Proc. of the 16th Workshop “From Object to Agents” (WOA15) June 17-19, Naples, Italy Simulations of Opinion Formation in Multi-Agent Systems using Kinetic Theory Stefania Monica, Federico Bergenti Dipartimento di Matematica e Informatica Parco Area delle Scienze 53/A, 43124 Parma, Italy Email: {stefania.monica, federico.bergenti}@unipr.it Abstract—In this paper we formulate the problem of opinion numerically the opinion evolution according to a model intro- formation using a physical metaphore. We consider a multi- duced in [9], which mandates that each agent can change its agent system where each agent is associated with an opinion own opinion because of two different reasons [10]. The first and interacts with any other agent. Interpreting the agents as reason is related to compromise between interacting agents. the molecules of a gas, we model the evolution of opinion in More precisely, the model assumes that both agents involved the system according to a kinetic model based on the analysis of interactions among agents. From a microscopic description in an interaction can change their respective opinions in favor of each interaction between two agents, we derive the stationary of that of the other agent. The second reason only involves profiles of the opinion under given assumption. Results show that, each single agent and it is related to the fact that an agent depending on the average opinion and on the parameters of the can change its opinion autonomously, giving rise to a process model, different profiles can be found, but all stationary profiles known as diffusion. are characterized by the presence just of one or two maxima. Analytic results are confirmed by simulations shown in the last This paper is organized as follows. In Section II, we part of the paper. describe the considered kinetic model from an analytic point of view. In Section III, we derive a stationary profile of opinion in a specific case. In Section IV, we show relevant simulation I. I NTRODUCTION results. Section V concludes the paper. In this paper we describe a model for opinion formation II. K INETIC F ORMULATION OF O PINION F ORMATION among agents. We assume that each agent is associated with an opinion v ∈ I ⊆ R and that it can change its opinion Kinetic theory of gases describes, from a microscopic each time it interacts with another agent. In the literature, point of view, the effects of interactions among molecules in various approaches that study opinion evolution in a society gases. By reinterpreting the molecules of a gas as agents, one of agents have been proposed, and many of them are based can use the kinetic framework to describe social interactions on Cellular Automata (CA) because CA describe well global among agents. While molecules are typically associated with effects of local phenomena (see, e.g., [1]). In order to overcome relevant physical properties, like their velocities, agents can be the synchronism that CA assume, in recent years the use of associated with attributes that represent some of their charac- microscopic models inspired from physics has been introduced teristics. In particular, since in this work we are interested in to describe asynchronous social interactions among agents in a modeling the evolution of the opinion, we assume that each society [2]. Such models are based on the idea that the laws of agent is associated with a scalar parameter v which represents kinetic theory, which are typically used to describe the effects its opinion and which is defined in a given interval I. In of interactions between two molecules of a gas, can also be the following, we consider I = [−1, 1], where ±1 represent used to model interactions between two agents. extremal opinions. Statistical mechanics and kinetic theory describe the details In order to use the kinetic approach, we need to define of each interaction between two molecules in a gas but they a function, denoted as f (w, t), which represents the density also allow finding classic laws which describe macroscopic of opinion w at time t, and which is defined for each opinion properties of gases [3]. Analogously, from the microscopic w ∈ I and for each time t ≥ 0. According to such a definition, Z laws which describe the details of each interaction between two agents, collective behaviour can be described from a macro- f (w, t)dw = 1. (1) I scopic point of view [4]. All the research challenges related to the application of kinetic and statistical formalisms to describe In order to formulate the problem of opinion evolution in multi-agent systems gave rise to new disciplines known as kinetic terms, we assume that the function f (w, t) evolves econophysics and sociophysics [5]. Such new disciplines have on the basis of the Boltzmann equation, which, under our been used to describe, e.g., wealth evolution [6] and market assumptions, can be written as economy [7], and they have also been adopted to characterize ∂f opinion evolution in a society [8]. = Q(f, f )(w, t) (2) ∂t In this paper, we focus on a model for opinion formation where Q is denoted as collisional operator. According to (2), based on kinetic theory of gases and we analyze the evolution the temporal evolution of the opinion density is governed by of the opinion in a society of agents. In particular, we analyze the collisional operator Q, whose explicit formulation depends 97 Proc. of the 16th Workshop “From Object to Agents” (WOA15) June 17-19, Naples, Italy on the details of binary interactions between any pairs of expression of the collisional operator Q used in (2), which is agents. Before deriving a formula for Q, let us describe the given by effects of each binary interaction. Z Z 0 1 0  Q(f, f ) = W f ( w)f (0 v) − W f (w)f (v) dvdηdη∗ Denoting as (w, v) the opinions of two agents before their B2 I J interaction, we assume that the following model holds  0 where B is the support of ϑ, 0 w and 0 v are the pre-interaction w = w + γ(v − w) + ηD(|w|) variables which generate w and v, respectively, 0 W is the (3) transition rate and J is the Jacobian of the transformation of v 0 = v + γ(w − v) + η∗ D(|v|) (0 w,0 v) in (w, v). The two addends in the previous equation where: (w0 , v 0 ) are the post-interaction opinions of the two represent the gain and the loss of agents in dw, respectively [9]. agents; γ is a constant defined in (0, 12 ); η and η∗ are two independent random variables with the same statistics; and In order to study the opinion evolution, we need to in- D(·) is a function that describes the impact of diffusion in the troduce the weak form of the Boltzmann equation. Generally considered interaction [9]. From (3) it can be observed that speaking, the weak form of a differential equation is obtained the post-interaction opinions of the two agents are obtained by multiplying both sides by a test function, namely a smooth by adding to their pre-interaction opinions two terms: the first function with compact support, and integrating. The weak form one is related to compromise, while the second one is related of the Boltzmann equation can then be found by multiplying to diffusion through function D(·). both sides of (2) by a test function φ(w) and integrating with respect to w. Using a proper change of variable in the integral, Observe that the contribution of the compromise is pro- the weak form of the Boltzmann equation can be written as: portional to the difference between the two pre-interaction Z opinions. Taking, for instance, the first equation in (3), we can d f (w, t)φ(w)dw = conclude that the second addend on the right hand side of (3) is dt I Z Z (6) positive if v > w, so that the opinion of the considered agent W f (w)f (v)(φ(w0 ) − φ(w))dwdvdηdη∗ . (whose pre-interaction opinion is w) increases if it interacts B2 I2 with an agent with greater opinion. At the opposite, if w > v the second addend is negative, so that the contribution of Setting φ(w) = 1 in (6) leads to compromise decreases the opinion of the considered agent Z (towards that of the agent it interacts with). Observe that the d f (w, t)dw = 0. (7) contribution of compromise is negligible if γ ' 0, while it dt I becomes relevant as γ increases. Such an equality corresponds to the fact that the number of Concerning the diffusion term, we assume that function agents is time invariant. This