Boltzmann and Fokker-Planck Equations Modelling Opinion Formation … · 2010-05-07 · Boltzmann...

ASC Report No. 10/2009

Boltzmann and Fokker-Planck EquationsModelling Opinion Formation in thePresence of Strong Leaders

Bertram During, Peter Markowich, Jan-Frederick Pietschmann,

Marie-Therese Wolfram

Institute for Analysis and Scientific Computing

Vienna University of Technology — TU Wien

www.asc.tuwien.ac.at ISBN 978-3-902627-02-5

Most recent ASC Reports

09/2009 Othmar Koch, Ewa WeinmullerNumerical Treatment of Singular BVPS: the New MATLAB code bvpsuite

08/2009 Irena Rachunkova, Svatoslav Stanek, Ewa Weinmuller, Michael ZenzLimit Properties of Solutions of Singular Second-order Differential Equations

07/2009 Anton Arnold, Jose A.Carillo, Chiara ManziniRefined Long-Time Asymptotics for some Polymeric Fluid Flow Models

06/2009 Samuel Ferraz-Leite, Christoph Ortner, Dirk PraetoriusConvergence of Simple Adaptive Galerkin Schemes Based on H-H/2 Error Esti-mators

05/2009 Gernot Pulverer, Svatoslav Stanek, Ewa B. WeinmullerAnalysis and Numerical Solutions of Positive and Dead Core Solutions of Sin-gular Sturm-Liouville Problems

04/2009 Anton Arnold, N. Ben Abdallah, Claudia NegilescuWKB-based Schemes for the Schrodinger Equation in the Semi-classical Limit

03/2009 Emmanuil H. Georgoulis, Edward Hall, Jens Markus MelenkOn the Suboptimality of the p-version Interior Penalty Discontinuous GalerkinMethod

02/2009 Samuel Ferraz-Leite, Jens Markus Melenk, Dirk PraetoriusReduced Model in Thin-film Micromagnetics

01/2009 Markus Aurada, Samuel Ferraz-Leite, Dirk PraetoriusConvergence of Adaptive Boundary Element Methods

38/2008 Roberta BosiClassical Limit for Linear and Nonlinear Quantum Fokker-Planck Systems

Institute for Analysis and Scientific ComputingVienna University of TechnologyWiedner Hauptstraße 8–101040 Wien, Austria

E-Mail: [email protected]

WWW: http://www.asc.tuwien.ac.at

FAX: +43-1-58801-10196

ISBN 978-3-902627-02-5

c© Alle Rechte vorbehalten. Nachdruck nur mit Genehmigung des Autors.

ASCTU WIEN

Boltzmann and Fokker-Planck Equations

modelling Opinion Formation in the

Presence of Strong Leaders

By Bertram During1, Peter Markowich2,

Jan-Frederik Pietschmann2, Marie-Therese Wolfram2

1Institut fur Analysis und Scientific Computing, Technische Universitat Wien,

1040 Wien, Austria2DAMTP, University of Cambridge, Cambridge CB3 0WA, UK

We propose a mathematical model for opinion formation in a society which is builtof two groups, one group of ‘ordinary’ people and one group of ‘strong opinion lead-ers’. Our approach is based on an opinion formation model introduced in Toscani(2006) and borrows ideas from the kinetic theory of mixtures of rarefied gases.Starting from microscopic interactions among individuals, we arrive at a macro-scopic description of the opinion formation process which is characterized by asystem of Fokker-Planck type equations. We discuss the steady states of this sys-tem, extend it to incorporate emergence and decline of opinion leaders, and presentnumerical results.

Keywords: Boltzmann equation, Fokker-Planck equation, opinion formation,

sociophysics

1. Introduction

Opinion leadership is one of several sociological models trying to explain forma-tion of opinions in a society. It is a concept that goes back to Lazarsfeld et al.

(1944). In the course of their study of the presidential elections in the USA in 1940,Lazarsfeld et al. (1944) found out interpersonal communication to be much moreinfluential than direct media effects. They formulated a theory of a two-step flowof communication where so-called opinion leaders who are active media users areselecting, interpreting, modifying, facilitating, and finally transmitting informationfrom the media to less active parts of the population. Later sociologists obtained anew view on opinion leader characteristics by developing the notion of public indi-viduation. Public individuation describes how people feel the urge to differentiatethemselves and act differently from other people (Maslach et al. 1985). This is anecessity for an opinion leader, because she or he must be willing to set herselfor himself apart from the ordinary people. Certain, typical personal characteristicsare supposed to characterize opinion leaders: high confidence, high self-esteem, astrong need to be unique, and the ability to withstand criticism. An opinion leaderis socially active, highly connected and held in high esteem by those accepting hisor her opinions. Although different from the others, opinion leaders are still relatedto their followers and not always easy to distinguish from them. This is becauseopinion leadership is specific to a subject and can change over time. Someone whois a strong opinion leader in one field may be a follower in another.

2 B. During, P.A. Markowich, J.-F. Pietschmann, M.-T. Wolfram

In the last decade, new communication forms like email, web navigation, blogsand instant messaging have globally changed the way how information is dissem-inated and opinions are formed in the society (cf., e.g. Rash 1997). Still, opinionleadership continues to play a critical role in all these processes, independent of theunderlying technology. Opinion leadership appears in such different fields as butnot limited to

• political parties and movements: a prominent example of the latter is AlGore’s initiative The Climate Project ;

• advertisement of commercial products: product reviewers in the media whohave a deeper knowledge and background than average consumers;

• dissemination of new technologies: early adopters play an important role,either as lighthouse customers that assist in the development or as individualsthat recommend a new product to others;

• pharmaceutical industry: companies engage with key opinion leaders, i.e.physicians who influence their colleagues’ prescribing behaviour.

