Nihar Gupte

After measuring the properties of 57 Binary Black holes, we found that three of them had signs of eccentric inspirals. Furthermore, we found a >99% probability that one of the 57 systems is measurably eccentric. This hints that binary black holes may form from chaotic interactions in dense stellar environments.

I'm a developer for "DINGO", a machine-learning code which quickly and accurately infers the properties of compact binaries which emit gravitational waves. This code is now being used by the LIGO-Virgo-KAGRA collaboration.

If the energy loss from two BHs approaching each other is enough to form a binary, it's a GW capture. We find this mechanism in nuclear star clusters can't explain all LVK binaries.

Using Geospatial data from NOAA, we localized fires in California. This was done using a real-time ML pipeline and updates authorities every 5 minutes. It has now become a successful startup extending to lighting detection and drone monitoring.

We developed an algorithm which covers polygons embedded on spheres using circles. This is applied to covering gravitational wave skymaps with electromagnetic telescopes.

In a project led by M. Dax, we developed a machine learning algorithm to accurately infer the properties of a binary neutron star within one second and even before the merger.