property is also found in classic D(·) depends on the absolute value of the opinion, meaning kinetic theory and it corresponds to mass conservation. that the propensity of changing opinion is symmetrical with Considering φ(w) = w as a test function and using (3) in respect to 0 (namely, with respect to the middle point of I). (6) gives Moreover, we assume that D(·) is non increasing with respect Z to the absolute value of the opinion, coherently with the fact d f (w, t)wdw = that, typically, extremal opinions are more difficult to change. dt I Z Z Finally, we assume that 0 ≤ D(|w|) ≤ 1 for all w ∈ I. According to such assumptions, the contribution of diffusion γ W f (w)f (v)(v − w)dwdvdηdη∗ (8) can be either positive or negative depending on the value of ZB2 ZI 2 η and η∗ . In the following, we denote the probability density + W f (w)f (v)ηD(|w|)dwdvdηdη∗ . function of η and η∗ as ϑ(·) and we assume that B2 I2 Z Z Denoting as u(t) the average value of the opinion at time t, ηϑ(η)dη = η∗ ϑ(η∗ )dη∗ = 0 namely Z Z Z (4) u(t) = f (w, t)w dw (9) η 2 ϑ(η)dη = η∗2 ϑ(η∗ )dη∗ = σ 2 . I the left hand side of (8) corresponds to the derivative u̇(t) Such a choice corresponds to considering η and η∗ as 0 mean of the average opinion. Moreover, observe that the right hand random variable with standard deviation σ. side term of (8) is 0. As a matter of fact, the first integral The effects of diffusion in the opinion evolution are taken is 0 for symmetry reasons, while the second integral is 0 into account through the transition rate, which is defined as because, according to (4), the average value of ϑ is 0. From (8) it can then be concluded that u̇(t) = 0, and, therefore, W (w, v, w0 , v 0 ) = ϑ(η)ϑ(η∗ )χI (w0 )χI (v 0 ) (5) the average opinion is conserved, namely u(t) = u(0) = u. This property corresponds to the conservation of momentum where χI is the indicator function of set I (equals to 1 if its in kinetic theory. argument belongs to I, and to 0 otherwise). The indicator func- tion in (5) is meant to impose that post-interaction opinions We are now interested in studying the asymptotic behaviour still belong to interval I [9]. of the distribution function f (w, t). For this reason, in order to simplify notation, let us define a new temporal variable Now that we have completed the definition of the law that describe each single interaction, we can write the explicit τ = γt (10) 98 Proc. of the 16th Workshop “From Object to Agents” (WOA15) June 17-19, Naples, Italy where γ is the coefficient which appears in (3) and it is related where C is a constant which is necessarily 0. As a matter of to compromise. Assuming that γ → 0, namely that each fact, by integrating (20) one obtains interaction causes small changes of opinions, the function Z Z v2 λ v2 ∂ 2  (1−|w|) g dw + (w −m)gdw = C(v2 +v1 ) g(w, τ ) = f (w, t) (11) 2 −v1 ∂w −v1 describes the asymptotic behaviour of f (w, t). In [9] it is from which, assuming that v1 → 1 and v2 → 1 one obtains shown that by substituting f (w, t) with g(w, τ ) in (6) and 0 + m − m = 2C using a Taylor series expansion of φ(w) around w in (6) the following equation of g can be derived which corresponds to C = 0. 2 dg λ ∂ ∂ Let us start by considering w > 0, so that (20) can be = (D(|w|)2 g) + ((w − u)g) (12) dτ 2 ∂w2 ∂w written as where λ ∂g (1 − w)2 + [(w − m) + λ(w − 1)] g = 0. (21) λ = σ 2 /γ. (13) 2 ∂w Equation (12) is known in the literature as the weak form of Dividing both sides by g one obtains the Fokker-Planck equation [11]. g0 2 2(m − w) = + (22) We are now interested in studying stationary solutions of g (1 − w) λ(1 − w)2 this equation, namely those which satisfy and observing that dg   = 0. (14) 2(m − w) d 2 2(m − 1) dτ 2 = − log(1 − w) + (23) λ(1 − w) dw λ λ(1 − w) In the following, we denote such solutions as g∞ . In next sec- equation (22) can be written as tion we analyze such solutions for different diffusion functions  0 D(|w|) and for different values of the parameter λ. 0 2 −2− λ 2(m − 1) (log g(w)) = log(1 − w) + (24) λ(1 − w) III. R ESULTS where we have used the facts that In this section we derive relevant stationary profiles for the g0 0 opinion density g. Such profiles are defined as solutions of (14) = (log g(w)) and, therefore, they depend on the parameters u and λ, which g (25) represent the average opinion and the ratio σ 2 /γ, respectively, 2 = −2(log(1 − w))0 . and on the choice of the diffusion function D. Recalling the (1 − w) initial assumptions on D, we develop our results considering Integrating (24) and applying the exponential function, one D(|w|) = 1 − |w| w∈I (15) finally obtains the following expression for the stationary profile which is symmetrical with respect to 0 and decreasing in |w|.   