In recent years, opinion formation has received growing attention from physicists(Deffuant et al. 2002; Galam & Zucker 2000; Sznajd-Weron & Sznajd 2000), open-ing an own research field termed sociophysics which goes back to the pioneeringwork of Galam et al. (1982). We refer also to Comincioli et al. (2009) and Galam(2005) and the references therein. Often, especially in numerical simulation studies,cellular automata are used. Another approach uses models of mean field type, whichlead to systems of (ordinary or partial) differential equations. This approach hasthe advantage that up to a certain extent they can be treated analytically and helpto get a deeper understanding of the underlying dynamics. A third approach is tointroduce kinetic models of opinion formation (Toscani 2006). The basic paradigmis that the behaviour of a sufficiently large number of interacting individuals in asociety can be described by methods of statistical physics just as well as the collid-ing molecules of a gas in a container. Exchange of opinion between individuals inthese models is defined by pairwise, microscopic interactions. In dependence on thespecification of these interactions, the whole society develops a certain macroscopic

opinion distribution.Independently of the approach chosen, the prevalent literature primarily has

focused on election processes, referendums or public opinion tendencies. With theexception of Bertotti & Delitala (2007) who propose a simple, discrete model forthe influence of strong leaders in opinion formation, less attention has been paid tothe important effect that opinion leaders have on the dissemination of new ideasand the diffusion of beliefs in a society. In this paper, we turn to this problem.

Our work is based on a kinetic model for opinion formation introduced in Toscani(2006). It is built on two main aspects of opinion formation. The first one is a acompromise process (Deffuant et al. 2002; Weidlich 2000), in which individuals tendto reach a compromise after exchange of opinions. The second one is self-thinking,

where individuals change their opinion in a diffusive way, possibly influenced byexogenous information sources like the media. Based on both, Toscani (2006) intro-duced a kinetic model in which opinion is exchanged between individuals throughpairwise interactions. In a suitable scaling limit, a partial differential equation of

Opinion Formation with Strong Leaders 3

Fokker-Planck type was derived for the distribution of opinion in a society. Simi-lar diffusion equations were also obtained recently in Slanina & Lavicka (2003) as amean field limit the Sznajd model (Sznajd-Weron & Sznajd 2000). Mathematically,the model in Toscani (2006) is related to works in the kinetic theory of granulargases (Cercignani et al. 1994). In particular, the non-local nature of the compromiseprocess is analogous to the variable coefficient of restitution in inelastic collisions(Toscani 2000). Similar models were used in the modelling of wealth and incomedistributions which show Pareto tails, cf. During et al. (2008) and the referencestherein.

Clearly, there are some limitations of our approach, which is —as often in appliedmathematics— a very simplified model of the complex reality. First, the statisticaldescription will be expected to be valid only if the number of individuals is ratherlarge. Second, we do not consider the structures of social networks that can playan important role in diffusing opinions. Mathematically, such networks can be ex-pressed as graphs (cf., e.g. Sood et al. 2008). However, it should be noted that alsoin sociological models which focus on such underlying structures of society, opinionleaders play an important role. They act as promoters of opinions across differentsub-groups of the society (Burt 1999), in which opinions are easily communicatedand spread as, e.g. in a group of colleagues at work, among friends and family ormembers of a social or sports club. Mathematically, this can be represented byscale-free networks with the opinion leaders as ‘hubs’, i.e. highest-degree nodes inthe graph. In our model, although we have abstracted from the underlying socialnetwork, we can model this fact by controlling the interaction frequencies betweenopinion leaders and their followers. In any case, it is important to obtain a bet-ter understanding of the influence of opinion leaders on normal people. The modelpresented in this paper is a first step to a quantitative study of opinion leadership.

The paper is organized as follows. In the next section we introduce the modelwhich leads to a system of Boltzmann equations. We derive and study an associatedsystem of Fokker-Plank type equations in Section 3. Numerical examples will bepresented in Section 4. Section 5 concludes.

2. Kinetic models for opinion formation

The goal of a kinetic model for opinion formation is to describe the evolution of thedistribution of opinion by means of microscopic interactions among individuals ina society. Opinion is represented as a continuous variable w ∈ I with I = [−1, 1],where ±1 represent extreme opinions. If concerning political opinions I can beidentified with the left-right political spectrum.

(a) Toscani’s model

The study of the time-evolution of the distribution of opinion among individualsin a simple, homogeneous society, has been recently studied by means of kineticcollision-like models in Toscani (2006). This model is based on binary interactions.When two individuals with pre-interaction opinion v and w meet, then their post-trade opinions v∗ and w∗ are given by

v∗ = v − γP (|v − w|)(v − w) + η1D(v),


w∗ = w − γP (|w − v|)(w − v) + η2D(w).

Herein, γ ∈ (0, 12 ) is the constant compromise parameter. The quantities η1 and

η2 are random variables with mean zero and variance σ2. They model self-thinking

which each individual performs in a random diffusion fashion through an exogenous,global access to information, e.g. through the press, television or internet. Thefunctions P (·) and D(·) model the local relevance of compromise and self-thinkingfor a given opinion. To ensure that post-interaction opinions remain in the interval Iadditional assumptions need to be made on the random variables and the functionsDj(·).

(b) A kinetic model with opinion leaders

In this section we propose a generalized model, where individuals from two differ-ent groups of individuals interact with each other. Human societies typically containa set of individuals who, empirically speaking, strongly influence opinion throughtheir strong personalities, financial means, access to media etc. The sociophysicalkinetic modelling of their effect on public opinion is based on the hypothesis thattheir own opinions are not changed through interactions with regular society mem-bers. Therefore, we consider two groups, one shall be identified with such ‘strongopinion leaders’ and the other with their followers, the ‘ordinary people’. We willadopt the hypothesis that all individuals belonging to one group share a commoncompromise parameter. This hypothesis can be further relaxed by assuming thatthe compromise parameter is a random quantity, with a statistical mean which isdifferent for the two groups.

