Share
Embed
Conservation of population size is required for self-organized criticality in evolution models
Yohsuke Murase
February 28, 2019
Transcript
32 slides
-
Conservation of population size is required for self-organized criticality in
evolution models Y. Murase1, P. A. Rikvold2,3 1RIKEN Center for Computational Science 2Florida State Univ., 3Univ. of Oslo ref. Y.Murase and P.A. Rikvold, New J. Phys. 20, 083023 (2018) -
The Bak-Sneppen model • a minimal model for biological coevolution
• shows SOC • critical avalanche of extinctions • intermittent dynamics • assumptions: • Darwinian competition • interspecies interactions Bak-Sneppen, PRL (1983) -
fitness (a scalar value [0,1]) loop { i = argmin(
f ) t += t_0 * exp( f[i] / f_0 ) f[i] = rand(0,1) f[i+1] = rand(0,1) f[i-1] = rand(0,1) } extinction of species i interspecies interaction Arrhenius' type function ⌧ext(fi) / exp (fi/f0) <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> -
spontaneous emergence of the threshold intermittent evolutionary dynamics (punctuated equilibria)
power-law avalanche size (mass extinctions) Bak-Sneppen, PRL (1993) -
Population dynamics models • population dynamics + species introduction •
e.g. dynamical system with variable system size loop { species.each {|i| x[i] = update_pop(x) } species.each {|i| extinction(i) if x[i] <= 0 } add_new_species() } e.g., Tangled-Nature model, scale invariant model, replicator equations, web world model, ... ref. H.J.Jensen, Eur. J. Phys. (2018) new species extinction -
Population dynamics models with migration rule • Migration rule :
interspecies interactions are randomly determined irrespective of the existing species • c.f. Mutation rule : new species are made based on the existing species Y.Murase et al., J. Theor. Biol. (2010) • 1/f2 fluctuations • exp. extinction sizes • skewed lifetime distribution non-SOC -
Dynamical graph model • A minimal model for the class
of migration models. • Population dynamics is replaced by a simple graph dynamics. • Species can survive as long as its incoming link weight ≥ 0. Y.Murase et al., NJP (2010) from T.Shimada, Sci.Rep (2014) -
• belongs to the same class as migration pop. dynamics
models • 1/f2 fluctuations • exp. extinction sizes • skewed lifetime distribution "Ising" model for population dynamics models -
SOC non-SOC BS model DG model population dynamics models Darwinian
competition & successive introduction of new species common assumptions -
Key differences b/w BS & DG 1. number of species
(N) • BS : N is fixed to a value given as a model parameter. • DG : N changes according to evolutionary process. -
Key differences b/w BS & DG 1. number of species
(N) • BS : N is fixed to a value given as a model parameter. • DG : N changes according to evolutionary process. 2. the extinction threshold (fth ) • BS : fth is self-organized as a result of evolutionary process. • DG : fth is predefined -
Key differences b/w BS & DG 1. number of species
(N) • BS : N is fixed to a value given as a model parameter. • DG : N changes according to evolutionary process. 2. the extinction threshold (fth ) • BS : fth is self-organized as a result of evolutionary process. • DG : fth is predefined 3. instant / regular migrations • BS : f-dependent τext followed by instant migration τimg =0 • DG : regular migration τimg =1 & f-dependent τext -
Key differences b/w BS & DG 1. number of species
(N) • BS : N is fixed to a value given as a model parameter. • DG : N changes according to evolutionary process. 2. the extinction threshold (fth ) • BS : fth is self-organized as a result of evolutionary process. • DG : fth is predefined 3. instant / regular migrations • BS : f-dependent τext followed by instant migration τimg =0 • DG : regular migration τimg =1 & f-dependent τext 4. fitness definition (f) • BS : node-based • DG : link-based, i.e., fi = ∑wji -
Model 1: link-based BS model • fixed N (as in
BS model) • eliminate minimum fitness species followed by an immediate introduction of new species (as in BS model) • increment time by τext ∝ exp(fmin /f0 ), • represented by a weighted directed network (as in DG model) • fi = ∑ wji • new species has new links drawn randomly (as in DG model) -
0 0.4 0.8 1.2 1.6 0 0.2 0.4 0.6 0.8
