Sid Mahesh

I am a graduate student at the Centre for Gravitational Waves and Cosmology at West Virginia University. I am passionate about researching black hole binaries and their environments.

I primarily work on developing analytical approaches to modelling the gravitational wave signals from the merger of black hole binaries. When analytically modeling such systems, the motion of the binary system is approximated as relativistic corrections to Newtonian gravity (post-Newtonian, or PN) during the inspiral part, or as effective perturbations about the remnant black hole during the ringdown part. Since the two regimes use different physical processes, it becomes crucial to develop a well-motivated, accurate prescription to transition from one regime to the other. This is where BOB (Backwards One Body) and I enter the picture. By modeling the gravitational wave propagation at merger as null divergences about the spacetime of the remnant black hole, the BOB formalism can create a way to evolve the end conditions of the PN inspiral to the quasi-normal mode emission of the remnant black hole. You can find out more about how this is accomplished in my paper. You can also access tools to generate waveforms using BOB at NRPy, where I have developed the seobnrv5_aligned_spin_inspiral example.

I also work on analytically modeling the impact of binaries on circumbinary accretion disks. When an accretion disk forms around the collective binary system, the gravitational force on the disk is no longer central and breaks angular momentum conservation. The resulting torque on the disk can make the interaction between the disk and the binary more complex, leading to interesting phenomena such as the clearing of a central region (or circumbinary gap). This interaction can be made even more complex when the binary system is eccentric, or if the spin angular momenta of the binary components are misaligned with the orbital angular momentum. An important consideration when modeling black hole binaries with accretion disks is whether or not the motion of the inspiralling binary is influenced by the disk. Earlier analytical studies had estimated that the influence of the disk vanishes when the timescale of viscous transport in the disk fluid exceeds the time to merger due to gravitational radiation. Since numerical hydrodynamic systems of binary-disk interactions show the disk moving in with the inspiralling binary, an alternative analytical picture becomes necessary to explain this behavior. In my first paper, I presented an analytical method to estimate the location of the circumbinary gap that was based on the characteristic timescale over which test particle orbits about the binary decay (also called the Lyapunov timescale). This method also proved to be accurate in predicting the size of the gap for misaligned disk and binary planes.

More recently, I have been combining my passion for studying dynamical systems and machine learning/AI techniques to further improve the transition between the inspiral and merger-ringdown regimes. Major contributors to the inaccuracy of post-Newtonian techniques close to merger are the breakdown of the slow motion approximation and the treatment of black holes as point particles. In light of this, I am designing a strategy to use Dissipative-Hamiltonian Neural Networks to learn the governing dynamical system at this regime. The Neural Networks act as mimickers for higher PN orders that are trained against numerical relativity waveforms. This strategy can quantify the relevant regions of the phase space where we see large discrepancies between PN and NR waveforms. More importantly, due to the success of BOB in modelling the merger through the properties of the remnant black hole, it also shows promise in analytically completing the effective PN dynamics to reflect perturbations around an emergent final black hole spacetime. This project is currently in development and progress can be found in this repo.