To some extent this can be seen as the analogue to the physical problem ofa mixture of gases, where the molecules of the different gases exchange momen-tum during collisions (Bobylev & Gamba 2006). However, a complete analogy fails,since the opinion leaders influence ordinary people in their opinion and main-tain their own. A maybe better analogy is with the solid state physics Boltz-mann equation, where charged particles collide with a fixed phonon background(Markowich & Poupaud & Schmeiser 1995). If two individuals from the same groupmeet, the interaction shall as in Toscani (2006) be given by (i=1,2)

v∗ = v − γiPi(|v − w|)(v − w) + ηi1Di(v), (2.1a)

w∗ = w − γiPi(|w − v|)(w − v) + ηi2Di(w). (2.1b)

If one individual from the group of ordinary people with opinion v meets a strongopinion leader with opinion w their post-interaction opinions are given by

v∗ = v − γ3P3(|v − w|)(v − w) + η11D1(v), (2.2a)

w∗ = w. (2.2b)

Again, γk ∈ (0, 12 ) (k = 1, 2, 3) are constant compromise parameters, which control

the ‘speed’ of attraction of two different opinions. This assumption can be furtherrelaxed by choosing the compromise parameters as random quantities, each with acertain statistical mean. In the following, we assume for simplicity that all individ-uals in the society share a common compromise parameter γ. The quantities ηij are


random variables with distribution Θ with variance σ2ij and zero mean, assuming

values on a set B ⊂ R.The functions Pi(·) (i = 1, 2, 3) and Dj(·) (j = 1, 2) model the local relevance of

compromise and self-thinking for a given opinion, respectively. The random variableand the function Dj(·) are characteristic for the particular class of individuals, andare the same in both types of interaction while the compromise function Pi(·) canbe different in the three types of interactions.

The first term on the right hand sides of (2.2a), (2.1a), (2.1b) models the com-promise process, the second the self-thinking process. Opinion leaders retain theiropinion in (2.2a), when interacting with ordinary people, which reflects their highself-confidence and ability to withstand other opinions. In our model, they canonly be influenced through their peers, by interactions in (2.1). The pre-interactionopinion v increases (gets closer to w) when v < w and decreases in the oppositesituation. We assume that the ability to find a compromise is linked to the dis-tance between opinions. The higher this distance is, the lower the possibility tofind a compromise. Hence, functions Pi(·) are assumed to be decreasing functionsof their argument. We also assume that the ability to change individual opinionsby self-thinking decreases as one gets closer to the extremal opinions. This reflectsthe fact that extremal opinions are more difficult to change. Therefore, we assumethat functions Dj(·) are decreasing functions of v2 with Dj(1) = 0. In (2.2), (2.1)we will only allow interactions that guarantee v∗, w∗ ∈ I. To this end, we assumeadditionally,

0 ≤ Pi(|v − w|) ≤ 1, 0 ≤ Dj(v) ≤ 1.

We now need to choose the set B, i.e. we have to specify the range of values therandom variables can assume. Clearly, it depends on the particular choice for Dj(·).Let us consider the upper bound at w = 1 first. To ensure that individuals’ opinionsdo not leave I, we need

v∗ = v − γiPi(|v − w|)(v − w) + ηikDj(v) ≤ 1

Obviously, the worst case is w = 1, where we have to ensure

ηikDj(v) ≤ 1 − v + γi(v − 1) = (1 − v)(1 − γi)

Hence, if Dj(v)/(1 − v) ≤ K+ it suffices to have |ηik| ≤ (1 − γi)/K+. A similarcomputation for the lower boundary shows that if Dj(v)/(1 + v) ≤ K− it sufficesto have |ηik| ≤ (1 − γi)/K−.

In this setting, we are led to study the evolution of the distribution functionfor each group as a function depending on the opinion w ∈ I and time t ∈ R+,fi = fi(w, t). In analogy with the classical kinetic theory of mixtures of rarefiedgases, the time-evolution of the distributions will obey a system of two Boltzmann-like equations, given by

∂

∂tf1(w, t) =

1

τ11Q11(f1, f1)(w) +

1

τ12Q12(f1, f2)(w), (2.3)

∂

∂tf2(w, t) =

1

τ22Q22(f2, f2)(w). (2.4)

Herein, τij are suitable relaxation times which allow to control the interactionfrequencies of opinion leaders and followers. The Boltzmann-like collision operators


are derived by standard methods of kinetic theory, considering that the change intime of fi(w, t) due to binary interaction depends on a balance between the gainand loss of individuals with opinion w. The operators Q11 and Q22 relate to themicroscopic interaction (2.1), whereas Q12 relates to (2.2).

Let 〈·〉 denote the operation of mean with respect to the random quantitiesηij . A useful way of writing the collision operators is the so-called weak form. Itcorresponds to consider, for all smooth functions φ(w),

∫

I

Qij(fi, fj)(w)φ(w) dw

=1

2

⟨∫

I2

(

φ(w∗) + φ(v∗) − φ(w) − φ(v))

fi(v)fj(w) dv dw

⟩

.

(2.5)

3. Fokker-Planck limit system

In general, it is rather difficult to describe analytically the behaviour of the evolutionof the densities. As is usual in kinetic theory, it is convenient to study certainasymptotics, which frequently lead to simplified models of Fokker-Planck type.By means of this approach it is easier to identify steady states while retainingimportant information on the microscopic interaction at a macroscopic level. To thisend, we study by formal asymptotics the quasi-invariant opinion limit (γ, σij → 0and σ2

ij/γ = λij), following the path laid out in Toscani (2006). Let us introducesome notation, analogous to Toscani (2006). First, restrict the test-functions φ toC2,δ([−1, 1]) for some δ > 0. We use the usual Holder norms

‖φ‖δ =∑

|α|≤2

‖Dαφ‖C +∑

α=2

[Dαφ]C0,δ ,

where

[h]C0,δ = supv 6=w

|h(v) − h(w)||v − w|δ . (3.1)

Denoting by M0(A), A ⊂ R the space of probability measures on A, we define

Mp(A) =

{

Θ ∈ M0

∣

∣

∣

∣

∫

A

|η|pdΘ(η) < ∞, p ≥ 0

}

, (3.2)

the space of measures with finite pth momentum. In the following all our probabilitydensities belong to M2+δ and we assume that the density Θ is obtained from arandom variable Y with zero mean and unit variance. We then obtain

∫

I

|η|pΘ(η) dη = E[|σY |p] = σpE[|Y |p], (3.3)

where E[|Y |p] is finite. The weak form of (2.3) is given by

d

dt

∫

I

f1(w, t)φ(w) dw =

∫

I

1

τ11Q11(f1, f1)(w)φ(w) dw

+

∫

I

1

τ12Q12(f1, f2)(w)φ(w) dw, (3.4a)


d

dt

∫

I

f2(w, t)φ(w) dw =

∫

I

1

τ22Q22(f2, f2)(w)φ(w) dw, (3.4b)

where the terms on right hand sides are given by (2.5). To study the situation forlarge times, i.e. close to the steady state, we introduce for γ ≪ 1 the transformation

τ = γt, gi(w, τ) = fi(w, t), i = 1, 2.

