磁性流体中的一些问题.ppt

How about the trajectories? Line shows the locations of endpoints as calculated by iterating closed set of equations to long time. With initial frequency of cooperation fc = 0. Time evolution of degrees fci=0.8 and r = 0.8 fci=0.1 Steady state mean degrees Lines are results of equations, symbols are simulation results Note the necessity in treating C and D nodes separately While simple theory captures all the main features, there are discrepancies between simple theory and numerical results! Can we improve the theory? Better moment closure schemes? Alternative ways to treat the second moments and to close the equations (2) Binomial distribution treatment (BINO) Picture: Assume that C-node (or D-node) of degree κ has a binomial distribution of links λCD. Comparing with SSC treatment, we have extra terms in BINO treatment! (3) Aside: Keeling-Eames treatment (KE) Fluctuations are included in a way that assumes the variance equals to the mean. Comparing with SSC treatment, there is an extra term in KE treatment A moment closure method used by Gross et al. in PRL 96, 208701 (2006) for an adaptive SIS model, based on Keeling and Eames PNAS 99, 13330 (2002) and J. R. Soc. Interface 2, 295 (2005) for epidemic models But KE approximation turns out to be a bad approximation for the present model. SSC treatment (dashed lines) BINO treatment (solid lines) – Worked much better in the disconnected state Comparing simple and modified theories with numerical results Ref: Oliver Graser, PhD Dissertation (CUHK 2010) Fixed Point Analysis Fixed Points: ?Nc = ?LCC = ?Lcd=0 Obvious candidate : LCC =κ/2, Lcd=0 (no dissatisfied links – disconnected state) Occurs for small r, large fci Fixed-point analysis is carried out based on MF equations using a binomial approximation Fixed point vs iteration of equations starting from some initial condiitions Taking advantage of having a set of mean-field equations Fixed Point Analysis Fixed Point: ?Nc= ?LCC = ?Lcd=0 Obvious candidate : LCC =κ/