Large scale simulations of two-dimensional bidisperse granular fluids allow us to determine spatial correlations of slow particles via the four-point structure factor $S_4(q,t)$. Both cases, elastic ($varepsilon=1$) as well as inelastic ($varepsilon
< 1$) collisions, are studied. As the fluid approaches structural arrest, i.e. for packing fractions in the range $0.6 le phi le 0.805$, scaling is shown to hold: $S_4(q,t)/chi_4(t)=s(qxi(t))$. Both the dynamic susceptibility, $chi_4(tau_{alpha})$, as well as the dynamic correlation length, $xi(tau_{alpha})$, evaluated at the $alpha$ relaxation time, $tau_{alpha}$, can be fitted to a power law divergence at a critical packing fraction. The measured $xi(tau_{alpha})$ widely exceeds the largest one previously observed for hard sphere 3d fluids. The number of particles in a slow cluster and the correlation length are related by a robust power law, $chi_4(tau_{alpha}) approxxi^{d-p}(tau_{alpha})$, with an exponent $d-papprox 1.6$. This scaling is remarkably independent of $varepsilon$, even though the strength of the dynamical heterogeneity increases dramatically as $varepsilon$ grows.
We test a hypothesis for the origin of dynamical heterogeneity in slowly relaxing systems, namely that it emerges from soft (Goldstone) modes associated with a broken continuous symmetry under time reparametrizations. We do this by constructing coars
e grained observables and decomposing the fluctuations of these observables into transverse components, which are associated with the postulated time-fluctuation soft modes, and a longitudinal component, which represents the rest of the fluctuations. Our test is performed on data obtained in simulations of four models of structural glasses. As the hypothesis predicts, we find that the time reparametrization fluctuations become increasingly dominant as temperature is lowered and timescales are increased. More specifically, the ratio between the strengths of the transverse fluctuations and the longitudinal fluctuations grows as a function of the dynamical susceptibility, chi 4, which represents the strength of the dynamical heterogeneity; and the correlation volumes for the transverse fluctuations are approximately proportional to those for the dynamical heterogeneity, while the correlation volumes for the longitudinal fluctuations remain small and approximately constant.
Dynamical heterogeneities -- strong fluctuations near the glass transition -- are believed to be crucial to explain much of the glass transition phenomenology. One possible hypothesis for their origin is that they emerge from soft (Goldstone) modes a
ssociated with a broken continuous symmetry under time reparametrizations. To test this hypothesis, we use numerical simulation data from four glass-forming models to construct coarse grained observables that probe the dynamical heterogeneity, and decompose the fluctuations of these observables into two transverse components associated with the postulated time-fluctuation soft modes and a longitudinal component unrelated to them. We find that as temperature is lowered and timescales are increased, the time reparametrization fluctuations become increasingly dominant, and that their correlation volumes grow together with the correlation volumes of the dynamical heterogeneities, while the correlation volumes for longitudinal fluctuations remain small.