This implies fi,0 = gi,0 and the evolution of the scaled densities gi(w, τ) follows

d

dτ

∫

I

g1(w, τ)φ(w) dw =1

γ

∫

I

1

τ11Q11(f1, f1)(w)φ(w) dw

+1

γ

∫

I

1

τ12Q12(f1, f2)(w)φ(w) dw, (3.5a)

d

dτ

∫

I

g2(w, τ)φ(w) dw =1

γ

∫

I

1

τ22Q22(f2, f2)(w)φ(w) dw. (3.5b)

Consider the first term on the right hand side of (3.5a). Due to the collision rule(2.1), it holds

w∗ − w = −γP1(|w − v|)(w − v) + η11D1(w).

Taylor expansion of φ up to second order around w in the first term of the righthand side of (3.5a) leads to

⟨ 1

γτ11

∫

I2

φ′(w) [−γP1(|w − v|)(w − v) + η11D1(w)] g1(w)g1(v) dvdw⟩

+⟨ 1

2γτ11

∫

I2

φ′′(w) [−γP1(|w − v|)(w − v) + η11D1(w)]2g1(w)g1(v) dvdw

⟩

=1

γτ11

∫

I2

φ′(w) [−γP1(|w − v|)(w − v))] g1(w)g1(v) dvdw

+⟨ 1

2γτ11

∫

I2

φ′′(w)[

γP1(|w − v|)(w − v) + η11D1(w)]2

g1(w)g1(v) dvdw⟩

+ R(γ, σ11)

= − 1

τ11

∫

I

φ′(w)K1(w)g1(w) dw

+1

2γτ11

∫

I2

φ′′(w)[

γ2P 21 (|w − v|)(w − v)2 + γλ11D

21(w)

]

g1(w)g1(v) dvdw

+ R(γ, σ11),

with w = κw∗ + (1 − κ)w for some κ ∈ [0, 1] and

R(γ, σ11) =⟨ 1

2γτ11

∫

I2

(φ′′(w) − φ′′(w))×

× [−γP1(|w − v|)(w − v) + η11D1(w)]2g1(w)g1(v) dvdw

⟩

.

Here, we defined

Ki(w, τ) =

∫

I

Pi(|w − v|)(w − v)gi(v, τ) dv, for i = 1, 2, (3.6)


K3(w, τ) =

∫

I

P3(|w − v|)(w − v)g2(v, τ) dv. (3.7)

Now we consider the formal limit γ, σ11 → 0 while keeping λ11 = σ211/γ fixed. We

will later argue that the remainder vanishes is this limit. Then, the first term onthe right hand side of (3.5a) converges to

− 1

τ11

∫

I

φ′(w)K1(w)g1(w) dw +1

2τ11

∫

I2

φ′′(w)[

λ11D21(w)

]

g1(w)g1(v) dvdw

= − 1

τ11

∫

I

φ′(w)K1(w)g1(w) dw +λ11M1

2τ11

∫

I

φ′′(w)D21(w)g1(w) dw.

For the second term on the right hand side of (3.5a) we obtain in the same way

− 1

2τ12

∫

I

φ′(w)K2(w)g1(w) dw +λ12M2

4τ12

∫

I

φ′′(w)D21(w)g1(w) dw.

Performing a similar analysis for the right hand side of (3.5b) we obtain afterintegration by parts the following system of Fokker-Planck equations

∂

∂τg1(w, τ) =

∂

∂w

((

1

τ11K1(w, τ) +

1

2τ12K3(w, τ)

)

g1(w, τ)

)

+

(

λ11M1

2τ11+

λ12M2

4τ12

)

∂2

∂w2

(

D21(w)g1(w, τ)

)

,

(3.8a)

∂

∂τg2(w, τ) =

∂

∂w

(

1

τ22K2(w, τ)g2(w, τ)

)

+λ22M2

2τ22

∂2

∂w2

(

D22(w)g2(w, τ)

)

. (3.8b)

subject to the following no flux boundary conditions (which result from the inte-gration by parts)

(

1

τ11K1(w, τ) +

1

2τ12K3(w, τ)

)

g1(w, τ)

+

(

λ11M1

2τ11+

λ12M2

4τ12

)

∂

∂w

(

D21(w)g1(w, τ)

)

= 0

on w = ±1, (3.8c)

1

τ22K2(w, τ)g2(w, τ) +

λ22M2

2τ22

∂

∂w

(

D2(w)2g2(w, τ))

= 0 on w = ±1 (3.8d)

and

φ′(w)D1(w)2g1(w) = φ′(w)D2(w)2g2(w) = 0 on w = ±1. (3.8e)

Note however that - if the solutions g1 and g2 are sufficiently regular - the thirdcondition holds automatically since D1(w) = D2(w) = 0 for w = ±1. What is left isto show that the remainder terms R(γ, σij) vanish in the above limit. We consideronly R(γ, σ11) as the argument is similar for all the remainder terms occurring inthe limit process of (3.5b). Note first that as φ ∈ F2+δ, by (2.1) and the definitionof w we have

|φ′′(w) − φ′′(w)| ≤ ‖φ′′‖δ|w − w|δ ≤ ‖φ′′‖δ|w∗ − w|δ


= ‖φ′′‖δ |γP1(|w − v|)(w − v) + η11D1(w)|δ .

Thus we obtain

R(γ, σ11) ≤‖φ′′‖δ

2γτ11

⟨

∫

I2

[−γP1(|w − v|)(w − v) + η11D1(w)]2+δ

g1(w)g1(v) dvdw⟩

.

