LECTURE NOTES ON MATHEMATICAL PHYSICS

Department of Physics, University of Bologna, Italy

URL: www.fracalmo.org

Lévy Stable Distributions

in the Theory of Probability
Francesco MAINARDI
Dipartimento di Fisica, Università di Bologna and INFN,
Via Irnerio 46, I-40126 Bologna, Italy
[email protected] [email protected]

Preliminary Version1 , Bologna, May 2, 2007

1 Introduction
Stable distributions are a fascinating and fruitful area of research in probability
theory; furthermore, nowadays, they provide valuable models in physics,
astronomy, economics, and communication theory, see e.g. .......
The general class of stable distributions was introduced and given this
name by the French mathematician Paul Lévy in the early 1920’s, see Lévy
(1923,1924,1925).
Formerly, the topic attracted only moderate attention from the leading
experts, though there were also enthusiasts, of whom the Russian mathemati-
cian Alexander Yakovlevich Khintchine should be mentioned first of all. The
inspiration for Lévy was the desire to generalize the celebrated Central Limit
Theorem, according to which any probability distribution with finite variance
belongs to the domain of attraction of the Gaussian distribution. The concept
of stable distributions took full shape in 1937 with the appearance of Lévy’s
monograph [63], soon followed by Khintchine’s monograph [46]2 .
1
LaTeX file fm− stable-new.tex
2
Nowadays F. Mainardi and S. Rogosin, with the aim of reevaluating the work of
Khintchine on limit theorems of sums of independent variables (not necessarily identically
distributed), are planning to translate into English (from the Russian, Italian, German and
French) the relevant papers by Khintchine and also a number of related papers by De Finetti,
2

The theory and properties of stable distributions have been systematically

presented by Gnedenko & Kolmogorov [26] and Feller [?]. These distribution
are also discussed in some other classical books in probability theory including
Lukacs (1960-1970), Feller (1966-1971), Breiman (1968-1992), Chung (1968-
1974) and Laha & Rohatgi (1979). Also treatises on fractals devote
particular attention to stable distributions in view of their properties of
scale invariance, see e.g. Mandelbrot (1982) and Takayasu (1990). It is only
recently that monographs devoted solely to stable distributions and related
stochastic processes have been appeared, i.e. Zolotarev (1983-1986), Janicki
& Weron (1994), and Samorodnitsky & Taqqu (1994), Uchaikin & Zolotarev
(1999), Nolan (????). In these books tables and/or graphs related to stable
distributions are also exhibited. Previous sets of tables and graphs have been
provided by Mandelbrot & Zarnfaller (1959), Fama & Roll (1968), Bo’lshev &
Al. (1968) and Holt & Crow (1973).
Stable distributions have three exclusive properties, which can be briefly
summarized stating that they 1) are invariant under addition, 2) possess their
own domain of attraction, and 3) admit a canonical characteristic function.
In the following sections let us illustrate the above properties which,
providing necessary and sufficient conditions, can be assumed as equivalent
definitions for a stable distribution. We recall the basic results without proof.

2 Invariance under addition

A random variable X is said to have a stable distribution P (x) = Prob {X ≤
x} if for any n ≥ 2 , there is a positive number cn and a real number dn such
that
d
X1 + X2 + . . . + Xn = cn X + dn , (2.1)
where X1 , X2 , . . . Xn denote mutually independent random variables with
d
common distribution P (x) with X . Here the notation = denotes equality in
distribution, i.e. means that the random variables on both sides have the same
probability distribution.
When mutually independent random variables have a common distribution
[shared with a given random variable X], we also refer to them as independent,
identically distributed (i.i.d) random variables [independent copies of X]. In
Lévy, Kolmogorov, Gnedenko, Feller, that have been source of inspiration for Khintchine
himself. The 1938 book by Khintchine has already been translated from the Russian by
Rogosin
 3

general, the sum of i.i.d. random variables becomes a random variable with
a distribution of different form. However, for independent random variables
with a common stable distribution, the sum obeys to a distribution of the same
type, which differs from the original one only for a scaling (cn ) and possibly
for a shift (dn ). When in (A.1) the dn = 0 the distribution is called strictly
stable.
It is known, see [20], that the norming constants in (2.1) are of the form

cn = n1/α with 0 < α ≤ 2 . (2.2)

