Error estimation and bounds for gravitational-wave surrogate models

The upcoming era of high-precision gravitational wave (GW) astronomy will allow us to more accurately probe the regime of strong-field gravity from massive compact object binary systems. As the sensitivity of GW detectors continues to improve, it is expected that waveform modeling errors will begin to dominate over statistical errors. Therefore, it will be important for GW models to provide uncertainty estimates across the parameter space, thus leading to improved parameter estimation studies and more confidence in our analysis. In this talk, I will present an error estimator for surrogate models built on non-spinning numerical relativity waveforms. Our estimator accounts for the impact of noisy training data and helps identify the steps of the surrogate modeling process responsible for noise amplification. We quantify the sharpness of our error estimator through numerical examples.