2026 Volume 35 Issue 4
Article Contents

Huaxin Chen, Wensheng Jia. Asymmetric model of the dynamic quantum Cournot duopoly game with asymmetric information and heterogeneous players[J]. Chinese Physics B, 2026, 35(4): 040307. doi: 10.1088/1674-1056/ae00af
Citation: Huaxin Chen, Wensheng Jia. Asymmetric model of the dynamic quantum Cournot duopoly game with asymmetric information and heterogeneous players[J]. Chinese Physics B, 2026, 35(4): 040307. doi: 10.1088/1674-1056/ae00af

Asymmetric model of the dynamic quantum Cournot duopoly game with asymmetric information and heterogeneous players

  • Received Date: 26/06/2025
    Accepted Date: 11/08/2025
    Available Online: 01/04/2026
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Figures(5)

Article Metrics

Article views(277) PDF downloads(2) Cited by(0)

Access History

Asymmetric model of the dynamic quantum Cournot duopoly game with asymmetric information and heterogeneous players

Abstract: Building on the existing symmetric quantization model of the dynamic Cournot duopoly game (CDG) with asymmetric information, we extend it to an asymmetric quantization model and study the stability of the quantum Bayesian Nash equilibrium (QBNE) under heterogeneous expectations. We analyze the influence of various parameters on the stability of QBNE, with a particular focus on the impact of the parameter α on system stability. The results show that when α < 1, under the same parameters, the quantum strategy of the asymmetric quantization model is more favorable for stabilizing the market. However, when α > 1, the quantum strategy of the symmetric quantization model is more conducive to stabilizing the market.

1.   Introduction
  • The integration of quantum theory and game theory has given rise to a new interdisciplinary field known as quantum game theory, which originated from Meyer’s pioneering work on a quantum coin-flipping game.[1] Not long thereafter, Eisert et al.[2] applied the concept of quantum entanglement (QE) to the classic prisoners’ dilemma game (PDG). The two-player quantum PDG was experimentally realized by Du et al. via NMR techniques.[3] Experimental realizations of the quantum PDG involving two to four players have subsequently been achieved via NMR[4] and linear optics techniques.[5,6]

    Oligopoly is a common market structure in real-world economies. Quantum game theory has recently seen growing application in the analysis and modeling of behavior in oligopoly markets. A quantum approach based on continuous-variable strategies was proposed by Li et al.[7] Building on this, Du et al.[8] applied the method to quantize the CDG under asymmetric information. Chen et al.[8] further generalized Du et al.’s framework[9] to broader scenarios. Later studies employed the Li–Du–Massar scheme to investigate Cournot, Bertrand, and Stackelberg games.[10–15] These quantum models have uncovered outcomes and strategic dynamics that are fundamentally absent in classical game theory. More recently, the two-turn game[16] involving continuous-variable quantum strategies has been studied. In addition, the quantum frameworks proposed by Li–Du–Massar and Frackiewicz were independently generalized by Wang and Yang[17] to accommodate multiple players.

    The above-mentioned research on quantum duopoly games has largely emphasized static analysis, which reveals outcomes but not the evolution of strategic interactions. Dynamic analysis is thus crucial for understanding firms’ adjustments and changes in system stability. Rand[18] was the pioneer in introducing dynamical systems theory into duopoly game models. Since then, this approach has been adopted by many researchers in the study of duopoly games.[19–25] The findings underscore the significance of studying duopoly games dynamically under bounded rationality, yet the Nash equilibrium often remains locally unstable. A number of recent works have explored the local stability of equilibria by integrating quantum game theory and complex dynamics. The stability of a quantum CDG was investigated by Zhang et al.[26] A quantum CDG with isoelastic demand was developed by Shi and Xu[27] to examine how QE affects system stability. The stability of equilibria in quantum Stackelberg duopoly models was analyzed by Hu et al.[28] These findings indicate that QE can delay the emergence of bifurcations and chaotic behavior.

    In addition, real-world markets rarely operate under conditions of complete information due to uncertainties arising from dynamic competition, evolving consumer preferences, and economic fluctuations. Consequently, exploring incomplete information scenarios becomes increasingly critical for practical market decision making. Harsanyi[29] proposed a Bayesian game to address the problem of incomplete information. Yu and Yu[30] constructed a dynamic CDG where firm 2 operates with two distinct marginal cost structures: one high and one low. Long and Zhao[31] explored equilibrium stability in dynamic Bertrand duopoly games influenced by cluster overflow and asymmetric information. Askar[21] studied a Cournot–Bertrand duopoly game involving firms with bounded rationality. Building on Yu’s model,[30] Wang et al.[32] generalized it to an n-player setting and examined the equilibrium stability. In the public goods game, Kang et al.[33] studied the impact of the cost of information acquisition for discreet investors on the level of cooperation.