Furthermore, we note that

[η11D1(w) − γP1(|w − v|)(w − v)]2+δ

≤ 21+δ(

|γP1(|w − v|)(w − v)|2+δ + |η11D1(w)|2+δ)

≤ 23+2δ|γ|2+δ + 21+δ|η11|2+δ.

Here, we used the convexity of f(s) := |s|2+δ and the fact that w, v ∈ I and thusbounded. We conclude

|R(γ, σ11)| ≤C‖φ′′‖δ

τ11

(

γ1+δ +1

2γ

⟨

|η11|2+δ⟩

)

=C‖φ′′‖δ

τ11

(

γ1+δ +1

2γ

∫

I

|η11|2+δΘ(η11) dη11

)

.

Since Θ ∈ M2+δ, and η11 has variance σ211 we have (see (3.3))

∫

I

|η11|2+δΘ(η11) dη11 = E

[

∣

∣

∣

√

λ11γY∣

∣

∣

2+δ]

= (λ11γ)1+δ2 E

[

|Y |2+δ]

. (3.9)

Thus, we conclude that the terms R(γ, σij) vanish in the limit γ, σij → 0 whilekeeping λij = σ2

ij/γ fixed.

(a) Stationary solutions of the Fokker-Planck system

Next we would like to analyze explicitly computable stationary state of theFokker-Planck system. We consider the special case Pi(|w − v|) ≡ 1 (i = 1, 2, 3),which implies conservation of the average opinion and the first momentum for(3.8b). For the sake of simplicity we choose

D1(w) = D2(w) = D(w) :=(

1 − w2)α

, (3.10)

with α > 12 as a model for the diffusion, which is consistent with the requirement

that post-collisional opinions have to be in I, at least when the ranges of therandom variables ηij are sufficiently small. This function has been introduced inToscani (2006) and includes that extremal opinions are more difficult to change thanmoderate ones. The choice of α is directly related to the regions where diffusion ofopinions is prevalent.

In the interest of readability the parameters τij , λij , and Mi are chosen in sucha way that all coefficients in (3.8) equal one. Then the steady state of (3.8) is asolution of

0 = (K1 + K3) g1(w) +(

D2(w)g2(w))

w, (3.11a)

0 = K2g2 +(

D2(w)g2

)

w. (3.11b)


We denote the masses of the opinion leaders and followers by ‘ =∫

gi dv withi = 1, 2 and their first order moments by mi =

∫

vgi dv, i = 1, 2.Equation (3.11b) can be written as

−wM2 − m2

D(w)2f2 =

d

dwf2,

with f2 = D(w)2g2. Therefore,

f2 = c2e−

∫ w

0vM2−m2

(1−v2)2α dvand hence g2 =

c2

(1 − w2)2αe−

∫ w

0vM2−m2

(1−v2)2α dv, (3.12)

where c2 is chosen such that the mass of g2 is equal to M2. Note that since |m2| < M2

and α > 12 , g2(±1) = 0. The solution of (3.11a) can be calculated using the same

arguments

g1 =c1

(1 − w2)2αe−

∫ w

0v(M1+M2)−(m1+m2)

(1−v2)2α dv. (3.13)

The mass and moment of (3.11b) are conserved. Integration of (3.11a) leads to

(M1 + M2)m1 − (m1 + m2)M1 = 0.

Therefore, we can fix m1 and rewrite (3.13) as

g1 =c1

(1 − w2)2αe−(M1+M2)

∫ w

0v

(1−v2)2α dvem2

“

1+M1M2

” ∫ w

01

(1−v2)2α dv.

¿From |m2| < M2 we conclude that if α > 12 then c1 can be determined such that

the mass of g1 equals M1.

(b) Emergence and decline of opinion leaders

Opinion leadership is not constant over time. Someone who is an opinion leadertoday may loose this role in the near future. Hence, the emergence and decline ofopinion leaders is a common process in a society, which we would like to includein the limiting Fokker-Planck system (3.8). The first assumption is that the overallmass of opinion leaders and followers remains constant in time, i.e.

d

dτ

∫

(g1(w, τ) + g2(w, τ)) dw = 0.

To model the emergence and decline of opinion leaders we assume that there isalways a certain percentage of leaders in a community, i.e. 5%. Leaders can emergeif the density of followers having the same opinion is greater than a certain thresholdand the overall mass of leaders is smaller than i.e. 5%. On the contrary leaders maydecline if the number of followers having sharing the opinion is below a certainthreshold.

Let µ ∈ (0, 1) denote the fraction of followers in the whole population, 1 − µaccounts for the number of opinion leaders. Then the extended model has the form


∂

∂τg1(w, τ) =

∂

∂w

((

1

τ11K1(w, τ) +

1

2τ12K3(w, τ)

)

g1(w, τ)

)

+

(

λ11M1

2τ11+

λ12M2

4τ12

)

∂2

∂w2

(

D2(w)g1(w, τ))

− a(g1)g1 + b(g1)g2,

(3.14a)

∂

∂τg2(w, τ) =

∂

∂w

(

1

τ22K2(w, τ)g2(w, τ)

)

+λ22M2

2τ22

∂2

∂w2

(

D2(w)g2(w, τ))

+ a(g1)g1 − b(g1)g2.

(3.14b)

The functions a and b are given by

a(g1) = 1{g1(w)≥c}e−

m21

√

2πσ , b(g1) = 1{g1(w)≤c}e−

(m1−µ)2√

2πσ ,

where 1A is the indicator function of the set A. If the density of normal people isgreater than certain threshold c ∈ R

+, the functions a models the rate of emergenceof opinion leaders. The function b behaves the other way around, if the density ofnormal people is below a certain threshold c opinion leaders decline. With thissimple extension of our model we try to get a first insight in the emergence anddecline of strong leaders.

