Anticipating Challenges of Propensity Measurements in Colloidal Systems

In order to study dynamics near the glass transition we simulate the Kob-Andersen binary glassformer in a series of isoconfigurational ensembles. This follows the work of Widmer-Cooper et al, (2004), where they used such an ensemble to define the propensity of each particle as the average motion of the particle, where the averaging is done across the ensemble. We seek to understand how sensitive the measurement of propensity is to the polydispersity of the particles. One could imagine attempting to measure propensity experimentally using a colloidal glass system, but these particles would have some distribution of size and there would always be the possibility for error when attempting to create an isoconfigurational copy of the system. We introduce polydispersity by splitting the population of each of the binary components into its own binary where half are increased in size and the other half are decreased in size by the same percentage. Experimental errors are then simulated by exchanging large and small variants each time a isoconfigurational copy is created. We find that the propensity signal is strengthened when polydispersity is increased, but decreased as more and more errors are introduced.