    Recently, researchers have incorporated quantum theory into duopoly games,[34–37] prisoners’ dilemma games,[38] and cooperation games[39] with asymmetric information. It is worth noting that although these quantum games involve asymmetric information, the quantization schemes applied are all symmetric, which creates a structural inconsistency. To address this, Qin et al.[40] proposed an asymmetric quantization scheme for the quantum Bertrand duopoly game and examined whether the asymmetric and symmetric quantum schemes can achieve optimal cooperative payoffs. Their results show that the asymmetric quantization scheme can achieve the optimal cooperative payoff, whereas the symmetric scheme fails to do so. Furthermore, Wang et al.[41,42] constructed a quantum Stackelberg duopoly game model using Qin et al.’s asymmetric quantization scheme[40] and studied the impact of changes in the asymmetry factor α and the QE γ on the Nash equilibrium profits. Similarly, Zhong et al.[43] used the asymmetric quantization scheme proposed by Qin et al.[40] to establish a quantum Stackelberg duopoly model under the condition of isoelastic demand and analyzed the impact of γ and α on the Nash equilibrium profits of the leader and the follower.

    A natural question is: Does the stability of the asymmetric model of a dynamic quantum duopoly with asymmetric information perform better than that of the symmetric model? To address this question, we construct an asymmetric quantization model of a quantum CDG under asymmetric information and investigate the stability of its QBNE in comparison to the symmetric quantization model proposed by Zhou et al.[34] The findings show that the effectiveness of market stabilization depends on the value of α, with the asymmetric quantization model performing better when α < 1, and the symmetric model being more effective when α > 1.

    The following sections are arranged as follows. An asymmetric quantization scheme for the quantum CDG under asymmetric information is detailed in Section 2. The stability of the QBNE in a dynamic model with heterogeneous expectations is investigated in Section 3. Theoretical results are confirmed by numerical simulations in Section 4. Concluding remarks are summarized succinctly in Section 5.

2.   The model
  • We introduce a CDG under asymmetric information,[30] where both firms produce homogeneous products. The market price is

    where qi (i = 1,2) is the output of firm i and a > 0. Let ci represent the marginal cost for firm i. Firm 2 is assumed to have two possible marginal cost levels: a high cost ch and a low cost cl. Let ch = c1 + ε, cl = c1 – ε and the probability distribution P(c2 = ch) = θ, P(c2 = cl) = 1–θ, where c1 > 0, ch > 0, cl > 0,ε > 0, and θ ∈ (0,1). Denote firm 2’s output as q2h and q2l when its marginal cost is ch and cl, respectively. The expected profit for Company 2 is

    The expected profit for Company 1 is

    Next, utilizing Qin et al.’s asymmetric quantization approach,[40] we construct a quantum CDG with asymmetric information. Following the full implementation of strategic operations and measurement in the asymmetric quantization model, the obtained values, denoted by q1 and q2, reflect the respective outputs of the two competing firms[42]

    where y1 and y2 denote the parameters through which the two firms carry out their strategic decisions; γ=γ1γ2, α=γ1/γ2, γ1 and γ2 are two independent entanglement factors. If γ1 = γ2, our quantization model is equivalent to the minimal quantization model proposed by Zhou et al.[34]

    Firm 1’s expected quantum profit can be derived by substituting Eq. (3) into Eq. (2):

    The quantum profits corresponding to firm 2’s choices of y2h and y2l are derived by substituting Eq. (3) into Eq. (1)

    To simplify the calculation, we define A = cosh2γ+sinh2γ+2αsinhγcoshγ and B = cosh2γ+sinh2γ+2α−1sinhγcoshγ. Below are the marginal profit functions for the two firms:

    To ensure the economic validity of the model, we impose that all relevant variables must take strictly positive values. Therefore, the following assumption holds.

    Assumption 1a-c1 > max{A+1,B}ε.

    Let ∂π2hQ∂y2h=0 and ∂π2lQ∂y2l=0, the following equations are derived:

    The mathematical expectation of y2 is Ey2(y1;c1,ε) = θy2h(y1;c1,ε) + (1 – θ)y2l(y1;c1,ε), then

    Let ∂π1Q∂y1=0, we have the following equation:

    Theorem 1 If Assumption 1 holds, then

    Proof If Ey2 = 0, based on Eqs. (7) and (8), y1≤(a−c1−ε)coshγA and y1=(a−c1)coshγA+1. However, since (a−c1)coshγA+1−(a−c1−ε)coshγA>0, then Ey2 = 0 does not hold.

