Tests of conservation laws in post-Newtonian gravity with binary pulsars

TL;DR

This paper tests conservation laws in alternative gravity theories using binary pulsar timing, deriving new constraints on the PPN parameter $$ and extending analysis methods to improve bounds.

Contribution

It introduces an extended analytical timing formalism for $$, applies Bayesian inference to combine multiple pulsar data, and tightens the constraints on $$ by a factor of three.

Findings

Upper limit on || is 1.3×10^{-5} at 95% confidence

Third derivative of spin frequency provides tighter constraints for some pulsars

Simulations show pulsar timing can effectively limit  in future observations

Abstract

General relativity is a fully conservative theory, but there exist other possible metric theories of gravity. We consider non-conservative ones with a parameterized post-Newtonian (PPN) parameter, $\zeta_2$. A non-zero $\zeta_2$ induces a self-acceleration for the center of mass of an eccentric binary pulsar system, which contributes to the second time derivative of the pulsar spin frequency, $\ddot{\nu}$. In our work, using the method in Will (1992), we provide an improved analysis with four well-timed, carefully-chosen binary pulsars. In addition, we extend Will's method and derive $\zeta_2$'s effect on the third time derivative of the spin frequency, $\dddot{\nu}$. For PSR B1913+16, the constraint from $\dddot{\nu}$ is even tighter than that from $\ddot{\nu}$. We combine multiple pulsars with Bayesian inference, and obtain an upper limit, $\left|\zeta_{2}\right|<1.3\times10^{-5}$ at 95% confidence level, assuming a flat prior in $\log_{10} \left| \zeta_{2}\right|$. It improves the existing bound by a factor of three. Moreover, we propose an analytical timing formalism for $\zeta_2$. Our simulated times of arrival with simplified assumptions show binary pulsars' capability in limiting $\zeta_{2}$, and useful clues are extracted for real data analysis in future. In particular, we discover that for PSRs B1913+16 and J0737$-$3039A, $\dddot{\nu}$ can yield more constraining limits than $\ddot{\nu}$.