Mo. Vlad et al., STATISTICAL FRACTALS WITH CUTOFFS, SHLESINGER-HUGHES RENORMALIZATION,AND THE ONSET OF AN EPIDEMIC, Physical review. E, Statistical physics, plasmas, fluids, and related interdisciplinary topics, 53(2), 1996, pp. 1382-1398
A method for constructing cutoff values of the probability densities a
ttached to statistical fractals is introduced for which only the begin
ning of the tail of a probability density is slowly decaying and given
by an inverse power law; the end of the tail, however, is short, deca
ying exponentially or faster. This method is illustrated by an example
from the theory of epidemics. The probability density (psi)over tilde
>(t)dt of the time interval t within which an infected individual is a
ble to spread an epidemic is evaluated based on the following assumpti
ons. (1) The spreading of the epidemic depends on the size of the germ
load carried by an infected individual; the total germ load is measur
ed in large units containing at least 10(5)-10(6) germs. It is assumed
that for each unit there is a constant probability alpha of infecting
other individuals. (2) The total number of germ units carried by an i
ndividual is a random variable; the corresponding distribution is eval
uated by assuming that there is a constant probability a: that an infe
cted individual carries a unit load of germs. (3) The encounters of an
infected individual with healthy individuals susceptible to receiving
the illness is a random event occurring with an average time-independ
ent contact frequency v. The probability density (psi)over tilde>(t)dt
determined by these three assumptions has a long tail characteristic
of an ideal statistical fractal behavior: (psi)over tilde>(t)dt simila
r to(vt)(-(H+1))Xi(ln(vt))d(vt) as t>>1/v, where H=ln(a)/ln(1-p) is a
positive fractal exponent and Xi(ln(vt)) is a periodic function of ln(
vt) with a period -ln(1-p). In this case all positive moments [t(q)] o
f the infection time with q greater than or equal to H are infinite. A
statistical fractal with a cutoff emerges if a fourth hypothesis is a
dded: (4) Due to the healing process the bacterial load of an individu
al decreases exponentially in time with a rate coefficient b. If the h
ealing process is slower than the encounter process of healthy individ
uals, v>b, then only the beginning of the tail of (psi)over tilde>(t)d
t obeys an inverse power law scaling (psi)over tilde>(t)dt similar to(
vt)(-(H+1)))Xi(ln(vt))d(vt) for 1/b>t>1/v; the end of the tail, howeve
r, is exponential and determined by the healing process (psi)over tild
e>(t)similar to const x exp(-bt) as t>>1/b. Due to the cutoff the mome
nts [t(q)] of the infection time, although finite, have an intermitten
t behavior characterized by the scaling law [t(q)]similar to const x b
(-(q-H)) as b-->0. A generalization of the epidemic model is given by
assuming that the size of the germ load carried by an infected individ
ual is an arbitrary random function of time with known stochastic prop
erties and that the encounters with healthy individuals can be describ
ed by a correlated random point process. An analytical expression is d
erived for the probability density (psi)over tilde>(t)dt of the infect
ion time in terms of the characteristic functionals of the germ load a
nd of the encounter process. A comparison is performed between the ons
et of an epidemic and the passage over a fluctuating energy barrier wi
th dynamical disorder. Some implications of the cutoff of a statistica
l fractal for the physics of fractal time are also analyzed.