    Then, if

    then (a−c1−ε)coshγA<y1 and Ey2≤(a−c1)coshγB. From Eqs. (7) and (8),

    However, y1>(a−c1−ε)coshγA is equivalent to

    namely (a–c1) ≤ ε(A+1)(B+1), which contradicts Assumption 1. So E(y2)≠(1−θ)(a−c1−ε)coshγ−Ay1B+1.

    Finally, E(y2)=(a−c1)coshγ−Ay1+(1−2θ)εcoshγB+1.

    Theorem 2 If Assumption 1 is satisfied, then the equilibrium strategy profile in the quantum CDG with asymmetric information is given by y*=(y1*,y2h*,y2l*), where

    Proof Based on Assumption 1 and Theorem 1, it follows that Ey2≤(a−c1)coshγB. Combining Eqs. (5), (6), and (8), the QBNE is determined to be y*=(y1*,y2h*,y2l*).

3.   The stability of QBNE
  • Here, firm 2 is assumed to follow adaptive expectations, and firm 1 adjusts its decision via a gradient method based on marginal profit.[19]

    Firms 1 and 2 adjust their strategies according to the following dynamic equations:

    where v ∈ (0,+∞) reflects firm 1’s quantum strategy adjustment speed; β serves as firm 2’s adaptive expectation.

    Let yi(t + 1) = yi(t) = yi, i = 1, 2h, 2l, yields the following two quantum equilibrium points of system (11):

    where y1*,y2h*,y2l* are determined from Eq. (10). The Jacobian matrix can be expressed as

    where b11 = 1 + v((a–c1)coshγ–2y1(A+1)–B(θy2h+(1–θ)y2l).

    We present the theorem on the boundary equilibrium E0 as follows.

    Theorem 3E0 is a saddle point if Assumption 1 is satisfied.

    Proof The Jacobian matrix J(E0) is

    Since Assumption 1 holds, the eigenvalues are λ1=1+vcoshγ(a−c1−(1−2θ)Bε)B+1>1, λ2 = λ3 = β ∈ (0,1).

    Regarding the QBNE E*, the theorem below holds.

    Theorem 4 If vy1*(2AB+(1−β)(A+B+1))−2(1+β)(B+1)<0 and Assumption 1 holds, then QBNE E* is locally asymptotically stable.

    Proof At the QBNE, the Jacobian matrix is given by

    Then the characteristic polynomial of J(E*) is

    Clearly, one of the eigenvalues is λ1 = β ∈ (0,1), and the two additional roots fulfill the equation below

    Thus, the two additional roots are

    where

    λ2 < 1 is equivalent to

    This is always holds.

    λ3 > −1 is equivalent to

    Since vy1*(2AB+(1+β)(A+B+1))−2(1+β)(B+1)<0, E* is locally asymptotically stable, completing the proof.

    The stability region S corresponding to the equilibrium E* is

    where

    Then

    Therefore, vG(β,γ) is strictly increasing and concave. The QBNE is locally asymptotically stable when (β, v) ∈ S. Conversely, when (β, v) ∉ S, the system may undergo bifurcation or even chaos. Based on Eq. (12), we derive the following proposition.

    Proposition 3 The stability region of QBNE expands with increasing β and c1, but contracts with increasing values of a and θ. Furthermore, if θ < 1/2, as ε grows, the stability region becomes larger; conversely, if θ > 1/2, it narrows with increasing ε. These findings are consistent with the results of Yu et al.,[30] Zhou et al.,[34] and Wang et al.[32]

4.   Numerical simulations
  • This section investigates the dynamics of the quantum CDG using numerical simulations. We employ a set of bifurcation plots, chaos diagrams, Lyapunov exponents, and strange attractors to assess the stability differences between the symmetric and asymmetric quantum schemes of the dynamic CDG with informational asymmetry and heterogeneous expectations. In this Section, let a = 6, c1 = 3, ε = 1, θ = 0.6, γ1 = 0.2, γ2 = 0.6, then y1 = 0.712, y2h = 0.26, y2l = 0.86. When γ1=γ2=0.12 and the remaining parameters are kept unchanged, y1 = 0.72, y2h = 0.23, y2l = 0.93. When γ1 = 0.6, γ2 = 0.2 and the remaining parameters are kept unchanged, y1 = 0.68, y2h = 0.15, y2l = 0.94. When γ1 = 0.9, γ2 = 0.3 and the remaining parameters are kept unchanged, y1 = 0.56, y2h = 0.06, y2l = 0.75.