4. Numerical simulations

In this section we illustrate the behaviour of the kinetic model and the limitingFokker-Planck system with various simulations. We assume that the diffusion ofopinion is given by (3.10) and the compromise propensity Pi(·) (i = 1, 2, 3) by

Pi(|v − w|) = 1{|v−w|≤ri}. (4.1)

The following parameters are fixed throughout this section, if not mentioned oth-erwise:

• Relaxation times: τ11 = τ12 = τ22 = 1;

• Ratio of normal people to opinion leaders: M1 = 0.95 and M2 = 0.05;

• Diffusion parameters: λ11 = λ12 = λ22 = λ := 5 × 10−3;

• Exponent of the diffusion function in (3.10): α = 2.

The initial distribution of normal people is given by a Gaussian

g1(w, 0) =1√

2πσ1

e−(w−σ1)2

2 (4.2)

with σ1 = 0.4. The initial distribution of opinion leaders is

g2(w, 0) =n

∑

i=1

qi√2πσi

e−(w−σi)

2

2 (4.3)

with weights∑n

i=1 qi = 1.


(a) Monte Carlo simulations

To illustrate the relaxation behaviour and to study the influence of the differentmodel parameters, we have performed a series of kinetic Monte Carlo simulationsfor the Boltzmann model presented in the previous section. Generally, in this kindof simulations, known as direct simulation Monte Carlo (DSMC) or Bird’s scheme,pairs of individuals are randomly and non-exclusively selected for binary collisions,and exchange opinion according to the rule under consideration. Let us denote byNi (i = 1, 2) the number of individuals in the groups we consider in our simulation.One time step in our simulation corresponds to N1 + N2 interactions. The averageof M = 10 simulations is used as an approximate steady opinion distribution. Tocompute a good approximation of the steady state, each simulation is carried out forabout 106 time steps, and then the opinion distribution is averaged over another 103

time steps. We choose N1 = 1900, N2 = 100, and γ = 0.02. The random variablesare chosen such that ηij assume only values ±ν = ±0.01 with equal probability.The initial distributions are chosen as discrete analogues of (4.2) and (4.3).

(b) Numerical solution of the Fokker-Planck system

To illustrate the long-time behaviour of the proposed model we discretize the non-linear Fokker-Planck system (3.8) using a hybrid discontinuous Galerkin (DG)method introduced by Egger and Schoberl (2008). This hybrid DG method wasinitially developed for convection diffusion equations and yields stable discretiza-tions for convection dominated problems as well as hyperbolic ones. In addition,the method is conservative, which is consistent with the assumption that the initialmass of the Fokker-Planck system is preserved in time.

We choose a partition of the time interval [0, T ], 0 = t0 < t1 < . . . < tj < . . . <tm = T , and define ∆tj = tj+1 − tj . We consider the following linearization of theFokker-Planck equations (3.8), which fits into the framework of Egger & Schoberl(2008),

gj+11 − gj

1

∆tj=

∂

∂w

((

1

τ11K1(g

j1;w, t) +

1

τ12K3(g

j2;w, t)

)

gj+11 (w, t)

)

+

(

λ11M1

2τ22+

λ12M2

4τ12

)

∂2

∂w2

(

D2(w)gj+11 (w, t)

)

,

(4.4a)

gj+12 − gj

2

∆tj=

∂

∂w

(

1

τ22K2(g

j2;w, t)gj+1

2 (w, t)

)

+λ22M2

2τ22

∂2

∂w2

(

D2(w)gj+12 (w, t)

)

.

(4.4b)

Here gji , i = 1, 2, denotes the solution at time t = tj . We choose an equidistant

mesh of mesh size h = 1400 to discretize the interval [−1, 1]. The time steps ∆tj are

set to 0.01.

(c) Numerical results

(i) Influence of interaction radii ri and distribution of opinion leaders

We choose a symmetric initial distribution of opinion leaders with wi = ±0.5, qi =0.5, and σi = 0.05, i = 1, 2. The interaction radii take the same value, ri = 0.5, for


i = 1, . . . , 3. The behaviour of both species is illustrated in figure 1. For the followersthe results of the numerical solution of the Fokker-Planck system and the MonteCarlo simulation agree well. For the opinion leaders, the peaked, high densitieswhich are obtained from the numerical solution of the Fokker-Planck system cannotbe as well resolved by the Monte Carlo simulation. The number of leaders is fixedto be 5% of the total population and hence the number of realizations is rathersmall. Increasing M and N1, N2 will lead to a better resolution but will render themethod to be computationally infeasible. Therefore, in our more involved exampleswe will rely on the numerical solution of the Fokker-Planck system.If we change the interaction radius to ri = 0.3, i = 1, . . . , 3, the formation of asmall group centered at w = 0, which is not attracted by the opinion leader, canbe observed (see figure 2). If r1 = 0.6 and r2 = r3 = 0.3, we observe that theinteraction of the normal people with each other dominate the opinion formationprocess and results in an aggregation at w = 0 (see figure 3).

Next we illustrate the opinion formation process with a non-symmetric initialdistribution of opinion leaders. We choose w1 = −0.7 and w2 = 0.5 with interactionradii ri = 0.5 for i = 1, 2, 3. The behaviour is illustrated in figure 4.

(ii) Understanding Carinthia

In our next example we would like to illustrate the behaviour of our model foropinion formation under extreme conditions, like in Carinthia. Carinthia is thesouthernmost state of Austria. Carinthia’s landscape of political parties shows aninteresting peculiarity. In 1999 the right-wing Freedom Party of Austria (FPO)became the strongest party in Carinthia. Since then their results in elections forthe state assembly (Landtagswahlen) continually improved, holding almost 45 % ofthe votes in 2008. This outcome was strongly influenced by the popularity of theirparty leader Jorg Haider. Haider, a controversial figure, was frequently criticized inAustria and abroad, being considered populistic, extreme-right or even antisemitic.On the other hand he was strongly acclaimed by his followers. Haider had beenelected Carinthian governor in 1989 but was forced to step down two years laterafter his remarks about a ‘proper employment policy’ in the Third Reich. He waselected again as Carinthian governor in 1999 and re-elected in 2004. Haider, whopractically lead the FPO single-handed, was able to unite the political spectrumfrom conservatives to extreme-right and establish a governing party whose successwas less founded on political ideologies than rather on the authority of one opinionleader — Haider himself.