2 2(m − 1) With this choice of the diffusion function, equations (3) g∞ (w) = c̃u,λ (1 − |w|)−2− λ exp (26) λ(1 − |w|) become  0 w = w − γ(w − v) + η(1 − |w|) where c̃u,λ is a normalization constant that depends on the (16) average opinion u and on λ. v 0 = v − γ(v − w) + η∗ (1 − |v|). In order to guarantee that post-collisional opinions still belong Let us now consider w < 0 so that equation (20) becomes to the considered interval I, we need to set the support of ϑ λ ∂g to (1 + w)2 + [(w − m) + λ(w + 1)] g = 0. (27) 2 ∂w B = (−(1 − γ), 1 − γ). (17) Dividing both sides by g leads to As a matter of fact, from (16) g0 2 2(m − w) 0 =− + (28) |w | ≤ (1 − γ)|w| + γ|v| + |η|(1 − |w|) g (1 + w) λ(1 + w)2 (18) ≤ (1 − γ)|w| + γ + |η|(1 − |w|) and by applying analogous calculation to the case with w > 0 and if, according to (17), |η| ≤ (1 − γ) then one obtains  0 |w0 | ≤ (1 − γ)|w| + γ + (1 − γ)(1 − |w|) = 1 (19) 0 −2− λ2 2(m + 1) (log g(w)) = log(1 + w) − . (29) λ(1 + w) Hence, we can conclude that if η ∈ B, then w0 ∈ I. Analogous results can be derived for v 0 . Integrating (29) leads to the following formula for the station- ary profile Now, if we substitute the expression of D defined in (15)   in (12) the stationary solution g∞ can then be found, according 2 −2(m + 1) g∞ (w) = ĉu,λ (1 − |w|)−2− λ exp (30) to (14), by solving the following partial differential equation. λ(1 − |w|) λ ∂  where w has been substituted by −|w| and ĉu,λ is a normal- (1 − |w|)2 g + (w − m)g = C (20) ization constant. 2 ∂w 99 Proc. of the 16th Workshop “From Object to Agents” (WOA15) June 17-19, Naples, Italy Since g∞ is the solution of a differential equation it must 1.4 be continuous. From (26) and (30) it is evident that g∞ is λ=1/3 continuous for w > 0 and w < 0. Imposing that g∞ is also λ=1 continuous in w = 0, the following equality needs to hold 1.2 λ=3     2u −2u c̃u,λ exp = ĉu,λ exp . (31) 1 λ λ Finally, the solution of (20) is g ∞ (w ) 0.8   2 −2− λ 2(1 − uw) g∞ (w) = cu,λ (1 − |w|) exp − (32) 0.6 λ(1 − |w|) where cu,λ is the quantity in (31) and it needs to be determined 0.4 in order to ensure Z 0.2 g∞ (w) = 1. (33) I 0 Observe that g∞ is piecewise C 1 and it is non-differentiable in −1 −0.8 −0.6 −0.4 −0.2 w 0 0.2 0.4 0.6 0.8 1 w = 0 (as the function D). Moreover, the solution is symmetric if we change w and u with −w and −u, namely Fig. 1. Stationary profiles g∞ for u = 0 and λ = 1/3 (blue line), λ = 1 g∞ (w; u, λ) = g∞ (−w; −u, λ). (34) (red line), and λ = 3 (green line) If u = 0, from (34) we can conclude that g∞ is an even function. Moreover, using a change of variable for negative Observe that this value is negative if and only if λ > u. Finally, values of w, the integral of g∞ can be written as the following cases can be considered Z 1 Z 1   2 −2− λ −2 • if u = 0 then g∞ 0 (w) = 0 in two points that are g∞ (w)dw = 2c0,λ (1 − w) exp dw symmetric with respect to 0, namely w = ± λ+1 λ −1 0 λ(1 − w) −2 • if u > 0 and, using the change of variable t = λ(1−w) , the previous integral can be expressed as - if 0 < λ < u then g∞ 0 (w) = 0 in a unique point, namely w = λ+1 u+λ ! λ2 +1 Z λ +∞ 2 - if λ > u then g∞ 0 (w) = 0 in two points, 2c0,λ t λ e−t dt. (35) namely w = u±λ λ+1 2 2 λ • if u < 0 Finally, introducing the incomplete gamma function defined as - if 0 < λ < −u then g∞0 (w) = 0 in a unique Z +∞ point, namely w = λ+1 u−λ Γ(x, a) = tx−1 e−t dt, (36) - if λ > −u then g∞ 0 (w) = 0 in two points, a namely w = u±λ λ+1 the value of c0,λ which satisfies (33) is then   2 −1 Observe that simple manipulations shows that λ λ +1  2 2 c0,λ = 2 Γ + 1, . (37) 