Extra! Extra! Critical Update on Life by Rik Blok for PWIAS Crisis Points March 18, 1998
40 min. Self-organized Criticality (SOC) critical point: under variation of a control parameter, an order parameter continuous change with discontinuous derivative order undergoes control many suitable choices for order parameter behaviour near critical point governed by power-laws, many divergent properties dominated by fluctuations no characteristic scale, can have events/structures on all scales universality: many details of the system are irrelevant to the critical behaviour, systems sharing the same few important properties (eg. dimensionality) belong to the same universality class self-organization: the system spontaneously approaches the critical point without tuning a control parameter + = SOC [1,2,3,4]
35 min. Game of Life Game of Life simulates the evolution of a colony of organisms and suggests that the theory of self-organized criticality can explain the dynamics of ecosystems. [4] cellular automaton 2D square lattice each site has 2 possible states, 8 nearest neighbors discrete time, all sites updated in parallel alive or dead Rules Births: dead and exactly three (live) neighbors at time t alive at t+1 Survivals: alive and either two or three neighbors at time t alive at t+1 Deaths: neither above condition at t dead at t+1 or ct+ 1 = ( 1 ct) δn, 3 + ct( δn, 2 + δn, 3) where c is state of cell (0=dead, 1=alive), n is count of live neighbors, and δ is the Kronecker delta function (returns 1 if the indices are equal, else zero) Sample steady-state configuration.
! 30 min. Synchronicity complicate matters by adding another parameter, s instead of updating all sites simultaneously, each site has a probability being updated in each time step call s synchronicity can adjust s to continuously vary between parallel (s=1) and Poisson ( one-at-a-time, s +0) updating s of Example s= ½ sites to be updated represented in gray Analysis Collect time-series of interesting statistics, such as: density, ρ fraction of live sites (eg. ρ = 5/25 in above) activity, a fraction of updated sites which have changed state (births and deaths, eg. a = 2/11 in above)
" " 25 min. GLAsynch: sample data for 128x128 periodic with s=0.83 5 0.35 density 0.3 0.25 0.2 " # transient steady state 0.15 0.1 0.05 0 0 500 1000 1500 2000 time GLAsynch: sample data for 128x128 periodic with s=0.97 5 0.35 density 0.3 0.25 0.2 " transient absorbed 0.15 0.1 0.05 0 0 500 1000 1500 2000 time time is rescaled by s (t st) $ qualitatively different evolutions suggest a phase transition
) (! % % & & 20 min. Critical Point 5 GLAsynch: Density exhibits critical phase transition for 128x128 periodic data power-law: s' c =0.908, β = 0.616 0.35 steady-state density 0.3 0.25 0.2 0.15 0.1 0.05 0 0 0.1 0.2 0.3 0.6 0.7 0.8 0.9 1 synchronicity, s 0.3 GLAsynch: Activity exhibits critical phase transition for 128x128 periodic data power-law: s' c =0.908, β = 92 0.25 steady-state activity 0.2 0.15 0.1 0.05 0 0 0.1 0.2 0.3 0.6 0.7 0.8 0.9 1 synchronicity, s fit to y y0 ( sc s) β recovered exponent and critical point: density: sc = 0.908 * ± 0.006, β = 0.616 * ± 0.031, + y0 0.0257 activity: sc = 0.908 * ± 0.025, β = 92 * ± 0.064, + y0 = 0 so GL is close to a dynamical critical point, which makes it look SOC
- # 1 15 min. Scaling ) runs on L L lattices with periodic boundaries want to minimize boundary effects 0.8 0.75 Scaling behaviour of density exponent, density analysis ( ) β =94 ± 0.018 0.8 0.75 Scaling behaviour of activity exponent, activity analysis β ( ) =90 ± 0.038 0.7 0.7 0.65 0.65 exponent 0.6 exponent 0.6 5 5 5 5 32 64 128 256 512 β ( ) = 94±0.018 L 32 64 128 256 512 β ( ) = 90±0.038 L 1 0.98 Scaling behaviour of s0 c density analysis activity analysis s0 2 3 ( ) c =0.9054 ± 0.0053 0.96 0.94 0.92 s c 0.9 0.88 0.86 0.84 0.82 0.8. 32 / 64 128 256 L 512 s ( ) c = 0.9054 * ± 0.0053
6 8 4 7 4 10 min. Directed Percolation uniform lattice of sites bonds between neighbors are randomly distributed with concentration p bonds are restricted to a certain direction (often related to time) space time critical point, pc, above which there is a nonzero probability of propagation to infinity critical behaviour independent of details like lattice structure in 2+1 dimensions β DP =β DP =86±0.014 [5] β DP =β DP =86±0.014 β=94±0.018 β =90± 5 0.038 GL( sc) appears to be in same universality class as DP class also includes Reggeon field theory (high energy physics) [6] surface reaction models [5] contact process [7 and Durrett, 1988] Domany-Kinzel cellular automaton [8]
6! = ; : : 4 < > : 9 5 min. Conclusions updating (parallel or sequential) important GL looks SOC because near critical point SOC is just criticality with an undiscovered control parameter [9] appears to belong to DP universality class References : [1] Bak P, Tang C and Wiesenfeld K (1987) Phys. Rev. Lett. 59 381-4 [2] Bak P, Tang C and Wiesenfeld K (1988) Phys. Rev. A 38 364-74 [3] Bak P, Chen K and Creutz M (1989) Nature 342 780-2 [4] Bak P and Chen K (1991) Sci. Am. 264 (1) 46-53 [5] Grinstein G, Lai Z-W and Browne D A (1989) Phys. Rev. A 40 > 4820-3 [6] Deutscher G, Zallen R and Adler J (1983) Percolation Structures and Processes (Adam Hilger, The Israel Physical Society and The American Institute of Physics) [7] Jensen I and Dickman R (1993) Phys. Rev. E 48 1710-25 [8] Dickman R and Tretyakov A Y (1995) Phys. Rev. E 52 3218-20...many more on DP... [9] Sornette D, Johansen A and Dornic I (1995) preprint de Physique I 5)? (to appear in J.