High-fidelity, error-controlled models of gravitational waves from generic binary black hole mergers

The National Science Foundation Division of Physics awarded the University of Massachusetts Dartmouth $209,943 on August 1, 2026, under the Mathematical and Physical Sciences program (CFDA 47.049) to develop fast, high-fidelity numerical models of gravitational waves produced by merging binary black holes with complex orbital dynamics and unequal masses.

The project will advance numerical relativity surrogate modeling to generate computationally efficient yet accurate waveform predictions that capture physical effects underrepresented in existing models. These models will enable more precise extraction of astrophysical information from current and future gravitational-wave detector observations, support precision tests of general relativity, and enhance understanding of the black hole population across the universe. The work includes development of numerical methods for modeling binary black hole mergers and their remnants, with resulting software and data released publicly to support ongoing scientific progress.

The award funds research, training of graduate and undergraduate students in computational science, mathematics, and astrophysics, mentored research opportunities, and public engagement on gravitational-wave astronomy. Performance occurs in North Dartmouth, Massachusetts, with a period of performance from August 1, 2026, through July 31, 2029. This award is classified as assistance type Project Grant.

Abstract

This award supports research in relativity and gravitational-wave astrophysics. Gravitational waves produced by violent collisions between black holes carry information about these objects and einstein's theory of general relativity across cosmic distances. As gravitational-wave detectors become more sensitive, they are observing stronger and more complex signals that reveal increasingly subtle physical effects. Extracting this information requires accurate models of the gravitational waves produced by merging black holes, but the most faithful computer simulations are too computationally expensive for direct use in many data-analysis applications. Building on previous NSF-supported research, this project will develop fast, highly accurate models for black hole systems with complex orbital motion and unequal masses, including physical effects not fully represented in existing models. These advances will increase the scientific return of current and future gravitational-wave experiments, support precision tests of general relativity, and improve our understanding of the black hole population in the universe. The project will also train graduate and undergraduate students in computational science, mathematics, and astrophysics, provide mentored research opportunities, and engage the public through talks and events on gravitational-wave astronomy. The resulting software and data will be made publicly available, supporting scientific progress and preparing students for careers requiring strong technical and computational skills. This award will support the development, implementation, and application of numerical methods for constructing fast, high-fidelity models of binary black hole gravitational waves and merger remnants. The project will build on data-driven numerical relativity surrogate modeling, which can reproduce computationally expensive numerical relativity simulations at arbitrary parameter values within a fraction of a second. The research will (I) extend surrogate models to cover systems with orbital eccentricity and spin-induced precession while moving toward intermediate-mass-ratio binaries, (II) incorporate nonlinear gravitational-wave memory, and (III) develop error-control and domain-decomposition techniques that substantially improve waveform accuracy during the merger and ringdown. Because these models are trained directly on numerical relativity simulations, they can capture the full strong-field dynamics represented in the underlying simulations without relying on approximate analytic descriptions of each physical effect. The resulting waveform and remnant models will be implemented efficiently in public software for use in parameter estimation, population studies, and precision tests of general relativity. The project will also investigate scientific machine-learning methods that may guide future gravitational-wave modeling efforts and develop broadly applicable techniques for other problems requiring repeated evaluation of expensive numerical simulations. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the foundation's intellectual merit and broader impacts review criteria.- subawards are not planned for this award.