0 lim g∞ (w) > 0 2 λ λ w→0+ 0 (38) The case with u = 0 is the only one where the value of cu,λ can lim− g∞ (w) < 0 w→0 be found analytically. Other cases can be studied numerically. so that w = 0, which is a non-differentiable point, can be We are now interested in studying the derivative of g∞ in considered as a point of minimum. order to find singular points which correspond to maximum or minimum points. Deriving (26) it can be shown that if w > 0 IV. N UMERICAL S IMULATIONS 0 g∞ (w) = 0 ⇐⇒ 2λ(1 − w) + 2(1 − w) + 2(u − 1) = 0. In this section, relevant numerical results are shown for Hence the (unique) singular point is stationary profiles for different values of u and λ. We focus on values of u ≥ 0 as the stationary profiles relative to u+λ negative values of u can be obtained by symmetry, according w= λ+1 to (34). The constant cu,λ , which appears in g∞ , is evaluated and it is positive if and only if λ > −u. Deriving (30), instead, numerically, using Newton-Cotes formulas [12]. it can be shown that if w < 0 First, we assume that u = 0 so that the average opinion 0 g∞ (w) = 0 ⇐⇒ 2λ(1 + w) + 2(1 + w) + 2(u + 1) = 0 corresponds to the middle point of I. As already observed in the previous section, if u = 0 then g∞ is symmetric with leading to the following singular point respect to 0 and it has two maxima at u−λ λ w= . w=± . λ+1 λ+1 100 Proc. of the 16th Workshop “From Object to Agents” (WOA15) June 17-19, Naples, Italy 2.5 3.5 λ=1/3 λ=1/3 λ=1 λ=1 λ=3 3 λ=3 2 2.5 1.5 g ∞ (w ) g ∞ (w ) 2 1.5 1 1 0.5 0.5 0 0 −1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1 −1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1 w w Fig. 2. Stationary profiles g∞ for u = 1/4 and λ = 1/3 (blue line), λ = 1 Fig. 3. Stationary profiles g∞ for u = 1/2 and λ = 1/3 (blue line), λ = 1 (red line), and λ = 3 (green line) (red line), and λ = 3 (green line) As λ increases, such points get nearer to the extremal values maxima correspond to w = ±5/8 and w = ∓3/8, as shown in of I, namely to extremal opinions. Observe that increasing the Fig. 2 (red line). Finally, Fig. 2 also shows the results obtained value of λ corresponds to incrementing the impact of diffusion with λ = 3 (green line). In this case the maximum points are with respect to compromise. w = ±13/16 and w = ∓11/16. As in the previous case, the Fig. 1 shows the stationary profiles g∞ (w) when u = 0 points of maximum get nearer to ±1, namely to the extremes for different values of λ. As expected from (34), the function of I, as λ increases. Moreover, observe that largest values of g∞ (w) is symmetric with respect to 0 and it has a minimum λ correspond to increasing the likelihood of negative opinions. in correspondence of w = 0 and two maxima whose values Let us now increase the value of the average opinion to depend on λ. The stationary profiles g∞ (w) in Fig. 1 corre- u = 1/2. The stationary profiles g∞ (w) are shown in Fig. 3 spond to λ = 1/3 (blue line), λ = 1 (red line), and λ = 3 for λ = 1/3 (blue line), λ = 1 (red line), and λ = 3 (green (green line). If λ = 1/3 the two maxima are in correspondence line). Observe that if λ = 1/3 the function g∞ (w) has only of w = ±1/4. Observe that in this case extremal distributions one maximum, namely the positive one. As a matter of fact, are associated with a very low probability. If λ = 1, instead, according the results in the previous section, the negative one the maxima correspond to w = ±1/2; while in λ = 3 they only exists if λ > u. The maximum point is w = 5/8. This correspond to w = ±3/4. value is greater than the one obtained for the same λ in the Therefore, it can be concluded that as λ increases the points case u = 1/4, accordingly with the fact that, in this case, we of maximum move towards the extremes of the considered in- consider a higher average opinion u. When considering λ = 1, terval I. Observe that, according to its definition, any increase the function g∞ (w) has two maxima since the condition λ > u of λ corresponds to assuming that the contribution of diffusion is satisfied. Such points are w = 3/4 and w = −1/4. As in is more relevant than that of compromise. Moreover, according Fig. 