1 P(f) fitness, f 10-6 10-5 10-4 10-3 10-2 10-1 100 0 5 10 15 20 25 30 35 exp(-s/s0 ) P(s) extinction size, s 0 0.04 0.08 0.12 0.16 0.2 -4 0 4 8 12 P(f) fitness, f 10-10 10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-2 100 101 102 103 104 exp(-(L/L0 )0.6) P(L) species lifetime, L 10-5 10-4 10-3 10-2 10-1 100 0 2 4 6 8 10 exp(-τ/τ0 ) P(τ) inter-event time, τ 10-8 10-7 10-6 10-5 10-4 10-3 10-2 10-1 100 100 101 102 103 s-1.5 P(s) extinction size, s 10-12 10-10 10-8 10-6 10-4 10-2 100 102 104 106 108 1010 L-1 exp(-L/L0 ) P(L) species lifetime, L 10-14 10-12 10-10 10-8 10-6 10-4 10-2 100 100 101 102 103 104 105 106 107 108 τ-1 P(τ) inter-event time, τ 0 0.04 0.08 0.12 0.16 0.2 -4 0 4 8 12 P(f) fitness, f 10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-2 10-1 100 100 101 102 103 104 s-1.5 P(s) extinction size, s 10-16 10-14 10-12 10-10 10-8 10-6 10-4 10-2 100 10-2 100 102 104 106 108 1010 1012 L-1 P(L) species lifetime, L 10-15 10-10 10-5 100 10-2 100 102 104 106 108 1010 1012 τ-1.2 P(τ) inter-event time, τ BS model DG model link-based BS model (a-1) (a-2) (a-3) (a-4) (b-1) (b-2) (b-3) (b-4) (c-1) (c-2) (c-3) (c-4) mean-field -
0 0.4 0.8 1.2 1.6 0 0.2 0.4 0.6 0.8
1 P(f) fitness, f 10-6 10-5 10-4 10-3 10-2 10-1 100 0 5 10 15 20 25 30 35 exp(-s/s0 ) P(s) extinction size, s 0 0.04 0.08 0.12 0.16 0.2 -4 0 4 8 12 P(f) fitness, f 10-10 10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-2 100 101 102 103 104 exp(-(L/L0 )0.6) P(L) species lifetime, L 10-5 10-4 10-3 10-2 10-1 100 0 2 4 6 8 10 exp(-τ/τ0 ) P(τ) inter-event time, τ 10-8 10-7 10-6 10-5 10-4 10-3 10-2 10-1 100 100 101 102 103 s-1.5 P(s) extinction size, s 10-12 10-10 10-8 10-6 10-4 10-2 100 102 104 106 108 1010 L-1 exp(-L/L0 ) P(L) species lifetime, L 10-14 10-12 10-10 10-8 10-6 10-4 10-2 100 100 101 102 103 104 105 106 107 108 τ-1 P(τ) inter-event time, τ 0 0.04 0.08 0.12 0.16 0.2 -4 0 4 8 12 P(f) fitness, f 10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-2 10-1 100 100 101 102 103 104 s-1.5 P(s) extinction size, s 10-16 10-14 10-12 10-10 10-8 10-6 10-4 10-2 100 10-2 100 102 104 106 108 1010 1012 L-1 P(L) species lifetime, L 10-15 10-10 10-5 100 10-2 100 102 104 106 108 1010 1012 τ-1.2 P(τ) inter-event time, τ BS model DG model link-based BS model (a-1) (a-2) (a-3) (a-4) (b-1) (b-2) (b-3) (b-4) (c-1) (c-2) (c-3) (c-4) mean-field -
SOC non-SOC BS model DG model link-based BS model population
dynamics models node-based / link-based is not the key factor -
Model 2: generalized model 1. fixed / variable N 2.
self-organized / fixed fth 3. instant / regular immigrations 4. node-based / link-based analogous to canonical / grand-canonical ensembles (fixed N vs fixed μ) DG model link-based BS model -
Model 2: generalized model 1. fixed / variable N 2.
self-organized / fixed fth 3. instant / regular immigrations 4. node-based / link-based analogous to canonical / grand-canonical ensembles (fixed N vs fixed μ) generalized model DG model link-based BS model -
model definition loop { i = argmin( f ) if
tau_ext(f[i]) < tau_mig(N) extinction(i) t += tau_ext(f[i]) else add_new_species t += tau_mig(N) end } ⌧ext(f) = exp (f/f0) <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> ⌧mig(N) = exp (µ(N N0)/f0) <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> μ : parameter to control the fluctuation of N around N0 ( N > N0 ) ⌧mig %, N & N < N0 ) ⌧mig &, N % <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> -
generalized model DG model link-based BS model µ = 0
<latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> µ ! 1 <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> ⌧ext(f) = exp (f/f0) <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> ⌧mig(N) = exp (µ(N N0)/f0) <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> t extinction migration t -
µ = 0 <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit
sha1_base64="(null)">(null)</latexit> µ = 0.1 <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> <latexit sha1_base64="(null)">(null)</latexit> -
10-10 10-8 10-6 10-4 10-2 100 100 101 102 103
104 s-1.5 P(s) extinction size, s µ = 0 10-4 10-3 10-2 10-1 0 0.1 0.2 -10 -5 0 5 10 P(f) fitness, f µ = 0 10-4 10-3 10-2 10-1 0 0.1 -0.3 0 0.3 10-16 10-12 10-8 10-4 100 10-2 100 102 104 106 108 1010 1012 L-1 exp(-(L/L0 )1/2) P(L) species lifetime, L µ = 0 10-4 10-3 10-2 10-1 10-20 10-15 10-10 10-5 100 10-2 100 102 104 106 108 1010 1012 τ-1.2 P(τ) inter-event time, τ µ = 0 10-4 10-3 10-2 10-1 (a) (b) (c) (d) The constraint on N significantly alters the model behavior. -
The system is under a high pressure of potential new
species trying to migrate into it. N=const critical point The constraint on N significantly alters the model behavior. The system goes to an off-critical state as N decreases, preventing critical avalanches of extinctions. Extremal dynamics + Constraints -> SOC -
Summary • We formulated and studied models that bridges the
DG model and the BS model in order to identify a key factor for generating SOC phenomena in a biological evolution model. generalized model DG model link-based BS model • The applicability of BS model is questionable as the conservation of the system size is not satisfied in general.