Bimodular Gravity: Unimodularising Bimetric Scalar-Tensor Gravity

TL;DR

This paper introduces bimodular gravity by unimodularising a disformal bimetric scalar-tensor theory, revealing two classically inequivalent formulations with distinct cosmological implications and proposing a covariant completion linking them.

Contribution

It develops the concept of bimodular gravity by formulating two inequivalent unimodular bimetric scalar-tensor theories and explores their cosmological and perturbative differences.

Findings

BUG fixes the relative volume element, leading to a bimodular cosmological constant.

BHT/BDUG allows a dynamical volume element, affecting scalar perturbations.

Distinct predictions for dark energy and expansion history emerge from the two formulations.

Abstract

It is the object of the present paper to unimodularise a disformal bimetric scalar-tensor theory, thereby defining what we call bimodular gravity. We impose one unimodular constraint per metric via multipliers $\lambda_{1,2}$ and show that two natural implementations-a dual fixed-determinant (BUG) and a dual diffeomorphism-invariant (BHT/BDUG) formulation-are classically inequivalent. In BUG the relative volume element $\nu=\sqrt{1-2BX}$ is fixed, enforcing a kinematic constraint on the biscalar and we derive the "bimodular cosmological constant" $\Lambda=\lambda_1+\nu\lambda_2$. In BHT/BDUG, $\lambda_{1,2}$ are individually constant but $\nu$ (hence $BX$) remains dynamical. Recasting the theory in an Einstein-frame form, we derive the biscalar sound speed and identify a subluminal domain $1+B(V+\lambda_2)>0$. At the background level, BUG admits constant-roll solutions governed by first-order flow, whereas BHT supports solutions with time-dependent roll. These structural differences yield distinct, in-principle testable predictions for the expansion history, the dark-energy equation of state, and the propagation of biscalar perturbations. Finally, we present a diffeomorphism-invariant completion that correlates the two HT volume forms, reproducing the $\Lambda$ of BUG on shell whilst maintaining full covariance.