2, the value of the maximum corresponding to the negative to the results in Fig. 1, any increase of λ leads to stationary value of w is less significant with respect to that relative to the profiles with small values in correspondence of opinions in the positive value of w. If λ = 3 the two maxima correspond to middle of the interval I. w = 5/8 and w = −7/8 and they are nearer to the extremes of I with respect to those obtained with lower λ. A comparison In Fig. 2 the stationary profiles g∞ (w) are shown when of the results in Fig. 2 with those in Fig. 3 shows that in the considering as average opinion the value u = 1/4. In this case, latter the values of the positive maxima are greater while those the function g∞ (w) is not symmetric and, as expected, it has a of the negative maxima are smaller. local minimum in w = 0. We consider the same values of λ as in the previous case. For each of these values, the number of Finally, we consider a greater value of the average opinion, maxima is two, since the condition λ > u is always satisfied. namely u = 3/4. This corresponds to considering an extremist If λ = 1/3 the positive maximum point is w = 7/16 and the society. Fig. 4 shows the stationary profiles for λ = 1/3 (blue negative one is w = −1/16. While the negative maximum is line), λ = 1 (red line), and λ = 3 (green line). As in the near the middle of the interval I, the positive one is farther. previous case, since λ < u, the profile g∞ (w) has only one In Fig. 2 the stationary profile g∞ (w) obtained with λ = 1/3 maximum if λ = 1/3. The point of maximum is w = 13/16 is shown (blue line) and it can be observed that the value of and it is closer to 1 than the other points of maximum obtained the maximum in w = 7/16 is far more significant than that with the same λ for lower values of the average opinion u. corresponding to w = −1/16, namely the positive opinions are From Fig. 4 it can be shown that, once again, the positive far more likely than the negative ones. This is in agreement maximum point moves towards the extreme 1 as λ increases, with the fact that the average opinion u is positive. If λ = 1 the since it corresponds to w = 7/8 if λ = 1 and to w = 15/16 101 Proc. of the 16th Workshop “From Object to Agents” (WOA15) June 17-19, Naples, Italy Further investigation on this subject is currently under 7 λ=1/3 development. In particular, we are interested in deriving the λ=1 explicit expressions of the stationary profiles with a different λ=3 6 choice of the diffusion function. We aim at studying the properties of such stationary profiles for different parameters of 5 the model. At the same time, we are studying the application of kinetic models to multi-agents systems also from a simulative point of view. More precisely, we are interested in comparing g ∞ (w ) 4 analytic results with simulation experiments and in studying 3 the number of iterations necessary to approximate to a certain degree an analytic stationary profile. 2 R EFERENCES 1 [1] S. Monica and F. Bergenti, “A stochastic model of self-stabilizing cellular automata for consensus formation,” in Proceedings of 15th 0 −1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1 Workshop “Dagli Oggetti agli Agenti” (WOA 2014), Catania, Italy, w September 2014. [2] L. Pareschi and G. Toscani, Interacting Multiagent Systems: Kinetic Equations and Montecarlo Methods. Oxford: Oxford University Press, Fig. 4. Stationary profiles g∞ for u = 3/4 and λ = 1/3 (blue line), λ = 1 2013. (red line), and λ = 3 (green line) [3] M. Groppi, S. Monica, and G. Spiga, “A kinetic ellipsoidal BGK model for a binary gas mixture,” EPL: Europhysics Letter, vol. 96, December if λ = 3. Moreover, Fig. 4 shows that an increase of the value 2011. of the average opinion significantly reduces the value of the [4] W. Weidlich, Sociodynamics: a systematic approach to mathematical modelling in the social sciences. Amsterdam: Harwood Academic negative maximum, which in this case is negligible. This result Publisher, 2000. is not surprising since if the average opinion is near the positive [5] B. K. Chakraborti, A. Chakrabarti, and A. Chatterjee, Econophysics and extreme, then