Assessing the waveform systematics from parameter estimation to population inference with eccentricity

Abstract

While masses and spins are routinely used to constrain compact binary formation channels, eccentricity provides an additional and potentially powerful diagnostic of binary origin, particularly for dynamically assembled systems. Recent advances in eccentric waveform modeling now make it possible to search for eccentric signatures in gravitational wave data; however, differences between waveform models can introduce systematic effects that may propagate into astrophysical population inference. In this work, we analyze 153 binary black holes, 2 binary neutron stars and 7 neutron star black hole binaries from the GWTC-4 catalog. We compare the source and population level inferences obtained with two eccentric waveform models, SEOBNRv5EHM and TEOBResumS-DALI, as well as with quasi circular waveform analyses. We find that the two eccentric models give broadly consistent source parameter estimates for most events, but some events exhibit subtle and coherent differences. These small, systematic offsets can accumulate in hierarchical population inference, leading to differences in inferred population properties, most notably in the redshift evolution and effective spin distribution. Because coherent event level biases can grow approximately as $\sqrt{N}$ for a catalog of N events, waveform systematics become increasingly important as gravitational wave catalogs expand. We also introduce a synthetic data framework that generates eccentric populations and corresponding RIFT posterior samples, enabling injection studies that test the recoverability of eccentric population properties.