Speaker
Description
First-order phase transitions in hidden sectors provide a well-motivated source of stochastic gravitational waves and offer a complementary probe of physics beyond the Standard Model. In this talk, I will discuss gravitational wave production from a minimal dark U(1) sector, with emphasis on a gauge-independent formulation of the finite-temperature phase transition based on the Nielsen identity and a controlled effective action expansion. This framework enables robust predictions in the low-temperature and supercooled regime, where the resulting signals can be relevant for pulsar timing arrays and future space-based gravitational wave observatories. I will also present recent progress in using reinforcement learning, based on Proximal Policy Optimization, to efficiently explore the dark sector parameter space and identify detectable gravitational wave benchmarks. The results illustrate how gauge-independent formulation and reinforement learning searches can jointly sharpen gravitational wave probes of hidden sectors. This talk is primarily based on arXiv:2602.14866 and arXiv:2606.26251.