is still open.
Echo templates. Searches need waveforms: discrete surface modes, a time delay that depends on compactness, and a template that disappears into detector noise the way a real echo would.
Electromagnetic tests. EHT polarimetry, photon-ring substructure, and possible light echoes. The exterior is Schwarzschild or Kerr at first look. The question is which of the next decimals is not.

Existing constraints

Chirenti and Rezzolla, Phys. Rev. D 94, 084016 (2016), arXiv:1602.08759, concluded that the measured ringdown of GW150914 cannot be modeled as a rotating gravastar. Echo claims in the early LIGO data sat at about 1.5σ. The collaboration treated them as noise. Searches in O3, and later model-agnostic work on loud events including GW150914, found no significant post-merger echoes and set amplitude limits. Those papers should be cited. A model that predicts a loud echo in the existing data is already in conflict with them.

The 2023 Universe review says the ringdown is not yet precise enough to test the discrete spectrum. That gap is the one worth writing into.

Email: emottola@unm.edu.