The parameter α is called the characteristic exponent or the index of stability

of the stable distribution. We agree to use the notation X ∼ Pα (x) to
denote that the random variable X has a stable probability distribution with
characteristic exponent α . We simply refer to Pα (x) , pα (x) := dPα (x)/dx
(probability density functions = pdf ) and X as α-stable distribution, density,
random variable, respectively.
Definition (2.1) with theorem (2.2) can be stated in an alternative version
that needs only two i.i.d. random variables. see also Lukacs (1960-1970).
A random variable X is said to have a stable distribution if for any positive
numbers A and B, there is a positive number C and a real number D such that
d
A X1 + B X2 = C X + D , (2.3)

where X1 and X2 are independent copies of X . Then there is a number α ∈

(0, 2] such that the number C in (2.3) satisfies C α = Aα + B α .
For a strictly stable distribution Eq. (2.3) holds with D = 0 . This implies
that all linear combinations of i.i.d. random variables obeying to a strictly
stable distribution is a random variable with the same type of distribution.
A stable distribution is called symmetric if the random variable −X has
the same distribution. Of course, a symmetric stable distribution is necessarily
strictly stable.
Noteworthy examples of stable distributions are provided by the Gaussian
(or normal) law (with α = 2) and by the Cauchy-Lorentz law (α = 1). The
corresponding pdf ′s are known to be
1 2 2
pG (x; σ, µ) := √ e −(x − µ) /(2σ ) , x ∈ R, (2.4a)
2π σ
where σ 2 denotes the variance and µ the mean, and
1 γ
pC (x; γ, δ) := , x ∈ R, (2.5a)
π (x − δ)2 + γ 2
4

where γ denotes the semi-interquartile range and δ the ”shift”. The

corresponding (cumulative) distribution functions are
" !#
1 x
PG (x; σ, ) := 1 + erf √ , x ∈ R, (2.4b)
2 2σ

and !
1 1 x
PC (x; γ, 0) = + arctan , x ∈ R. (2.5b)
2 π γ

3 Domain of attraction
Another (equivalent) definition states that stable distributions are the only
distributions that can be obtained as limits of normalized sums of i.i.d. random
variables. A random variable X is said to have a domain of attraction,i.e.
if there is a sequence of i.i.d. random variables Y1 , Y2 , . . . and sequences of
positive numbers {γn } and real numbers {δn }, such that

Y1 + Y2 + . . . Yn d
+ δn ⇒X . (3.1)
γn
d
The notation ⇒ denotes convergence in distribution.
If the random variable X has a stable distribution and all Yi are taken to
be independent and distributed like X , then clearly (2.1) implies (3.1), and
so trivially every random variable with a stable probability distribution has a
domain of attraction. The converse is also true, namely that every random
variable with a domain of attraction has a stable probability distribution, see
[26]. Therefore we can alternatively state that a random variable X is said to
have a stable distribution if it has a domain of attraction.
When X is Gaussian and the Yi′ s are i.i.d. with finite variance, then (3.1)
is the statement of the ordinary Central Limit Theorem. The domain of
attraction of X is said normal when γn = n1/α ; in general, γn = n1/α h(n)
where h(x) , x > 0 , is a slow varying function at infinity3 The function
h(x) = ln x , for example, is slowly varying at infinity: it enters in a general
3
Definition: We call a (measurable) positive function a(y), defined in a right
neighbourhood of zero, slowly varying at zero if a(cy)/a(y) → 1 with y → 0 for every
c > 0. We call a (measurable) positive function b(y), defined in a neighbourhood of infinity,
slowly varying at infinity if b(cy)/a(y) → 1 with y → ∞ for every c > 0. Examples: (log y)γ
with γ ∈ R and exp (log y/log log y).
 5

theorem on the domain of attraction of the Gaussian law for distributions with
infinite variance (in particular densities decaying as |x|−3 ), as formerly shown
independently by Khintchine, Lévy and Feller, see e.g. [46, 26, 20].

4 Canonical forms for the characteristic func-

tion
Another definition specifies the canonical form that the characteristic function
(cf ) of a stable distribution of index α must have. Let us recall that the cf
is the Fourier transform of the pdf . If we denote by pα (x; ·) the pdf for a
generic stable distribution of index α (the · stands for additional parameters)
the corresponding cf reads in our notation
Z +∞
p̂α (κ; ·) := hexp iκXi = eiκx pα (x; ·) dx ÷ pα (x; ·) ,
−∞

where ÷ denotes the juxtaposition of a function with its Fourier transform.

We note that the cf of the most popular stable distributions, the Gaussian
(2.4) and the Cauchy-Lorentz (2.5), turn out to be
2 2
pbG (κ; σ) = e −(σ /2) |κ| , (4.1)

pbc (κ; γ) = e−γ|κ| . (4.2)

Let first consider the (simplified) canonical form for strictly stable
distributions adapting from that formerly introduced by Feller [19, 20] and
Zolotarev [92, 90]. Using our notation this canonical forms reads
 
p̂α (κ; θ) := exp −|κ| e i(sign κ) θ π/2 ,
α
(4.3)

were θ is a real parameter whose domain is restricted to the following region

(depending on α) 
α, if 0 < α ≤ 1 ,
|θ| ≤ (4.4)
2 − α , if 1 < α ≤ 2 .
We recognize that pα (x, θ) = pα (−x, −θ) , so the symmetric stable distributions
are obtained if and only if θ = 0 ; θ is called the asymmetry parameter, or
simply (but improperly) skewness.
In the plane {α , θ} the allowed region for the parameters α and θ {0 <
α ≤ 2 , |θ| ≤ min (α, 2 − α)} turns out to be a diamond with vertices in the
6