    Initially, the stability domain of the QBNE is investigated. Setting γ=0.12, while keeping all other parameters constant. To investigate how parameters β and α influence the stability region, the stability region diagrams for α = 1/3, α = 1 and α = 3 with respect to β are generated, as shown in Figs. 1(a), 1(b), and 1(c). Clearly, an increase in α leads to a contraction of the stability region, while it expands as β increases. Comparing Figs. 1(c) and 1(d), as the parameter γ increases, the stability region becomes smaller, which is consistent with the findings of Zhou et al.[34]

    To explore how the parameters v and α influence QBNE stability, bifurcation diagrams are plotted for α = 1/3, α = 1, and α = 3 with respect to v, as shown in Fig. 2(a). As v increases, the firm’s output evolves from a stable equilibrium to bifurcation, and ultimately enters a chaotic regime. To more clearly illustrate how variations in α influence QBNE stability, the associated maximal Lyapunov exponent is plotted in Fig. 2(b). Obviously, a larger α causes bifurcation and chaos to occur earlier. In other words, when α < 1, under the same parameters, the stability region of the asymmetric quantization model system is larger than that of the symmetric quantization model system. That is, the quantum strategy of the asymmetric quantization model is more favorable for stabilizing the market. However, when α > 1, the quantum strategy of the symmetric quantization model is more conducive to stabilizing the market. Additionally, figures 2(c) and 2(d) further demonstrate that an increase in γ advances the emergence of bifurcations and chaos.

    To explore how the parameters θ and α influence QBNE stability, bifurcation diagrams are plotted for α = 1/3, α = 1 and α = 3 with respect to θ, as shown in Fig. 3(a). As θ increases, the firm’s output evolves from a stable equilibrium to bifurcation, and ultimately enters a chaotic regime. To more clearly illustrate how variations in α influence QBNE stability, the associated maximal Lyapunov exponent is plotted in Fig. 3(b). Obviously, a larger α causes bifurcation and chaos to occur earlier. Additionally, figures 3(c) and 3(d) indicate that a larger α delays bifurcation and leads to an earlier emergence of chaos.

    Finally, we use strange attractors to further illustrate the impact of α on the system’s stability. A hallmark of these strange attractors is the existence of at least one positive and one negative Lyapunov exponent, accompanied by a Lyapunov fractal dimension. The definition of the Lyapunov dimension is given by[44]

    where Li (i = 1,2,…,n) is the Lyapunov exponent, and L1>L2 > ⋯ > Ln; k is the largest positive integer such that ∑i=1kLi≥0 and ∑i=1k+1Li<0.

    Figure 4 is the attractor corresponding to Fig. 2(a) with v = 1.32. Using the same parameter settings as in Fig. 2(a), the Lyapunov exponents in Fig. 4(a) are L1 = −0.072, L2 = −0.48, and L3 = −0.5108, indicating that the system reaches fixed points. The Lyapunov exponents in Fig. 4(b) are L1 = 0.3741, L2 = −0.4073, and L3 = −0.5108. The Lyapunov dimension is d(L)=1+L1|L2|≈1.918, indicating that the system exhibits a strange attractor. As α increases, the attractor transitions from fixed points to chaotic attractors.

    Figure 5 is the attractor corresponding to Fig. 2(a) with v = 1.195. Using the same parameter settings as in Fig. 2(a), the Lyapunov exponents in Fig. 5(a) are L1 = −0.8161, L2 = −0.4173, and L3 = −0.5108, indicating that the system reaches fixed points. The Lyapunov exponents in Fig. 5(b) are L1 = 0.3742, L2 = −0.4045, and L3 = −0.5108. The Lyapunov dimension is d(L)=1+L1|L2|≈1.925, indicating that the system exhibits a strange attractor. As α increases, the attractor transitions from fixed points to chaotic attractors.

5.   Conclusion
  • This paper proposes an asymmetric quantization model for dynamic quantum CDG with asymmetric information, mainly comparing the stability of the system between the quantum strategy of the asymmetric quantization model and that of the symmetric quantization model.[34] The results show that when α < 1, under asymmetric information and heterogeneous expectations, the stability region of the asymmetric quantization model system for the dynamic quantum CDG is larger than that of the symmetric quantization model system. That is, the quantum strategy of the asymmetric quantization model is more favorable for stabilizing the market. However, when α > 1, the system’s stability region under the symmetric quantization model is larger than that under the asymmetric quantization model. That is, the quantum strategy of the symmetric quantization model is more conducive to stabilizing the market. Moreover, Proposition 3 indicates that the effects of the parameters β, θ, c1, and ε on the stability of the system are consistent with the results of Zhou et al.[34]

Figure (5)  Reference (44)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return