Back to Table of Contents
45987-AC6
Non-Perturbative Effects of Dynamical Inhomogeneities in Viscous Fluid Mixtures
Jeremy Schofield, University of Toronto
Over the last year, the PRF funds have been applied towards
the salary of post-doctoral Research Associate Ramses van Zon.
This financial support has enabled us to continue our collaboration
on the dynamics of glassy systems. The PRF support has enabled work
to proceed along one of the central lines of interest in my research
group, the role of collective motion in systems exhibiting
rugged-energy landscapes. In addition, thanks to these funds,
Dr. van Zon has been able to continue his work in the group while
looking for a permanent position in Canada.
During the funding period, work has been initiated on combining a
model composed of infinitely thin, hard crosses with a mesoscopic model of
a fluid. In model solvent, the fluid molecules are point
particles with punctuated rotational interactions that lead
to hydrodynamic motion of the fluid on a macroscopic scale.
A number of computational techniques to extend the time scale
accessible to simulation have been developed, including the
extension of the hard interactions between crosses to include
attractive interactions and softer repulsions while maintaining
the advantages of event-driven simulations. Simulating irregularly
shaped objects, such as long needles, is facilitated by using
rigid bounding boxes to limit the number of interaction partners
of a particular needle. The bounding boxes can be utilized to
identify analytically the earliest time at which a given pair of
needles can interact and thereby improve the efficiency of the
simulation and glassiness of the systems studied. We have
conducted investigations into the signature of slow orientational
relaxation of long needles near the nematic ordering transition
and observed similarities with simple glass formers in the form
of the stretched-exponential relaxation of the auto-correlation
function of the long-axis director of the needle. The orientational
dynamics near the nematic transition is characterized by slow
and highly cooperative collective motion of local ordering fields
of the director,
Of particular future interest are extensions of the cross model
to include an arbitrary number of arms. The limit of rotors with
infinitely many arms is especially intriguing, since the molecules
would, on one hand, become a solid sphere, but on the other
hand still would not have any excluded volume.
The simulation of glassy systems requires efficient means of
computing the time evolution of constituents of the system.
In situations where the dynamics is not amenable to event-driven
algorithms, symplectic sequential integration schemes offer superior
stability and accuracy. Utilizing operator splitting algebra, we
have developed the first methods of evolving rigid body systems that
rigorously conserve a shadow Hamiltonian. The integration method is
based on combining the exact free propagation of an arbitrary rigid
body with updates of momenta and angular momenta to construct
integrators of arbitrary order. We have demonstrated that while a
second-order Verlet-like scheme is optimal if only static
information is required, a simple variant of the Verlet scheme is
optimal at accuracy levels typically required in the computation of
dynamical correlations, as is the case in glassy systems.
Furthermore, if very accurate trajectories of the system are desired,
an advanced gradient-like fourth order scheme offers the most
efficient implementation of the evolution.
Along analytical lines, we have examined kinetically constrained spin models, in
particular the Frederickson-Andersen model. We have been working towards
formulating simple schemes, both numerical and analytical, of
incorporating an arbitrary number of multilinear modes
in order to get nonperturbative results, starting from the complete
basis set in terms of down-spin domains. Formulating such schemes is
facilitated by the fact that the coupling vertices are easily obtained
analytically for these models. For instance, incorporating one
down-spin domain of arbitrary size (nonperturbatively) in the theory,
the spin-spin time auto-correlation function can be predicted in an
analytic form up to fairly low values of the mobile spins. This can be seen as a
resummation of the contributions of a specific infinite set of basis vectors.
Improvements can be made by including several down-spin domains after
one another. We are currently in the process of developing
resummation schemes that include an arbitrary number of
domains to produce non-perturbative results for the Frederickson-Andersen model.
Back to top