1
θ

0.5

0 2
0.5 1 1.5

−0.5

α
−1

Figure 1: The Feller-Takayasu diamond

points (0, 0) , (1, 1) , (2, 0) , (1, −1) , that was formerly depicted in Takayasu’s
book, see Fig. 1. Honouring both Feller and Takayasu, we call the described
region the Feller-Takayasu diamond.
We note that in his original and pioneering paper [19], Feller used a
skewness parameter δ different from our θ ; in fact his characteristic function
turns out to be
  α 
2 π θ
p̂Fα (κ; δ) := exp − |κ| e −i (sign κ) δ , so δ = −
θ = − αδ . ,
π 2 α
(4.5)
In his two books [92, 90] Zolotarev used a notation which is confusing and
misleading due to irritating misprints: as matter of fact one can recognizes that
two different canonic forms are used for strictly stable distributions, namely
  
 exp −|κ| e −i(sign κ) δ1 α π/2 ,

 α

p̂Zα (κ; δ1,2 ) := 

  (4.6)
 exp −|κ|α e −i(sign
 κ) δ2 π/2 .

Our notation is in agreement with that adopted by Schneider (1986) [he,

however, uses the letter β instead of θ] and Takayasu (1990) [see (5.23)-(5.24),
p. 124], but all of them neglect the case α = 1.
Feller has proved that, assuming 1/2 < α < 1 and x > 0 ,
1
p1/α (x−α ; θ) = pα (x; θ∗ ) , θ∗ = α(θ + 1) − 1 . (4.7)
xα+1
 7

A quick check shows that θ∗ falls within the prescribed range, |θ∗ | ≤ α ,
provided that |θ| ≤ 2 − 1/α .
Stable distributions with extremal values of the skewness parameter are
called extremal. One can prove that all the extremal stable distributions with
0 < α < 1 are one-sided, the support being R+ −
0 if θ = −α , and R0 if θ = +α .
For 0 < α < 2 the stable distributions exhibit heavy tails in such a way
that their absolute moment of order ν is finite only if ν < α . In fact one can
show that for non-Gaussian, not extremal, stable distributions the asymptotic
decay of the tails is
 
pα (x; θ) = O |x|−(α+1) , x → ±∞ . (4.8)

For the extremal distributions this is valid only for one tail, the other being
of exponential order. For 0 < α < 1 we have one-sided distributions which
exhibit an exponential left tail (as x → 0+ ) if θ = −α , or an exponential right
tail (as x → 0− ) if θ = +α . For 1 < α < 2 the extremal distributions are
two-sided and exhibit an exponential left tail (as x → −∞) if θ = +(2 − α) ,
or an exponential right tail (as x → +∞) if θ = −(2 − α) .
Consequently, the Gaussian distribution is the unique stable distribution
with finite variance. Furthermore, when α ≤ 1 , the first absolute moment
h|X|i is infinite as well, so we need to use the median to characterize the
expected value.
However, there is a fundamental property shared by all the stable
distributions that we like to point out: for any α the corresponding stable
pdf is unimodal and indeed bell-shaped, i.e. its n-th derivative has exactly n
zeros, see Gawronski (1984).
From Lévy’s times it is usual to adopt a more general canonical form for
stable distributions that takes into account of a scale parameter γ > 0, of a
shift parameter δ ∈ R in addition to a skewness parameter β ∈ R restricted to
be |β| ≤ 1. For this class of stable distributions, partly following the notation
of Samorodnitsky & Taqqu (1994) and denoting by Y the random variable, we
write Y ∼ Qα (y; β, γ, δ) with characteristic function

q̂α (κ; β, γ, δ) = exp {iδκ − γ α |κ|α [1 + i (sign κ) β ω(|κ|, α)]} , (4.9)

where 
tan (α π/2) , if α 6= 1 ,
ω(|κ|, α) = (4.10)
−(2/π) ln |κ| , if α = 1.
Consequently a random variable Y is said to have a stable distribution if there
are four real parameters α, β, γ, δ with 0 < α ≤ 2 , −1 ≤ β ≤ +1 , γ > 0 , such
8