Table 1 shows the results of the state elections in 2004 and 2009, respectively.We set the initial distribution of normal people to

g1(w, 0) =0.07

σ1

√2π

e−

(w+0.75)2

2σ21 +

0.385

σ1

√2π

e−

(w+0.25)2

2σ21

+0.115

σ1

√2π

e−

(w−0.25)2

(2σ21) +

0.45

σ1

√2π

e−

(w−0.8)2

2σ21 ,

(4.5)

where the weights of the Gaussian distributions are chosen in accordance to theresults of the Landtagswahlen in 2004 (see table 1). Here, w = −0.75 correspondsto the Greens (Grune), w = −0.25 to the Social Democratic Party of Austria


time

opinion

-1 0 1 2 3 4 5 6 7

0

200

400

600

800

1000

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

(a) Evolution of normal people

time

opinion

-2 0 2 4 6 8

10 12 14 16 18 20

0

200

400

600

800

1000

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

(b) Evolution of opinion leaders

-1

0

1

2

3

4

5

6

7

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

opinion

(c) Stationary solution: normal people

-2

0

2

4

6

8

10

12

14

16

18

20

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

opinion

(d) Stationary solution: opinion leaders

−1 −0.5 0 0.5 10

2

4

6

8

opinion

(e) Stationary solution: normal people

−1 −0.5 0 0.5 10

1

2

3

4

5

6

opinion

(f) Stationary solution: opinion leaders

Figure 1. Numerical solutions of the Fokker-Planck system ((a)-(d)) and densityhistograms of the Monte Carlo simulation ((e)-(f)) with r1 = r2 = r3 = 0.5

(SPO), w = 0.25 to the Austrian People’s Party (OVP) and w = 0.8 to the FPO.We assume that there are several opinion leaders present in the system associatedwith the different parties, but with different weights representing their influence.The initial distribution of opinion leaders is given by

g2(w, 0) =0.1

σ2

√2π

e−

(w+0.75)2

2σ22 +

0.15

σ2

√2π

e−

(w+0.2)2

2σ22

+0.3

σ2

√2π

e−

(w−0.25)2

2σ22 +

0.45

σ2

√2π

e−

(w−0.8)2

2σ22 .

(4.6)


time

opinion

0 1 2 3 4 5 6 7

0

500

1000

1500

2000

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1


time

opinion

-2 0 2 4 6 8

10 12 14 16 18 20

0

500

1000

1500

2000

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1


0

1

2

3

4

5

6

7

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

opinion


-2

0

2

4

6

8

10

12

14

16

18

20

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

opinion


−1 −0.5 0 0.5 10

2

4

6

8

opinion


−1 −0.5 0 0.5 10

1

2

3

4

5

opinion


Figure 2. Numerical solutions of the Fokker-Planck system ((a)-(d)) and densityhistograms of the Monte Carlo simulation ((e)-(f)) with r1 = r2 = r3 = 0.3

We choose the following parameters

α = 1.5, λ = 3 × 10−3, r1 = r2 = 0.2, r3 = 0.45,

τ11 = τ12 = 1, τ22 = 10, σ1 = 0.1, σ2 = 0.05.

The behaviour of the solution is depicted in figure 5. We observe that in presence ofthe stronger OVP leader, people move from the SPO to the OVP, while the peoplewith an extreme opinion accumulate around the strong leader at w = 0.8. Note thata small group of people splits from the initial density at w = 0.8 (initially attractedby the strong leader at w = 0.5) and form a new group at w = 0.7. This formation


time

opinion

0 1 2 3 4 5 6 7 8

0

200

400

600

800

1000

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1


time

opinion

0 2 4 6 8

10 12 14 16 18 20

0

200

400

600

800

1000

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1


0

1

2

3

4

5

6

7

8

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

opinion


0

2

4

6

8

10

12

14

16

18

20

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

opinion


−1 −0.5 0 0.5 10

2

4

6

8

opinion


−1 −0.5 0 0.5 10

1

2

3

4

5

6

opinion


Figure 3. Numerical solutions of the Fokker-Planck system ((a)-(d)) and densityhistograms of the Monte Carlo simulation ((e)-(f)) with r1 = 0.6 and r2 = r3 = 0.3

can be interpreted as the separation of two parties associated with the formationof a new opinion leader. This is an interesting similarity with the real situation inCarinthia. In April 2005 Haider formed a new party, the Alliance for the Future ofAustria (BZO), with himself as leader, thereby de facto splitting the FPO into twoparties. Haider died in a car crash in October 2008. In the elections in March 2009the BZO, strongly referring to its deceased leader, managed to enlarge its share ofvotes to 44.9 %, while the FPO failed to enter the Landtag.


time

opinion

-1 0 1 2 3 4 5 6 7 8 9

10

0

200

400

600

800

1000

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1


time

opinion

-5 0 5

10 15 20 25 30 35 40

0

200

400

600

800

1000

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1


-1

0

1

2

3

4

5

6

7

8

9

10

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

opinion


-5

0

5

10

15

20

25

30

35

40

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

opinion


−1 −0.5 0 0.5 10

2

4

6

8

10

opinion


−1 −0.5 0 0.5 10

2

4

6

8

10

opinion


Figure 4. Numerical solutions of the Fokker-Planck system ((a)-(d)) and density his-tograms of the Monte Carlo simulation ((e)-(f)) with non-symmetric initial data for theopinion leaders and with r1 = r2 = r3 = 0.5

(iii) Emergence and decline of opinion leaders

In our final example we would like to show the emergence and decline of opinionleaders. We choose (4.2) as initial guess and set g2(w, 0) = 0. The equilibrium ratiois chosen as µ = 0.95. The interaction radii are ri = 0.3 for i = 1, 2, 3, the upperand lower threshold are given by

c = 1.0 and c = 10−3.


