We have performed a very accurate computation of the nonequilibrium fluctuation-dissipation ratio for the 3D Edwards-Anderson Ising spin glass, by means of large-scale simulations on the specialpurpose computers Janus and Janus II. This ratio (computed for finite times on very large, effectively infinite, systems) is compared with the equilibrium probability distribution of the spin overlap for finite sizes. Our main result is a quantitative statics-dynamics dictionary, which could allow the experimental exploration of important features of the spin-glass phase without requiring uncontrollable extrapolations to infinite times or system sizes.spin glasses | fluctuation-dissipation relation | glasses | statics-dynamics equivalence | out-of-equilibrium dynamics T heory and experiment follow apparently diverging paths when studying the glass transition. On the one hand, experimental glass formers (spin glasses, fragile molecular glasses, polymers, colloids, and . . .) undergo a dramatic increase of characteristic times when cooled down to their glass temperature, Tg (1). Below Tg, the glass is always out of equilibrium and "aging" appears (2). Consider a rapid quench from a high temperature to the working temperature T (T < Tg), where the system is left to equilibrate for time tw and probed at a later time t + tw. Response functions such as the magnetic susceptibility turn out to depend on t/t µ w , with µ ≈ 1 (2-4). The age of the glass, tw, remains the relevant time scale even for tw as large as several days. Relating the aging experimental responses to equilibrium properties is an open problem.A promising way to fill the gap is to establish a statics-dynamics dictionary (SDD) (5-8): nonequilibrium properties at "finite times" t, tw, as obtained on samples of macroscopic size L → ∞, are quantitatively matched to equilibrium quantities computed on systems of "finite size" L [the SDD is an L ↔ (t, tw) correspondence]. Clearly, in order for it to be of any value, an SDD cannot strongly depend on the particular pair of aging and equilibrium quantities that are matched.Some time ago, we proposed one such a SDD (6-8). However, this SDD was unsatisfactory in two respects. First, L was matched only to tw (irrespectively of the probing time t + tw). Second, our SDD matched spatial correlation functions whose experimental study is only incipient (9, 10).One could think (5) of building an SDD through the generalized fluctuation-dissipation relations (GFDRs) first introduced in ref. 11 (for related developments, see refs. 12-19). The GFDRs are correct at very large times. However, on time scales that can be investigated in experiments, glassy systems are not fully thermalized because the approach to equilibrium is very slow. Strong corrections pollute GFDRs at finite times.
SignificanceThe unifying feature of glass formers (such as polymers, supercooled liquids, colloids, granulars, spin glasses, superconductors, etc.) is a sluggish dynamics at low temperatures. Indeed, their dynamics are so slow that thermal equilibrium is n...