that its characteristic function has the canonical form (4.9)-(4.11). We note
that the parameter β appears with different signs for α 6= 1 and α = 1 . This
minor point has been the source of great confusion in the literature, see Hall
(1980) for a discussion. The presence of the logarithm for α = 1 is the source
of many difficulties, so this case has often to be treated separately.
An interesting problem is related to the inclusion of the canonical form
(4.3)-(4.4), which is valid only for strictly stable distributions in the more
general canonical form (4.9)-(4.10) We first note that the two parameters γ and
δ in (4.9), being related to a scale transformation and a translation respectively,
are not so essential since they do not change the shape of the distribution. If
we take γ = 1 and δ = 0 , we obtain the so-called standardized form of the
stable distribution and the corresponding random variable Y ∼ Pα (y; β, 1, 0) is
referred to as the α-stable standardized random variable. On the other hand,
keeping δ = 0, we can choose the scale parameter γ in such a way to get the
simplified canonical form for strictly stable distributions. As a matter of fact
the relation between the two classes X and Y for stable random variables can
be explored if we compare the corresponding canonical forms not only keeping
δ = 0 but also assuming α 6= 1, as shown below.
We easily recognize
     
π π π
γ α = cos θ , tan θ = β tan α , (4.11)
2 2 2
so the case α = 1 must be excluded. Thus, the Feller-Takayasu canonical
form for strictly stable distributions with index α 6= 1 and skewness θ , can
be obtained from the Lévy canonical form if we (implicitly) select the shift
parameter δ = 0, and the scale parameter γ and the skewness parameter β
related to α and θ according to (4.11). We note that necessarily 0 < γ ≤ 1. We
also note that for α = 1 we have identity between the two canonical forms in
the limiting case θ = β = 0 corresponding to the (symmetric) Cauchy-Lorentz
pdf . By the way for α = 1 with δ = 0 and β 6= 0 the general Lévy canonical
form yields not strictly stable distributions.
Specifically, the random variable X ∼ Pα (x; θ) turns out to be related to
the standardized random variable Y ∼ Qα (y; β, 1, 0) by the relations:
X = Y /γ , qα (y; β, 1, 0) = γ pα (x = γy; θ) , (4.12)
with
γ = [cos (θπ/2)]1/α , (4.13)
and
tan (θπ/2)
θ = (2/π) arctan [β tan (απ/2)] , β= . (4.14)
tan (απ/2)
 9

We note that for the symmetric stable distributions we get the identity between
the standardized and the Lévy canonical forms, since in (4.14) β = θ = 0
implies in (4.13) γ = 1 . A particular but noteworthy case is provided
by p2 (x; 0) = q2 (y; 0, 1, 0) corresponding to the Gaussian distribution with
variance σ 2 = 2 . The identity is valid also in the limit for α = 1, for which
p1 (x; 0) = q1 (y; 0, 1, 0) corresponding to the Cauchy-Lorentz distribution with
semi-interquartile γ = 1 .
The extremal stable distributions corresponding to β = ±1 and α 6= 1 are
obtained in the Feller-Takayasu representation for θ = ±α if 0 < α < 1 , and
for θ = ∓(2 − α) if 1 < α < 2 . For these cases, the scaling parameter turns
out to be γ = [cos (|α| π/2)]1/α .
It may be an instructive exercise to carry out the inversion of the Fourier
transform when α = 1/2 and θ = −1/2 . In this case we obtain the analytical
expression for the corresponding extremal stable pdf , known as the (one-sided)
Lévy-Smirnov density,
1
p1/2 (x; −1/2) = √ x−3/2 e−1/(4x) , x ≥ 0. (4.15)
2 π

The corresponding standardized form for this distribution can be easily

obtained from (4.15) using (4.12)-(4.14) with α = 1/2 and θ = −1/2 . We
get γ = [cos (−π/4)]2 = 1/2 , β = −1 , so
1 1
q1/2 (y; −1, 1, 0) = p1/2 (y/2; −1/2) = √ y −3/2 e−1/(2y) , y ≥ 0 , (4.16)
2 2π
in agreement with Holt & Crow (1973) [§2.13, p. 147].
10

References
[1] R. Bartels, Generating non-normal stable variates using limit theorems
properties, J. Stat. Comp. Simulation 7 (1978) 199–212.
[2] J.-P. Bouchaud and M. Potters, Theory of Financial Risk: from Statistical
Physics to Risk Management, (Cambridge Univ. Press, Cambridge, 2000.
[English enlarged version of Théories des Risques Financiers, CEA, Aléa, Saclay, 1997.]

[3] Breiman, L. : Probability, SIAM, Philadelphia 1992.