Table 1. Results of the state elections in Carinthia

Grune SPO OVP FPO BZO

2004 6.7% 38.4 % 11.6 % 42.5 % —

2009 5.2% 28.8 % 16.8 % 3.8 % 44.9 %

time

opinion

-2 0 2 4 6 8

10 12 14 16 18

0 200

400 600

800 1000

1200

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1


time

opinion

-5 0 5

10 15 20 25 30 35 40

0 200

400 600

800 1000

1200

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1


-2

0

2

4

6

8

10

12

14

16

18

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

opinion


-5

0

5

10

15

20

25

30

35

40

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

opinion


Figure 5. Opinion formation in Carinthia

The evolution of the normal people and the emergence of an opinion leader at w = 0is illustrated in figure 6. Note that no opinion leaders emerge at w = ±0.65 becausethe density of normal people does not exceed the threshold.

5. Conclusions

We introduced and discussed a nonlinear kinetic model for a society which is builtof two social groups, a group of strong opinion leaders and a group of ordinarypeople. The evolution of opinion is described by a system of Boltzmann-like equa-tions in which collisions describe binary exchanges of opinion and self-thinking. Weshowed that at suitably large times, in presence of a large number of interactionsin each of which individuals change their opinions only little, the nonlinear systemof Boltzmann-type equations is well-approximated by a system of Fokker-Plancktype equations, which admits different, non-trivial steady states which depend on


time

opinion

0 0.5

1 1.5

2 2.5

3 3.5

4

0

200

400

600

800

1000

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1


time

opinion

-0.001 0

0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009

0

200

400

600

800

1000

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1


0

0.5

1

1.5

2

2.5

3

3.5

4

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

opinion


-0.001

0

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

0.009

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

opinion


Figure 6. Emergence of opinion leaders

the specific choice of the compromise and self-thinking functions and parameters.We extended this model by allowing for emergence and decline of opinion leaders.

This publication is based on work supported by Award No. KUK-I1-007-43, made by KingAbdullah University of Science and Technology (KAUST) and by the Leverhulme Trustthrough the Research Grant entitled KINETIC AND MEAN FIELD PARTIAL DIFFER-ENTIAL MODELS FOR SOCIO- ECONOMIC PROCESSES (PI Peter Markowich). P.Markowich also acknowledges support from the Royal Society through his Wolfson Re-search Merit Award. Bertram During is partly supported by the Deutsche Forschungs-gemeinschaft (DFG), grant JU 359/6 (Forschergruppe 518). Bertram During thanks theDepartment of Applied Mathematics and Theoretical Physics of the University of Cam-bridge, where a part of this research has been carried out, for the kind hospitality.

References

Bertotti, M.L. & Delitala, M. 2008 On a discrete generalized kinetic approachfor modeling persuadors influence in opinion formation processes. Math. Comp.

Model. 48, 1107-1121.

Bobylev, A.V. & Gamba, I.M. 2006 Boltzmann equations for mixtures of Maxwellgases: exact solutions and power like tails, J. Stat. Phys. 124, 497-516.

Burt, R.S. 1999 The social capital of opinion leaders. Annals Am. Acad. Pol. Social

Science 566, 37-54.


Cercignani, C., Illner, R. & Pulvirenti, M. 1994 The mathematical theory of dilute

gases, Springer Series in Applied Mathematical Sciences, Vol. 106, Springer.

Comincioli, V., Della Croce, L. & G. Toscani 2009 A Boltzmann-like equation forchoice formation, Kin. Rel. Models 2(1), 135-149.

Deffuant, G., Amblard, F., Weisbuch, G. & Faure, T. 2002 How can extremismprevail? A study based on the relative agreement interaction model, JASSS J.

Art. Soc. Soc. Sim 5(4).

During, B., Matthes, D. & Toscani, G. 2008 Kinetic equations modelling wealthredistribution: a comparison of approaches, Phys. Rev. E 78(5), 056103.

Egger, H. & Schoberl, J. 2008 A hybrid mixed discontinuous Galerkin method forconvection-diffusion problems, accepted for publication in IMA J. Numer. Anal.

Galam, S. 2005 Heterogeneous beliefs, segregation, and extremism in the makingof public opinions, Phys. Rev. E 71(4), 046123.

Galam, S. & Zucker, J.D. 2000 From individual choice to group decision-making,Physica A 287(3-4), 644-659.

Galam, S., Gefen, Y. & Shapir, Y. 1982 Sociophysics: a new approach of sociologicalcollective behavior, J. Math. Sociology 9, 1-13.

Lazarsfeld, P.F., Berelson, B.R. & Gaudet, H. 1944 The people’s choice: How thevoter makes up his mind in a presidential campaign. New York: Duell, Sloan &Pierce.

Markowich, P.A. & Poupaud, F. & Schmeiser, C. 1995 Diffusion approximation ofnonlinear electron phonon collision mechanisms Math. Mod. Num. Anal. 29(7),857-869.

Maslach, C., Stapp, J. & Santee, R.T. 1985 Individuation: Conceptual analysis andassessment. J. Personality Social Psychology 49(3), 729-738.

Toscani, G. 2006 Kinetic models of opinion formation, Commun. Math. Sci. 4(3)481-496.

Toscani, G. 2000 One-dimensional kinetic models of granular flows, Math. Mod.

Num. Anal. 34, 1277-1292.

Rash, W. 1997 Politics on the nets: wiring the political process, New York: Freeman.

Slanina, F. & Lavicka, H. 2003 Analytical results for the Sznajd model of opinionformation, Eur. Phys. J. B 35, 279-288.

Sood, V., Antal, T. & Redner, S. 2008 Voter Models on Heterogeneous Networks,Phys. Rev. E 77, 041121.

Sznajd-Weron, K. & Sznajd, J. 2000 Opinion evolution in closed community, Int.

J. Mod. Phys. C 11, 1157-1165.

Weidlich, W. 2000 Sociodynamics - A Systematic Approach to Mathematical Mod-

elling in the Social Sciences, Harwood Academic Publishers.

Boltzmann and Fokker-Planck Equations Modelling Opinion Formation … · 2010-05-07 · Boltzmann...

Documents

Transcript of Boltzmann and Fokker-Planck Equations Modelling Opinion Formation … · 2010-05-07 · Boltzmann...