the number of agents with negative opinion has sociophysics: Trends and perspectives. Berlin: Wiley, 2006. to be small. [6] F. Slanina, “Inelastically scattering particles and wealth distribution in an open economy,” Physical Review E, vol. 69, pp. 46–102, 2004. V. C ONCLUSIONS [7] S. Cordier, L. Pareschi, and G. Toscani, “On a kinetic model for simple market economy,” Journal of Statistical Physics, vol. 120, pp. 253–277, In this paper we discussed the opinion dynamics in a 2005. multi-agent system through a kinetic approach. We considered [8] K. Sznajd-Weron and J. Sznajd, “Opinion evolution in closed commu- an opinion evolution model inspired from the interactions of nity,” International Journal of Modern Physics C, vol. 11, pp. 1157– molecules in a gas and we studied, from an analytic point 1166, 2000. of view, the asymptotic behaviour of the opinion distribution. [9] G. Toscani, “Kinetic models of opinion formation,” Communications in Assuming that the opinion of an agent can change each Mathematical Sciences, vol. 4, pp. 481–496, 2006. time it interacts with any other agent because of compromise [10] E. Ben-Naim, “Opinion dynamics: Rise and fall of political parties,” and diffusion, we showed that the average opinion of the Europhysics Letters, vol. 69, pp. 671–677, 2005. system is conserved. The stationary profiles can have different [11] G. Toscani, “One-dimensional kinetic models of granular flows,” ESAIM: Mathematical Modelling and Numerical Analysis, vol. 34, pp. characteristics, depending on the parameters of the model and 1277–1291, 2000. on the explicit expressions of the function which represents [12] G. Naldi, L. Pareschi, and G. Russo, Introduzione al Calcolo Scientifico. the diffusion. For a particular diffusion function, we showed The Mcgraw-Hill Company, 2001. that the asymptotic distribution is characterized by one or [13] F. Bergenti, G. Caire, and D. Gotta, “Agents on the move: JADE for two maxima, depending on the parameters of the model. The Android devices,” in Procs. Workshop From Objects to Agents, 2014. stationary profiles are shown for different values of the average [14] F. Bergenti, G. Caire, and D. Gotta, “Agent-based social gaming with opinion and for different parameters of the model. AMUSE,” in Procs. 5th Int’l Conf. Ambient Systems, Networks and Technologies (ANT 2014) and 4th Int’l Conf. Sustainable Energy We recognize in kinetic models the possibility of describing Information Technology (SEIT 2014), ser. Procedia Computer Science. complex and decentralized systems that can exhibit interesting Elsevier, 2014, pp. 914–919. emergent behaviours. In particular, we are mainly interested in [15] F. Bergenti, A. Poggi, and M. Somacher, “A collaborative platform using kinetic models as a conceptual framework that captures for fixed and mobile networks,” Communications of the ACM, vol. 45, essential characteristics of opinion formation in multi-agent no. 11, pp. 39–44, 2002. systems, and to adopt it in the design of mobile scenarios [16] F. Bergenti, G. Caire, and D. Gotta, “Large-scale network and service management with WANTS,” in Industrial Agents: Emerging Applica- that would eventually use general-purpose industrial strength tions of Software Agents in Industry. Elsevier, 2015, pp. 231–246. technology (see, e.g., [13], [14]). Moreover, we recognize that [17] S. Monica and G. Ferrari, “Accurate indoor localization with UWB kinetic models can be effectively used to model agent-based wireless sensor networks,” in Procs. 23rd IEEE International Con- cooperation (like the ones discussed in, e.g., [15]), and that ference on Enabling Technologies: Infrastructure for Collaborative they can be used to study large scale systems (like the ones Enterprises (WETICE 2014), Parma, Italy, June 2014, pp. 287–289. discussed in, e.g., [16]). Finally, we are interested in modeling [18] S. Monica and G. Ferrari, “Swarm intelligent approaches to auto- the emergent behaviors of wireless sensor networks used to localization of nodes in static UWB networks,” Applied Soft Computing, vol. 25, pp. 426–434, December 2014. support accurate localization (see, e.g., [17], [18]). 102