[The SIAM edition is an unabridged, corrected republication of the 1st edn, Addison-
Wesley, Reading Mass. 1968]

[4] P. Butzer and U. Westphal, Introduction to fractional calculus, in:

H. Hilfer (Editor), Fractional Calculus, Applications in Physics, World
Scientific, Singapore, 2000, pp. 1-85.
[5] J.M. Chambers, C.L. Mallows and B.W. Stuck, A method for simulating
stable random variables, J. Amer. Stat. Assoc. 71 (1976) 340–344.
[6] Chao Chung-Jeh, Explicit formula for the stable law of distribution, Acta
Math. Sinica 3 (1953) 177-185. [in Chinese with English summary]
[7] P.G. Christoph and W. Wolf, Convergence Theorems with a Stable Limit
Law, Akademie Verlag, Berlin, 1992.
[8] K.L. Chung, A Course in Probability Theory, 2nd edn, Academic Press,
New York, 1974. [1st edn, Harcourt Brace Jovanovich, 1968]
[9] N. Cressie, A note on the behaviour of the stable distributions for small
index α, Z. Wahrsch. verw. Gebiete 33 (1975) 61-64.
[cited by Uchaikin-Zolotarev and by Shanbhag-Sreehari (1977)]

[10] A.R. Dabrowski and A. Jakubowski, Stable limits for associated random
variables, The Annals of Probability ?? (1993-94) to appear?
[Quoted by Samorodnitsky-Taqqu book (1994)]

[11] R.A. Davis, Stable limits for partial sums of dependent random variables,
The Annals of Probability 11 (1983) 262-269.
[Quoted by Samorodnitsky-Taqqu book (1994)]

[12] G.C. De Vries, On the relation between GARCH and stable processes, J.
Econometrics 48 No 3 (1991) 313–324.
[Quoted by SHIRYAEV]
 11

[13] L. Devroye, A triptych of discrete distributions related to the stable law,

Stat. Prob. Lett 18 (1993) 349-351.
[quoted by Erdogan, 1999.]

[14] C. Di Castro and G.Jona-Lasinio, On the microscopic foundation of

scaling laws, Physics Letters A 29 (1969) 322-323.

[15] C. Di Castro and G. Jona-Lasinio, The renormalization group approach to

critical phenomena, in: C. Domb and M.S. Green, eds., Phase transitions
and Critical Phenomena, Academic Press, London, 1976; Vol. 6, Ch. 8,
pp. 507-558.
[Biblioteca Fisica BO: 3Q.11.30 (non suggerito da JL) see APPENDIX A: A
Connection with Probability Theory (via stable pdf, pp 552-554)]

[16] W. Du Mouchel, Stable distributions in statistical inference, I: Symmetric

stable distribution compared to other symmetric long-tailed distributions,
J. Amer. Statistical Association 68 No 342 (1973) 469-477.
[Xerocopied in AARHUS, Aug 2001. Part II is cited in preparation. PhD thesis at Yale
Univ. 1971]

[17] A. Erdélyi. (Editor), Tables of Integral Transforms, Vols. 1-2, McGraw-

Hill, New-York 1953-1955.

[18] E. Fama and R. Roll, Some properties of symmetric stable distributions,

J. Amer. Statist. Assoc. 63 (1968) 817-836.

[19] W. Feller, On a generalization of Marcel Riesz’ potentials and the semi-

groups generated by them. Meddelanden Lunds Universitets Matematiska
Seminarium (Comm. Sém. Mathém. Université de Lund), Tome suppl.
dédié à M. Riesz, Lund (1952), 73-81.

[20] W. Feller, An Introduction to Probability Theory and its Applications, Vol.

2, 2-nd edn, Wiley New York, 1971. [1-st edn. 1966]

[21] Y. Fujita, Cauchy problems of fractional order and stable processes, Japan
J. Appl. Math. 7 No 3 (1990) 459-476.

[22] W. Gawronski, On the bell-shape of stable distributions, Ann. Probab. 12

(1984), 230-242.

[23] W. Gawronski, Asymptotic forms for the derivatives of one-sided stable

distributions, The Annals of Probability 16 No 3 (1988) 1348-1364.
12

[24] W. Gawronski and M. Wiessner, Asymptotics and inequalities for the

mode of stable laws, Statistics & Decisions 10 (1992) 183-197.

[25] B.V. Gnedenko, On the theory of domains of attraction of stable laws,

Uchenye Zapiski Moskov. Gos. Univ. Matematkia 45 (1940) 61-72. [in
Russian]
[quoted in the book by Mantegna & Stanley, 2000]

[26] B.V. Gnedenko and A.N. Kolmogorov, Limit Distributions for Sums of
Independent Random Variables, Addison-Wesley, Cambridge Mass., 1954.
[English Transl. from the Russian edition (1949), with notes by K.L.
Chung, revised (1968)]

[27] R. Gorenflo, A. Iskenderov and Yu. Luchko, Mapping between solutions

of fractional diffusion-wave equations. Fractional Calculus and Applied
Analysis 3 (2000) 75–86.

