DI Herculis as a test of internal stellar structure and general relativity

Abstract

For the past three decades, and until recently, there has been a serious discrepancy between the observed and theoretical values of the apsidal motion rate dw/dt of the eccentric eclipsing binary DI Her, which has even been interpreted occasionally as a possible failure of General Relativity (GR). Recent observations of the Rossiter-McLaughlin effect have shown convincingly that the reason for the anomaly is that the rotational axes of the stars and the orbital axis are misaligned, which changes the predicted rate of precession significantly. Although as a result of those measurements the disagreement is now drastically smaller, it remains formally at the level of 50%, possibly due to errors in the measured apsidal motion rate, outdated stellar models, or inaccuracies in the stellar parameters. Here we address each of these issues in order to improve the agreement further. New times of minimum have been collected in order to redetermine the apsidal motion rate. We have computed new stellar evolution models with updated physical inputs, and derived improved apsidal motion constants for the components. We have performed Monte Carlo simulations to infer the theoretical distribution of dw/dt, including the contributions from GR as well as tidal and rotational distortions. All observational errors have been accounted for. Our simulations yield a retrograde apsidal motion rate due to the rotationally-induced oblateness of -0.00056 deg/cycle (mode of the distribution), a GR contribution of +0.00068 deg/cycle, and a tidal contribution of +0.00034 deg/cycle, leading to a total predicted rate of +0.00046 deg/cycle. This is in excellent agreement with the newly measured value of +0.00042 deg/cycle. The formal difference is now reduced to 10%, a small fraction of the observational uncertainties. (abridged)