[28] R. Gorenflo and F. Mainardi, Fractional calculus: integral and differ-

ential equations of fractional order, in A. Carpinteri and F. Mainardi
(Editors), Fractals and Fractional Calculus in Continuum Mechanics.
Wien and New York, Springer Verlag (1997), 223-276. [Reprinted in
http://www.fracalmo.org]

[29] R. Gorenflo and F. Mainardi, Random walk models for space-fractional

diffusion processes, Fractional Calculus and Applied Analysis 1 (1998)
167–191.

[30] R. Gorenflo and F. Mainardi, Fractional calculus and stable probability

distributions, Archive of Mechanics 50 (1998) 377–388.

[31] R. Gorenflo and F. Mainardi, Approximation of Lévy-Feller diffusion by

random walk, J. Analysis and its Applications (ZAA) 18 No 2 (1999)
231–246.

[32] R. Gorenflo and F. Mainardi, Random walk models approximating

symmetric space-fractional diffusion processes, in: J. Elschner, I.
Gohberg and B. Silbermann (Editors), Problems in Mathematical Physics
(Birkhäuser Verlag, Basel, 2001), pp. 120-145. [Series ”Operator Theory:
Advances and Applications” No 121]

[33] R. Gorenflo and F. Mainardi, Fractional Diffusion Processes, Lecture

Notes of the MaPhySto Centre, University of Aarhus, (2001), in
 13

preparation. To be submitted to World Scientific, Singapore, Advanced

Series on Statistical Science & Applied Probability, Vol. ?, edited by
OEBN.

[34] R. Gorenflo, F. Mainardi, D. Moretti, G. Pagnini, and P. Paradisi,

Fractional Diffusion: Probability Distributions and Random Walk
Models, Physica A 305 No 1/2 (2002) 106-112.

[35] R. Gorenflo, F. Mainardi, D. Moretti, G. Pagnini, and P. Paradisi :

Discrete random walk models for space-time fractional diffusion, Chemical
Physics, special issue on Strange Kinetics Guest Editors: J. Klafter, R.
Hilfer, R. Metzler

[36] - R. Gorenflo and F. Mainardi, Fractional diffusion processes: probability

distributions and continuous time random walk, in: G. Rangarajan and M.
Ding (Editors): Processes with Long Range Correlations, Springer Verlag,
Berlin, 2003, pp. 148-166. [Lecture Notes in Physics, No 621]

[37] V.V. Gorodetskii, On convergence to semi-stable Gaussian processes,

Theory Probab. Appl. 22 (1977) 498-508.
[quoted by D. Marinucci & P.M. Robinson, on fBm (1999)]

[38] P. Hall, A comedy of errors: the canonical form for a stable characteristic
function, Bull. London Math. Soc. 13 (1980), 23-27.

[39] P. Hall, On unimodality and rates of convergence for stable laws, J.

London Math Soc. (Ser. 2) 30 (1984) 371-384. [?]
[Quoted in Gawronski & Wiessner, 1992]

[40] J. Hoffman-Jørgensen, Stable densities, Theory of Probability and its

Applications 38 No 2 (1993) 350-355.

[41] D.R. Holt and E.L Crow,. : Tables and graphs of the stable probability
density functions, J. Res. Nat. Bureau Standards 77B (1973) 143-198.

[42] I. Ibragimov and K.E. Cernin, On the unimodality of stable laws, Theor.
Prob. and its Appl. 4 No. 4 (1961) 417-419
[Quoted in Shiryaev’s book on Stochastic Finance]

[43] I. Ibragimov and Y. Linnik, Independent and Stationary Sequences of

Random Variables, Addison-Wesley, Reading, Massachusetts, 1968.
14

[44] Janicki, A and Weron, A simulation and Chaotic Behavior of α–

Stable Stochastic Processes. Marcel Dekker, New York, 1994.

[45] J-P. Kahane, Definition of stable laws, infinitely divisible laws, and Lévy
processes, in: M.F. Shlesinger, G.M. Zaslavsky and U. Frisch (Editors),
Lévy Flights and Related Topics in Physics, (Springer Verlag, Berlin,
1995), pp. 99-109 [Lecture Notes in Physics, No 450]

[46] , A. Ya. Khintchine, Limit Laws for Sums of Independent Variables,

ONTI, Moscow, 1938. [in Russian]
[A provisional translation by S. Rogosin into English is available]

[47] A. Ya. Khintchine and P. Lévy, Sur le lois stables, C.R. Acad. Sci Paris
202 (1937) 374-376.

[48] L.B. Klebanov, G.M. Maniya and I.A. Melamed, A problem of Zolotarev
and analogs of infinitely divisible and stable distributions in a scheme or
summing a random number of random variables, Theory Prob. Appl. 29
(1984) 791–794. [quoted by Erdogan, 1999 and Pillai, 1990]

[49] A.N. Kolmogorov, Sulla forma generale di un processo stocastico

omogeneo (Un problema di Bruno De Finetti), Atti Accad. Naz. dei Lincei,
Rendiconti (Ser VI), 15 (1932), 805-808, 866-869.

[50] T.J. Kozubowski and S.T. Rachev, The theory of geometric-stable

distributions and its use in modeling financial data, European J. Oper.
Res. 74 (1994) 310–324.

[51] W.P. Kurenok, On weak convergence of random walks to symmetric stable

processes, Minsk Trudy Instituta Matematiki TOM 6, pp. 109-112 (1999).
[in ?? (Editors), Proceedings AMADE 1999] [Gorenflo 8 sept 00: Kurenok
proves weak convergence of our Gruenwald-Letnikov random walk and
the Bordeaux wine (Paris) random walk by means of the Levi-Khintchine
formula.]

[52] R.G. Laha. and V.K. Rohatgi, Probability Theory, Wiley, New York, 1979.

[53] J.W. Lamperti, Semi-stable stochastic processes, Trans. Amer. Math. Soc.
104 (1962) 62-78.

[54] J.W. Lamperti, Probability. A Survey of the Mathematical Theory

(Benjamin, New York and Amsterdam, 1966), pp. 149.
 15

[55] J.W. Lamperti, Semi-stable Markov processes, Z. Wahrsch. verw. Gebiete

22 (1972) 205-235. [

[56] J.W. Lamperti, Stochastic Processes, Springer Verlag, Amsterdam, 1977.

[Series of Appl. Math. Sciences, No 23]

[57] P. Lévy, Sur le rôle de la loi de Gauss dans la théorie des erreurs,
C.R. Acad. Sci. Paris, 174 (1922), 855-857. (Séance du 27 mars 1922).
[Reprinted in Oeuvres de Paul Lévy edited by D. Dugué, Vol 3 (Eléments
aléatoires) No 88 pp. 9-11, Gauthier-Villars, Paris, 1976).

[58] P. Lévy, Sur une application de la la dérivée d’ordre non entier au calcul
des probabilités, C.R. Acad. Sci. Paris, 176 (1923) 1118-1120 (Séance du
23 avril 1923). [Reprinted in Oeuvres de Paul Lévy edited by D. Dugué
Vol 3 (Eléments aléatoires) No 115 pp. 339-341, Gauthier-Villars, Paris,
1976.]

[59] P. Lévy, Sur les lois stables en calcul des probabilités, Compt. Rendus
Acad. Sci. Paris 176 (1923), 1284-1286. (Séance du ?? ??l 1923).
[Reprinted in Oeuvres de Paul Lévy edited by D. Dugué, Vol 3 (Eléments
aléatoires) No 116 pp. 342-344, Gauthier-Villars, Paris, 1976.]

[60] P. Lévy, Théorie des erreurs. La Loi de Gauss et les lois exceptionelles,
Bull. Soc. Math. France 52 (1924) 49-85. [Premiere Partie: La loi de
Gauss, pp. 49-68, §1-9] [Seconde Partie: Les lois exceptionnelles, pp. 68-
84, §10-15] [Reprinted in Oeuvres de Paul Lévy edited by D. Dugué, Vol
3 (Eléments aléatoires) No 90 pp. 14-49, Gauthier-Villars, Paris, 1976.]

[61] P. Lévy, Calcul des Probabilités (Gauthier-Villars, Paris, 1925), pp.

viii+350. Reprinted in 2003 by Editions Jaques Gabay (Les Grands
Classiques Gauthier-Villars)

[62] P. Lévy, Sur un theoreme de M. Khintchine Bull. Sci. Math. (Ser. 2) 55

(1931) 1-16 (145-160). [Reprinted in Oeuvres de Paul Lévy edited by D.
Dugué, Vol 3 (Eléments aléatoires) No 94 pp. 86-101, Gauthier-Villars,
Paris (1976)]

[63] P. Lévy, Théorie de l’addition des variables aléatoires, 2nd Edn, Gauthier-
Villars, Paris, 1954. [1st Edn (1937)] [Reprinted in 2003 by Editions
Jaques Gabay (Les Grands Classiques Gauthier-Villars)]
16

[64] P. Lévy, Remarques sur un probleme relatif aux lois stables, Studies
in Math. Anal. and Related Topics (Essays in Honor of George
Pólya) Stanford Univ. Press, Stanford, California (1962), pp. 211-218.
[Reprinted in Oeuvres de Paul Lévy edited by D. Dugué, Vol 4 (Processus
stochastiques) No 170bis pp.410-???. Gauthier-Villars, Paris, 1980.]
[quoted by ZOLOTAREV’s book on Stable Distributions, ref 52]

[65] E. Lukacs, Characteristic Functions, 2nd edn, Griffin, London 1970. [1st
edn, 1960]

[66] S. Lung and F.J. Molz, How well are hydraulic conductivity variations
approximated by additive stable processes? Advances in Environmental
research 5 (2001) 39-45. [X]

[67] E. Lutz, Fractional transport equations for Lévy stable processes, Phys.
Rev. Lett. 86 No 11 (2001) 2208-2211

[68] F. M a i n a r d i, Fractional calculus: some basic problems in

continuum and statistical mechanics, in A. Carpinteri and F. Mainardi
(Editors), Fractals and Fractional Calculus in Continuum Mechanics.
Wien and New York, Springer Verlag (1997), 291-348. [Reprinted in
http://www.fracalmo.org]

[69] F. M a i n a r d i, Yu. L u c h k o, G. P a g n i n i , The

fundamental solution of the space-time fractional diffusion equation.
Fractional Calculus and Applied Analysis 4 (2001), 153–192. [Reprinted
http://www.fracalmo.org]

[70] F. M a i n a r d i and G. P a g n i n i, Salvatore Pincherle: the pioneer

of the Mellin-Barnes integrals. J. Computational and Appl. Mathematics
153 (2003), 331-342.

[71] F. Mainardi and M. Tomirotti, Seismic pulse propagation with constant

Q and stable probability distributions, Annali di Geofisica 40 (1997),
1311-1328.

[72] B.B. Mandelbrot, The Fractal Geometry of Nature, Freeman, San

Francisco 1982.

[73] B.B. Mandelbrot, Fractals and Scaling in Finance, Springer-Verlag, New

York, 1997.
 17

[74] B.B. Mandelbrot. and F. Zarnfaller Five place tables of certain stable
distributions, Technical Report RC-421, IBM Thomas J. Watson Research
Center, Yorktown Heights, New York, Dec 31, 1959.

[75] R.N. Mantegna, Fast, accurate algorithm for numerical simulation of Lévy
stable stochastic processes, Phys. Rev. E 49 (1994) 4677–4683.

[76] R.N. Mantegna and H.E. Stanley, An Introduction to Econophysics,

Cambridge University Press, Cambridge, 2000.

[77] O.I. M a r i c h e v, Handbook of Integral Transforms of Higher

Transcendental Functions, Theory and Algorithmic Tables. Chichester,
Ellis Horwood (1983).

[78] A.M. M a t h a i and R.K. S a x e n a, The H-function with Applications

in Statistics and Other Disciplines. New Delhi, Wiley Eastern Ltd (1978).

[79] Meerschaert, M. M., Benson, D. A. and Scheffler, H.-P.

Stochastic solution of space-time fractional diffusion equations. Physical
Review E, 65 (2002), 041103/1-4.

[80] E.W. Montroll and M.F. Shlesinger, On the wonderful world of random
walks, in: J. Leibowitz and E.W. Montroll, eds., Nonequilibrium
Phenomena II: from Stochastics to Hydrodynamics (North-Holland,
Amsterdam, 1984), pp. 1-121.

[81] E.W. Montroll and B.J. West, On an enriched collection of stochastic

processes, in: E.W. Montroll and J. Leibowitz, eds., Fluctuation
Phenomena (North-Holland, Amsterdam, 1979), pp. 61-175.

[82] Samorodnitsky, G. and Taqqu, M. S. Stable non-Gaussian Random

Processes. Chapman & Hall, New York, 1994.

[83] Sato, K. Lévy Processes and Infinitely Divisible Distributions. Cam-

bridge University Press, Cambridge, 1999.

[84] W.R. S c h n e i d e r, Stable distributions: Fox function representation and

generalization, in S. Albeverio, G. Casati and D. Merlini (Editors), Sto-
chastic Processes in Classical and Quantum Systems. Berlin-Heidelberg,
Springer-Verlag (1986), 497-511. [Lecture Notes in Physics, Vol. 262]

[85] M.D. S p r i n g e r, The Algebra of Random Variables. New York, Wiley

(1979).
18

[86] H.M. S r i v a s t a v a, K.C. G u p t a and S.P. G o y a l, The H-

Functions of One and Two Variables with Applications. New Delhi and
Madras, South Asian Publishers (1982).

[87] Stanislavski, A. A. Black-Scholes model under subordination. Physica

A, 318 (2003), 469-474.

[88] H. T a k a y a s u, Fractals in the Physical Sciences. Manchester and New

York, Manchester University Press (1990).

[89] C. Tsallis, Lévy Distributions, Physics World 10 No 7 (1997) 42–45.

[90] V.V. U c h a i k i n and V.M. Z o l o t a r e v, Chance and Stability.

Stable Distributions and their Applications. Utrecht, VSP (1999).

[91] Zolotarev, V. M. Mellin-Stieltjes transform in probability theory.

Theor. Prob. Appl., 2 (1957), 433–460.

[92] V.M. Z o l o t a r e v, One-dimensional Stable Distributions. Providence,

R.I., Amer. Math. Soc. (1986). [Translation from the Russian edition
(1982)]